The mechanics of claims and permissible killing in war 9780190872045, 0190872047

Table of contents :
Basic premises and method --
A theory of rights --
Putting the mechanics of claims in perspective --
Avoiding a misplaced emphasis on intentions --
Innocent threats versus innocent bystanders --
From innocent threats to noncombatants innocently aiding unjust combatants --
Negative agent-claims and the agent-patient inference --
Intervening agency and the right of non-sacrifice.

Citation preview

The Mechanics of Claims and Permissible Killing in War

The Mechanics of Claims and Permissible Killing in War Alec D. Walen

3

3 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2019 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-​in-​Publication Data Names: Walen, Alec, author. Title: The Mechanics of Claims and Permissible Killing in War / by Alec D. Walen. Description: New York, NY : Oxford University Press, 2019. | Includes bibliographical references and index. Identifiers: LCCN 2018035701 | ISBN 9780190872045 (hc : alk. paper) Subjects: LCSH: Self-defense (Law)—Philosophy. | Self-defense (International law)—Philosophy. Classification: LCC K5087. S4 W35 2019 | DDC 172/.42—dc23 LC record available at https://lccn.loc.gov/2018035701 9 8 7 6 5 4 3 2 1 Printed by Sheridan Books, Inc., United States of America

To my mother, for her inspiring love of art and craft, and to my father, for his joie de vivre.

CONTENTS

Preface  xiii

1 Introduction  1

1 A Brief History of Recent Work on the Right of Self-​Defense  2



2 My Position on the Right to Defend Against Threats  8



3 The Thesis in Practice  13



4 The Structure of the Argument  19

2 Basic Premises and Method  21

1 Reflective Equilibrium and the Relative Balance of Theory and Case Intuitions  21



2 Knowledge and Justification  23



2.1 The Use of Cases with Unrealistic Assumptions of Knowledge  23



2.2 Two Audiences and Two Kinds of Justification  23



2.3 The Case for a Qualified Fact-​Relative Framework  26



3 Reductive Individualism  29



3.1 A Defense of Reductive Individualism  29



3.2 Warfare-​Based Objections to Reductive Individualism and Replies  34



4 The Distinctive Causal Structure of Eliminative Killing  37

3 A Theory of Rights  43

1 The Purpose and Basic Structure of Rights  43



1.1 The Basic Function of Rights  44



1.2 The Three Principles That Ground the Space of Rights  47



1.2.1 The Autonomy Principle  48



1.2.2 The Equality Principle  49



1.2.3 The Welfare Principle  50



1.3 The Further Relevance of the Fundamental Principles  55



1.3.1 Rights as Grounding only a Subset of Possible Wrongs  55



1.3.2 The Autonomy Principle and the Topology of Rights  58





1.3.3 The Autonomy Principle and Owned Luck  60

2 Formalizing the Mechanics of Claims  65



2.1 Balancing Agent-​Claims and Basic Patient-​Claims  65



2.1.1 Overview of Balancing Claims, Organized in



2.1.2 Agents’ Negative Agent-​Claims and Positive



2.1.3 Some Scope for Positive Agent-​Claims  71



2.1.4 The Argument Against Agent-​Centered

Terms of Partiality and Impartiality  66 Patient-​Claims  68

Prerogatives  73

2.2 Including Special Patient-​Claims  77



3 Points of Contrast with the Infringement Model  79



4 The Mechanics of Claims and the Restricting Claims Principle  82



4.1 Introduction to the Restricting Claims Principle  83



4.2 The Property Baseline for the Restricting Claims Principle  87



4.3 Objections and Replies  90



4.4 Refining the Restricting Claims Principle with



4.5 Two Final Clarifications Regarding the Restricting

Empowering Claims  93 Claims Principle  97

viii  | Contents

4 Putting the Mechanics of Claims in Perspective  100

1 A Fresh Approach to Compensation  100



1.1 The Infringement Model’s Non-​Unique Ability to



1.2 Why the Mechanics of Claims Is Better on

Account for Compensation  101 Compensation  103



2 Other Problems for the Infringement Model  108



2.1 Intrinsic Oddness of the Infringement Model  108



2.2 A Substantive Objection Regarding an Inappropriate



2.3 Occam’s Razor and the Rights–​Consequentialism



2.4 Other Forms of Explanatory Power in the Mechanics of

Normative Gulf  111 Hybrid  113 Claims  114



3 Threshold Deontology  115



3.1 A Positive Account of Threshold Deontology  115



3.2 Overcoming Problems with Threshold Deontology Using This Account  119



4 Final Comment on the Infringement Model  123

5 Avoiding a Misplaced Emphasis on Intentions  124

1 Introduction to Lazar’s Argument  125



2 The Limited Significance of Mixed Motives  125



3 Objection: Respect as a Subjective State  130



4 Application of the Preceding Argument to Just War Theory  134



4.1 Application to Lazar’s Argument  134



4.2 Lessons for Just War Theory About the Relevance of Intentions  136

6 Innocent Threats Versus Innocent Bystanders  137

1 Introduction  137



2 Innocent Victims and Innocent Bystanders  138



3 Innocent Threats and Innocent Victims  143



3.1 Stating the Problem  143



3.2 The Hybrid Argument: Its Insights and Its Failures  144

Contents  |  ix



3.2.1 Insights  145



3.2.2 Failures  146



3.3 The Dual Significance of Being a Threat  149



3.4 A Final Review of the Hybrid Argument and the Infringement Model  150



4 Objections and Replies  151



4.1 The Nature of the Claims of Innocent Threats  151



4.2 A Temporal Objection  152

7 From Innocent Threats to Noncombatants Innocently Aiding Unjust Combatants  154

1 Overview of the Terms and the Argument  154



2 Non-​Culpable Aggressors  155

3 Direct Innocent Threateners  157



4 Indirectly Threatening Harm  159



4.1 An Initial Argument That Indirectness Does Not Matter  160



4.2 Objection and Reply  161



4.3 A Way That Indirectness May Matter  164



5 Threatening Only in Numbers  165



6 Conclusion  168

8 Negative Agent-​Claims and the Agent–​Patient Inference  171

1 From Ducking to Actively Causing Harm  172



1.1 Allowing Harm by Ducking  172



1.2 Causing Harm That Is Like Ducking  177



2 From Agent-​Rights to Patient-​Rights  178



2.1 The Problem with Moving from Agent-​Rights to



2.2 Responsiveness  179



2.3 Justified Competition  181



2.4 The Bear and the Plank  184

Patient-​Rights  178



3 Justified Acts and the Agent–​Patient Inference  187



3.1 Good Agents and the Agent–​Patient Inference  187



3.2 Culpable Agents and the Agent–​Patient Inference  188



3.3 Third Parties and the Agent–​Patient Inference  189



x  | Contents

4 Application to Just War Theory  191

9 Intervening Agency and the Right of Non-​Sacrifice  195

1 Introduction to the Puzzle  195



2 Two Unsuccessful Approaches to the Puzzle  197



2.1 Misplaced Use of Counterfactual Baselines  197



2.2 A Misplaced Appeal to the Importance of Rights Violations  200



3 Intervening Agency and Causal Roles  203



4 Application of the Claim of Non-​Sacrifice  209

10 Conclusions  212

1 Conclusions About Eliminative Killing  212



2 Conclusions About Rights Theory  216 Table of Cases (with page of first use)  219 Glossary of Terms  221 Bibliography  225 Index  233

Contents  |  xi

PREFACE

T

he overarching aim of this book is to articulate a new theory of rights—​new except for its precursors in my own prior work. This is a theory that aims to explain why rights are a central and non-​ eliminable part of morality. It presents an account of rights that is fundamentally different from the dominant view, according to which rights can be overridden by a range of competing moral considerations, though always leaving some residue behind. It is also different from the standard alternative account, according to which rights are mere conclusory labels identifying what is or is not permissible after the real moral reasoning has been done. According the view developed here, rights are the conclusions of a distinctive sort of rights-​based reasoning that involves balancing pro tanto rights, or claims, against each other. The nature of the competing claims and the way they compete against each other defines a “space of rights,” a normative space that has its own distinctive normative structure. The original ambition of this book was not to articulate a new theory of rights; it was a more modest ambition: to provide a better account of the right to defend against threats than currently exists. By “threats,” I meant persons or things that are likely to cause harm unless action is taken; I did not mean verbal statements intending to coerce. I took it that there’s no deep problem explaining why it is permissible for an agent to defend herself or others against culpable aggressors:1 they have forfeited their right  To avoid using “they” all the time, or the awkward “he or she,” or treating the generic person as either male or female, I have adopted a convention to treat agents as women and those who are affected by an agent’s acts as men—​though sometimes those roles swap around and I have to simply pick a gender. 1

not to be attacked. There are interesting details to be worked out, such as when innocent victims must suffer a small amount of harm because the only way to avoid that is to cause much more harm to a culpable aggressor.2 But I did not want to focus on such details. Rather, I wanted to focus on three more basic questions: First, what may be done to those who non-​culpably threaten to do what they have no right to do? Second, what may be done to people who have a right to threaten others? Third, what difference does it make if the threat they pose is indirect, that is, if it is mediated by the directly threatening acts of others? I was drawn to engage these questions because, when I  started to think about just war theory, I realized that the way I framed certain basic distinctions in the space of rights was inadequate. Bit by bit, my attempts to expand what I had done before to handle the issues that were arising when thinking about defending against threats took me to new, interesting theoretical territory. And at each step along the way, it seemed as if the ideas were revealing their own inner logic. Thus, focusing on questions about the right to defend against threats set me on the road to formulating a more satisfying general theory of rights. I shifted my focus to a theory of rights more generally because, as I wrestled with how to make sense of the right to defend against threats, I came to realize that I had a fundamental disagreement with almost all the other just war theorists. They were almost all working on trying to articulate conditions under which a person who poses a threat to another is liable to be killed. That is, they were almost all working to articulate when and why being a threat could cause the threatening person to forfeit his right not to be killed. But in my view, the fundamental justifiability of killing threats does not depend on them forfeiting a right not to be killed. Forfeiture remains, I believe, relevant. But even if threats have not forfeited their right not to be killed, the very fact that they pose a threat without having a right to do so undermines their right not to be killed. Thus, I found myself working to articulate why one can get most of what one needs to understand the justifiability of defensive killing without worrying about liability to be harmed.3

 See Christopher Wellman, 2012, pp. 375–​376, for a list of seven distinct problems for any theory of forfeiture. See also Kimberly Kessler Ferzan, 2016, for a thorough discussion of these in the context of self-​defense. 3  The one person I have read whose work does not focus on forfeiture is Victor Tadros. He still uses the concept of liability, but he cuts it free from forfeiture. See Tadros, 2016a. In doing so, he rightly points out that in the law and common parlance liability is not connected with forfeiture. Consider, for example, liability to pay taxes. Ibid., p. 113. 2

xiv  | Preface

The deeper issue this revealed concerns the basic model of rights one uses. Almost everyone in the field uses what I call the infringement model.4 According to this model, the rights people have do not settle what may be done to them. Even if they have in no way alienated their rights, if the cost of respecting a right is great enough, then it can permissibly be infringed. Rights violations are impermissible, but mere rights infringements are permissible.5 Infringing a right wrongs the right holder, but it is justifiable overall as long as it brings about a significantly lesser evil or respects a more stringent right. I found this model of rights problematic in a number of ways. First, it is unnecessarily conceptually obscure to talk about permissibly or rightfully wronging another or, more precisely, to represent this as anything other than an extraordinary moral phenomenon.6 More importantly, there are ways in which the infringement model proves misleading or unhelpful on substantive issues: 1. It oversimplifies when compensation is due. 2. It opens up an inappropriate normative gulf between those who have forfeited claims and those who have not, making it too easy to justify doing harm to those who have and too hard to justify doing harm to those who have not. 3. It makes it unnecessarily difficult to make sense of how concerns with rights and welfare are to be weighed against each other. 4. It lacks the resources to explain a range of moral phenomena, from why it is that we should not treat people as a means without their consent to why claims not to be harmed as a side effect should be slightly, but only slightly, stronger than claims for aid, all else equal.7

 The one person I have found whose work is at least marginally concerned with just war theory who is equally skeptical of the infringement model is Michael Moore. Tadros is, in a way, also skeptical of it; but he is skeptical of the importance of rights generally. 5  This way of using the term infringing is arguably a piece of jargon that has arisen only recently in the field of analytic normative philosophy. The more familiar meaning is synonymous with a rights violation. The first non-​obsolete definition in The Compact Oxford English Dictionary, New Edition is as follows: “To commit a breach or infraction of (a law, obligation, right, etc.); to violate or break (an oath, pledge, treaty, etc.); to transgress, contravene.” 6  In Chapter 4, § 3, I argue that there may be occasions when it is, all things considered, morally appropriate to wrong another; but I think such occasions must be extraordinary. 7  I discuss all of these in Chapter 4, § 2. 4

Preface  |  xv

Developing these problems with the infringement model and arguing that they can be avoided with a better model, one that provides us with a better understanding of the moral landscape, became the deep core of this book. The alternative model that I argue for is one that I call the mechanics of claims. To be clear, the mechanics of claims is not offered as an analysis of what it means to have rights. It is offered as a model of rights that helps us understand what is substantively morally true. According to the mechanics of claims, people have claims that protect their interests, but whether a particular claim corresponds to a right depends on how all the competing claims interact. To work out how they interact, the mechanics of claims distinguishes fundamentally different types of claims, such as claims of agents versus claims of patients (those affected by the choices of agents), diffent types of patient-​claims reflecting the different causal roles a patient might play in an agent’s acting to achieve some end, and different types of agent-​claims depending on whether they are claims to act or to refrain from acting. The mechanics of claims takes the strength of claims to depend in part on the kind of claim it is. It also provides a distinct structure for weighing different kinds of claims against each other and for explaining how the presence of some kinds of claims impacts the strength of other kinds of claims. Moreover, these distinctions and structural features are not simply designed to produce morally attractive results; they are grounded in three fundamental moral principles. The resulting picture explains in a deeply coherent way why we have the rights we have. This coherent explanatory account cannot be provided—​at least I cannot see how it could be done—​using the infringement model. I am indebted to many people for helping me to write this book. First, it is only because I was fortunate enough to be able to present, in various fora, the various inadequate papers that I have written on these topics and to get good feedback that I was able to realize at each step along the way that more work had to be done. My gratitude along this first dimension starts with Seth Lazar, who invited me to comment on his book. I am next grateful to Jeff McMahan, Victor Tadros, and Helen Frowe for inviting me to present my first attempts at providing a positive account of when eliminative killing is justifiable—​the patient-​focused part of the book—​ at their respective home universities. After that, my gratitude goes out to David Rodin and the organizers of the Ethics, Law and Armed Conflict workshop for yet another chance to present and get feedback on the next generation of that part of the book. It was there, when I got an especially probing challenge from Adil Haque, that I realized that I had a book to

xvi  | Preface

write. With regard to the agent-​focused part of this book, I am grateful to Larry Alexander, Kim Ferzan, Stephen Galoob, and Yuan Yuan for inviting me to present that material. And finally, I  am grateful to Johann Frick who agreed to co-​teach a graduate seminar called Rights and Risks, jointly offered at Princeton and Rutgers Universities, where I was able to teach a draft of the book. I would now like to thank those whose conversation and comments on papers or drafts of the book were especially helpful. At the top of the list is Kim Ferzan, whose encouragement and collaboration have been among the most important and fruitful I’ve ever had. Second is Ketan Ramakrishnan, who read the whole book in an early draft and gave me many useful comments. Third, I would like to offer special thanks to three people who participated in the Rights and Risks seminar and offered especially helpful comments: my co-​teacher, Johann Frick, and two Rutgers students who audited the course: Benjamin Bronner and James Goodrich. Fourth, I would like to thank Alexander R. Cohen for his work editing the book before I turned it in to Oxford University press. In addition, I  would like to thank all of the following people who have given me helpful comments along the way:  Larry Alexander, Richard Arneson, Yitzhak Benbaji, Joseph Bowen, Lars Christie, Russell Christopher, Michael Deigan, Tom Dougherty, Helen Frowe, Christopher Frugé, Stephen Galoob, Adil Haque, Lisa Hecht, Douglas Husak, Tyler John, Savannah Kinkaid, Gerald Lang, Seth Lazar, Adam Lerner, S.  Matthew Liao, Todd May, Jeff McMahan, Michael Moore, Armando Jose Perez-​Gea, Juan S. Piñeros Glasscock, Ajay Ravichandran, Massimo Renzo, Samuel C. Rickless, David Rodin, Richard Schoonhoven, Re’em Segev, Adam Slavny, Uwe Steinhoff, Victor Tadros, Danny Underwood, Leif Wenar, Peter Westen, and Yuan Yuan. My apologies to anyone whose name I have inadvertently left out. Finally, I would like to thank my wife, Rachel Somerville, for occasionally expressing disbelief when I ran an idea or supposed insight past her; her sharp sense for BS has helped me avert errors on more than one occasion. Additionally, I would like to thank her and my son, Amitav Walen, for their encouragement as I wrote this book. I don’t think I’ve been too much of a grump or a derelict partner or parent. But writing a book is still a big, time-​consuming endeavor; and it certainly helps to have family who appreciate the project and cheer you on.

Preface  |  xvii

The Mechanics of Claims and Permissible Killing in War



1 Introduction

W

ars of conquest were never consistent with respect for rights:  the conquered were killed, raped, maimed, and enslaved; their homes, their lives, and often their civilizations were destroyed. And now, in a world shaped by the horrors of two world wars, one punctuated by the dropping of atomic bombs, the idea of war as a path to glory has been largely rejected.1 It is now widely accepted that justice and respect for individual rights require strict limits on the right to wage war (jus ad bellum) and the things that may be done in war (jus in bello). The current law of armed conflict makes it clear: wars may be fought only in self-​or other-​defense.2 It is no surprise, then, that much contemporary just war theory would be modeled on the theory of self-​defense. In this book, I offer a new account of how this modeling should work. But my focus will be more foundational than is normal. My reason is that I think the dominant account of rights used by people interested in the right of self-​defense is doubly problematic: its internal tensions undermine the plausibility of the views people try to defend, and it is in some ways morally distorting. Thus, I will be primarily concerned with developing and then deploying a new account

 Reference to the two world wars and the need “to save succeeding generations from the scourge of war” form the opening words of the Preamble to the United Nations Charter. The global surge in nationalist rhetoric, as I write these words, leads me to worry, however, that war is once again coming to be seen as a path to glory, horrors be damned. 2   The United Nations Charter limits the use of military force to two conditions:  (1) actions the Security Council considers necessary to respond to “any threat to the peace, breach of the peace, or act of aggression” (Article 39) and (2) exercises of the “right of individual or collective self-​defence if an armed attack occurs against a Member of the United Nations, until the Security Council has taken measures necessary to maintain international peace and security” (Article 51). 1

of rights, one that I  hope will put the discussion of the right to defend against threats on more solid footing. This introductory chapter covers four themes. First, to explain why I think the current state of just war theory calls for a new account of rights, I trace a brief history of recent discussions of the right of self-​defense.3 Second, I describe my position on when it is permissible to defend against threats. Third, I explain how this relates to current law and to other philosophical work on the topic. Finally, I  sketch an outline of the rest of the book.

1. A Brief History of Recent Work on the Right of Self-​Defense I start with Judith Jarvis Thomson’s 1991 article “Self Defense,”4 which set the table for contemporary moral discussions of the right of self-​ defense. Her aim was to understand why it seems permissible for an agent to engage in self-​defense against not only culpable aggressors but also innocent threats—​people who threaten to cause harm to another not by any voluntary acts they perform but simply by the involuntary motion of their bodies—​even though it seems impermissible for an agent to kill a bystander to save herself. Her key move was to suggest that if a threat, even an innocent threat, would kill an innocent victim if he were not killed first, then he would “violate [her] right that he not kill [her].”5 This is not true of innocent bystanders. This contrast provided the basis for Thomson’s claim that threats lack, while bystanders retain, the right not to be killed. In response, both Jeff McMahan and Michael Otsuka argued that Thomson was mistaken to think that an innocent threat can violate a victim’s rights. As McMahan put it: A moral constraint can apply only to the action of a responsible agent. Neither a falling boulder nor a charging tiger can be subject to a moral constraint; thus neither can violate a right. Since . . . an Innocent Projectile is

 I return to give a more positive treatment of the themes in this section in Chapter 6.  Thomson was not concerned with addressing just war theory. She “bypass[ed] the question how the fact of war affects questions about self-​defense” on the ground that it was “too hard.” Thomson, 1991, p. 298. 5  Ibid., p. 300. 3 4

2  |  The Mechanics Of claims And Permissible Killing in war

no more an agent than a falling boulder, it seems that [such a person] cannot violate rights and hence cannot forfeit them.6

This argument left McMahan and Otsuka taking the position that self-​ defense against a fairly broad range of innocent agents is normally impermissible. The only condition under which it would be permissible is that the harm to them would be substantially outweighed by the harm they would impose on others if not stopped, in which case a lesser-​evil justification could be invoked. The range of innocent agents who retain their right not to be killed includes not only an innocent projectile but a person suffering an epileptic seizure whose motions could kill another, a person who has started to attack another because she drank coffee laced with a behavior-​altering drug, and a person who is about to press a button on her phone that, for reasons she could not have anticipated, will cause a bomb to explode that will kill an innocent victim.7 On their view, a potential innocent victim (I leave the word potential implicit from this point forward) is no more free to kill any of these innocent threats in self-​defense than she would be to kill an innocent bystander in self-​preservation. McMahan and Otsuka also proposed a strategy to limit the implication of their view so that it did not imply that only culpable threats could forfeit their right not to be killed.8 They focused on the responsible choice to impose a risk on another, giving rise to a situation in which someone must die. Here, in essence, is McMahan’s example: Conscientious Driver: Connie, who always keeps her car well maintained and always drives carefully and alertly, decides to drive to the cinema. On the way, a freak event that she could not have anticipated occurs that causes her car to veer out of control in the direction of Peter, a pedestrian.9

 McMahan, 1994, p. 276. Otsuka offered a nearly identical argument, identifying a falling human with “a chunk of granite.” Otsuka, 1994, p. 80. Otsuka also convincingly argued that the view that an innocent threat causes an innocent victim’s rights to be violated—​a view taken, for example, by Frances Kamm—​likewise makes no sense (p. 80n17). 7  I take the epileptic case and the laced-​coffee case from Rodin, 2002, p. 80; the phone-​button case comes from McMahan, 2009a, p. 165. To be clear, in all cases we must assume that the person had no warning that she might have a seizure, drink drugged coffee, or become a projectile. If she had a warning and ignored it, then she knowingly or negligently imposed a risk on others and her case moves into the other category, in which she may have forfeited her right not to be killed. 8  McMahan came to his solution later than Otsuka. 9  This case is almost identical to the one described by McMahan, 2009a, p. 165. 6

Introduction  |  3

McMahan argued that Connie has forfeited her right not to be killed in defense of Peter. As he wrote, she knows “that her action carries a small risk of causing great though unintended harm.”10 He continued: “because she knowingly imposed this risk for the sake of her own interests[,]‌she is morally liable to defensive action to prevent her from killing an innocent bystander.”11 Otsuka similarly offered a case in which self-​defense was, in his view, justifiable because the person against whom it is used has non-​culpably chosen to put another’s life at risk. Hologram: Polly projects a holographic gun onto Howard’s hand as he extends his hand to greet Diane. Diane, taking the gun to be real, reaches for her own gun to shoot Howard in self-​defense. Howard, seeing the holographic gun too, realizes that Diane will shoot him in self-​defense if he does not shoot her first with the real gun he is carrying in a holster. If he acts quickly he can shoot her and save himself—​i.e., they will not both get shot and die.12

Diane’s mistake is reasonable, and she acts non-​culpably, given what she reasonably believes. But given the facts, she threatens to kill Howard without having any right to do so. Otsuka argued that Howard can justifiably kill Diane because Diane “takes a gamble” by choosing to use violence against him.13 In essence, he says that her choice to use force makes her liable to be killed if it turns out, as it does in this case, that it is not in fact a justified choice. The main problem with these responsible-​choice accounts arises from the fact that McMahan and Otsuka accept that there is a strong presumption against intentional killing such that to justify it, as McMahan put it, “there must be some important moral difference between the Threat and [the Victim].”14 The problem is that the responsible choices the threats make in Conscientious Driver and Hologram do not seem to provide an “important moral difference” between the threats and their victims. There’s a Coasian symmetry in the situations that McMahan and Otsuka both ignore that makes it, in fact, unclear who took the responsible choice to create it.  Ibid., p. 166.  Ibid. 12  This case traces back to Otsuka, 1994, p. 91. The formulation I use is modeled closely on one from Frowe, 2014, p. 81. 13  Otsuka, 1994, p. 91. Otsuka, 2016, defends this position again. 14  McMahan, 2009a, p. 169. 10 11

4  |  The Mechanics Of claims And Permissible Killing in war

Seth Lazar seems to be the first person to make this point, and, following Lazar, I focus first on McMahan’s case and then turn to Otsuka’s. Lazar’s central insight is that the “Pedestrian’s voluntary choices also contribute to the forced choice between their lives coming about.”15 Indeed, it is clear that if Peter took unreasonable risks, say walking into traffic, then he would have forfeited his right to defend himself against a conscientious driver bearing down on him. We can assume that Peter took normal precautions not to get hit, but then we are also assuming that Connie took normal precautions not to drive out of control. Why is Connie liable to be killed for the forced choice and not Peter? It might be tempting to recall McMahan’s point that Connie “knowingly imposed this risk for the sake of her own interests.” But we can assume that Peter knowingly assumed the risk of being near traffic for his own interests. Moreover, it is not merely the case that he put himself at risk. If he has the means to defend himself, then he puts drivers at risk too. Given these similarities, there is no “important moral difference” between Connie and Peter. There is a difference: Connie is a threat to Peter. But this difference, by itself, does not seem to be the kind of “important moral difference” that would explain why Connie has forfeited her right not to be killed. One might think that Hologram works better for the assumption of risk argument. After all, Diane chooses not only to put Howard at risk but to try to kill him. But risk is still the key notion in the argument. As Otsuka put it, Diane “takes a gamble  .  .  .  when she engages in such avoidable risky activity” as pulling her gun on Howard.16 But, again, she could not threaten him if he were not there. Moreover, Howard assumes that Diane is attacking him just as she assumes that he is attacking her, and they both pull guns on each other on the basis of that assumption. The only difference between them is that Polly projected the hologram into Howard’s hand and, thus, that Diane is the one who made a mistake about the existence of a threat (at least before he pulls his real gun). But this is just a matter of who had the bad luck to be mistaken. It does not seem to be “an important moral difference.” As Lazar puts it, if we assume that the right to life can be forfeited only because of “something distinctive about us as moral subjects,” then presumably it must be “something beyond mere bad luck.”17

 Lazar, 2009, p. 717.  Otsuka, 1994, p. 91. 17  Lazar, 2009, p. 701. See also McMahan, 2009a, p. 177: whether someone is liable “cannot be merely a matter of luck” (original emphasis). 15 16

Introduction  |  5

Summing up, if we start with the assumption that killing is particularly hard to justify because people have robust rights not to be killed, then it looks like self-​defense will be justifiable only against culpable threats or when rights can be overridden because of strong consequentialist considerations giving rise to a lesser-​evil justification.18 Self-​defense against innocent threats and innocent aggressors seems just as much a rights violation as self-​preservation that kills innocent bystanders. But this has profound implications for just war theory, for, as Lazar points out, “many unjustified combatants are morally innocent of the threats that they pose, excused either by duress or because they reasonably believe their cause to be justified, in conditions of uncertainty and high risk.”19 Thus, we seem to face the following options: 1. Reject the idea that we can learn anything useful about the justifiability of war by thinking about the rights of individuals in the context of self-​defense.20 2. Accept that individual rights limit the justifiability of war and adopt a high threshold for rights forfeiture, accepting that individual rights therefore limit the justifiability of war in a way that implies a strong presumption of pacifism, one that can be overridden only if the result of waging war is so much better in consequentialist terms than not waging war that it justifies infringing individual rights.21 3. Accept that individual rights limit the justifiability of war but adopt a low threshold for rights forfeiture, accepting that unjust combatants will therefore generally have forfeited their right not to be killed, and that the same will often be true of unjust noncombatants who contribute to an unjust war effort.22 4. Embrace the idea that people enjoy strong agent-​ centered prerogatives that permit them to kill in self-​defense even when the attacker has done nothing to forfeit her right not to be killed.23

 Rodin, 2002, p. 97, embraced that position.  Lazar, 2009, p. 701. 20  See, e.g., Benbaji, 2008. 21  Lazar, 2009, p. 728, tentatively embraces this position. 22  Frowe, 2014, adopts this position. She thinks that liability for posing an unjust threat depends on having had “a reasonable opportunity to do otherwise” (p. 73) and that even noncombatants who aid unjust combatants are generally liable to being killed. 23  See, e.g., Quong, 2012. 18 19

6  |  The Mechanics Of claims And Permissible Killing in war

But I reject all of these. I reject the first option because I think restrictions in war should reflect the imperative to respect individual rights.24 I  reject the second option because I think the right not to be killed is not as restricting as the present discussion suggests. I  reject the fourth option because I think agent-​centered prerogatives would wrongly imply that an agent may kill an innocent bystander in self-​preservation.25 This would seem to leave the third option:  accepting a low threshold for forfeiture. But I reject that option too. Appealing to forfeiture still requires us to draw too sharp a line between those who make responsible but non-​culpable choices and the sorts of innocent threats and aggressors who simply cannot be taken to have done anything to forfeit any rights.26 The key to getting past these four bad options is understanding why the right not to be killed is not as restricting as the present discussion suggests. At a superficial level, my solution is to say that merely being a threat who lacks the right to threaten undermines the threat’s claim not to be harmed in defense of the victim. This applies to innocent threats and to minimally responsible threats like Connie and Diane. Though there is no reason to think that they have forfeited their right not to be killed, there is good reason to think that they have weakened claims not to be killed and that it is therefore at least plausible that they may be killed.27 To see why this position on the rights of threats is plausible, I think it helps to go back to the basic ideas about rights that inform the debate. I believe the debate has been going in the wrong direction ever since Thomson suggested that the explanation for why it would be permissible to kill innocent threats has to appeal to their violating the rights of their victims. The mistake lay in thinking of the right to life as something an agent has unless she waives or forfeits it.28 This thought is not easily put aside as it reflects the intuitively compelling idea that rights must provide a robust sort of normative protection. Nonetheless, I think it is misplaced. It is possible for

 I argue this in Chapter 2, § 3.  I argue against agent-​centered prerogatives and for a narrower notion of negative agent-​claims in Chapter 3, § 2.4. 26  Frowe tries to get past this problem by adopting a hybrid of duty-​based reasoning and a lesser-​evil justification first proposed by Tadros. I argue in Chapter 6 that this proposal fails. 27  I argue this point in Chapters 6 and 7. 28  The notion of waiver can be read broadly to cover voluntary renunciation of a claim for the rest time or narrowly to cover only renunciation for a particular instance or set of instances. If read narrowly, then we need to introduce the notion of alienation to describe voluntary renunciation of a claim for the rest of time. I will generally use the broad concept of waiver. 24 25

Introduction  |  7

rights to provide a robust normative protection and for a threat not to have a right not to be killed simply because he is a threat. I will argue for a fundamentally different model of rights, according to which they are the result of the balance of a set of competing claims. Claims are pro tanto rights. That is, they have a normative force that pushes to establish an unqualified right. An unqualified right (I leave the word unqualified implicit from this point forward) emerges from the balance of claims as a normative conclusion, indicating, if it is a patient-​right, that one or more agents has a corresponding duty or, if it is an agent-​right, that the agent has the freedom to do or refrain from doing something. For example, P’s claim not to be killed pushes to establish that P has a right not to be killed. If it is undefeated in the balance, then P has that right; if it is outweighed, then P does not have that right. I will argue that the weight of the claims in the balance reflects more than responsibility for choices made (the stuff of forfeitures and waivers, and of contractual or promissory acquisition of property or duties); it also reflects how a claimant relates to others. Moreover, unlike those who emphasize responsible choice as the key to taming the threat posed by luck, I think we need to embrace ownership of one’s luck as key to making sense of rights. Luck is pervasive, and often what distinguishes how people relate to each other is just a matter of luck. I will lay out this alternative model of rights, which I call the mechanics of claims, in Chapter 3. Before doing so, I want to take care of two preliminary matters. First, I will spend the rest of this chapter describing my position on the right to defend against threats. My purpose in doing so is in part to make explicit what I assume and what I argue for and in part to identify and own what I think the implications of my view are. Second, in Chapter 2, I will describe and defend my methodology for doing moral philosophy, my understanding of moral justifications, my conception of what it means to eliminate a threat, and my view of the relationship between individual rights and just war theory.

2. My Position on the Right to Defend Against Threats My ultimate goal in the following discussion of the right to defend against threats is to formulate an account of the justifiability of killing threats to eliminate the danger they pose, that is, eliminative killing. I see no reason why my position cannot be extended to harms less grave than death; I focus on killing simply because it is the greatest harm we should be prepared to

8  |  The Mechanics Of claims And Permissible Killing in war

try to justify29 and because my discussion is complex enough as is. I also use a relatively narrow understanding of the concept of a threat. I mean a person or thing that is likely to cause harm to a victim if action is not taken. One could say that a person who presents a physical or moral obstacle to a victim seeking to avoid a threat is a threat in a secondary sense.30 But I treat such obstacles as having a different kind of moral significance.31 I should clarify how I understand eliminative killing. One could identify eliminative killing with acting on the intention to eliminate a threat by killing the person who poses it. But, for reasons of analytic clarity and to avoid unduly emphasizing the moral significance of intentions, I  want to frame it, instead, through the lens of a potential justification for the act. I define eliminative killing as (a)  killing one or more other persons in a particular context, one in which (b) the persons being killed pose a threat (whether directly or indirectly) to others and (c) doing what kills them would eliminate, or at least help to reduce the magnitude of, the risk posed by that threat. What interests me is whether and when such acts can be justified by reference to the fact that killing them eliminates or reduces the threat they pose. If an act is justified, then it is permissible to perform it. One can then ask whether an agent must be acting on an intention that corresponds to a successful justification to perform a permissible action, where an action is defined as an act performed for a particular reason.32 Taking acts as basic in this way enables us to distinguish wrongmaking factors that would make it impermissible for any agent to perform the act in question, no matter what her reasons for doing so, from the intentions and subjectively embraced reasons for action that might also be wrongmaking factors. It is particularly important to be able to separate those two kinds of wrongmaking factors because even if an agent’s reasons for action can make an otherwise permissible act impermissible, they are neither the only nor the most important set of factors that can do so.33  I assume that torture is out of the question and that those cases in which causing severe suffering is both worse than killing and not the result of torture are sufficiently rare that they don’t affect the overall analysis. 30  This is the position taken by Frowe, 2014, especially ­chapter 2. 31  I argue in Chapter 3 that there are two kinds of obstacles: those who have relatively weak claims not to be killed because they are obstacles to the agent doing what she proposes to do and those who have relatively strong claims not to be killed because they have prior claims to a resource the agent needs to do what she proposes to do. 32  As far as I know, this distinction between acts and actions was introduced by Ross, [1930] 1988, p. 7. This distinction is also sometimes described as the difference between acts thinly and thickly described. See Wedgwood, 2011, p. 388. 33   I  focus on this point in Chapter  5, where I  argue that intentions are generally not the sort of wrongmaking features of an action that matter for just war theory. 29

Introduction  |  9

I now offer a set of six conditions that together describe my position, which I label permissible eliminative killing. To describe these conditions, I consider a schematic in which an agent, A, can eliminatively kill a threat, T, to protect some victim, V, possibly thereby endangering some bystander, B. To be clear, any of these characters could represent more than one person. In addition, A could be identical to V. I discuss them separately on the assumption that other-​defense and self-​defense are justifiable under the same conditions. 1.  T is a threat to some possible victim, V, and killing T will eliminate T’s threat to V or at least reduce the magnitude of the harm to V or the risk that V will be harmed. This condition sets the context for the discussion of eliminative killing. There are two things to say about this condition. First, I use the success verbs eliminate and reduce because it is important to distinguish the person who thinks her act will eliminate or reduce the threat but who is mistaken and is at best excused from the person who is correct on that score and who would be justified in killing T if the other requirements are met. Second, the risk clause is important because sometimes reducing risk is the best that an agent can hope for, and eliminative killing is sometimes justified if the risk reduction is sufficient. This may seem to sit uneasily with the idea of conditioning the justifiability of eliminative killing on successfully eliminating or reducing the threat, but I will argue in Chapter 2, section 2, that these ideas can be reconciled. 2. V’s claim to avoid the harm that T would have caused him is strong enough to outweigh T’s claim not to be killed and B’s claim not to be harmed by the eliminative killing of T. This is the first of the two proportionality clauses. I call it baseline proportionality as it compares the harm caused by eliminative killing to the baseline of doing nothing. A’s act of eliminative killing is permissible only if it is baseline proportional.34   This condition is, strictly speaking, problematic and unnecessary. It is problematic because in complicated cases it is often unclear what it means to say that A “does nothing.” (I am grateful to Benjamin Bronner for making this clear to me.) It is unnecessary because one can get its content out of condition 4, which deals with comparative proportionality. One can simply set A’s act to “allowing the threat to take place” (in however many meaningfully different ways A might do that) to see whether among the acts that pass the test of comparative proportionality is one that involves 34

10  |  The Mechanics Of claims And Permissible Killing in war

I want to highlight three things about this condition. First, proportionality is not just a matter of the relative harms that T, V, and B might suffer; it reflects the extent to which T, V, and B might have waived or forfeited their claims not to be killed or harmed and the way in which different kinds of claims register even when at full strength. I  argue, starting in Chapter 3, that they do not all register the same. Second, I focus on harms that might be caused to T or B and on harms that T might cause V if not killed. One might also consider weighing other benefits that might be enjoyed by V, B, or A  herself (e.g., one of these people will now inherit wealth that would have gone to T). But these sorts of side benefits cannot help to justify killing T.35 Likewise, one might consider weighing the good things that T would have done had T been left alive or the good things that V, B, or A herself would have done if A had chosen not to eliminatively kill T.  But these sorts of side benefits carry very little weight in helping to justify the thought that it would be disproportional to kill T to save V.36 Third, I leave A’s interest out of the balance because the point is to ask whether it would be proportional for her to kill T on the assumption that she is willing to do so. Even if it is proportional, she often will have the right to choose not to kill T if she finds doing so too costly for her. But that is a separate issue, which comes up under comparative proportionality. Conversely, if T’s death is too great a harm to inflict on him for the sake of saving V from some lesser harm, then A’s interest in killing T will not make it permissible.37 3.  There is no alternative way to protect V to the same degree that is Pareto superior in terms of the claims of V, T, B, or A. This is the traditional necessity condition, which also sets a necessary condition for permissible eliminative killing. It is of fairly limited use as

A eliminatively killing T. Nonetheless, I think it is important to describe this condition independently because it captures what people intuitively mean by “proportionality” when they discuss “necessity and proportionality” as two tests that must be met for an act of eliminative killing to be justified.   See McMahan, 2013–​14, p.  4. I  say a bit more about benefits and why some might not count against harms in Chapter 3, § 1.2.3. 36  I think this limit reflects the idea that V cannot be asked to sacrifice for the benefits others would enjoy. I return to this idea of a right of non-​sacrifice in Chapter 9. 37  I argue for this point in Chapter 3, § 2.1.4. Note: I assume that if A is a corporate agent, like the state, then it has no agent-​claims. The duties it owes to its stakeholders should not be taken to give it agent-​claims; rather, they should be framed as special claims on it. I  define special claims in Chapter 3, § 2.2. 35

Introduction  |  11

the conditions in which it forbids an act are unlikely to occur. If, for example, some alternative act to the one under consideration would cause far less harm to T but cost A a little more effort, then we cannot rule out that act on the ground that it is unnecessary. We would need to look to the second of the proportionality clauses. Before moving on to this second proportionality clause, it is worth linking necessity and imminence. Traditional doctrine in both criminal law and the law of armed conflict requires that a threat be imminent before it is permissible to use defensive force against it. But I agree with David Rodin that the imminence requirement “is simply the application of the necessity requirement subject to epistemic limitations. The point is that we [normally] cannot know with the required degree of certainty that a defensive act is necessary until the infliction of harm is imminent.”38 Given that imminence is at best an imperfect proxy for necessity and that I take rights to turn on fact-​relative justifications, not evidence-​relative justifications,39 I will put imminence to the side and focus directly on necessity and the related proportionality notion. 4.  If A can choose between two alternative acts, act 1 and act 2, and, compared to act 2, act 1 would harm T less but would provide less protection to V or impose more cost on A or B, then act 2 is permissible only if the claims of V to more protection or of A and B to have to bear less cost outweigh T’s claim to suffer less harm. This condition describes comparative proportionality. It fills in when the necessity clause fails to apply. Any option that fails this test is ruled out just as it would be if it failed the necessity test. All of the preceding comments about condition 2 apply to this condition, with this exception: A gets to consider her own interest as part of the comparative proportionality balance and take an option off the table if it is too costly to her, though she may also waive her right to do so.40 As it is awkward to refer to “baseline proportionality and either necessity or comparative proportionality,” I will use the familiar phrase “necessary and proportional,” with the understanding that necessity refers to necessity or comparative proportionality.

 Rodin, 2002, p. 41  See Chapter 2, §§ 2.2 & 2.3. 40  I say more about an agent’s right to take options off the table in Chapter 3, § 2.1.4. 38 39

12  |  The Mechanics Of claims And Permissible Killing in war

5.  T lacks the right to threaten or contribute to the threat to V in the way that he does. This condition represents what I call the agent–​patient inference: if T has the right to threaten V, then A  may not eliminatively kill T to save V. This is a controversial idea, especially in just war theory, but I argue for it in Chapter 8. 6.  There are no other moral reasons (such as the importance of adhering to the law) that prohibit A from killing T. This final condition is a catch-​all for other moral reasons, such as the importance of obeying the law or problems that would arise from undermining important social norms, that would make it impermissible to eliminatively kill a threat. I return to consider this condition in Chapter 10. In listing and describing these six conditions, I have tried to flag the particularly controversial ideas that I will argue for and where I will argue for them. But the most basic ideas, such as that baseline proportionality has to refer to the balance of claims, are ones that fall out of the model of rights that I will introduce in Chapter 3.

3. The Thesis in Practice What does permissible eliminative killing mean in practice? It does not mean that terrorism is permissible. To justify (or attempt to justify) an act as an act of terrorism is to appeal to the political or military utility of conspicuously harming other humans. Usually the definition is restricted to conspicuously harming noncombatants for political purposes41 as it is normally thought part of just warfare to seek to harm combatants and thereby not only eliminatively kill them but also discourage other combatants so that they surrender.42 But to avoid simply assuming that this is permissible and to keep a parallel with eliminative killing, which can be directed at  There is no standard legal definition of terrorism, and the term is sadly politicized because so many want to use it to justify their attacks on others as “terrorists,” while at the same time insisting that the morally dubious acts that they support are the acts of freedom fighters or some such thing. My definition seeks to cut through these politicized uses. 42  See Lazar, 2015, pp. 70–​71: “Victory is achieved not by killing every individual who contributes to a threat against us, but by communicating to the enemy armed forces and their leadership . . . that the costs of continuing to fight outweigh the expected advantages of doing so. . . . Killing in war is always pour encourager les autres.” 41

Introduction  |  13

combatants and noncombatants, I want to include all humans as possible targets of terrorism. My goal in doing so is to highlight the contrast between what it takes to justify (or attempt to justify) an act as an act of terrorism and what it takes to justify (or attempt to justify) an act as an act of eliminative killing. To justify an act as an act of eliminative killing is to appeal to the fact that the killing directly eliminates or reduces a threat. The word directly is important. Eliminative killing, as such, does not reduce a threat by way of deterring potential future threats. Eliminative killing may have the effect of influencing others and thereby reducing future threats. But the act can be justified as an instance of eliminative killing if and only if the direct reduction in the threat justifies the killing.43 So much for what permissible eliminative killing does not entail in practice. What does it entail? One thing it strikingly entails is that a wide range of civilians might permissibly be targeted for eliminative killing. Here, I use the word civilian, a legal term meaning a person who is not formally a member of the military, rather than noncombatant. I do so to make room for the possibility that certain civilians function as combatants; in general, I  will be focused on the functional distinction between combatants and noncombatants, but in this context, for reasons that will immediately be made clear, I need to use the legal term. Consider, for example, civilians who design or manufacture weapons; who manufacture parts for weapons; who provide other equipment needed by combatants, such as reconnaissance drones or hardened vehicles; who do linguistic analysis of information intercepted from the enemy; or who raise money to provide resources for combatants. Assuming that one is fighting a just war, that it will take some time to win the war, and that the efforts of these civilians are prolonging both the war and the killing of innocents, then it may be permissible to eliminatively kill them. Moreover, the point is not simply that they may be killed in the effort to prevent them from performing specific instances of the relevant threatening acts. Permissible eliminative killing holds that they may be targeted wherever they are, whatever they are doing, as long as doing so is necessary and

 I am not here arguing that targeted killing for terroristic (or, more broadly, instrumental) purposes is necessarily ruled out. For warfare to be permissible, it had better be true that unjust combatants have generally forfeited their right not to be targeted for terroristic purposes. Inquiring whether this is true is beyond the scope of my book, which focuses on eliminative killing, not terrorist killing. I do, however, touch on the theme again in Chapter 2, § 3.2. And in Chapter 7, I criticize Frowe for proposing a theory of forfeiture with such a low threshold that it seems noncombatants would too often be liable to terrorist killing. 43

14  |  The Mechanics Of claims And Permissible Killing in war

proportional to reducing the threat posed by an unjust enemy and is not for some other reason morally prohibited. It is probably true that one would be more certain that an agent would kill only those civilians who can justifiably be eliminatively killed if she targets the threatening acts themselves where they normally take place, like the proverbial (in writing on just war theory) munitions factory. But if such a factory is hard to reach (say it is buried deep underground) and the most effective option is to kill the workers at home, that may be permissible too.44 This may seem like a radical thesis. It may seem to conflict with a core principle in the law of armed conflict: the principle of distinction,45 according to which combatants must try to distinguish between civilians and combatants on the other side and must direct their attacks only against combatants. It is tempting to view this as a fundamental moral requirement as well. That is, it is tempting to say that as long as combatants adhere to the rules governing conduct in war, they are morally permitted to attack opposing combatants but not opposing civilians. Civilians may be killed, on this view, only as a side effect of targeting legitimate military targets and only to the extent that their deaths are unavoidable if a just military campaign is to achieve sufficiently important military objectives.46 In truth, however, permissible eliminative killing is not a “radical” departure from the principle of distinction. It can be understood as a moral extension of a well-​established exception to the principle of distinction, an exception that permits attacks on civilians insofar as and for so long as they act as combatants.47 Interpretation of this exception is not well-​settled law, but one of the more influential discussions of it, found in an opinion of the Israeli Supreme Court, called Public Committee Against Torture in Israel v. Israel,48 states  Some writers who are sympathetic with the claim that noncombatants may be eliminatively killed have shied away from the full implications of their reasoning. See, e.g., Draper, 2016, p. 200. 45  The principle of distinction is listed as “rule 1” by the International Committee of the Red Cross in its database on customary international humanitarian law:  “The parties to the conflict must at all times distinguish between civilians and combatants. Attacks may only be directed against combatants. Attacks must not be directed against civilians” (https://​www.icrc.org/​customary-​ihl/​ eng/​docs/​v1_​cha_​chapter1_​rule1). 46  Rule 14 encodes the proportionality limit: “Launching an attack which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage anticipated, is prohibited” (https://​ihl-​databases.icrc.org/​customary-​ihl/​eng/​docs/​v1_​cha_​chapter4_​rule14). 47  Rule 6 qualifies the protection of rule 1: “Civilians are protected against attack, unless and for such time as they take a direct part in hostilities” (https://​www.icrc.org/​customary-​ihl/​eng/​docs/​v1_​ cha_​chapter1_​rule6). 48   HCJ 769/​ 02 (2005) (http://​www.haguejusticeportal.net/​Docs/​NLP/​Israel/​Targetted_​Killings_​ Supreme_​Court_​13-​12-​2006.pdf). For a more restrictive interpretation of the exception that allows 44

Introduction  |  15

[A]‌civilian taking a direct part in hostilities one single time, or sporadically, who later detaches himself from that activity, is a civilian who, starting from the time he detached himself from that activity, is entitled to protection from attack. He is not to be attacked for the hostilities which he committed in the past. On the other hand, a civilian who has joined a terrorist organization which has become his “home,” and in the framework of his role in that organization he commits a chain of hostilities, with short periods of rest between them, loses his immunity from attack “for such time” as he is committing the chain of acts. Indeed, regarding such a civilian, the rest between hostilities is nothing other than preparation for the next hostility.49

Permissible eliminative killing has exactly the same structure with regard to the temporal element, though placing no emphasis, other than epistemic, on joining any sort of organization.50 The way permissible eliminative killing differs from the Israeli court’s opinion is with regard to the significance of taking “a direct part in hostilities.” The court says, on the one hand, that “a civilian bearing arms (openly or concealed) who is on his way to the place where he will use them against the army, at such place, or on his way back from it, is a civilian taking ‘an active part’ in the hostilities.”51 On the other hand, “a civilian who sells food or medicine to unlawful combatants is . . . taking an indirect part in the hostilities.”52 In the gray area in between, the court suggests using a broad interpretation of “directly” engaging in hostilities, so “civilians are encouraged to stay away from the hostilities to the extent possible.”53 In practice, the court takes that to mean [T]‌he following cases should also be included in the definition of taking a “direct part” in hostilities: a person who collects intelligence on the army,

the targeting of civilians insofar as and for such time as they take a direct part in hostilities, see Melzer, 2009.  Ibid., ¶ 39.  I think court president Aharon Barak, the author of the opinion, would agree that membership in an organization is significant merely for providing evidence of someone’s ongoing activities. 51  Ibid., ¶ 34. The court cheats a bit when it includes the civilian on the way back with the arms. If this is his one act of supporting hostilities and he is on his way home, he may be liable to arrest but not to targeted attack. The court is really relying on the thought that when he is so closely connected to a hostile act, he must demonstrate, in some way, that he is done with engaging in hostilities to regain his protected status. 52  Ibid. 53  Ibid. 49 50

16  |  The Mechanics Of claims And Permissible Killing in war

whether on issues regarding the hostilities . . . or beyond those issues . . . a person who transports unlawful combatants to or from the place where the hostilities are taking place; a person who operates weapons which unlawful combatants use, or supervises their operation, or provides service to them, be the distance from the battlefield as it may.54

The bottom line for the court is that “[a]‌ll those persons are performing the function of combatants. The function determines the directness of the part taken in the hostilities.”55 Without meaning to gainsay the Israeli court’s interpretation of the law, my defense of permissible eliminative killing implies that, from a purely moral point of view, its emphasis on the function of combatants is misplaced. If the aid civilians provide makes it possible for unjust combatants to attack people whom they have no right to attack, such that interrupting that aid would reduce the threat posed by those unjust combatants, then we have passed the threshold question for justifying eliminative killing; we now have to engage the other moral considerations that might nonetheless make it impermissible. It might turn out to be impermissible, all things considered, but the directness or indirectness of the threat and its connection or lack of connection to any traditional activities of combatants are, I argue, morally irrelevant, except in two secondary ways:  (1) the legal prohibition on targeting civilians gives some moral weight to the pro tanto duty not to target civilians,56 and (2) the indirectness or legality of their harmful activity might tend to make them less morally culpable, and thus less likely to have forfeited their right not to be killed, which, in turn, affects the conditions under which targeting them is permissible.57 In sum, my thesis is not as radical a departure from the principle of distinction as it seems at first blush. In addition, the permissive aspects of permissible eliminative killing break no new ground in the literature

 Ibid., ¶ 35.  Ibid. For a broader treatment of this topic, see Melzer, 2009. According to Melzer, some experts on the law of armed conflict think that activity should count as direct if its impact is large and unique: “in extreme situations, namely where the expertise of a particular civilian was of very exceptional and potentially decisive value for the outcome of an armed conflict, such as the case of nuclear weapons experts during the Second World War” (p. 57n122). Tadros offers a slightly different reading of the law. According to him, the law uses some notion of causal proximity to harm to define what it means to be directly taking part in hostilities. See Tadros, unpublished manuscript, pp. 2–​4. In the end, these different accounts do not matter for my point. 56  I return to this point in Chapter 10, § 1. 57  I return to this point in Chapter 4, § 2.1. 54 55

Introduction  |  17

on just war theory. Something similar has been defended, for example, by Jeff McMahan, who writes “[T]‌he account of liability to attack in war that I have defended cannot rule out the possibility that civilians may be liable to intentional attack.”58 Helen Frowe adds “that non-​combatants are often liable to [defensive] killings, and that this conclusion follows naturally for those who think that moral responsibility for an unjust threat renders a person liable to defensive harm.”59 My account differs from theirs, in terms of the extent to which I  justify the eliminative killing of noncombatants, only by being slightly more permissive, and it has that effect only because I reject their reliance on forfeiture (i.e., liability). Importantly, the implications of permissible eliminative killing are not only permissive; they are also restrictive. The theory holds that eliminatively killing just combatants is impermissible. This too may seem to be a radical thesis, given that the law of armed conflict imposes no such restriction. But, in truth, this is not a radical thesis in the history of just war theory.60 It is also not a part of the thesis that I will spend much time on.61 What I  will spend more time on, with regard to restrictions, and what is slightly more radical is the broader claim that it is impermissible to eliminatively kill, or even to interfere with, the threatening acts of those who have a right to engage in such acts. In doing so, I take a more hard-​line position on the impermissibility of responding to those who have a right to act than any others of whom I am aware. McMahan comes close when he argues that bystanders who would be killed as a side effect of the acts of justified combatants may not kill them in response.62 But even he thinks that the bystanders may still try to defend themselves by jamming the weapons of justified combatants. I argue, in Chapter 8, that even that is not permissible. If the acts of combatants are justified, they may not be resisted. And if the acts of noncombatants that contribute to

 McMahan, 2009a, p. 221.   Frowe, 2014, p.  187. See also Draper, 2016, pp.  198–​202 (discussing how munitions workers might be liable to defensive [i.e., eliminative] killing). 60  See McMahan, 2009a, p. 35. McMahan acknowledges that other major figures in the just war tradition have also taken this position: Francisco de Vitoria and Francisco Suárez in the sixteenth and seventeenth centuries and G. E. M. Anscombe in the twentieth (p. 34). Both Reichenberg, 2008, and Steinhoff, 2012, emphasize that this view has deep roots, going back to Augustine, Thomas Aquinas, and Hugo Grotius. Steinhoff characterizes the contrary view as a modern aberration, due mostly to Michael Walzer, 1992. But Reichenberg (p. 193) traces the contrary view to the fifteenth century, with “roots in ancient Rome.” 61  I do discuss it in Chapter 8, §§ 3.3 and 4. 62  McMahan, 2014b, p. 110. 58 59

18  |  The Mechanics Of claims And Permissible Killing in war

the threat posed by others are justified, they too may not be targeted for interference.

4. The Structure of the Argument The argument in this book has four main parts. This chapter and Chapter 2 are the introductory chapters. Chapters 3 through 5 comprise the second part of the book, laying out the case for the mechanics of claims63 and for rejecting two competing theories. One of these competitors is the currently dominant model of rights, which I  call the infringement model. This is the model according to which there are two ways to fail to respect rights:  to impermissibly violate them or to permissibly infringe them.64 The other competitor theory competes at a different level. It is a competitor theory not of rights but of the underlying normative foundation for just war theory. It is the view that the intentions of agents are of fundamental moral significance. Chapters 6 and 7 form the third part of the book, the part most directly concerned with justifying eliminative killing. Chapter 6 makes the argument that eliminatively killing innocent threats is easier to justify than killing innocent bystanders as a side effect—​the latter presenting the issue of killing as collateral damage. Chapter 7 then builds on this foundation to argue that the eliminative killing of noncombatants who contribute to the threats posed by unjust combatants is also permissible under similar conditions. Chapters 8 and 9 constitute the last part of the book. They examine the last substantive condition in permissible eliminative killing, condition 5, which prohibits eliminatively killing anyone for doing what he has a right to do. In Chapter 8, I argue that it provides an agent the right not to be eliminatively killed. In Chapter 9, I argue that indirectness, which reflects intervening agency, can be relevant to having the right to act. But I also conclude that it has limited relevance to just war theory because intervening agency provides an agent the right to act only when it creates a causal structure in which she is exercising the right not to sacrifice herself for others. This sort of transformation is important for cases of self-​defense, but its relevance to just war theory is more or less limited to cases in which people face the choice of submitting to an invasion or resisting and thereby

 I introduced this model in Walen and Wasserman, 2012.  This model’s most influential expositors are arguably Feinberg, 1980, and Thomson, 1990.

63 64

Introduction  |  19

bringing it about that an invader harms others. In other words, generally, intervening agency will not provide a normative basis for blocking the justification of eliminative killing. Lastly, because I  invent a lot of new jargon in the course of developing my argument, I include at the back of the book, in addition to the index and a table of cases, a glossary providing brief definitions of the key theoretical ideas.

20  |  The Mechanics Of claims And Permissible Killing in war



2 Basic Premises and Method

T

his chapter covers four themes in two groups. The first group concerns my general approach to moral theory. It includes the role of theory and cases in my work and the role of evidence and facts in justification. The second group concerns two topics specifically relevant to just war theory: reductive individualism and the distinctive causal structure of eliminative killing.

1. Reflective Equilibrium and the Relative Balance of Theory and Case Intuitions The reigning method in moral philosophy is the search for reflective equilibrium. As John Rawls wrote, one searches for a theory that “as a whole articulates our more firm considered convictions . . . at all levels of generality, after due examination, once all adjustments and revisions that seem compelling have been made.”1 In practice, this means that one seeks to develop a theory that is grounded in high-​level principles that are morally plausible and relevant, keeping in view how they work to explain mid-​ level principles as well as judgments in particular situations. Insofar as one cannot match one’s considered judgments in particular situations with a theory consistent with one’s high-​level principles, one has to adjust one or both to try to find a better fit. An interesting feature of contemporary moral philosophy is how much weight most theorists put on matching intuitions in test cases. My own

 Rawls, 1993, p. 28.

1

inclination is to lean the other way, to put more weight on finding plausible, relevant, high-​level moral principles that can be used to generate mid-​level principles and judgments in particular situations. This is for two reasons. First, there are good reasons to distrust intuitions in particular cases. Intuitions are known to vary depending on influences as diverse as cultural prejudices, the mood of the person offering a judgment, and the order in which questions are posed (i.e., framing effects).2 This is not to say that intuitions in particular cases are morally irrelevant or useless. Insofar as a certain result is clearly morally implausible, that can serve as a reason to reject a theory that implies it. In addition, moral intuitions can “usefully prod moral theorists into developing richer, more nuanced theories, [which] would then have to be tested to see whether they stand up to scrutiny in terms of other basic moral commitments.”3 Thus, I do not shy away from discussing intuitions in particular cases. But I also try not to put too much weight on matching them in seeking reflective equilibrium. Second, I think it makes sense to put more weight, relatively speaking, on good theoretical work. I have more trust in theory that articulates plausible, relevant, and generative moral principles, as long as those principles do not produce clearly implausible results, than I do in raw intuitions. My reason is simple: I think we have a better grasp of the kinds of reasons that could count as fundamental moral principles than we have of what counts as a morally sound judgment in particular cases. The fundamental principles have to make sense when asking the question, What is the point of taking moral reasons seriously? This is a project we can engage with in a rich way. By contrast, it is at best mysterious why we can rely on an intuition about whether, for example, A killing T in a particular artificial and unusual situation is morally permissible. Maybe most (or some) people have reliable, inchoate insights into such truths; but it is unclear why we would think so without being able to ground those insights in a deep, plausible moral theory. Thus, my discussion tilts in favor of focusing on theoretical plausibility. This methodological tilt will be evident whenever I  embrace a counterintuitive result because the theoretical reasons for doing so seem to me to outweigh the appeal of widely shared moral intuitions.4

 See Walen, 2014, p. 461.  Ibid. 4   For example, I  reject the widely shared intuition that people whose rights are infringed are “wronged” (Chapter 4), and I remain agnostic about the common intuition that an agent may use lethal force to defend herself against an innocent threat (Chapter 6). 2 3

22  |  The Mechanics Of claims And Permissible Killing in war

2. Knowledge and Justification 2.1. The Use of Cases with Unrealistic Assumptions of Knowledge Having mentioned my reason for being wary of putting too much weight on cases, I want to take this opportunity to say a few things about and on behalf of the kinds of cases that I and many other moral philosophers use. My first point concerns the worry that my cases unrealistically attribute knowledge to agents. I discuss cases in which people face threats and an agent has to consider making trade-​offs, like whether to turn a trolley from a track where it would run over and kill five people to a track where it would kill only one or whether to save another at some cost (e.g., a broken foot) to herself. Some think that it is problematic to use cases like these, presupposing that the agents know exactly what their options are and what the consequences of choosing each option would be. Real agents don’t operate in the world like that. How should a person at a trolley switch know that the five would all be killed if she does not turn the trolley? Why can’t they get out of the way? Why can’t the one on the other track get out of the way? How can an agent know that she will break her foot if she tries to save another? Might she not come out fine or suffer worse harm? These sorts of questions are all perfectly fair if we want truly realistic cases. But they also miss the point. The point of using cases as I and most other moral philosophers do is that we want to establish what one ought to do in various circumstances. We stipulate circumstances to test theories. Of course, a real agent facing the choice of deflecting a threat from five onto one would normally not know all sorts of relevant facts, nor would she have time to think much about what seems to be true. Nonetheless, moral philosophers can still hope to provide guidance by making certain factual assumptions and working out, in advance, the sorts of considerations that are truly relevant. This guidance, we may hope, might prove to be useful to people who confront difficult choices in the messy circumstances of real life.

2.2. Two Audiences and Two Kinds of Justification To appreciate how rights work, it will help to distinguish two fundamentally different audiences for moral guidance:  an agent herself and those who have the potential to interfere with her or help her. The distinction between these audiences arises because a third party might know more

Basic Premises And method  |  23

than an agent. Even if an agent does what she ought to do, given what she knows and can readily come to know, a third party who knows more might realize that she does not have a right to do what she does. In such a case, she may be blameless, but that does not mean that he should treat her as though she really has the right to do what she reasonably but mistakenly thinks she has the right to do. To capture that distinction between different points of view on an agent’s act, I think it best to judge the agent using an evidence-​relative account of how to justify her acts. She ought to use the best evidence available to her, and if she does so, she is blameless. But a third party, operating from a different epistemic perspective, should appeal to a fact-​relative justification for her acts. This enables him to justify interfering in cases in which he realizes that she has made a mistake about the morally relevant facts.5 To illustrate how evidence-​relative and fact-​relative justifications can come apart, consider the following: Mine Shafts6: A hundred miners are trapped underground, with floodwaters rising. Amy is a rescuer on the surface trying to save these people. There are two mine shafts, and Amy knows that all the miners are in the same one, but she doesn’t know which. There are three floodgates that she could close by remote control. The results would be as follows:

Gate 1 Amy closes Gate 2 Gate 3

The miners are in Shaft A She saves 100 lives She saves no lives She saves 90 lives

Shaft B She saves no lives She saves 100 lives She saves 90 lives

In a fact-​relative sense, the best thing to do depends on where the miners are. If the miners are in shaft A, it would be best if Amy closed gate 1; if they are in shaft B, it would be best if she closed gate 2. But given that to the best of Amy’s knowledge they are equally likely to be in either shaft, the expected benefit from her point of view of closing either gate is only a 50 percent chance of saving 100, or 50 lives saved. The expected benefit of closing gate 3 is much greater: 90 lives saved. Given the evidence

  For the distinction between evidence-​relative and fact-​relative justifications, see Parfit, 2011, pp. 150–​58. 6  This is based very closely on the case by Parfit (ibid., p. 159). I believe the first person to present such a case was Regan, 1982. 5

24  |  The Mechanics Of claims And Permissible Killing in war

available to her, then, closing gate 3 is the only justifiable choice; it is what she ought to do. Now suppose that the miners are in shaft B.  And suppose that Amy is sitting alone in the control booth. She learns of the danger to the 100 miners and does all she can to find out where they are. But she has only five minutes to act; after that the system will switch into default mode and close gate 2. That means that if she does nothing, all will be saved. But as the five-​minute mark approaches, Amy prepares to do what she ought to do; she prepares to close gate 3. Now let’s add a third party, Ben, who knows where the miners are, can see what Amy is about to do, but cannot communicate with Amy. Ben can, however, press a button that will cause the control panel to become electrified; if and when Amy touches it, she will be electrocuted, which will prevent her from closing gate 3 but will also kill her. May Ben press that button? Amy is doing what she ought to do, yet if Ben does not press that button, 10 extra miners will die. It seems to me that Ben may press that button. One way to make sense of that—​the way I adopt—​is to say that Amy in fact has no right to close gate 3.7 Another straightforward way of describing this case is to say that Amy has a right to close gate 3 and that Ben may nonetheless kill her to prevent her from doing so. In other words, one might take the position that third-​party interference is something to be justified on its own terms; it should not, on this view, be conflated with or tied to Amy’s having a right to act. There is nothing inherently confused about this way of talking, but it deprives us of the ability to stake out a useful position in rights. I will argue in Chapter 8 that there is good reason to say that if an agent has a right to do X, then a third party may not interfere unless there is some reason to think that competition is justified in that realm of activity. As I think competition is not justified in Mine Shafts, I conclude that Amy’s act is one that she has no right to perform. Of course, Ben too might be mistaken about the facts. Ben can only guide himself by his best assessment of the evidence, and that determines what he ought to do. But if, in his best assessment, he concludes that he knows something that Amy does not know, that Amy is about to do something that she has no right to do, and that the miners’ claims to be saved from Amy’s choice

 Foreshadowing: I think Amy may not be held to have forfeited her right not to be killed for doing what she ought to do. She may be killed only because, as a matter of fact, she poses a threat to 10 people’s lives and her claim not to be killed is weaker than their claims to be saved from her. I return to discuss the conditions of forfeiture in Chapter 4, § 2.1. 7

Basic Premises And method  |  25

are stronger than Amy’s claim not to be killed, then Ben should take himself to have a right to save the miners, even at the cost of killing Amy. If some further third party, Cindy, realizes that Ben is mistaken, then Cindy may have a right to harm Ben to save Amy. And so on. The point is not that the fact-​relative point of view is somehow magically available to third parties. Rather, the points are these. First, the evidence-​relative point of view of one agent, in this case Amy, does not necessarily govern the acts of a third party, such as Ben, who is considering interfering with that agent on the basis of different evidence. Second, the fact-​relative account of rights enables us to usefully gain critical normative distance on an agent’s acts, one that gives us the ability to take into account facts not available to the primary agent.

2.3. The Case for a Qualified Fact-​Relative Framework I ended the previous section with the suggestion that the need to take into account facts that an agent does not know implies that we should adopt a fact-​ relative account of rights. But there may often be esoteric facts such that, if only they were known, then we would recognize that what seems necessary is not in fact necessary. This seems to imply that we are often wrong when we think that we have a right to do harmful things. This seems to imply, in turn, that the notion of a right to cause harm is actually not very useful as we often have no way of knowing if it applies. But given that such practical notions exist to be useful,8 that seems to provide a reductio ad absurdum of the claim that the right to cause harm is to be justified in a fact-​relative way. The way out of this problem, I suggest, is to say that esoteric facts are irrelevant to fact-​relative justification. What matters are only facts that are sufficiently accessible to the agents in a situation. To make this concrete, consider the following generic scenario: Preventing Murder: Mark is planning to murder Victor, and Andrea reasonably thinks that the only way to stop Mark, given how close he is to killing Victor, is to shoot Mark.

Here are three sorts of facts that show that Andrea is mistaken. Startled Prevention: Mark is easily startled and very sensitive to loud noises. All Andrea needs to do to prevent him from killing Victor is shoot her gun in the air and Mark will drop to the ground, quivering in fear. 8

 See Scanlon, 2008, pp. 47–​48.

26  |  The Mechanics Of claims And Permissible Killing in war

Spooky Prevention: Mark would be unnerved if Andrea could reveal to him that she knows all sorts of intimate facts about him that she should not know, such as what he had for breakfast. If she shared those facts with him, he would be concerned that he could not get away with his crime and would desist.9 Future Tech Prevention: Andrea has nearby her all the materials necessary to create a chemical that would temporarily paralyze Mark. They could, in fact, be assembled in a moment. But no one yet knows how to combine these materials in this way; it is a fact that will not be discovered for another 100 years.

These cases are presented in order from fairly accessible facts to facts that are completely esoteric. I’ll assess them in the same order. The relevant fact in Startled Prevention is the kind of fact that makes it easy to say that Andrea does not have a right to shoot Mark. We can easily imagine a third party, Tom, who knows how easily Mark is startled and who would want to try to use that knowledge to save Mark, say by firing his gun in the air first, hoping to startle both Andrea and Mark into not firing. Whether or not such a third party exists, the fact that the information that Andrea is missing is not particularly esoteric means that we would hope that someone would do what Tom might do. That is, we readily conceive of Andrea as being merely excused, rather than as acting as she has a right to act. The relevant fact in Spooky Prevention is more esoteric. If Tom had that knowledge and could share it either with Andrea or with Mark in such a way that it would prevent Mark from murdering Victor, then we would have reason to say that she has no right to shoot Mark. But this sort of information is inherently esoteric:  that’s why it might work; it would be spooky if someone not intimate with Mark had that knowledge. If Tom were not fairly certain that he could wield this information to save Mark, the more natural thing to say is that the mere existence of this fact—​that there are intimate details of Mark’s personal life that he would not expect others to know—​does not undermine Andrea having the right to shoot Mark.  I base this case on one that Steinhoff, forthcoming, gives about how Hitler could have been stopped. Steinhoff also introduces an example involving a woman about to shoot another to prevent him from murdering a third person, and he suggests that her act is not necessary because she could shoot his bullet out of the air, rather than shoot him. I find that a less interesting example because, even if it is physically possible to shoot a bullet out of the air, a human agent cannot reliably do so and, thus, it is irrelevant to reasonable decisions about how to act. 9

Basic Premises And method  |  27

The relevant fact in Future Tech Prevention is, by hypothesis, not available to anyone at present. It is completely esoteric. Such facts in no way undermine Andrea’s right to shoot Mark. This rejection of the relevance of esoteric facts to claims that certain actions are not in fact necessary is especially important for handling cases with risk. Consider the following: Risky Vaccine: Deborah is vaccinating a number of children against a new, deadly strain of the flu. The odds of dying from the flu are 1 in 1,000. The odds of dying from the vaccine are 1 in 100,000. For all Deborah knows, it is much better for each child to get the vaccine than not. However, if Deborah knew that a particular child was likely to have an adverse reaction to the vaccine, she would be obliged not to give it to him.

Suppose that there is some genetic condition that makes the vaccine 100 times more dangerous than the flu for any child with that condition. And suppose that one of the children Deborah is about to treat, Carl, has that condition. But suppose that the test for this condition has not yet been developed. We can say, on the one hand, that it is factually unjustifiable for Deborah to give Carl the vaccine. But until the test has been developed and made available to doctors like Deborah, this fact is esoteric. We should say that she has the right to give Carl the vaccine because, for all she and anyone in a position to interfere know, she does what is in fact justifiable, given the risks. This notion of an esoteric fact is obviously a scalar one; facts are more or less esoteric. It is hard to say exactly how esoteric a fact has to be in order to say that it could not undermine the claim that an agent is justified in causing harm. But the normative idea, I  hope, is clear enough:  there has to be a practical link between the existence of facts and the normative guidance of rights. If the facts are too esoteric to be of practical value for the agent or third parties, then they should be counted too esoteric to matter for fact-​relative justification.10 Looked at another way, facts should matter as their particular relevance becomes known or knowable, not as a class that merely highlights what we already knew: that we are far removed from being gods.

 What about the retrospective judgment that an act that caused harm was unnecessary because of facts revealed at a later date? I think retrospective judgments should track contemporaneous ones. If the fact was sufficiently knowable, then it makes sense for the agent to regret not having known it and to regret having unnecessarily caused harm; if the fact was sufficiently esoteric and she did what 10

28  |  The Mechanics Of claims And Permissible Killing in war

Before closing out this section, I want to make one more point about necessity. One might object that I’ve overlooked a reason why Andrea’s shooting Mark is not necessary to prevent him from shooting Victor, a reason that has nothing to do with esoteric facts. It is unnecessary, one might say, as long as one accepts the common assumption that Mark could simply choose not to shoot Victor. We should conclude, then, that Andrea is justified in shooting Mark to save Victor only if Mark lacked free will and could not help but kill Victor if not shot himself first. My response is to say that even if we assume that Mark has the ability to choose not to kill Victor, this does not show that Andrea’s shooting Mark is unnecessary; it shows that we need to add an important clarification regarding which facts are held constant and which can vary across counterfactual space. When we ask whether Andrea’s act is necessary, we hold the intentions and projected acts of others fixed unless Andrea does something to change them. We want to know what Andrea would have to do to save Victor given that Mark will continue on his path toward killing Victor. If Tom could save Victor by doing something less harmful to Mark and is willing to do so, then Andrea’s act would be unnecessary. Or if there is some non-​esoteric fact that would enable Andrea to save Victor without killing Mark if only she knew it, then again we have reason to say that Andrea’s killing of Mark was unnecessary. But if no such non-​esoteric fact exists and no third party exists with the intention and ability to stop Mark in a less harmful way, then Andrea has a right to kill Mark to save Victor. The fact that her act would not be necessary if other things, including the acts of others, were different is irrelevant.11

3. Reductive Individualism 3.1. A Defense of Reductive Individualism The dominant methodology in just war theory these days assumes that we can inquire into what a state and state agents (in particular, combatants) may do, morally speaking, to respond to a threat by asking how an individual with no special authority may respond to threats either to herself or

she ought to have done, given what she knew, then she should retain her judgment that she did what was necessary and feel no regret.  If she were acting in joint agency with others, then their acts should vary together; and it would be only the acts of those not acting with them that should be held fixed. I return to the issue of joint agency in the next section. 11

Basic Premises And method  |  29

others. This methodology is known as “reductive individualism.”12 The assumption that reductive individualism is methodologically sound is central to my argument for permissible eliminative killing. I clarify what I mean by it, and defend it so understood, here. There are two dimensions to reductive individualism: an agent dimension and a patient dimension. On the agent side, the fundamental thought is that as individuals, or groups of individuals, come together to form a state, they gain no fundamentally new right to kill (or impose other serious harms on people).13 The state and its agents—​for brevity I  will use the state to cover both—​are, with limited exceptions, no freer to use another as a means of promoting some social goal than private individuals are. Likewise, the state is no freer to cause harms to some as a side effect of protecting others than private individuals. And the state is no freer to engage in eliminative killing than private individuals. On the patient side, the fundamental thought, to quote Jeff McMahan, is that people cannot “be liable to attack, or immune from attack, merely by virtue of their membership in a group, particularly when membership is involuntary, or largely involuntary.”14 I accept this position and extend it from liability and immunity to the proposition that a person cannot be deemed a threat merely because she is a member of a group that, as a whole, operates in a threatening manner. If she does not contribute to the group’s threatening activity, she is not a threat. One can distinguish stronger and weaker versions of reductive individualism. The strongest version holds that group agents are fundamentally fictions, in the sense that all responsibility for choices has to reduce to the responsibility of individual agents.15 A somewhat weaker version accepts that group agents can make choices distinct from the choices of any individuals in them and be responsible in ways that are not reducible to the choices of any individuals within them.16 An even weaker version accepts that states have the authority to do things that individuals do not have the authority to do but insists that these unique powers are limited in ways that do not undermine reductive individualism with regard to eliminative killing or other ways of inflicting serious harm on people. I argue here only

 The term was coined by Rodin, 2002. It has been embraced as methodology by, among others, McMahan, 2009a; Fabre, 2012; Frowe, 2014; Lazar, 2015; Draper, 2016; and Haque, 2017. 13  See McMahan, 2009a, p. 82. 14  Ibid., p. 209. 15  McMahan’s position is often taken to be the stronger one. But his embrace of the agent and patient sides of reductive individualism seems consistent with the next weaker position. 16  I take this point to have been decisively made by Pettit, 2007. 12

30  |  The Mechanics Of claims And Permissible Killing in war

for the weakest version. I accept that group decision procedures can make groups responsible in ways that no individual member is and that states have powers that no individual has. Nevertheless, I will argue that neither undermines the core truth of reductive individualism. The most basic reason to doubt reductive individualism relies on the forms of authority the state has that individuals lack. The first thing to say about this objection is that it gets a grip only in the domestic context; in the international context, there is no reason to recognize a form of moral magic by which a group of people, by getting together to form a community, gets rights that it otherwise would lack to use force against people who are not part of its community. But in the domestic context, it has to be admitted (unless one is an anarchist) that states have special authority. They have the authority to tax, to regulate behavior, and to punish legal violations that private individuals lack. They also have the authority to settle disputes that private individuals lack. And they have a normative monopoly on the use of force in most situations, giving them rights to use force that private individuals lack. Nonetheless, I  don’t think these facts tell against reductive individualism with regard to serious harms even in the domestic context. The reasons are threefold. First, the special authority that states have is, I say, immanent in the rights individuals have to come together to solve collective action problems. Second, the state’s authority is limited by individual rights that states must respect, as well as by other norms of justice. Those limits are especially strong when it comes to inflicting harm. Third, the residual difference between the state’s authority to use force and an individual’s authority to use force reflects the way individuals in a state relinquish their right of self-​help, rather than the state having some right to use force that individuals fundamentally do not have. I take these points in turn. We accept states as at least potentially legitimate primarily because they help us form communities in which behavior can be coordinated for the common good, resources can be gathered through taxation and spent on public goods, rights such as property rights and rights of contract can be made determinate and protected, other social norms can be set, and disputes can be settled in a fair and impartial way.17 In line with those functions, other activities, such as punishment, take on a special meaning

 See Renzo, 2018 (discussing how states have authority because of how they “enable us to discharge some of our most important duties of justice”). 17

Basic Premises And method  |  31

in the state: punishment becomes a public act of holding a wrongdoer to account.18 One might say that none of these are activities that individuals have a right to perform on their own. For example, individuals may not tax each other for the common good. Thus, in some sense, the state clearly has authority that cannot be reduced to the authority of individuals. But I think this departure from reductive individualism is a minimal one because, as I said, I think the authority of the state is immanent in the right and duty of individuals to come together to solve collective action problems. That is to say that the state’s authority is a reflection of the claims individuals have on each other that they come together and form a state that will enable them to live in a well-​organized way in a social setting. Indeed, the space of rights is ultimately problematically indeterminate unless and until the state sets rules and procedures fixing the contours of certain terms, like property rights. Thus, while the state does have authority that individuals lack, it has that authority only because it is serving a set of basic social and political ends of individuals. To be clear, I am not here taking the traditional social contract line that the state’s authority derives from the consent of the governed. But I am stating a close cousin idea, that the state derives its authority from the right and duty of individuals to cooperate and coordinate. In a democracy the majority will set the terms of cooperation,19 and those terms will be binding on all. The reason the minority is bound to respect those terms is that it lacks the right to block reasonable efforts to set the ground rules for coordination and collective action. But it is bound to respect only the sort of state action that reflects reasonable efforts to find reasonable terms of cooperation. The state’s authority remains tethered to its grounding in the original right and duty to come together to form it. Regarding the second reason not to reject reductive individualism, the rights of individuals not only give the state its authority; they also limit it. For example, the state is not free to punish at will. It may punish only to a proportionate degree and only after providing criminal defendants with due process. Similar points can be made about the use of force. As I have said, the state, with limited exceptions, is no freer to use a person as a means, harm a person as a side effect, or eliminatively kill a person than

 See Duff, 2001. In that sense, I think theorists like John Locke are wrong to suggest that punishment in the state simply takes over what an individual could do in the state of nature. See Locke, (1690) 1980, §§ 8 & 123–​27. 19  Of course, that is an oversimplification. In practice, well-​organized minorities often set the terms, though a majority retains the ability to retake that power. 18

32  |  The Mechanics Of claims And Permissible Killing in war

a private individual.20 Even the authority to tax is fundamentally limited. It arises from the need to provide for public goods and is limited by that purpose. The state may not simply take property; it may take property only as part of a just punishment, as part of a general tax scheme that applies to all, or as a means to achieving a public good, with just compensation provided. Arguably, it can do more in circumstances of extreme emergency; but in those circumstances, private individuals can do more too.21 These matters of fundamental justice constrain the state and show the deep truth of reductive individualism. Regarding the third reason, the residual difference between the authority of the state to use force and the authority of individuals to use force is to be explained by the fact that individuals in a state relinquish their right to use self-​help. Returning to the example of punishment, there may be a core Lockean right to punish that individuals possess in the state of nature that they mostly give up in civil society in order to avoid bias in judging, excessive punishment based on emotion, and the inequities that would result if only the strong could punish. Likewise, people give up self-​help in settling disputes and avoiding threats in order to avoid bias, cycles of violence, and the inequities that would result if only the strong could engage in self-​help. But that does not imply that the state has some power that individuals fundamentally lack. It implies only that, when there is a better way to meet certain ends, individuals are morally obliged to use it. When this way is unavailable—​as happens in self-​defense cases when the threat is imminent and the state cannot respond in time to protect a victim—​the right of self-​help is recovered. Thus, again, we see that reductive individualism—​as a framework for evaluating how the state may permissibly use force—​seems morally sound. The arguments so far considered serve to defend reductive individualism from the agent side. One might also question it, however, from the patient side. If we accept that states can be responsible even though no individuals carry that responsibility, then it would seem that any attack on a state, meant to hold it responsible for wrongfully threatening others (or even its own citizens), would have to involve attacking people who are not responsible for its actions. But that objection moves too quickly. Take, for example, a case in which the citizens of country Z vote for leaders who

 The limited exceptions concern the ways the state may demand money and labor from people for the sake of the common good. But these demands must still respect the idea that each person has her own life to lead. See Chapter 3, § 1. 21  See the discussion of threshold deontology in Chapter 4, § 3. 20

Basic Premises And method  |  33

choose to wage war on country Y. In a straightforward sense, Z is a threat to Y. And in a straightforward sense, all citizens of Z are members of the group of people who comprise Z. But that does not give Y the right to attack each and every citizen of Z. Those who are merely minding their business may not be taken to be threats to Y, except insofar as their going about their business helps Z to wage war. The contribution a typical schoolteacher, for example, makes to Z’s ability to wage war is negligible, so a typical schoolteacher may not be treated as a threat to Y. All citizens of Z may owe something to Z, and that may give them duties both to try to pull Z back from war and, if Z is defeated, to do their part to pay reparations. But that sort of duty does not extend to being liable to attack simply because Z engaged in an unjust war, nor does it extend to being considered a threat merely because Z as a whole is a threat.

3.2. Warfare-​Based Objections to Reductive Individualism and Replies Even if weak reductive individualism is true in general, there are reasons to worry that war is somehow morally different enough from most killing outside of war that the rules regarding the former cannot be usefully derived from the rules regarding the latter. I will here discuss what I consider to be the four strongest objections along that line.22 My responses to these objections will be quick and may not be fully satisfying, but they are, I hope, sufficient to make the project of the rest of the book seem at least reasonable. The first objection is that wars are full of unjust acts on both sides, making it hard to say who is fighting a just war. This complexity can be interpreted metaphysically, as the claim that no one is fighting justly, or it can be interpreted epistemically, as the claim that it is especially hard to know who is fighting justly. Neither interpretation, however, calls for rejecting reductive individualism. Whatever is called for by metaphysical complexity, it is neither a free-​for-​all nor pacifism. Such complexity must instead call for an ongoing commitment to act more justly in the future and a willingness to negotiate to escape a cycle of wrongdoing if others are willing to do so as well. And in the meanwhile, it would seem that all sides should do their best to respect the individual rights of all affected people.  For a good list of objections that motivate just war theorists who reject reductive individualism, along with a description of which theorists adopt which objections, see Lazar, 2016, § 4. 22

34  |  The Mechanics Of claims And Permissible Killing in war

Moreover, I will assume in what follows that even if no side in a conflict is blameless, it is possible to identify unjust aggressors and, in the local context of a particular dispute, just defenders. That suffices to talk meaningfully about unjust combatants and those who help them unjustly threaten others. As for epistemic confusion, it generally calls for caution so as not to unintentionally wrong another. But it is also important not to make oneself a victim simply because one cannot be sure who is in the right. And one should reason about how to strike that balance in fundamentally the same way, whether the conflict is private or international. The second objection has to do with problems respecting the necessity and proportionality limits on just killing and the idea of just war. Seth Lazar raises this argument by comparing the effectiveness of killing to resist small-​scale threats and the effectiveness of killing in war. He correctly observes that, in war, “killing any individual combatant is much less likely to be effective in averting the threat.”23 To be effective, killing in war must aim at causing a deterrent (terroristic) effect. In other words, it cannot simply be eliminative killing; it has to be terroristic killing.24 This could pose a problem for reductive individualism because, on an individualist account of rights, terroristic killing can be justified only if the targets are liable to being used in that way.25 Lazar presses the problem in terms of a dilemma: either reductive individualists set the bar for liability to killing too high and cannot account for the degree to which killing in war is permissible, or they set the bar too low and must exaggerate the extent to which killing is permissible in the context of private individuals. If it worked, this would be a powerful argument against reductive individualism. But there are at least two ways to resist it. First, one could become a pacifist and reject the idea that wars are ever justified as a whole (given that they involve killing some to terrorize others). I want to reject that option, so I turn to the second way to resist the argument, which is to challenge the thought that unjust combatants are not liable to being killed for terroristic purposes. Arguing for that claim is outside the scope of this book, but I look to Lazar’s own work on why combatants are more likely to be culpable for an unjust threat than noncombatants as providing a promising start.26 If they are generally more culpable, then enough of

 Lazar, 2012, p. 28.  See Chapter 1, § 3. 25  The high bar to using another as a means, without his consent, will be argued for in Chapter 3, § 4. 26  See Lazar, 2015, pp. 89–​90. 23 24

Basic Premises And method  |  35

them may forfeit their right not to be used as a means of deterring others, while noncombatants would in general retain that right. The third war-​specific objection is that wars are started by states that have distinctive interests at stake that do not have analogues on the scale of individuals engaged in self-​defense, including, most importantly, political self-​determination. The fact that states have distinctive interests affects not just the decision to go to war but questions of what counts as a proportional response in war. But that does nothing to undermine the fundamental structure of reductive individualism. We simply need to allow extra considerations to weigh in the moral scales. And, as I argued in the previous section, these extra considerations are grounded, ultimately, in the right and duty individuals have in coming together to form a state and solve their collective action problems. Finally, some object that wars take place within a set of rules that have evolved to try to constrain and minimize the damage they cause and that people have roles within those wars set by the structure of international and domestic law in which they take place. This objection implies that it is at least possible that the rules that have evolved to regulate war are not very much like those that have evolved to regulate conflict between individuals and that it might be a mistake to try to make the former conform to the latter. I will respond to this objection in two steps. First, this objection makes more sense if one is a consequentialist, especially a rule-​consequentialist, than if one is a deontologist. If one is a consequentialist, then minimizing damage might be the most important thing to do; and if different contexts call for different rules, then so be it. But if one is a deontologist, then one thinks that the ends are not all that matter; the means to getting there must also respect the rights of the individuals involved. So, my first response is to say, not so fast; rights still have to be respected. Second, within a deontological framework one might still take up this objection by saying that the law is morally important. We cannot have people substituting their own judgment of what may be done for the judgment of the relevant military and political leaders because otherwise chaos will ensue.27 This is an important cautionary point. Coordination is important; governmental authority plays a crucial role in achieving it, and that authority is undermined if too many people act as if the government has no authority. But as an argument against reductive individualism, this is a non

 Renzo, 2018, makes this point in terms of the presumptive duty to obey the law.

27

36  |  The Mechanics Of claims And Permissible Killing in war

sequitur. It may be that it is morally preferable, under most circumstances, for state agents to follow an unjust order or law than to engage in individual acts of resistance. This, however, is just a question about what is most just to do in a complicated situation where matters of first-​order justice to those immediately affected must be balanced against considerations of downstream effects and political obligation. Recognizing this complexity does not undermine the moral relevance for state action and actors of the basic principles of justice governing killing, principles that can be gleaned from thinking about individual cases. One might argue that my arguments overlook some other reasons why the law must not follow the dictates of my reductive individualistic analysis. This is an interestingly contested position. Jeff McMahan, who has moral views similar to mine, thinks that there may nonetheless be sufficient reason for the law to take a position on the eliminative killing of noncombatants that is different from the moral position established by a reductive individualist analysis.28 Victor Tadros, who also has moral views similar to mine, argues that these reasons are not as strong as McMahan thinks and that there is good reason to think that the law should be reformed.29 Addressing this last dispute is more or less outside the scope of this book. My concern is primarily with understanding the first-​order morality of war, and of eliminative killing more broadly, where by first-​order morality I mean the moral norms that apply before taking into account the complexities that arise from the politics of legislation, adjudication, and enforcement. In that more limited enterprise, I am a reductive individualist about justifiable killing in war. The state has to respect the prior fundamental rights of individuals, just as other individuals do. Whether the law may justifiably depart from these limits, how far, and to what effect I leave to others (at least for now).

4. The Distinctive Causal Structure of Eliminative Killing As I noted in Chapter 1, I treat eliminative killing as a kind of act that is justified in a particular way, rather than as an action defined by the intention of the agent. As an act, it takes its particular identity from its role in

 McMahan, 2008.  Tadros, unpublished manuscript.

28 29

Basic Premises And method  |  37

a causal pattern:  the person killed must be threatening others—​whether directly or indirectly, alone or with others—​and the killing of that person must reduce, if not eliminate, that threat.30 Once one sees it in these causal terms, one can see fairly easily that it is different from the two more familiar causally identified moral categories of causing harm to others: using as a means and causing harm as a side effect. It is not unusual for those who frame eliminative killing in terms of intentions to hold that there are three types of killing:  intentional-​ opportunistic (aiming to exploit the target’s presence), intentional-​ eliminative (aiming to eliminate a threat posed by the target), and nonintentional (with foresight or by accident).31 But they tend to treat eliminative killing as harder to justify than nonintentional killing because, for them, the intentional feature carries significant weight. I will argue that this is a mistake in Chapter 5. Here I want to look at the two familiar causal notions—​using as a means and harming as a side effect—​and argue that we need a third causal category to handle eliminative killing. I start with the thought that eliminative killing is a form of using another as a means. This position was once adopted by Michael Otsuka, who wrote that “when you vaporize the body of the falling Threat, your killing of an innocent person is a means to save your life rather than merely a foreseen consequence.”32 Otsuka’s claim, however, seems to trade on the fallacy of the excluded middle. The fact that the causal role of a threat is not that of someone harmed as a mere side effect does not imply that his role must be that of someone whom an agent uses as means of saving herself. There are two reasons to conclude that he is not a means of saving her: linguistic and moral. Linguistically, the more natural thing to say is that her means are the tools she uses to kill him and stop the threat, be it a gun, a missile, or a pole. The threat himself is not the means she uses to stop him from being a threat to her. That would be like saying that if she stops a boulder from crushing someone by blowing it up, then she uses the boulder as a means of preventing the boulder from crushing someone. That is simply not what we mean by using a person or a thing as a means. Morally, the point, as

  What if the person is threatening only in her intention but not in reality? That is, suppose the person is making an impossible attempt, such as trying to kill someone with a gun that she does not realize has no bullets. She may then be liable to some opportunistic uses that would help others. That is, her attempt may have caused her to forfeit some of her claims not to be harmed for the sake of others. But conceptually speaking, she cannot be eliminatively killed. 31  See Chapter 1, § 3. 32  Otsuka, 1994, p. 87, emphasis in the original. 30

38  |  The Mechanics Of claims And Permissible Killing in war

we will see in the chapters that follow, is that a proper moral account of eliminative killing requires us to recognize that agents who engage in eliminative killing are not simply using threats as means. The argument that eliminative killing is somehow related to killing someone as a side effect is a more interesting and instructive one to refute as it brings out an important point about the use of counterfactuals. The argument can be found in Seth Lazar’s writing on this topic. He tries to elucidate the meaning of eliminative killing by saying, “When I  kill you eliminatively, I  am aiming only at ends that I  could just as easily have realized had you been absent.”33 Lazar uses that understanding of eliminative killing to make the link to killing as a side effect: If we bomb a military target to advance our just cause and foreseeably kill some nearby civilians as a result, then we derive no benefit from killing them that we would not have had in their absence. If I  save the five by diverting the trolley down the side track, where it kills one, then I (and the five) have gained nothing from killing the one that I would not have had if the track had been empty. These are species of eliminative harm, which helps explain why they are easier to justify than opportunistic harms.34

In other words, for Lazar, eliminative killing is the genus, and there are two species: side effect eliminative killing and targeted eliminative killing. The suggestion here seems to be that both species of eliminative killing are fundamentally on a par because of this shared counterfactual significance. I  will argue, however, that this kind of counterfactual analysis, and the equivalence it suggests, is morally unsound.35 To appreciate how and why, consider a case Lazar introduces to illustrate (targeted) eliminative killing: Bayonet: Suppose I kill an enemy soldier, who is rushing at me with a bayonet. I am no better off by killing him than I would have been had he not been there.36

 Lazar, 2015, p. 60. This definition seems to involve an intention element as well because it uses the idea of “aiming.” I think that can safely be disregarded as just a way of drawing attention to the distinctive counterfactual possibility that Lazar thinks is morally important. 34  Ibid. 35  Lazar is hardly alone in appealing to this sort of counterfactual to make moral arguments along these lines. I did that myself in Walen, 2014. Others who have done the same include Øverland, 2014, and Otsuka, 1994. I only recently repented of this sort of argument, in Walen, 2016, § II.A. For another good account of why counterfactual baselines cannot properly be used as I and others have used them, see Ramakrishnan, 2016, pp. 144–​46. 36  Lazar, 2015, p. 60. 33

Basic Premises And method  |  39

The second sentence in this description seems plausible enough, but if we explore the nature of the counterfactual implicit in the statement, the claim that the agent is no better off if he kills his attacker turns out to be contingent. There are variations on the case that show that the counterfactual analysis is at best a misleading heuristic, one that obscures the difficult moral questions that need to be asked. To see the problem, one must first be sufficiently clear about the relevant counterfactual. As Lazar rightly points out, it cannot involve asking what would happen if the person with a claim “never existed.” That could introduce all sorts of differences that are not only irrelevant but radical, including making it impossible for the agent to be in a situation in any way like the one at issue. Rather, it involves imagining that “your potential victim disappears. Vanishes.”37 But, of course, he cannot vanish in a way that disturbs anything else; there should be no thunderclaps as the air rushes in to fill the void; no one should faint from the shock of seeing another simply (magically) disappear. Rather, as much as possible, we are to imagine that everything else goes on as it would if the person simply, non-​shockingly, ceased to be a threat. Given that conception of the relevant counterfactual, we can introduce the following variation on Bayonet, in which it is not the case that “I am no better off by killing [the soldier rushing at me with a bayonet] than I would have been had he not been there.” Sniper: A sniper was about to shoot me with a pellet containing a horrible disease. If shot, I would die, but only after I first had a chance to spread the disease to my comrades in arms, and then it would kill many of them too. Moreover, the death would be a gruesome death for all of us. But as it turns out, one of the sniper’s fellow soldiers is charging me with a bayonet. If he would just trip and get out of the way (the counterfactual), then the sniper would shoot me. As it is, however, he blocks the sniper’s line of fire. Once I shoot him, the recoil from my gun will push me back and to the side, out of the sniper’s line of fire.

In Sniper, I  would be worse off if the soldier with the bayonet were to “vanish” before I shot him than if he remained on his present course. Indeed, I would be better off even if I did not kill him and he killed me. But I would be made much better off by killing him than I would have been had

 Ibid.

37

40  |  The Mechanics Of claims And Permissible Killing in war

he not been present. If the counterfactual account were the correct account of eliminative killing, then in this case killing him would not count as eliminative, and it would be much harder to justify killing the man with the bayonet than in the case in which there is no sniper in the background.38 Clearly, this is wrong. By killing him, I eliminate the threat he poses. Moreover, while it is true that I am better off because he is present until the moment I  kill him, that fact cannot ground some special claim he would have not to be killed by me. It is not as if I owe him any debt of gratitude for blocking the sniper. He is not doing so on purpose, nor am I “using” him for that purpose; I simply happen to benefit from his presence. I  would, presumably, owe him a debt of gratitude if he were not trying to kill me. But as it is, that is what he is trying to do, and insisting on gratitude would be quite bizarre. And even if, in some attenuated sense, I owe him a debt of gratitude, it would certainly not be so great as to make killing him wrong in a way resembling the wrongness of an opportunistic harming (e.g., using a human shield). The obvious move to make is to say that what matters is not the counterfactual but the causal structure of the case. That is, because my killing him eliminates the threat he poses, my act is an eliminative killing. This switch to a causal account will handle the kind of deviant counterfactual that my counterexample turns on. But it raises the problem of assessing what the moral significance of that causal role is. That is the topic I turn to in Chapters 6 and 7. For now, I simply want to emphasize that the causal role is clearly distinct from that of killing as a side effect. To do so, I need to remind the reader of a point I made in Chapter 139: that a causal role is identified within a particular justification that is offered for an act. To say that A kills B as a side effect is to say that B’s death, or more precisely the impact on B that leads to his death, does not serve to help justify the act within a particular justification. This is not to say that the impact on B, or B’s death, must have no positive value, instrumentally or intrinsically (if B deserves to die); it is only to say that any positive value derived from the impact on B plays no essential role in the particular justification for the act under consideration. By contrast, if the act is justified as an eliminative killing, then the death, or more precisely the impact on  Johann Frick suggested to me that one might respond to this case by appealing to their collective agency, saying that I am worse off with both of them operating than I would be by killing one of them; thus, the killing is still eliminative. But this response is unsatisfactory. I could vary the case again so that the sniper is not another agent but some background threat that the soldier with the bayonet is unwittingly keeping at bay. 39  See Chapter 1, § 2, where I first defined eliminative killing in terms of a justification. 38

Basic Premises And method  |  41

B that leads both to neutralizing the threat that B poses and to B’s death, must figure as part of the justification of the act. These are different justificatory structures, turning on different causal roles. That said, I want to wrap up this section by repeating the lesson that counterfactual analysis that imagines a person not to be present is to be avoided. It is what made it seem that eliminative killing is a genus that includes side effect killing and targeted eliminative killing as two species of that genus. That picture of their relationship makes room for the possibility that the two species are morally different, but it is set up to emphasize, instead, their similarity. It does so, however, in a way that cannot be relied on. As the Sniper case showed, it fails to capture some cases that should count as eliminative killing. And it invites us to take a moral shortcut in thinking about how eliminative killing should be justified as just a variation on side effect killing. We need, instead, to look at it as its own causally distinct type of killing, with its own distinct kind of justification.

42  |  The Mechanics Of claims And Permissible Killing in war



3 A Theory of Rights

T

his chapter covers the theory of rights that sits at the core of this book. I start with an account of the purpose of rights, the three principles that ground what I call the space of rights, and the basic functional structure of rights, which I call the mechanics of claims. Next, I give a more formal account of what I mean by the mechanics of claims. I  then contrast it with the infringement model. Finally, I  introduce, explain, and defend the restricting claims principle, which, in turn, can be used to defend a key principle in deontology: the means principle, that is, the principle that it is particularly hard to justify using another as a means if doing so imposes some cost on him, unless he has waived or forfeited his right not to be so used.

1. The Purpose and Basic Structure of Rights Rights constitute a substantial part of the normative space in which justice sets the limits of what we may do. On the patient side, rights are the coin in which justice is measured when it is the impact of an agent’s choice on particular people that determines whether her choice is permissible. In that sense, rights fundamentally exist to ensure that individuals enjoy the respect they are due. They mark out a sense in which the individual has a status that is morally inviolable.1 On the agent side, rights mark out the bounds of the permissible. An agent has a right to act if and only if her

 See Rawls, 1971, p. 3: “Each person possesses an inviolability founded on justice that even the welfare of society as a whole cannot override.” See also Nagel, 1995, p. 89, identifying our distinctive moral status with “a certain kind of inviolability, which we identify with the possession of rights.” 1

action does not violate the rights of others and is otherwise consistent with the demands of justice.2 In what follows I spell out these ideas more fully, arguing for a Kantian conception of the space of rights—​I define what I mean by that at the end of the next section—​and articulating the three principles on which it rests. As I noted in the Preface, the model of rights I offer here is not meant to be an analysis of what it means to have rights. I offer it as a model of rights that helps us understand what is substantively morally true.

1.1. The Basic Function of Rights The basic function of rights is to provide a normative space in which we can each pursue our own ends while treating each other as free but responsible, equal fellow agents whose welfare morally matters.3 The theory I offer develops this substantive conception of what rights do. The theory is grounded in three deeply attractive moral principles that I articulate with some care in section 1.2. Before articulating the grounding principles, I  want to contrast this basic idea with the currently dominant theories of rights, the will and interest theories. The will theory emphasizes the importance of powers that give rights-​holders the ability to exercise their will; the interest theory emphasizes the importance of promoting the welfare or interests of the right-​ holder. Each runs into well-​known problem cases or counterexamples. But those problem cases are not my concern here.4 My concern is to say that these theories present overly narrow bases for ascribing rights. The bases are narrow not just in the sense that they exclude each other but more fundamentally in the sense that they don’t conceive of rights as constituting a space in which to articulate the limits of permissible action generally; they

 This sentence reflects my fundamentally liberal belief that agents are at liberty unless there is a sufficient reason to restrict their liberty. I intend a capacious notion of justice here, one that allows for duties to obey reasonably just laws that are passed for the general welfare. See § 1.3.1 for more detail. 3  This is a distinctly Kantian conception of the function of rights, drawing in particular from his doctrine of right: “Any action is right if it can coexist with everyone else’s freedom in accordance with a universal law.” Kant, (1797) 1991, p. 56. I depart from Kant most fundamentally insofar as I incorporate welfare into the space of rights. 4  See Leif Wenar, 2013, for a general discussion of the will and interest theories and the problems they face and for a creative alternative that he calls the kind-​desire theory. In the end, however, I think the question of who we say is a rights-​holder has been given much more attention than it deserves; when disputed, it is not really of great moment who is the right-​holder, and the question has led to the focus on the will and interest theories, which, as I argue in the text, has been distorting for rights theory as a whole. 2

44  |  The Mechanics Of claims And Permissible Killing in war

see rights as more or less directly reflecting and highlighting a subset of ethical concerns of particular relevance to the right-​holder. The dominance and the misguided narrowness of the will and interest theories explain two related problems with the current state of rights theorizing. First, too much of normative significance is left out of determining who has what rights. Saying that A has a right to X does not come close to determining what may be done in the final analysis. The common view is that after recognizing that something about a person gives her a right, one has to deal with possible conflicting rights and conflicts between rights and consequentialist considerations. And this latter conflict will have to be resolved on grounds that are not specified inside the space of rights. The incompleteness of the normative picture that results from recognizing that a person has a right leads, naturally, to the infringement model, according to which it is often morally permissible to act against a person’s right. I will articulate, mostly in Chapter 4, a host of reasons not to accept that model of rights. To date, the dominant way to avoid the infringement model is to embrace specificationism, the view that the better thing to say, when some would want to say that it is permissible to infringe a right, is that the right contains an exception that makes that act permissible.5 For example, while property rights normally give owners a right to exclude others from their property, in emergency situations it may be acceptable for others to use an owner’s property without her permission. Rather than treat such cases as infringing on the owner’s right to exclude, we could simply say that the right to exclude does not cover such cases. Many object that specificationism cannot make sense of the moral residue owed the owner in a case like that; I return to that problem in Chapter 4, section 1. The problem I want to raise now is that specificationism does not help bring more normative considerations into the space of rights. Rather, it too assumes that one often has to look outside of those narrow considerations that ground a right to understand why a right might have to be specified and limited in a particular way to accurately track our judgments of permissible action.6 The second problem with contemporary rights theory is the related view that rights can be eliminated from our normative discourse without

 See Shafer-​Landau, 1995; Oberdiek, 2004. I say a bit more about specificationism in Chapter 4, § 1.  Liberto, 2014, argues in favor of a partial or “restricted” version of specificationism. On her view:

5 6

if the justification for an exception is the same as the justification for rights within the broader ethical theory, then the exception is specified in the right; however, if the justification for the exception is different from the justification for rights, then the exception is not specified by the right, and acting in accordance with this exception constitutes an infringement of the right. (p. 177)

A Theory Of rights  |  45

any substantive loss, that the loss would be only in terms of rhetorical convenience and the hope of agreeing on certain mid-​level principles without having to agree on the deeper foundations.7 This is a mistake. Rights are not theoretically eliminable “intermediate conclusions” in the overall landscape of normative argument.8 Rights are the conclusions of a specific form of normative reasoning, one that reflects the principles underlying the space of rights. This is a normative “space,” one that I model on what Wilfred Sellars called “the space of reasons”: “in characterizing an episode or a state as that of knowing, we are not giving an empirical description of that episode or state; we are placing it in the logical space of reasons, of justifying and being able to justify what one says.”9 The space of rights is a subspace in the space of reasons; to move in it, one has to engage in a specific form of normative reasoning, using a particular structure and operating with a particular set of intermediate principles that describe how normative considerations are to be weighed in that structure. If one thinks that rights are just rhetorical conveniences, mere intermediate conclusions, one will overlook the way the space of rights shapes our normative discourse, and the distinctive features of rights discourse will seem more mysterious and more ad hoc than they should.10 The idea of a space of rights has another significance as well, beyond being a special subspace in the space of reasons. The metaphor of a space is crucial. It is frequently thought that rights have a distinctive bilateral form: “A has a right that B do X” implies that and is implied by “B owes A a duty to do X.” This corresponds to what Thomas Nagel called justification “to a subject”11: B is obliged to do X because she cannot justify not doing X to A, who has a right over her. A similar structure underlies

This suggestion helpfully flags the fact that, on the standard theories of rights, one often, but not always, has to look outside of the narrow justification for a particular right to appreciate its contours in practice. But that does not help with the problem that too much normative work is being done outside of the space of rights. 7  This is exactly the view taken by Raz, 1986, p. 181. See also Oberdiek, 2008, p. 133 (emphasizing that “rights are not themselves normative bedrock, rather, practical reasons are”). 8  Again, see Raz, 1986, p. 181. 9  Sellars, 1997, p. 76. 10  One feature of rights discourse that I do not discuss in this book, but that I hope to explore in future work, is the categorical nature of the duties that arise out of it. If one thinks merely in terms of practical reasoning, it is a bit mysterious why duties to respect rights should be accepted in the categorical way in which they present themselves. If one thinks about agents operating in a space of rights, a space that has the function of determining what is and is not permissible, a space that an agent cannot dip into at will—​one either operates in it or does not—​then the categorical nature of duties to respect rights becomes fairly straightforward. 11  Nagel, 1979b, pp. 66–​68.

46  |  The Mechanics Of claims And Permissible Killing in war

T. M. Scanlon’s version of contractualism.12 The alternative, according to followers of these sorts of models of rights and deontology more broadly, is justification that appeals to some version of impersonal value. The notion of a space of rights is meant to call attention to the idea that this is a false dichotomy. Though A having a right against B is a bilateral relationship, it occurs in a larger space in which the facts that establish that A has this right refer to all the parties who have a claim on B. Instead of operating in a one-​dimensional line connecting A and B, the space of rights operates in many dimensions. But it also does not merely appeal to impersonal value. It has its own distinctive structure and rules and is grounded in the claims of individuals, not in impersonal value. This justificatory structure too is lost if we fail to mark the distinctiveness of a space of rights.

1.2. The Three Principles That Ground the Space of Rights The space of rights rests on three principles. These principles should be understood not as values—​though the third does concern the integration of value into the space of rights—​but as axioms, analogous to those in geometry. 1. Autonomy principle: we all have our own lives to lead; the space of rights must give us substantial freedom to do so, while also holding us accountable for our choices. 2. Equality principle: we are all fundamentally equal as members or “citizens” of the space of rights. 3. Welfare principle: our welfare matters, and that value affects the rights we have. I take these or similar principles to be familiar. The autonomy principle (modified by dropping the accountability clause) by itself grounds ethical egoism. Combine the autonomy principle as written with the equality principle and one gets the basis for libertarianism. Combine the equality principle (modified by substituting “the moral community” for “the space of rights”) and the welfare principle (modified by substituting “what may be done” for “the rights we have”) and one gets utilitarianism. Combining all three principles, however, grounds and explains the space of rights and much of its content.

 Scanlon, 1998, p. 229. I say more about Scanlon’s model in § 1.2.3.

12

A Theory Of rights  |  47

In what follows, I explain the content of each principle and how each helps give rise to a structured space of rights.13 In subsequent sections, I articulate more carefully how to operate in that space of rights. 1.2.1. The Autonomy Principle

The autonomy principle, as I wrote in “Transcending the Means Principle,” is a broadly Kantian principle; but it does not exactly reflect “Kant’s idea of autonomy.”14 It departs from Kant because “the emphasis is not on using one’s own reason [to legislate a universal law].”15 But it picks up on the Kantian idea that each person is to be treated as an end in herself in the sense that she can rightfully be especially concerned with her own ends and her own welfare.16 It also picks up on the Kantian idea of agents as capable of regulating their pursuit of their own ends by reflecting on the demands of morality. These demands require them to recognize themselves as just one among a larger community of people who are all equally free to pursue their own ends, all of whom deserve respect and all of whom can be held responsible for their choices. This will prove to be the most significant of the three principles, so I will say much more about it soon. But I want, first, to be clear how to distinguish the different meanings of autonomy so that the meaning picked up by the autonomy principle is not confused with other meanings. Autonomy can be considered a capacity, a value, and a status. The autonomy principle is fundamentally concerned with the last of these three, status, though the capacity is relevant to having the status and has a value. Autonomous capacity is the ability to make decisions. In a weak sense, it is possessed by any self-​directing organism that has its own subjective point of view—​thus a computer is not autonomous, but a normal chicken, fish, or dog is. In a stronger sense, it is a capacity to make decisions in such a way that the agent is morally responsible for her decisions—​in this

 This is a broadly constructivist project, in line with the way Rawls describes constructivism, 1993, pp. 93–​95. In embracing constructivism about rights, I mean in no way to reject normative realism as regards other aspects of morality. My claim is only that the space of rights is particularly well accounted for by appealing to the implications of working with this particular set of fundamental principles. 14  Walen, 2014, p. 436. 15  Ibid. Kleingeld, 2018, argues that the later Kant, in the Metaphysics of Morals, dropped his appeal to the principle of autonomy because his politics had changed: he came to view legislating as involving real consent and thus dropped the idea that the moral law reflects real legislative activity. 16   I  take it that Kant’s thought that morality was the condition for being “worthy of happiness” reflects a basic idea that each can rightfully be primarily concerned with her own happiness. Kant, (1785) 1993, p. 393. 13

48  |  The Mechanics Of claims And Permissible Killing in war

sense a dog or a child is not autonomous, but most adults are. As a value, the weaker notion of autonomy can ground claims that the welfare of the being in question has a distinctive moral value, a kind of value not possessed by a tree or a river. But the value of autonomy is more importantly connected to the strong sense because there is a special value in helping a person achieve and maintain her autonomy—​a point with both first-​person and third-​person significance. As a status, autonomy can be understood as the normative claim that an autonomous agent, in the strong sense, should be treated as such, which means that she is not only held responsible for her choices but accorded the fundamental right to make choices primarily for her own ends. As already noted, the autonomy principle articulates autonomy in the status sense. It says, in essence, that the space of rights must recognize that autonomous people, in the strong sense, have the status of beings who may rightfully pursue their own ends, as long as they do so within the bounds set by the space of rights. 1.2.2. The Equality Principle

For the equality principle, the idea that we are fundamentally equal is key. Of course, we do not all share the exact same rights. But in virtue of achieving a threshold level of autonomous capacity (in the strong sense), all those who have crossed that threshold have fundamentally equal status as “citizens” in the space of rights. Insofar as people have the same sorts of claims for the same sorts of interests, one person’s claims count as strongly as another’s. Claims have weight in determining who ultimately has a right to what in ways that reflect not just the interest protected but the nature of the claim—​a point I explore in sections 2 and 4 in this chapter. The important point to make at this juncture is that insofar as different people have the same interests and those interests are carried by the same sorts of claims, those claims will count the same. Again, it is important to be clear that equality in this sense is a principle, not a value. It is not a state to be achieved, possibly by leveling down. It is a principle affecting how people’s interests register in the space of rights and how their claims get balanced against one another. They register and get balanced in a way that reflects the fundamental equality of people as moral agents, as citizens in the space of rights. Animals present an obvious worry for this notion of fundamental equality. I accept that animals have rights, but they are not fundamentally equal to most humans. The reason is that fundamental equality in the space

A Theory Of rights  |  49

of rights is for citizens in the space of rights, and citizens are people who can navigate the space of rights as autonomous agents, in the strong sense. Citizens possess not only claims but also the capacity to recognize and respect their duties. They can also justly be held responsible if they act wrongfully. And they can exercise Hohfeldian powers,17 thereby changing their rights by such acts as promising and trading. Small children are not yet citizens in the space of rights, but we treat them as having claims not to be harmed and to be aided on a par with full citizens because we expect them to become full citizens.18 Former citizens who have lost their ability to act as moral agents have claims on us via the citizens they once were as well as via the sentient, or at least living, beings they still are. But nonhuman animals are trickier. I  suggest we treat them as beings with claims not to be harmed and claims to be helped, with immunities that protect against others stripping them of certain claims, but without liberties, duties, or powers. They are not even second-​class citizens; they are something more like guests in the space of rights. I do not mean to denigrate their claims on us, but it is a secondary issue, which I put to the side to focus on the claims of normal moral agents. It is they who are fundamentally equal in the space of rights. 1.2.3. The Welfare Principle

I turn now to the welfare principle, which, unlike the other two principles, fundamentally expresses a value. There are five important points to make about this principle. First, the welfare principle straightforwardly explains why claims add up. Each person’s welfare matters, and it matters as a good thing, such that the more people do better, the better. This should not be taken to mean that we should create more people so that there is more value in the world; its implications along that dimension are surely complex, contested, and beyond the scope of this book.19 For my purposes, it need mean only that insofar as people do exist, their having better lives is morally valuable.20  See Hohfeld, 1913, where he introduced his analysis of rights into four pairs (two first-​order and two second-​order); powers are second-​order rights. 18  I take no stand on when those claims attach: whether at birth or conception or somewhere in between. There is too much to say on that topic to say anything quickly. 19  On the ethics of creating people, see Parfit, 1986; Temkin, 2012. 20  I do accept an important limit on how claims add up, the limit known as “limited aggregation.” The idea is that if patient-​claims are of vastly different strengths, then no number of weaker claims outweighs a stronger claim. For example, an agent should save a life rather than prevent any number of people from having a headache. I discuss this idea in some depth in Walen, unpublished manuscript. 17

50  |  The Mechanics Of claims And Permissible Killing in war

This utilitarian idea might seem to be the sort of thing that should be alien to a deontologist. But I think the thought that deontologists should disavow the aggregation of claims should be rejected as a confusion.21 It would take me too far afield to explore that claim in depth, but I will address one reason for rejecting the thought that the numbers should be taken to add up: the contractualist view that, as T. M. Scanlon puts it, “the justifiability of a moral principle depends only on various individuals’ reasons for objecting to that principle or alternatives to it.”22 According to Scanlon, the implication of this view is that the numbers can count only insofar as having more on one side than another can break a tie. For example, if A  could go left and save three people or go right and save five people, she has reason to go to the right because three on the right offset three on the left, and each of the two remaining people on the right makes a difference.23 But there are two problems with this approach to the relevance of numbers: an intuitive problem and a misguided theoretical assumption. The intuitive problem is that Scanlon’s tiebreaking model fails to capture the way numbers intuitively matter. It may account for why one should rescue a group of n + 1 people rather than a group of n people, all else equal. But it cannot explain the common intuition on the following chestnut of deontological ethics24: Trolley Switch: Brenda, a bystander at a switch, can throw the switch and thereby turn a trolley that is hurtling down a hill, out of control, away from five innocent people and towards one innocent person on a sidetrack (the “sidetrack man”), thereby killing the sidetrack man and saving the five.25

To see why Scanlon’s tiebreaking model runs into trouble, consider a variation in which Brenda’s choice is to save two by turning the trolley onto and killing one. Many think it is not so clear that she may proceed. Most people intuitively judge that killing is more than a little harder to justify

 The original objection to letting the numbers count was posed by Taurek, 1977. I believe Parfit’s (1978) response to Taurek is decisive. I will say a bit more about Taurek’s view in Chapter 3, § 2.2. 22  Scanlon, 1998, p. 229, emphasis in the original. 23  Ibid., p. 232. 24  To be fair, Scanlon’s discussion of tiebreaking was in the context of save–​save cases, not kill–​save cases. But his insistence on the individualist restriction seems quite general. 25  Foot, 1978, first formulated a trolley case. But Foot left a driver in control and saw it as a choice between killing one or killing five. It was Thomson, 1976, who first put the responsibility on the shoulders of a bystander at a switch, thereby raising the problem of justifying the killing of one to save five. 21

A Theory Of rights  |  51

than allowing to die; thus, it seems to many to take more than a two-​to-​one advantage to justify killing. That is why the case was originally designed to have five on one track and one on the other. Let us stipulate, for the sake of argument, that the “magic” ratio for justifiably turning the trolley is three to one. Scanlon cannot account for this with his tiebreaking move. First, there is no tie; the claims to be saved are weaker than the claim not to be killed. Second, there is no way to avoid the problem by saying that one claim not to be killed is stronger than one claim to be saved but that adding one more claim to be saved shifts the balance. By hypothesis, it takes two extra claims to be saved to shift the balance. Thus, we have to allow claims to be counted or to add up.26 The deeper problem with Scanlon’s objection to letting the numbers count is that it is based on a misguided theoretical assumption. He assumes that if we allow the numbers to count, then we cannot keep our grip on the idea that morality has to be justified to individuals. This is a mistake. Respect for our individuality is provided by the autonomy principle. It guarantees that each person is given adequate respect as an individual. As we will see shortly, the autonomy principle grounds differences in types of claims. These differences limit the way the numbers can matter. Paradigmatically, patients have strong claims not to be used as a means without their consent. Given that we can rely on the autonomy principle to ensure respect for the individual, we can also allow the welfare principle to ground the common-​sense thought that it is morally proper to let the numbers count, as long as we take into consideration the kinds of claims that are being counted.27 In stating my second objection to Scanlon I  have already raised the second important point about the welfare principle:  insofar as it makes sense to attach welfare interests to the claims of particular individuals, the way the welfare interests count reflects the nature of the claims of the persons whose welfare is at stake. If a patient has waived or forfeited a claim not to be harmed, then his welfare in that regard will count for much less than if he had not done so. In particular, the claims of people who have forfeited their right not to be harmed do not seem to aggregate; it is

 For a related argument, see Saunders, 2009, p. 283; Woodward, 2014, pp. 634–​37.  Some, going back to Anscombe, 1967, also worry that there is something paradoxical in saying that some number of people could have a right only because they together cross some threshold number. See also Scanlon, 1998, p. 230. But my point is that this is a mistaken prejudice. Rejecting it does not require accepting any funny notion of groups having interests above and beyond those of the individuals who comprise them. It requires only that claims, which inhere in individuals, can aggregate in the balance that determines who has what rights. 26 27

52  |  The Mechanics Of claims And Permissible Killing in war

permissible to kill any number of culpable aggressors to save an innocent victim. There is a sense, however, in which a patient’s claim to have his welfare count is inalienable: it still counts for necessity and proportionality. It is impermissible to kill a person who poses a threat to some innocent victim if it is possible to stop the threat without killing him and at very little extra cost to the agent, the victim, or bystanders.28 Nevertheless, if a patient has forfeited his claim to have his welfare count, then his welfare will register in the balance of claims via a much weaker claim. Importantly, there are also a variety of other ways, to be made clear in sections 2 and 4 of this chapter and in Chapter 6, in which the way one’s welfare weighs in determining who has a right to what depends on the kind of claim that picks it up. For example, if it is picked up by a claim not to be harmed, it will register more strongly, all else equal, than if it is picked up by a claim to be aided. It is important to be clear that this second point is not that people’s interests count more if they are protected by rights. That is a common but deeply confused thought. One way it is confused is that it is simply false. As Scanlon writes, it seems wrong to say that “murder is a much worse thing to have happen” than an accidental killing.29 Another way it is confused is that it wrongly suggests that the importance of not violating rights is to be cashed out in value terms. To the contrary, it is a basic fact about rights that they are to be respected. This is not to say that nothing can be said about why rights must be respected. My conception of the space of rights is that it provides people an important normative space within which to live. The mistake is to reduce the importance of rights to something like consequentialist value. The third important point to make about the welfare principle is that all welfare that can register through competing claims on agents does register that way. It is a mistake to treat rights as limited to a certain subset of welfare claims, such as claims not to be harmed and claims based on things like promises, leaving other welfare interests to somehow register in the overall moral balance but not through the filter of the space of rights.30 This just leads to confusion. Rather, all welfare claims that matter to what

 See Chapter 1, § 2, in particular the comparative proportionality principle (principle 5) in permissible eliminative killing. 29  Scanlon, 1998, p. 83. I come back to this issue in Chapter 9, § 2.2 (discussing an argument of Helen Frowe’s). 30  See Thomson, 1990, p. 162. 28

A Theory Of rights  |  53

an agent is permitted to do register, as far as possible, as claims in the space of rights.31 Fourth, the welfare interests of a patient that matter for a particular claim are not the full set of his interests affected by an agent accommodating or “contravening” his claim; they are the fairly narrow interests tied directly to that claim.32 For example, if A threatens to cause harm H to V, we should say that the strength of V’s claim not to suffer H is a reflection just of his interest tied directly to his not suffering H. If his not suffering H would also cause him to enjoy some benefit B, that is picked up by a separate claim to enjoy B. The reason for this restriction is that it is necessary to respect V’s autonomous capacity to decide whether to make such trade-​ offs or not. That is, it is up to V to decide if he would like to accept H as the price of enjoying B; it is not up to A to force that package deal on him. One might challenge this point by introducing an example like this: A patient arrives in a hospital unconscious, without a healthcare directive and without the company of anyone with power of attorney. He needs an operation; otherwise, he will soon die. A surgeon, we may assume, is permitted to cut into him, thereby harming him in contravention of his claim not to be cut but responding to his more significant claim to have his life saved. But I don’t take this to present a real counterexample. We think the surgeon may cut into the patient because, absent information about what he would want, she has to guess what he would want, and being saved is more likely to be the correct guess. In other words, this is a case of imputed consent. The default assumption in this case in no way undermines the thought that the surgeon may not operate on someone who positively refuses his consent. Fifth and finally, the word benefit has two importantly distinct meanings. One meaning is simply a direction of effect. To save someone from harm is to benefit that person. In that sense, benefits are as central to my project as harms. But there is another meaning with respect to which the reasons to benefit a person are substantially weaker than the reasons to save him from harm, and this is so even when an agent might reasonably choose to suffer the harm to gain the benefit.33 This second meaning of the term involves

 I am agnostic about whether the welfare interests of people who do not yet exist, but who do come to exist, can register in the space of rights. Some, for example, Woodward, 1986, extend claims to such people. I am unsure whether we should treat nonexisting people as having rights, rather than merely having morally relevant interests that could limit what an agent is permitted to do. 32  As far as I know, Woodward, 1986, pp. 809–​11, was the first person to make this point. 33  For a useful discussion of this point and the preceding point about two meanings of benefit, see Shiffrin, 2012. 31

54  |  The Mechanics Of claims And Permissible Killing in war

some sort of baseline state of well-​being: falling below that baseline is suffering a harm; rising above it is enjoying a benefit. It is hard to draw this line in a principled way, and I will not try to do so. All the cases I focus on, however, will fall clearly on the side of harms caused or harms averted.34

1.3. The Further Relevance of the Fundamental Principles 1.3.1. Rights as Grounding only a Subset of Possible Wrongs

Having introduced three fundamental principles, I want now to make three more points about what I aim to do by invoking these principles. The first is a negative point, about what I am not saying; the next two are positive points about the significance of the autonomy principle. The negative point is this: the three principles and the space of rights they ground do not offer a complete framework for making sense of what is and is not “wrong” to do. This is so for two reasons. First, the notion of moral wrongness is broader than the space delimited by the space of rights. For example, it may be at least pro tanto wrong for an agent not to develop her talents, to take pleasure in the pain of others, and to be ungrateful for the good things others have done for her. One might be tempted to object that the last two wrongs are involuntary and thus fundamentally not the sort of thing that she can do at all. But they each contain a voluntary dimension: she can choose, for example, to indulge or avoid the pleasure she would experience in perceiving the pain of others (not only by not inflicting or by helping to end their pain but by directing her attention), and she can choose likewise to cultivate at least the awareness of why she has reason to be grateful to others. This allows us to say that she might act wrongly in all of these cases. Nevertheless, I think agents have the right to commit these sorts of wrongs. This sort of right to do wrong is not merely a claim-​right not to be forced to avoid these sorts of wrongs.35 It is also a right with first-​person significance: an agent need not address herself with the command that she must not do these things.36 She should strive not to do these things, but these sorts of moral failures are wrong in a way that is not connected with the imperative to respect the rights of others. The second reason these principles and the space of rights they ground do not offer a complete basis for making sense of what is and is not “wrong”

 This point was first raised in Chapter 1, § 2.  This is the interpretation of the right to do wrong in Waldron, 1993; I believe it is too narrow. 36  For a similar view, see Harman, 2016. 34 35

A Theory Of rights  |  55

to do is that even within the space of rights not all instances in which an agent has no right to do something reflect the limitations imposed on her by the rights of others. I have said something about this, but I want to say a bit more before putting this sort of exception to the side. The key idea is that there may be duties of justice that cannot be explicated by reference to particular claimants who have a right that the agent perform her duty. I  offer three examples to illustrate the point. First are duties of fairness, such as the duty not to freeride on the sacrifices of others. Often, in such cases, none of one’s fellow citizens can claim the relevant standing to say that the duty is owed to her.37 Moreover, often, there is no clearly identifiable group of others who are, collectively, the rights-​holders. Take, for example, cheating on one’s taxes. Is this duty owed only to others who do not cheat on their taxes? What about those who have cheated a little but not much or often? It seems better to say that one simply has no right to be a tax cheat. If we want to say that the corresponding duty is owed to someone or something, it is best to say that it is owed to an abstraction: one’s fellow citizens. But it may be clearer to say simply that one has a duty of justice to pay one’s fair share. Second, the duty to do retributive justice might also be a duty owed to no one in particular. It is not owed to the criminal: pardoning or choosing not to prosecute criminals does not violate their rights.38 Nor is it owed in particular to the victims; even if they would choose to pardon the offender, a retributivist would think there is nevertheless some reason for the prosecution to go forward. Again, if this duty to punish the guilty exists, it seems to be owed to the abstraction of one’s fellow citizens. Third, something similar can be said about many duties of office holders more broadly. The president’s duty to faithfully execute the law is owed to her fellow citizens but to no one in particular. In addition to these three kinds of duties of justice, agent-​rights may be limited by duties to respect that which has agent-​neutral value out of proportion to the value particular agents happen to attach to it. Consider, for example, the preservation of certain species of animals, awesome

 Standing to demand respect of one’s rights may not be essential to having patient rights in all cases, but there is good reason to think that they generally go hand in hand. See Gilbert, 2018. For an exception, see the discussion of voluntary “punishment” in Chapter 4, § 1.2. 38  There may be something to the Hegelian notion that wrongdoers have a right to be punished, but I interpret Hegel’s point to be that they have a right to be treated as responsible beings, for whom punishment is a fitting response, where the contrast is being treated as merely a dangerous animal. See Morris, 1968. 37

56  |  The Mechanics Of claims And Permissible Killing in war

natural formations, or historical artifacts. The duty to protect or preserve such things may call for devoting resources to that end rather than to satisfying the claims of particular individuals, and it arguably does so to a degree not reflected in the claims of particular individuals who may happen not to care as much as they morally should care about particular goods of that nature. For the sake of completeness, it is worth mentioning another way in which the limits on agent-​rights reflect something other than the claims of others: moral rights may be grounded, in particular societies, in legal decisions to give certain people, or perhaps all people, legal rights not because doing so reflects or delimits prior moral rights, but because it promotes certain other values. Consider, for example, the particular contours of free-​speech rights. They may reflect, to some degree, the individual interests of speakers and listeners; in that way the legal rights would reflect the moral claims that correspond to those interests. But to a substantial degree they reflect a range of other considerations as well, such as the extent to which it is believed that granting them will promote the search for truth (an agent-​neutral value). When the law creates legal rights that do not simply track or specify prior moral rights, it creates a framework of practices and expectations in which the legal right–​holder acquires a moral expectation that she will be free to enjoy her legal rights. Thus, the law gives birth to moral claims both for liberty and for noninterference. Something similar can be said of the moral rights of patients that are at least partly grounded in legitimate expectations established by social and legal convention, such as property rights. Indeed, as property rights are central to the space of rights, this is a profound way in which legal rights impact the contours of moral rights. Nonetheless, despite all of these ways in which the content of rights reflects something other than the moral claims of individuals, many rights, governing both individuals and the state, do reflect first and foremost the balance of the conflicting moral claims of individuals, as grounded in the three principles. Going forward, I will focus almost exclusively on those cases, ones in which it is the balance of moral claims on an agent, and the agent’s own moral claims, that determines who has a right to what.39

 For the reader well versed in Hohfeldian terminology, it may seem jarring to use the term claim as the generic; Hohfeld used it to designate one species of right, that of a patient, corresponding to a duty in an agent. But I think that claim serves as the best generic term in common parlance, so I beg the reader’s indulgence in my using it to cover the pro tanto rights of both patients and agents. 39

A Theory Of rights  |  57

1.3.2. The Autonomy Principle and the Topology of Rights

The first positive point about the significance of the autonomy principle is that it explains the topological structure of the space of rights. What I mean is that rights are built around the claims of patients on an agent. This is quite unlike consequentialism, in which the topology is much simpler: good things and bad things simply add up on different sides of a single agent-​neutral balance—​like positive and negative charges in an electrical field.40 In the space of rights, balancing is also in play; but it is claims, not values, that weigh in the balance, and the balances always operate on individual agents. Two agents, confronting the same set of patients, may have different claims on them because of the different ways they relate to the various patients. Rather than a single balance, the topology of rights is one of multiple nodes, with the claims of all patients coming in separate lines to each agent. This relational topology—​where rights arise from the balancing of patient-​claims on agents—​is essential for making sense of many aspects of rights. I will focus, however, on one particular feature to show that it must be accepted: the need to respect a fundamental distinction between two kinds of patient-​claims, positive claims for aid and negative claims not to be harmed.41 This distinction has to be marked if we are to make sense of the autonomy principle for people not only as agents but also as patients. That is because, to have one’s own life, one needs not only to be free to make one’s own choices but also to live with them and with the way that luck inevitably filters into them. This does not mean that the space of rights should treat people’s bad luck as something that others have no duty to address; the welfare principle is inconsistent with that. It does mean, however, that claims to be helped to overcome harms already inflicted or already heading one patient’s way have to be weaker, all else equal, than competing claims not to be harmed. And to make sense of that notion, rights have to be relational.   This is not necessarily true. Consequentialism can be agent-​relative in a variety of ways. See Sen, 1988. But consequentialism has traditionally been a fundamentally agent-​neutral moral theory. There are moral theorists who use the label differently. For example, Dreier, 2011, uses the idea of good consequences to denote a property that is morally worth aiming at. Whatever is the right thing to do would be, in that way of framing “good consequences,” at least adequately good. This, however, is a revisionist conception of consequentialism. I refer to the original use of consequentialism, according to which the notion of value is capacious; but value has to be measured from an agent-​neutral point of view. (For a compelling argument that Dreier’s revisionist consequentialism is implausible, see Lazar, 2017.) 41  The argument in the next two paragraphs recapitulates a set of arguments I first made in Walen, 1995, pp. 188–​93. I return to the significance of luck in the next section. 40

58  |  The Mechanics Of claims And Permissible Killing in war

To see why that is, consider this case: Murderer: Martin, a would-​be murderer, is threatening to kill two people by running them over. Ronda can save them, but only by ramming Martin’s car with her car. However, if she does that, she will drive his car into a bystander, Bill, killing him.

Now consider the claims in this case. The two whom Martin is threatening to kill have negative claims not to be killed. Bill, whom Ronda would kill if she rammed Martin’s car, also has a negative claim not to be killed. If claims simply added up like agent-​neutral goods, then clearly she should ram Martin’s car so that fewer negative claims are contravened. But that analysis misses an important feature of the case. The two have negative claims on Martin, that he not kill them. Their claims on Ronda, however, are positive claims that she aid them by saving them from Martin. Meanwhile, Bill has a negative claim on her that she not kill him by driving Martin’s car into him. If negative claims are significantly stronger than positive claims, then Ronda may not kill Bill, not even if that means that she cannot save the other two. We can make sense of that only by supposing that what matters are their claims on her, not their claims in some free-​floating claim space that allows them to be added up from an agent-​neutral point of view.42 As noted, other features of the space of rights—​including my explanation for why patients generally have a right not to be used as a means for the welfare of others43—​also presuppose this relational topology for rights. Indeed, a lot of what I will say about rights presupposes it. But this argument from the inability to distinguish negative from positive claims without it suffices to establish it. And it is the autonomy principle that fundamentally requires this topology. Without the autonomy principle, we should be consequentialists. To accommodate it, we need to fundamentally reconfigure the moral space into the space of rights, with its agent-​ relative topology.

 Accepting this relational structure of rights is thus key to avoiding the “paradox of deontology.” See Scheffler, 1982, ­chapter 4. This is the supposed paradox of thinking that it is impermissible to violate some right, R, when doing so is the only way to prevent more violations of R from occurring. This supposed paradox arises only if one thinks of rights as important because of their consequentialist value. If one sees rights as normatively relevant on their own terms and as having the relational topology described in the text, the “paradox” easily dissolves. 43  See § 4. 42

A Theory Of rights  |  59

1.3.3. The Autonomy Principle and Owned Luck

In describing why the autonomy principle requires an agent-​relative topology, I mentioned that people have to own their luck. This means, among other things, that when an agent contemplates acting, it matters whether the patients she confronts are already threatened or harmed and in need of help or are not yet threatened or harmed. If a patient brought a threat or harm upon himself through action that was negligent or worse (reckless, knowing, or intentional), then it should be no surprise that he would have a weaker claim than one who did no such thing. That is a straightforward application of the idea that autonomy and responsibility for one’s choices go hand in hand. But I want to emphasize here that my point is more radical: even if it is just dumb luck that caused one person, and not another, to have suffered a harm, to be under threat, or even to be a threat, that is something that rights accord significance by distinguishing the strength of positive and negative claims and, as I will argue in Chapter 6, the claims of threats and nonthreats. This too is an implication of the autonomy principle: people have to own their luck. I will make the case for thinking that people have to own their own luck first by way of considering an example and then by examining a theoretical argument.44 The most obvious sort of case concerns ownership of one’s body and the connection between that and luck. Some people, through no fault of their own, contract diseases, often fatal ones, that could be cured if and only if they receive new organs from others. Sometimes an organ can be made available at relatively minor cost to the donor: we all have two kidneys, and some, people who are particularly taken with altruism, think that they should donate their second kidney to those in need. But what about hearts? We each have only one. Should all hearts be part of a common pool so that if someone comes to need a heart, through no fault of his own, he gets to insist that a heart lottery be run, the result of which will determine who has to lose a heart? There are two good reasons to reject the heart lottery. First, it is presumably very costly. And what we get in the end is not an extra life being saved; we only get a new distribution of who has a good heart and who has to die. Second, there is good reason to say that the heart lottery was run by nature, and the person who lost that lottery has no cause to call for

 A whole book could easily be written on this topic, so my treatment will necessarily ignore many details. But I think the main point can be made clearly and convincingly in a short space. For a good overview of the issues that arise in the context of luck and justice, see Lippert-​Rasmussen, 2018. 44

60  |  The Mechanics Of claims And Permissible Killing in war

a new lottery. In the traditional literature on running lotteries, there is always some indivisible good to allocate or some indivisible evil to avoid, and there is no default position; a choice has to be made.45 But in this case, there is a clear default position: leave everyone with their original heart. But when we add age and longevity into the picture, we can see that if we are committed to cancelling the effects of dumb luck, there is a reason to run a heart lottery. Everyone has an interest in having a reasonably full life, one that includes getting the chance to grow up and to have a family of her own, a career, or whatever other basic life projects she wants to pursue. But some people, through no fault of their own, will be born with defective hearts that will prevent them from ever reaching adulthood, much less old age. We could address that with a life–​heart lottery. Here is an example: we could set up a system in which anyone whose heart fails, through no fault of her own, below the age of 20 has a claim for a heart from someone 50 or older and anyone whose heart fails, through no fault of her own, between the ages of 20 and 50 has a claim on a heart from someone 30 years her senior. Everyone in the 50 and older set would have to register his heart in the heart pool. Anyone whose number comes up would have to turn himself in to have his heart removed and transplanted to a younger person who has not yet had the chance to lead a full life. If we really want to give all people a fair chance to lead a full life, we should seek to set up such organ–​life lotteries, not only for the heart but for every vital organ of the body. Of course, the costs might be too high, and setting up such a system might create perverse incentives. But if we object to giving brute luck a substantial role in making some better off than others, then we have at least a strong reason to seek to set up such lotteries. I take this to be a reductio of the idea that we should be seeking to negate the role of brute luck in causing some people to fare better and other worse. Part of what it is to have one’s own life is to have one’s own body. If an agent has not voluntarily agreed to join a life–​heart risk pool, others may not treat her, or her body, as a mere resource for equalizing luck. Each person has to accept that the luck of the draw, in terms of how well her body works, is first and foremost her problem or her good fortune. To be clear, I  acknowledge that there is a reason of fairness to run lotteries when deciding how to allocate indivisible goods among people who are in the same pool, with roughly equal claims on those goods. But lotteries should only be run among people whose claims are roughly

 See Broome, 1984.

45

A Theory Of rights  |  61

equal.46 Those who seek to be saved and those who seek not to be tools by which to save others do not have roughly equal claims. I am also not rejecting the importance of human welfare. The welfare principle calls for making sacrifices for the sake of others. But those sacrifices have to be small enough to leave each person the basic freedom to live her own life. A person’s ill health is, again, first and foremost her problem. Others might have to pitch in through taxes to help support medical care for the needy, but the duty to donate one’s heart if one’s number comes up does not count as a small sacrifice consistent with being free to live one’s own life. One might be tempted to turn my reductio around on me and to say that my intuitive rejection of the heart–​life lottery reflects nothing more than status quo bias: it just seems too unusual given that we don’t yet have such lotteries. Just as it was a mistake when people used to think that we have to put up with some people being slaves, so, one might say, it is a mistake to think that we have to accept the brute luck of having healthy or sickly bodies. This resistance comes from the following theoretical commitment: we can be held responsible for how things turn out when we take “deliberate and calculated gambles,” but we should not be held responsible for how things turn out when good or bad things just happen by brute luck, in ways that reflect no prior choice to run a risk, as when “hit by a falling meteorite” out of the blue.47 I will argue, however, that while there is something to this distinction between choice luck and brute luck, I think it is hopelessly misguided to try to seek to eliminate the effects of brute luck on our lives. First, even being hit by a meteor is something we can take more or less care to avoid, for example, by trying to stay deep underground, as opposed to walking out in the open or sitting under flimsy roofs. Second, and more fundamentally, brute and choice luck are thoroughly intermixed into our lives and cannot meaningfully be disentangled. It is not merely the case that two people can get different results when making equally risky choices (result luck). In addition, people bring different skills, resources, and dispositions to the choices they confront, which to a substantial degree reflects the luck of their genetic endowment and the environment in which they were raised (constitutive luck). And people come across different options in life:  some will get the chance to do horrible things and will  See Broome, 1984, p. 628.  Dworkin, 2000, p. 73. What I call “choice luck,” Dworkin called “option luck.”

46 47

62  |  The Mechanics Of claims And Permissible Killing in war

do them, and others will never even face the temptation but would have succumbed if they had; some will get the chance to do great things and will do them, and others will never encounter similar options but would have done as well if they had; some will face nothing but a collection of costly choices, while others will have many happy, low-​cost options (circumstantial luck).48 As Thomas Nagel wrote, the portion of an agent’s life that reflects her choice, unaffected by these sorts of luck, “seems to shrink . . . to an extensionless point.”49 One response to the pervasive presence of brute luck in our lives is to give up the idea of being responsible for what one does and what one achieves. But I think that’s a panicky overreaction. The better thing to say is that we are thrown into the world and then start to weave a self together, dealing with the hand we are dealt and making choices that lead to the next hand. If we achieve the status of normally functioning adults, we become fully responsible for our choices. But to do so we must own the luck that filtered into them. It is a fool’s errand to try to tease out the effects of brute luck; there would be nothing left; it is baked into our lives like flour in a cake. To have one’s own life, as the autonomy principle supposes we do, is to own both sorts of luck—​choice luck and brute luck—​as they have been woven together. This is not to say that we have no reason to distinguish brute luck from choice luck. To hold people responsible, we have to allow them to make choices by which they can forfeit claims that they might otherwise have. The calculated gamble to buy stocks includes the assumption of risk that one will lose money as the flip side of the claim to enjoy new wealth if the stocks one has bought perform well. It makes sense to tell stock investors that they put their money at risk in a way that bank depositors do not. Likewise, as already noted, those who act negligently or worse become liable to paying costs (e.g., through tort suits) that others who cause similar harms do not. And the criminal law, though normally requiring at least recklessness, also normally punishes people who cause harms more seriously than those who do not.50 These are all ways that the autonomy principle supports distinguishing choice luck from brute luck. But that should not be taken to mean that we have reason to seek to do the impossible: to root out and negate the influence of brute luck on our lives.

 These are three of the four kinds of moral luck mentioned by Nagel, 1979a.  Ibid., p. 35. 50  Alexander and Ferzan, 2009, argue that one should only be responsible in criminal law for the risks one takes, not the outcomes. I criticize their view in Walen, 2010. 48 49

A Theory Of rights  |  63

What I  have said should also not be taken to imply that we have no reason to restrict the impact of brute luck on the lives of agents. Distributive justice, I  believe, requires us to seek to design institutions so that brute luck has no more impact on how one can benefit from the fruits of social cooperation than necessary to allow for the kind of incentive structure that allows free people to choose satisfying, productive lives.51 This rules out discrimination in many realms of public activity, such as employment and housing, based on race, sex, sexual orientation, and other similar difficult-​ to-​change characteristics. It also rules out a system that allows an unlimited market-​based reward for talents. In addition, as I already noted with regard to health problems, the welfare principle implies that when people suffer harms due to brute luck, they have reasonably strong claims for aid. The fundamental limit on these claims is only that they should not be so demanding that others have too little normative space to lead their own lives. My point then is not to deny that brute luck can be distinguished from choice luck or that the negative effects of brute luck should be mitigated. My point is only that it is pointless, foolish, and even morally inappropriate to try to strip out the effects of brute luck on our lives. We are too much the product of the brute luck that is woven into our lives. And thus, while we have claims to limit the impact of brute luck in certain ways, we also must own it as ours. I close this section with two final thoughts. First, I have to acknowledge the difficult problem of determining when luck “vests.” It seems intuitively that a threat headed to A  is A’s problem, even if it could be redirected onto B. But it also seems intuitively that if A and B are in need of rescue and a resource like a rescue boat is headed to A but the boat captain then discovers that B is also in need, it would be perfectly reasonable for her to flip a coin. There may come a point when A has a claim on the boat—​for example, when A has been given reason to rely on its coming to him. But until then, the mere fact that the boat would have gone to A if the captain had done nothing does not give A a stronger claim on it than B. It is hard to say just when luck vests and why, and trying to do so would take me too far afield.52 But I think it is generally clear enough for the cases I consider. Second, there is a lesson in all of this for what it is permissible to do to threats. The fundamental assumption in the self-​defense literature that

 I take this to be essentially a restatement of one of Rawls’s arguments for the difference principle. See Rawls, 1971, pp. 72–​73. 52  I will say a bit more about rescue cases in Chapter 8, when discussing competition. 51

64  |  The Mechanics Of claims And Permissible Killing in war

looks to forfeiture to explain why threats can be killed is that the right not to be killed is too robust to be affected by the mere bad luck of being a threat.53 I hope it is now clear why that assumption is in principle unsound. Working out the concrete implications of owning luck for the right to defend against threats is something I come back to in Chapters 6 and 7.

2. Formalizing the Mechanics of Claims Having laid the foundational principles for the space of rights, I turn now to explaining how I  think it operates. I  invoke the mechanics of claims to explain how rights emerge as outputs based primarily on claims as inputs.54 This occurs in two stages. In the first stage, rights emerge from first weighing the competing basic patient-​claims55 on an agent and then taking into account any agent-​claims the agent might have to act contrary to the balance of patient-​claims on her.56 In the second stage, special patient-​claims must be taken into account. I discuss these stages in turn. To ward off confusion, let me start with this clarification: agent-​claims are not claims on anyone; they are claims to be free to act or refrain from acting. In that way, they are fundamentally unlike patient-​claims. If an agent has a right to X, that does not, by itself, imply that anyone has any duties to her. It means simply that no one has a right that she not X and that there are no other considerations that make it the case that she may not X.

2.1. Balancing Agent-​Claims and Basic Patient-​Claims Basic patient-​claims are patient-​claims that presuppose no special relationship between the patient and the agent on whom the claim registers. All patients who have an interest in what an agent does have basic patient-​claims on that agent. Even those who have waived or forfeited the normal claims associated with an interest normally have some weaker residual claim. As noted, even culpable aggressors have a claim not to be killed if doing so is unnecessary to stop their aggression, that is, if their

 See Chapter 1, § 1.  As noted in § 1, claims are not the only inputs, but they will be the only inputs in most of the cases I discuss. 55   In Walen and Wasserman, 2012, we used the label impartial patient-​claims rather than basic patient-​claims. I now think that was a poor choice as these claims function, in some ways, to ground justified partiality. 56  As will be seen shortly, some of the agent’s claims register in the balance as basic patient-​claims, but the agent also has agent-​claims. 53 54

A Theory Of rights  |  65

aggression can be safely stopped with the use of less harmful means and no substantial cost to anyone else.57 The weight or strength (I use those terms interchangeably) of basic patient-​claims reflects the magnitude of the interests in play, whether the claim is for help or not to be harmed, and other ways the patient potentially relates, causally, to the agent and to other patients whom the agent might affect by her choice. They do not, however, reflect the history of the relationship between the agent and the patient, insofar as that history can give rise to special patient-​claims. They do not, for example, concern the claims of promisees on promisors or the claims of family on family, friend on friend, or victim on tortfeasor. Those are all special patient-​claims, which have to be addressed at stage 2. In what follows in this section, I  do four things. First, I  present an overview of the main features of the mechanics of claims and the stage 1 balancing it assumes. Then, I  focus on how an agent’s interests enter the balance, explaining that they sometimes enter as negative agent-​ claims and sometimes as positive patient-​claims. Next, I argue that there is some scope for positive agent-​claims. Finally, I end by arguing against those who endorse “agent-​centered prerogatives” by arguing that, when it comes to causing significant harms, agents enjoy no substantial positive agent-​claims. 2.1.1. Overview of Balancing Claims, Organized in Terms of Partiality and Impartiality

The three main features of the mechanics of claims, at stage 1, can be stated as follows: • If one or more patients, P1, have basic patient-​claims that A do X, one or more patients, P2, have basic patient-​claims that A not do X, and P1’s claims are stronger than P2’s, then A must do X unless A has an agent-​claim not to have to do X that is strong enough to free her from the duty of responding to the balance of basic patient-​claims. • If P1’s claims ground a duty for A to do X, then, unless and until P1 waives or forfeits those claims or some new situation arises in which P1’s claims become weaker or competing claims come to outweigh P1’s claims, P1 has a right that A do X, and A owes a corresponding duty to P1.

 See § 1.2.3.

57

66  |  The Mechanics Of claims And Permissible Killing in war

• If A’s agent-​claim gives her the liberty not to do X, then she has a pro tanto right not to do X. But if the balance of basic patient-​claims calls on her to do X, then she would also have a pro tanto right to do X.58 I propose to unpack these bullet points by focusing on the way the mechanics of claims handles impartiality and partiality. Basic patient-​claims embody the norm of impartiality insofar as they reflect the equality principle: we are all fundamentally equal in the space of rights. Thus, if an agent confronts two sets of competing basic patient-​claims, the normative force on her reflects the impartial balance of those claims. It doesn’t matter who has basic patient-​claims. Basic claims embody partiality insofar as they reflect the autonomy principle: each person has her own life to lead. The normative idea that enables the mechanics of claims to make proper room for partiality in the balance of basic claims is property. Property demarcates what each person can rely on as hers as she seeks to pursue her own life plan. These tools for living are, first, our bodies, our time, and our energy, which constitute our first property, and, second, the rest of what we lay claim to, which constitutes our second property.59 I assume, throughout, that property owners have an immunity to being dispossessed of their property by private individuals unless they forfeit it, transfer it, or give others rights to use it or others have what I will call “empowering claims” that give them temporary and limited-​use rights.60 Obviously, this assumption presupposes a more or less just property regime. If the regime is not sufficiently just, then second property, and arguably even bits of first property, can be appropriated more liberally.61 Additionally, most legal orders will tax property and will occasionally take property for public use—​though in the latter instance owners should be fairly compensated.62 Property has an important function in the mechanics of claims for both agents and patients. For agents, it registers in two types of claims: claims not to have to use their property for ends other than their own and claims to be free to dispose of their property as they wish. For patients, it is slightly more complicated, but for now we can say that it also registers in two types

 The pro tanto rights A has at stage 1 may be restricted by a special patient-​claim at stage 2.  I take the notion of first and second property from Thomson, 1990, p. 226. 60  I describe empowering claims in § 4.4. 61  On the possibility of appropriating first property in an unjust regime, see Stanczyk, 2012. 62  For an enlightening overview of the relationship between property claims and liability claims, see Calabresi and Melamed, 1972. 58 59

A Theory Of rights  |  67

of claims63: claims not to have their property used without their consent and claims on agents—​special claims—​that are property-​like claims over an agent’s liberty. I discuss primarily agent-​claims here in section 2.1. In section 2.2, I discuss the second role of property in patient-​claims: special claims. And in section 4.1, I discuss the first sort of patient-​claim in more depth. 2.1.2. Agents’ Negative Agent-​Claims and Positive Patient-​Claims

Turning, then, to the claims of agents, it is important to be clear that agents are never merely agents. Agents have agent-​claims insofar as they are acting, but they also have patient-​claims that register how they will be affected by what they do. There is no double counting here; this division reflects a division of conceptual labor. I argue in this section that agents have very strong negative claims, claims not to have to sacrifice themselves. I argue in section 2.1.4 that they lack substantial positive agent-​ claims, claims to be free to do things to favor themselves at a cost to others. This denial of substantial positive agent-​claims is not overly burdensome, however, especially considering that their interest in benefiting from their acts is captured by their patient-​claim to that benefit. An illustration will help. Recall Trolley Switch: Brenda confronts the choice of whether to turn a trolley from five on one track onto one on a sidetrack. She is confronted with six basic patient-​claims, and she also gets to consider her own agent-​claim. The patient-​claims break down into two groups: the five have positive claims to be saved; the one has a negative claim not to be killed. Brenda has a negative agent-​claim not to have to turn the trolley if she does not want to and logically could have—​though, as just noted, I will argue that this is not the case—​a positive agent-​claim to be free to turn it if she wants to. Now suppose Brenda is also one of the five toward whom the trolley is headed. For the purpose of counting patient-​claims, she gets to count her life as one of the five. Her claim to be saved weighs, along with the other four, against the claim of the one not to be killed. If there is some minimal number of lives that must be saved before the positive patient-​claims outweigh the negative one—​say it is five—​then her patient-​claim would tip the balance of patient-​claims.

 The complication is that, as I will argue in Chapter 7, patients also sometimes have property-​like claims that reflect their claims as agents. 63

68  |  The Mechanics Of claims And Permissible Killing in war

What about Brenda’s agent-​claims? Assume for now that she has no meaningful positive agent-​claims. She would then be permitted to turn the trolley if and only if the patient-​claims in favor of turning it outweigh the patient-​claim against turning it. We are supposing that that is the case, so she may turn the trolley and save herself along with the four others. What about her negative agent-​claims? They protect her freedom not to have to serve as a tool of the balance of patient-​claims. As that is a fundamental function of the autonomy principle, an agent has to have a strong claim not to have to use her property—​in this case, her body, energy, and time—​for ends other than her own. Small sacrifices can be required to prevent large harms to others, but if the sacrifices are large, the agent may choose not to sacrifice herself for the sake of others. Importantly, the measure of the sacrifice the agent has a claim not to make has to take into account whether such situations repeat; repeated small duties erode an agent’s ability to lead her own life, and thus an agent has a stronger claim against small duties if they repeat than if they arise only once or rarely. To illustrate the significance of negative agent-​claims using a trolley switch case, we cannot continue to imagine that Brenda is one of the five who will be hit; in that case, her permission not to turn it onto the one might reflect her Hohfeldian power to waive her patient-​claim to be saved. To avoid that complication, we should return to the original Trolley Switch, in which Brenda is a bystander confronting five positive patient-​claims to be saved and one negative patient-​claim not to be killed. We can continue to assume that the balance of patient-​claims favors the five. The question is, How does her negative agent-​claim fit into the balance? Suppose that her primary interest is in not being a killer. That could reasonably be taken to be a fairly significant interest: even if killing is morally permissible, it is, for most people, a traumatic thing to have to do. Given this significant interest and that the balance of patient-​claims is not completely one-​sided,64 her strong negative agent-​claim would plausibly give her a right not to turn the trolley. By contrast, if her concern were simply that she did not want to expend the energy to turn the trolley, she might be required to turn it. What if she would be better off making the choice to turn the trolley? The autonomy principle, which grounds a basic commitment to anti-​ paternalism, implies that that would not undermine her right not to turn it.

 I think it would make a difference if the sidetrack man had wrongfully set the trolley on the other five and thereby forfeited his claim not to be killed. It is then more plausible that Brenda would have a duty to turn the trolley onto him, despite the trauma that this might cause her. 64

A Theory Of rights  |  69

It is her life, and she gets to decide what to make of it, even if her decisions are not optimal. This gives us reason to respect her choice as one that she has a right to make, even if it is not the one that she should make. Note that an interesting complexity is revealed in this case, one concerning how we should weigh the agent’s interest. The strength of her claim depends on both the magnitude of her interest and the nature of her claim. Her interest is a compound of her autonomy interest, to choose according to her own lights, and an objective assessment of what would be good for her. Anti-​paternalism implies that unless the welfare assessment is very great—​putting the claims of others aside for the moment—​an agent should be allowed to choose by her own lights. But the net weight of her interest would be slighter if the two component parts pull in opposite directions than if they pull in the same direction. Thus, if they pull in opposite directions, it will be easier for the balance of patient-​claims to outweigh her negative agent-​claim, and thus easier to establish that she has a duty to act. It might be tempting to say that an agent’s objective welfare must count as a patient-​claim for her, regardless of whether her agent-​claims are negative or positive. But that split—​leaving her autonomy interest to register as a negative agent-​claim, while registering her objective welfare as a patient-​claim—​is implausible. If an agent’s interests could be separated into an autonomy interest and an objective welfare interest, then it would be hard to make sense of how serious the autonomy interest is. But surely an agent’s autonomy interest is more important if she is making a choice about a matter that impacts her welfare significantly than if she’s making a choice about a trivial matter. Thus, the welfare interest and the autonomy interest must be linked. This has an important implication for how to count interests for an agent: when considering if an agent has a right to act, she gets to count her interest in performing the act as a patient-​claim to benefit from the act; but when considering whether she has a right not to perform an act, she does not get to count her interest in not performing it as grounding a patient-​ claim that weighs in favor of not performing it. Rather, both components of her interest in not performing it are fully picked up in her much stronger negative agent-​claim not to have to sacrifice herself for the sake of the balance of patient-​claims. Thus, for example, in a variation on Trolley Switch in which the agent is the person on the sidetrack (i.e., she would have to turn it onto herself to save the five) her interest in not being hit counts as giving weight to her negative agent-​claim not to have to turn the trolley. But if she is with four others on the track toward which the trolley

70  |  The Mechanics Of claims And Permissible Killing in war

is headed, her interest in not being hit counts as one of the patient-​claims that push for her to turn it. Taking stock, the general picture is that agents function as nodes on which patient-​claims compete. To a first approximation (I introduce a small qualification immediately below), insofar as an agent wants to perform an act that some patients have a claim that she not perform, she can justify doing so only if the balance of patient-​claims (including her own) pulls in the direction of her doing so. Insofar as she wants to avoid acting in the service of the balance of patient-​claims, however, she has a strong claim to be free to avoid so acting and would have a right to avoid so acting unless her interests are greatly outweighed by the balance of patient-​claims.65 2.1.3. Some Scope for Positive Agent-​Claims

I have been assuming that positive agent-​ claims carry essentially no weight, that agents are free to act only insofar as they have positive patient-​ claims that support their doing so. But one could object that this picture gives agents too little freedom to use their property—​again, understood to include their bodies, their time, and their energy—​as they would like in order to lead lives they would like to lead. To see why, take, for example, A, who wants to paint her house yellow but who lives next to B, who hates yellow houses. Putting aside zoning ordinances that might restrict A  in certain ways, A should be free to do that sort of thing. The mere fact that B would find A’s choice disagreeable is not a sufficient reason to give B veto power. Yes, if B would suffer a serious harm if A painted her house yellow—​for example, B would have convulsions or B’s house would lose 20 percent of its sale value—​then B might have a moral veto. But A ought to be presumptively free to paint her house the color she wants to paint it; it’s her house. Yet it may seem that my picture would not give A that freedom. After all, if we weigh B’s negative patient-​claim not to suffer aesthetic distress caused by the yellow color against A’s positive patient-​claim to the aesthetic enjoyment she would get from having a yellow house, and if we keep in mind that negative patient-​claims are generally stronger than positive patient-​claims, all else equal,66 then B’s claim would seem to win. To handle that case, we need to recognize a basic freedom to use one’s property in ways that cause only minor annoyance or offense to others. One might try to make this case, as John Stuart Mill did, in utilitarian  Strictly speaking, this is just a pro tanto right, one which could be undermined at stage 2 by a special patient-​claim on her liberty. 66  See § 1.3.2. 65

A Theory Of rights  |  71

terms, appealing to the overall welfare. But Mill’s argument was distinctly focused on limiting the policing of acts (vices) that cause only minor harm.67 I want to argue directly for the moral right to act, where any right of non-​interference is derivative of the right to act. Nevertheless, I think we can do that in fundamentally similar welfarist terms. If we take into account the importance for individuals of being presumptively free to use their property as they wish, as long as they do not cause significant harm to others, we find—​I believe and assume—​that we have a better system for human flourishing than if we allow the offense and annoyance of others to morally restrict the freedom of agents. Of course, the contours of this freedom can and must be specified and adjusted in various ways to further promote the general welfare. A more determinate framework will normally be provided by a mix of cultural and legal norms, such as zoning laws, that set legitimate expectations both for what agents should be free to do and for what they should refrain from doing in any particular cultural and legal setting. This is not to say that whatever the local culture and law say is what morality should require. Both may be defective in various ways. For example, they may wrongly permit or require discrimination against certain kinds of people, in violation of the equality principle. In addition, it is often quite proper for the law to grant more legal rights than should morally be exercised. Consider, for example, free-​speech rights. The law might have good reason to leave them broader than morality should allow. The reason is that the law has to concern itself with institutional matters that are not relevant to first-​ order morality, such as who would censor speech and what tests should be used. In the domain of free speech, then, we seem to have good reason to allow custom and moral judgment to mark the limits of the permissible exercise of freedom without support from the law.68 But the general point remains: law, custom, and morality should allow people a basic freedom to use their property as they see fit, without having to worry about every minor harm and annoyance they might cause. There is something deeply right in the basic liberal conceit that what consenting adults do with their own lives and property is and ought to be primarily their business. To that extent and only to that extent—​as I  now argue—​do agents enjoy positive agent-​claims. From here on, I will make no further reference to these minor positive agent-​claims; their existence can be presumed as part of

 See Mill, (1859) 1975, c­ hapter 4.  In saying this, I take no position on just how much speech can be properly regulated by the state, only that the state probably should not seek to censor all morally impermissible speech. 67 68

72  |  The Mechanics Of claims And Permissible Killing in war

the background, a background that will have no relevance when what is at stake are significant harms that an agent might cause. 2.1.4. The Argument Against Agent-​Centered Prerogatives

Negative agent-​claims are claims not to have to do things for the sake of others who have positive claims for aid; positive agent-​claims are claims to be free to do things despite their negative impact on others who have negative patient-​claims not to be harmed. The idea of agent-​centered prerogatives, as the term is standardly used, fails to mark that distinction. Such prerogatives simply allow agents to give preference to their own ends over the competing ends of others.69 They are like excuses in that, unlike standard justifications, they do not extend from the agent to third parties. But they are more affirming than excuses. An agent may not plan to take advantage of an excuse70; she may plan to act on an agent-​centered prerogative. Many people think that we enjoy agent-​centered prerogatives. They are mistaken.71 Negative agent-​claims must be strong because they are essential for giving agents the moral freedom to live their own lives, as opposed to being morally required to dedicate their lives to serving the balance of basic patient-​claims on them. I, along with David Wasserman, have argued elsewhere that there are three reasons why agents do not also need, and should not be taken to have, positive agent-​claims to complement their strong negative agent-​claims.72 First, such positive claims are not necessary for agents to avoid having to treat themselves as tools for the greater welfare; that task is achieved by negative agent-​claims. Second, an agent’s interest in being free to do what she wants to do is fairly taken into account by her patient-​claims to deny others the use of her property, the freedom to use it herself as long as she causes no significant harm in doing so, and her positive patient-​claims to enjoy the benefits of her own acts. Once these three factors are taken into account, the autonomy principle is satisfied; she has the freedom to lead her own life and  The idea was first popularized by Scheffler, 1982. It has since been influentially taken up in the self-​defense literature by many people, perhaps most notably by Quong, 2009. 70  She may not, for example, arrange to be coerced so that she can claim the benefit of duress. Nor may she choose not to know something relevant and claim ignorance: that’s willful blindness. She may, however, sometimes plan to act and plan to get the benefit of an excuse if the act is sufficiently important and the excuse too hard to avoid. She may know, for example, that she suffers seizures and yet choose to engage socially. If she then has a seizure and falls on those near her, she will be excused for doing so. 71  Tadros, 2011, c­ hapter 9, makes essentially the same argument. 72  Walen and Wasserman, 2012, pp. 555–​56. 69

A Theory Of rights  |  73

needs no additional positive agent-​claim to be free to pursue her ends even to the detriment of others. I can make the third point no better now than before, so I  quote our earlier text at some length: [G]‌iving [her] positive agent-​claims significant weight would allow [her] to choose to harm others for the benefit of [herself] or those [she] care[s] about in ways that seem unjust. Consider, again, the case in which [an agent] could turn the trolley from [herself] onto another. [Her] patient-​claim is the positive claim to be saved from the trolley; the [sidetrack] man’s patient-​claim is the negative claim not to be hit by the trolley. If we assume that negative patient-​claims outweigh otherwise similar positive patient-​claims, then the balance of patient-​claims would (in the absence of other considerations) prohibit [her] from turning the trolley. If [her] positive agent-​claim had substantial weight, it might tip that balance. But that seems unjust. [She] may not, we believe, turn the trolley from [herself] onto another. To capture the sense that it would be unjust for [her] to do so, we would say that [her] positive agent-​claim does not have the capacity to tip the balance here. Indeed, we think it never has the capacity to tip the balance between lives or other weighty interests. Even to let a claim to do what one likes break a tie between negative claims not to be killed would fail to respect the gravity of the latter claims, and the importance of impartiality when it comes to respecting patient-​claims. In sum, if [her] agent-​claim is positive, then [she is] required to defer to the balance of patient-​claims.73

One might object that this rejection of positive agent-​claims cannot handle the following example.74 Suppose that a trolley is hurtling down the hill toward 10 people and Brenda could turn it onto one of two tracks: either onto the left track, where it will kill her son, or onto the right track, where it will kill a stranger. The balance of patient-​claims calls on her to turn the trolley away from the 10. It also calls on her to be impartial between turning it onto her son and the stranger. If she has no positive agent-​claim to choose her son over the stranger, then she must do something to treat them fairly, such as flip a coin.75 But imposing on her son a 50 percent risk of being killed so that she can save 10 seems inconsistent with the special  Ibid.   Steinhoff, forthcoming, raises this example against Tadros’s 2011 position rejecting an agent-​ centered prerogative. 75  This is one of two positions that Wasserman and I considered plausible. See Walen and Wasserman, 2012, pp. 562–​63. 73 74

74  |  The Mechanics Of claims And Permissible Killing in war

duty she owes her son. If the case were one in which she had to choose whether to turn the trolley from 10 onto her son, his special claim on her would remove from her the liberty she would otherwise have to save the 10; she would owe him the duty of not turning the trolley onto him—​a point I explore more in the next section. It is no different if she would be imposing on him a 50 percent chance of dying; that is much more than she may rightfully impose on him for the sake of strangers. But then it seems that she must either allow the 10 to die or invoke a positive agent-​claim to favor her son and turn onto the stranger. Given the importance of saving the 10, the normative pressure here is to acknowledge a positive agent-​ claim to favor her son. Indeed, we could make this pressure even stronger by changing the case such that there are two strangers on the other track. If she wants to save the 10, then she must sacrifice her son as his basic patient-​claim not to be killed is weaker than that of the two strangers on the other track. She would have a right, based on a negative agent-​claim, not to have to sacrifice her son for the sake of the 10 strangers. But that means that 10 must die rather than the two onto which she could turn the trolley. That, however, seems wrong. It might seem, then, that the only solution is to say that she must be free to exercise a positive agent-​claim to favor her son over two strangers. The problem with this argument is that it presupposes that the three-​ way choice has to be broken down as follows: first, decide whether to save the 10, then decide which of the others to kill. But there is another way to approach the three-​way choice that makes at least as much sense, avoids the problems, and retains the idea that agents enjoy no positive agent-​ claims to act contrary to significant patient-​claims. Start with the thought that Brenda has a strong negative agent-​claim that grounds a right not to sacrifice herself or her son for the sake of others. She can then take that option off the table. Having done so, she should then look at the remaining options: either allow 10 to die or turn the trolley onto one or two strangers. Assuming that the balance of basic patient-​claims favors five or more positive claims over one negative claim, all else equal, she may turn the trolley. At no point must she invoke a positive agent-​claim to justify doing so.76 Given that this enables us to simultaneously satisfy the autonomy principle’s protection of her freedom not to have to sacrifice or risk sacrificing her son and the welfare principle’s concern for the lives of the 10 and that it also protects patients from the overly broad privileges  This is the other position that Wasserman and I considered plausible. Ibid., p. 563. I now conclude that it must be the better position. 76

A Theory Of rights  |  75

that would result from according agents positive agent-​claims, this seems the better approach. It does lose the appeal of the thought that if Brenda decides to save the 10, then her son and the others are in some sense in the same boat and should be treated equally—​an appeal that draws on the equality principle. But as the objectors would reject that equality in the next move, by giving her a positive agent-​claim that would allow her to turn the trolley onto the others to save her son, respect for the equality principle provides no real reason to accept the framing which is the basis for the objection. The simpler solution, and the one that coheres with the overall commitments that frame the mechanics of claims, is to accept the framing of the situation according to which she gets to first withdraw the option of sacrificing her son and then consider the other options.77 In sum, agent-​claims are negative when they are claims not to have to make a sacrifice for the benefit of others. Agents get to give substantial priority to their negative agent-​claims. If they protect relatively trivial interests, the balance of patient-​claims can tip the overall balance and impose on an agent a duty that contravenes her agent-​claim. But if her agent-​claim protects an important interest of hers and she owes no one a special duty that effectively negates her agent-​claim with respect to that claimant, then she must be free not to serve the welfare of the patients whose interests ground the positive patient-​claims that compete with her negative agent-​claim. If, however, her interest is in doing something that runs contrary to the balance of patient-​claims, when that balance is tipped in favor of negative claims not to be significantly harmed, then she must yield. She has no right to favor herself in such a case. In such a case, her property interest in denying others the use of her body, time, energy, or second property does not come into play. There is no general right to be partial to oneself, and there are no general agent-​centered prerogatives.

 Kimberly Ferzan asked me what I would say about a double-​switch case in which Brenda could save 10 by turning the trolley onto her son but then could save her son by turning the trolley onto two others. Each stage seems impermissible, but the outcome seems permissible. I am inclined to say that she may make the double switch. It is part of a single plan to save 10 at the cost of two, and at no stage along the way is anyone unjustly harmed. Yes, her son is briefly threatened, but if she knows that she can throw the second switch, that does not violate his right not to be endangered by her. And the two are not really killed to save one; they are killed to save 10. It would be different if, say, she turned the trolley from 10 onto someone and then discovered that she had turned it onto her son. She may not save her son from her own mistake by killing two. (Note, this is a case in which the permissibility of the act is indeterminate; it depends on the action. This is an exception to the general rule I articulated in Chapter 1, pp. 9-​10: that the permissibility of acts should be established first.) 77

76  |  The Mechanics Of claims And Permissible Killing in war

2.2. Including Special Patient-​Claims Special patient-​claims on agents reflect the prior history particular patients have with particular agents. For example, the claims of promisees on promisors, children on their parents, and tort victims on tortfeasors are all special patient-​claims. They enable a patient to make a stronger claim on an agent than others can make. However, they do not—​I believe—​enable patients to empower agents to do things that they could not rightfully do given the balance of basic patient-​claims on them. For example, Brenda is not permitted to turn a trolley from one onto two or more people, and this, I believe, is true even if the one is her own child. The claim a child has on his parent does not give the parent permission to kill two others for the sake of the child. Special patient-​claims function only as a property-​ like claim on the liberty of an agent, a claim to limit the freedom an agent would otherwise have, given the balance of basic patient-​claims on her and her own agent claims. Consider a parent confronted with a number of children all stuck in a burning building. She has a special duty to try to save her own child first, even if she could save more children if she tried to save other children first. This is for two reasons. First, she has a right not to have to save the greater number of children if doing so would cost her the life of her own child. Having a right not to devote herself to saving the others, she retains the property right over her time and energy to save her own child. Second, her child can then demand of her that she save him, a demand grounded in the child’s special patient-​claim on her liberty.78 Her child’s claim essentially overrides the second-​order property right, the Hohfeldian power, that the parent normally would have to waive her right to use her time and energy for the rescue of her own child. But this ability to override an agent’s power to waive her own rights does not grant the child, the holder of the special claim, an ability to augment the parent’s first-​order liberty. If the parent does not have the liberty to save her child on the basis of the balance of the basic patient-​claims and her agent-​claim not to serve the balance of basic patient-​claims, then the child’s special claim fails to attach to

 According to this account, contra Taurek, 1977, one cannot choose to save one stranger over, say, three others, simply because one likes something about him. The loss to the agent of not saving the one has to be great enough, judged in some objective way, to give the agent the right not to save the others despite the fact that they have, collectively, a stronger set of basic patient-​claims. This comment presupposes that Kieran Setiya’s view of love based on nothing more than another’s humanity is mistaken. See Setiya, 2014. 78

A Theory Of rights  |  77

any power the agent might have had. A special claim cannot override the power of an agent if the agent lacks that power in the first place. My claim that a parent may not save her child if the cost of doing so is that she kill two others as a side effect may not go down easily. I confess: it gets worse. If special patient-​claims cannot empower an agent to do what she could not do in their absence, then a parent may not turn a trolley from her child onto only one other person as the balance of basic patient-​claims would not allow a bystander to turn a trolley from one person onto another. I readily admit this is difficult to swallow. But there are two things that one might say about this case. First, one might suggest that we are misled into thinking that a parent may turn the trolley from her child onto another because we are misled into thinking that turning a threat from one onto another is morally more innocuous than it is. I argue in Chapter 6, section 2, that there is no moral difference between (a) turning a trolley from A onto B and (b) blowing up a trolley to save A, thereby killing B with shrapnel from the explosion. But it is not so intuitively difficult to accept that a parent may not set off a bomb to save her child if she will thereby kill a stranger. If that is not so hard to accept and if turning a trolley from A onto B is morally on a par with blowing up the trolley to save A, thereby killing B with shrapnel from the explosion, then we should perhaps simply accept that a parent has no right to turn a trolley from her child onto a stranger. Second, the function of rights is to enable a real harmonization of the different lives and moral interests involved. In some cases, it seems that that harmonization breaks down. A parent killing a stranger as a side effect of saving her child seems to be one such case: the importance of intimate relationships may cause a breakdown in the harmonization of lives that the space of rights aims to provide. The demands of love and loyalty, in this kind of case, tear at the idea that rights provide a coherent space in which all can live together in mutual respect; these demands may thus be irreconcilable with the demands of rights. An agent has no right to kill for the sake of her child, but she also has a special duty to her child such that she wrongs him if she does not save him in a situation like this. I return to the idea of the breakdown of the harmonization that the space of rights aims to provide in Chapter 4, section 3, when I discuss threshold deontology, and again in Chapter 8. As for this possibility of a breakdown in the possible harmonization of different lives in a system of rights, there is one point in particular that I want to emphasize here: I think it is a virtue of my view of rights that it does not easily devolve into the right of mutual aggression. I admit that morality cannot rule out mutual aggression in the case of a parent who

78  |  The Mechanics Of claims And Permissible Killing in war

has a duty to save her child even though it would kill others, while others, or third parties acting on their behalf, have a right, or even a duty, to use force to try to stop the parent from doing so. Many theorists accept that situations like that—​in which people have a right, if not a duty, to try to kill one another—​arise easily enough in the context of self-​defense. I believe this view is deeply mistaken. It arises from a view of rights as some sort of given, as if it is some sort of fundamental moral truth that one has a right to engage in self-​defense. But that is a bad way to look at rights; rights are more contextual than that. Moreover, conflicting interests and competing claims are the condition of life in a shared space, but conflicting rights, outside of the limited space of justified competition (a topic I discuss in Chapter 8), should be seen as exceptional cases of a breakdown in the normative order that the space of rights aims to establish. Under normal conditions, they should hardly ever arise. In sum, in my view, rights are grounded in the three principles—​the autonomy principle, the equality principle, and the welfare principle—​and the object of a space of rights is to harmonize those principles into a normative space in which all can live together respectfully. It may not be possible to achieve complete harmonization, but breakdowns should be rare. And as long as a breakdown has not occurred, special claims should be seen as operating only to limit the freedom an agent otherwise would have had.

3. Points of Contrast with the Infringement Model The mechanics of claims is offered as an alternative to the infringement model.79 While the mechanics of claims treats rights as conclusions about what may be done, the infringement model treats rights as inputs into further moral deliberation about what may be done. Sometimes the conclusion is that a right must be respected; sometimes the conclusion is that it need not be respected. Failure to respect a right that must be respected constitutes a rights violation; failure to respect a right when such failure is permissible constitutes a rights infringement. The stringency of a right, along with other considerations, such as a consequentialist assessment of the good that might be done by not respecting it, determine whether the right may permissibly be infringed or must instead be respected. If

 I am not the only theorist who is critical of the infringement model. For example, Tadros, 2011, p. 200, writes that he is “skeptical of the significance of the idea that rights are infringed.” But the argument that follows is the most detailed criticism of which I am aware. 79

A Theory Of rights  |  79

an agent infringes the rights of another, she permissibly (or rightfully) wrongs him; and such wronging normally grounds a right to compensation.80 But people can waive or forfeit their rights. If they do, then an act that would otherwise be a rights violation or rights infringement is simply permissible, as long as nothing else makes it impermissible. The way the infringement model allows rights to be overridden by other, stronger considerations fits our common usage of rights language.81 Additionally, it captures something important that the mechanics of claims cannot capture: the rhetorical force of saying “people have a right to do X.” For example, if people are suffering from polluted air, it is rhetorically more powerful to insist that people have a right to clean air, even if that right is not absolute, than to insist that people have a claim to clean air. But this sort of political rhetoric is distorting when speaking morally about what particular acts may or may not be performed. Therefore, I argue in the next chapter, and in passing throughout the book, that the infringement model is inferior to the mechanics of claims for the purpose not of rhetoric but of moral reasoning. Before making that argument, I want to highlight two more points of contrast between the mechanics of claims and the infringement model. First, claims, like rights, can be waived or forfeited; but they cannot be infringed. They can only be weighed against one another. If a claim is outweighed, no right corresponds to it. An agent does not “wrong” a patient if she “contravenes” a patient-​claim that fails to ground a right. This can seem counterintuitive in two ways. First, there are cases in which an agent rightfully harms another and the horror of what is done to the other makes it seem that the agent must have wronged the other. But I will argue in the next chapter that this is just a confusion. Second, it might seem to imply that the mechanics of claims cannot pick up an important insight captured by the infringement model, namely that compensation is owed when rights are, as the infringement model would put it, infringed. But the  It is noteworthy that while philosophers who use the infringement model seem to assume that if A infringes B’s right not to be harmed, then A owes B compensation, the law is not so clear. Tort law in the United States allows B to recover from A if A is a private actor acting to preserve herself or her property from a greater harm than she inflicts on B, but English law does not. Moreover, courts in the United States are divided about whether compensation is owed if a public actor invokes public necessity to justify infringing B’s rights. And it is not clear whether a private actor inflicting harm for the sake of others should count as a public actor or a private one. See Simons, 2018. I am not suggesting that tort law is doing a better job of tracking when compensation is owed than moral philosophers. I am suggesting, however, that the intuitions of moral philosophers about when compensation is owed should be taken with a grain of salt. See also Tadros, 2016a, p. 110n2 (doubting “that the relationship between liability and compensation is simple”). 81  See the Preface, note 7, for more on common usage of the term. 80

80  |  The Mechanics Of claims And Permissible Killing in war

fact that a claim is contravened is not necessarily the end of the story; it is consistent with finding that related claims ground secondary rights, such as the right to compensation. As I argue in the next chapter, we do not need the idea of infringing a right and wronging the right-​holder to account for claims for compensation; indeed, the infringement model oversimplifies when compensation is owed, and we can account for claims for compensation better without it. The second point of contrast between the mechanics of claims and the infringement model concerns the range of interests that are taken to register in the space of rights. As already noted, the mechanics of claims assigns claims to all welfare interests that could sensibly be registered that way. This contrasts with the infringement model, according to which welfare interests frequently register not in terms of rights but in terms of consequentialist considerations that somehow weigh against rights. The idea of “lesser-​evil justifications” brings this contrast into high relief. The infringement model uses those as a sort of complement to rights; the mechanics of claims eschews such talk. One result of weaving all interests that can sensibly be registered by claims into the mechanics of claims is that the interests that are picked up by different kinds of claims count differently from how they would if simply counted as items with consequentialist value. To be fair, many people who use the infringement model use a sophisticated conception of consequentialist value. Seth Lazar, for example, in discussing “the necessity constraint” on self-​defense, says that “it should not weigh [all relevant harms] equally”; an “Attacker’s interests [should] be discounted relative to [a]‌Defender’s [and] should also be discounted relative to [a] Bystander’s interests.”82 In other words, “The necessity constraint should enjoin avoiding all unnecessary morally weighted harm, whoever the victim.”83 I  recognize that this notion of morally weighted harm is an improvement over a more simpleminded consequentialism that looks at harms without any context. But even this sophisticated consequentialism will not catch all of the morally relevant nuances picked up by the mechanics of claims. It won’t, for example, pick up the relational nature of rights claims or the particular way that different kinds of claims register in their particular role in the mechanics of claims.

 Lazar, 2012, p. 6. I do not mean to suggest that Lazar is a consequentialist. But he, like every person who uses the infringement model, is partially wedded to consequentialism. 83  Ibid., pp. 6–​7, emphasis added. 82

A Theory Of rights  |  81

Before proceeding, I want to state that the two points of contrast are deeply related. The notion that infringing a right gives rise to a claim for compensation puts pressure on a theory of rights to carve out interests that are not protected by rights:  those interests the nonsatisfaction of which gives rise to no claim for compensation. To appreciate how this comes up, consider again a variation on Trolley Switch. Imagine a version in which the trolley is headed toward one person and that to save that person the agent at the switch would have to turn it onto another track where there are five people. She may not kill five to save one. But it would be quite counterintuitive to suggest, in such a case, that the agent or the five owe the one (or his heirs) compensation for not saving him. To avoid implying that implausible implication, the infringement model holds that he has no right to be saved at all.84 All he has is an interest that would be picked up by consequentialist considerations. The implication for the original Trolley Switch, where the trolley is headed to five and may be turned onto one, is that the five too lack a right to be saved; their interest, too, registers only as a good with consequentialist value. Again, I will argue in the next chapter that this is not the right way to handle claims for compensation. My point here is simply that this is not the only way to handle the thought that those who are rightfully not rescued are owed no compensation. One could say, as I  do say, that something else determines when compensation is owed and that all interests that can sensibly be picked up by claims are picked up and represented by claims.

4. The Mechanics of Claims and the Restricting Claims Principle A theory should be judged not only by its internal coherence but by its fruit, that is, the benefits it provides in terms of being able to make sense of important moral principles. One principle the mechanics of claims enables us to make sense of is the means principle: the principle that it is particularly hard to justify using another as a means if doing so imposes some cost on him, unless he has waived or forfeited his right not to be so used. Other benefits will become clear in later chapters. The means principle sits at the heart of deontological ethics. But it has been notoriously hard to defend. Some have tried to cash it out in terms of how problematic it is for an agent to intend to use another as a means,  See Thomson, 1990, pp. 160–​63.

84

82  |  The Mechanics Of claims And Permissible Killing in war

but this explanation runs into numerous problems, the most basic of which is that it seems implausible to put so much weight in deontology on the intentions of agents, rather than more directly on the claims of patients.85 Others have tried to cash it out in terms of the causal role of being a means to an end.86 But they have been unable to explain why T.  M. Scanlon was wrong to write that “being a means in this sense—​being causally necessary—​has no intrinsic moral significance.”87 The mechanics of claims enables us to frame another principle, the restricting claims principle,88 that can answer Scanlon’s challenge. And the value of reviewing here how it does so is that it provides a vehicle for substantially developing the mechanics of claims, which is valuable both in itself and for the way it lays the groundwork for the discussions later in the book of the right to eliminatively kill threats and the right not to sacrifice oneself to stop threats.

4.1. Introduction to the Restricting Claims Principle The principle is easy to state, though unpacking how it works and why it provides a ground for the means principle will take the rest of the chapter. Restricting Claims Principle: restricting claims are substantially weaker, all else equal, than non-​restricting claims in the balance of patient-​claims.

This principle accounts for the means principle by first flipping it on its head, asking why patient-​claims not to be harmed as a side effect of pursuing some goal or simply being allowed to suffer are substantially  This point is the central theme of Chapter 5. I also make this point in Walen, 2014, pp. 428–​29. The point is made compellingly and with great detail, distinguishing a variety of intention-​based views, in Ramakrishnan, 2016, §§ 1.A, C, D, & F. 86  Frances Kamm is the most influential philosopher to take this line. See Kamm, 2007 and 2015. I discuss her view and explain why I think it fails, in Chapter 6, § 2. 87  Scanlon, 2008, p. 118. 88  I first developed what I now call the restricting claims principle in Walen, 1995; I returned to it and gave it its current name in Walen, 2014. Both versions were essentially identical to the “moral obstacles” principle in Øverland, 2014. But I soon realized that Øverland and I were wrong to rely, as we did, on counterfactuals; and I came up with a new, property-​based foundation for the theory in Walen, 2016. That version closely tracks the “means principle” offered in Alexander, 2016. I say, on the basis of an extended email exchange, that it also functions exactly like Ramakrishnan’s “utility” principle, with one basic exception: his principle does not apply to second property. We do have a deep disagreement, however: Ramakrishnan thinks that the relevant principle “probably needs to be accepted as a primitive moral proposition.” See Ramakrishnan, 2016, p. 136. The account that I offer here is new, a fourth generation of the account, one that develops more clearly the role of empowering claims and helps further explain why the principle is not a “primitive moral proposition” but one that follows naturally from the mechanics of claims and the three fundamental principles of rights. 85

A Theory Of rights  |  83

weaker than claims not to be used as a means and thereby harmed.89 It also explains why claims not to be used as a means should count as strong relative to competing claims. But its foundational insight is that claims not to be harmed as a side effect should be seen as having a strength in the same ballpark as claims to be saved, all else equal. And it explains why they are both relatively weak by explaining why both are restricting claims. In that way, they contrast with claims not to be harmed or allowed to suffer harm as a means to some end. Those are non-​restricting claims, and they are much stronger, all else equal. To be clear, by saying that claims have a strength “in the same ballpark” I do not mean “equal to.” Negative restricting claims are still stronger than positive restricting claims, all else equal. The point is that the difference in their strengths is small enough that a few positive restricting claims (most people’s intuitions are that the number is somewhere between two and five) can outweigh a single negative restricting claim, all else equal. Now the first obvious question is, Why call certain claims “restricting”? The answer is that some patient-​ claims have a normative force that “pushes” to restrict an agent relative to her baseline freedom. By pushing to restrict an agent relative to her baseline freedom, these claims push to make other patients worse off than if the agent retained her baseline freedom. A  restricting claim need not successfully restrict. The point is that restricting claims have a normative force that pushes to make other patients, with competing restricting claims, worse off. The most intuitively accessible way to make sense of the idea of restricting claims is to set a patient’s baseline freedom with reference to what she could do if the patients with the restricting claims were not present. Consider Trolley Switch again. If the sidetrack man were not present, Brenda could justifiably turn the trolley away from the five onto the empty track. Indeed, unless doing so would impose some large cost on her, she would have a duty to do so. His presence with a claim not to be killed pushes to make the five substantially worse off. If his claim had to be respected as a right, they would be transformed from people who have a right to be saved into people who may not be saved—​a clear hit to their normative standing. Conversely, their claims push to impose a similar normative cost on him. If they were not present, Brenda would obviously have

 From this point forward, the patient-​claims I discuss are almost exclusively basic patient-​claims, so I leave the word basic implicit unless the context calls for distinguishing basic from special patient-​ claims. I  also often drop the word patient when the context makes it plain that I  am discussing patient-​claims. 89

84  |  The Mechanics Of claims And Permissible Killing in war

no right to turn the trolley onto him. If their claims had to be respected as rights, they would transform him from a person who had a right not to be killed to a person that Brenda had a duty to kill (not as her aim but as the expected side effect of saving the five). Both sets of claims push to restrict Brenda relative to her baseline freedom, either to save the five or not to kill the sidetrack man. The contrast case involves using another as a means: Massive Man: Bertha, a bystander at a switch, can throw the switch and thereby cause a massive man to topple into the path of a trolley that is hurtling down a hill out of control, thereby killing him and saving five people whom the trolley would otherwise kill.

Using a counterfactual test, it is clear why the massive man’s claim is not restricting. If he were not present, Bertha could do nothing to save the five. If his claim not to be used as a means of saving them has to be respected as a right, it would not make the five any worse off than if he were not present. Thus, his claim does not push to make them worse off than they would be if the agent had her baseline freedom. And that explains why his claim is much stronger than the claim of the sidetrack man.90 The explanation works this way: just as agents should not be free to act in ways that impose excessive costs (negative externalities) on others, so patients should not have claims that impose excessive costs on others. It is a matter of basic fairness that the strength of restricting claims should be limited so that they do not impose excessive costs on others with competing claims. Even if negative restricting claims should still be stronger than positive ones, all else equal, we must recognize that negative restricting claims do not simply push to restrict an agent from harming the claimant. They also push to make others with competing positive claims worse off. This competition, via the agent, puts them in the same ballpark, in terms of strength. The notion of establishing that claims are rivals in the same ballpark might suggest that the restricting claims principle increases the strength of positive restricting claims just as much as it reduces the strength of negative restricting claims. But there is nothing about the way restricting claims compete that suggests strengthening. Rather, the effect is to weaken the strength of negative restricting claims, relative to what one might expect if one focused merely on the fact that these are claims not to be harmed.  The explanation is defective insofar as the account of the agent’s baseline freedom is misguided; but it points in the right general direction, as will be made clear shortly. 90

A Theory Of rights  |  85

These claims not to be harmed are weaker than they might otherwise appear to be because they are not to be weighed in isolation against the agent; their strength has to reflect as well the “externality” they impose on other patients with conflicting restricting claims. In contrast with the patient-​claims in Trolley Switch, the massive man’s claim imposes no “externality” on others. It does not push to make the five worse off relative to the agent’s baseline freedom. As a result, the massive man’s claim should be at full strength. That explains why his claim should be stronger than the sidetrack man’s claim. This provides an account of the key thought in the restricting claims principle, namely that restricting claims should be weaker than non-​restricting claims, because the former and only the former push to impose something like a negative externality on others. But the account of what makes a claim restricting or not fails. As I pointed out in Chapter 2, when discussing the counterfactual analysis of what it means to perform an eliminative killing, counterfactual accounts are susceptible to counterexamples.91 The same is true in this context. We can imagine a variation on Trolley Switch in which, but for the presence of the sidetrack man, the trolley would, if turned, hit 10. Yet the sidetrack man would still be killed, if Brenda turns the trolley, as a side effect of turning it away from the five. It is impermissible to turn a trolley from five onto 10, yet his claim still should be understood as a claim not to be killed as a side effect.92 Conversely, we can imagine cases in which, but for the presence of the massive man, Bertha would be able and permitted to topple something else in the path of the trolley, yet the massive man’s claim still should be understood as a claim not to be used as a means.93 These counterexamples simply demonstrate the familiar truth—​familiar from cases of causal overdetermination—​that causal roles cannot be well understood with a counterfactual account. And what the restricting claims principle tracks is really causal role. More important than these counterexamples, the counterfactual account of an agent’s baseline freedom skirts the deeper questions about how we should understand an agent’s baseline freedom. This is fundamentally a moral notion. To truly make sense of the restricting claims principle, we need a moral account of why an agent should not be taken to have a baseline freedom to use another for good ends, a baseline freedom that the other’s claim not to be used pushes to restrict. That is what I offer in the next section.

 See my discussion of Bayonet and Sniper in Chapter 2, § 4.  I discuss such a case, Sidetrack Man Protecting Others, in Walen, 2016, pp. 222–​23. 93  I discuss such a case, Massive Man Displacing a Moose in ibid., pp. 224–​25. 91 92

86  |  The Mechanics Of claims And Permissible Killing in war

4.2. The Property Baseline for the Restricting Claims Principle Restricting claims should be understood as claims not to be harmed, or to be helped, when the claimant’s relevant interest is not based on ownership of the means the agent would use to achieve whatever good is proposed to justify her act. The question is, Why use the absence of competing prop­ erty claims regarding the means the agent would use to achieve her goal as a baseline for agent freedom? Indeed, given that the essence of patient-​ claims is to have some normative force on agents, one might wonder how a claim could be anything other than restricting. The answer to both questions is that some claims function to establish an agent’s baseline freedom, while other claims come into play to restrict an agent relative to her baseline freedom. As I now understand the restricting claims principle, the distinction between restricting and non-​restricting claims reflects the fact that there are two ways we see ourselves in relationship to others. First, there is the agent–​agent way of dividing up the world into that which is available to each agent for her own use, that which is reserved for others, and that which is available for all (or none). This is, obviously, a division along property lines. Then there is the way we, as patients, confront agents who seek to use their property to promote their ends. We have an interest in limiting their freedom so that it does not inflict undue harm and so that they act in ways that benefit us. I call these, respectively, the agent–​agent division and the agent–​patient frontier. The agent–​agent division is important because the essence of agency is the ability to choose how to act. For an agent to be free to act, she needs to have the means by which to act. The means she is presumptively free to use include her body, her time, her energy, and any extensions of her body that she needs to act effectively (i.e., her second property). These comprise her toolkit, so to speak. What is reserved for others in the agent–​agent division is not in her toolkit. A fellow agent with a claim to deny her the use of some part of the world contributes to defining the limits of her toolkit and, with that, the limits of her baseline freedom. Such a claim is not “restricting” because its moral effect is not to restrict an agent relative to her baseline freedom; it occurs at a morally prior stage, that of setting her baseline freedom. In past writing, I have called such claims “non-​restricting”94; I now think it is more illuminating to call them “property claims.” And it is obvious enough why property claims, at least concerning first property, concern not being used as a means. It should also be clear why second property has the same  See Walen, 2014 and 2016.

94

A Theory Of rights  |  87

functional role of establishing the limits on an agent’s freedom and should thus have a similar normative significance. In contrast to property claims, restricting claims operate in the space of patient–​patient conflict and reach across the agent–​patient frontier to push to restrict an agent relative to her baseline freedom. These contrasts enable us to answer a basic question for the restricting claims principle: why should property claims be stronger than restricting claims that protect the same sort of interest? The answer has two aspects, both of which have already been suggested. The first aspect relates to the difference made by patient–​patient competition, as opposed to setting the baseline freedom of agents; the second refers to the different principles the two kinds of claims express. Turning to the first aspect, compare the claim of the sidetrack man with the claim of the massive man. The sidetrack man’s claim not to be killed and the claims of the five to be rescued compete to restrict Brenda with regard to her baseline freedom. The massive man’s claim does not compete with the claims of the five in the same way. His property claim over his body does not compete to restrict Bertha relative to her baseline freedom; rather, it serves to limit her baseline freedom before the claims of the five can even come into play. Thus, his claim not to be killed does not impose anything like a negative externality on the five. Because property claims do not impose anything like negative externalities, they are freed from the sort of leveling that limits the strength of negative restricting claims. Before continuing, it is important to emphasize that property claims are not limited to claims over the body. Consider this case: Drug Owner: To save the lives of five people, Dr. Aspire needs to give them a certain drug. But all the readily available supply of that drug is owned by one person, Bryson. Moreover, Bryson is not simply hoarding the drugs; he bought what he owns because he needs all of it to live.95

It seems to me clearly impermissible for Dr. Aspire to take Bryson’s drugs, and it seems to me that the impermissibility of doing so has the same basis as the impermissibility of toppling the massive man in front of the trolley: both have property claims that protect their interest in their lives. The second reason the massive man’s and drug owner’s claims are stronger than restricting claims protecting the same interest is that property claims are essential to respecting the autonomy principle from the patient side of the balance, while restricting claims are essential to respecting the  This case comes from Walen, 2014, p. 453.

95

88  |  The Mechanics Of claims And Permissible Killing in war

welfare principle from the patient side of the balance. Fleshing out this aspect of the answer requires establishing two points: first, that the different kinds of patient-​claims respond to different principles and, second, that the autonomy principle has some priority over the welfare principle. Property claims are essential to respecting the autonomy principle because they provide agents the means with which to lead autonomous lives. As expressed in their patient form, they protect against other agents using the owner’s property in any way that interferes with the owner’s interest in having an exclusive right to use it. The claim to deny other agents use rights serves the same fundamental autonomy interest as negative agent-​ claims. Just as negative agent-​claims protect agents from having to serve as a tool for the balance of restricting patient-​claims (if that balance calls on the agent to act), so property claims function to protect a patient’s tools (including her body) so that they serve the interests of the patients who own them. Property claims function, therefore, as the patient analogue of negative agent-​claims. This is what the agent-​agent divide represents. By contrast, restricting claims are fundamentally concerned with protecting the welfare of patients when agents seek to exercise their baseline freedom to act. Of course, patients who die lose their autonomy. But in this context, once the agent-​agent divide has set an agent’s baseline freedom, autonomy is relevant only secondarily, as a contributor to welfare and as a basis for holding patients responsible for their own choices (waiver and forfeiture) and their own luck (positive versus negative claims). What matters fundamentally is welfare. The second part of the argument is that the autonomy principle has to have a limited priority over the welfare principle. It needs such a priority to function at all. The space of rights must ensure that welfare does not matter to such an extent that it turns agents into tools for the promotion of the greater welfare. Different people have different sensibilities about how much priority to give the autonomy principle. Libertarians will give it infinite priority; utilitarians incorporate it into welfare and thus give it no priority. I think the truth is in the middle: a reasonable balance gives substantial priority to the autonomy principle, all else equal, but holds that negative agent-​claims and property claims are outweighed when the interests they protect are relatively minor and the competing welfare claims are much greater. The key is this: the autonomy principle must have enough priority over the welfare principle for it to be true that agents have a meaningful normative space in which to lead their own lives for their own ends. In sum, the restricting claims principle is a natural outgrowth of the autonomy principle. The autonomy principle provides that the space of rights

A Theory Of rights  |  89

must give people the normative space in which to lead their own lives. To do that, it must let them own property and have priority over its use. That is the basis for the agent–​agent division, which sets the baseline freedom for each agent. When claims are restricting, they register after that divide has already been taken into account. It is only restricting claims that compete as ballpark rivals—​holding interests to be on a par and assuming that no one has waived or forfeited a claim—​for it is only restricting claims that push to restrict agents relative to their baseline freedom.

4.3. Objections and Replies I now address two cases that might be offered as raising problems for the restricting claims principle. Both call for refining what it means to have a restricting or a property claim. The first case challenges the notion that non-​property claims should be seen as a relatively weak restricting claims. Rescue Ahead: Renee is on a train rushing down the line to save five people who are in imminent danger of death. In the distance ahead, she sees someone trapped on the track (the “trapped-​ahead man”). She sees no reason to think that he has recklessly caused himself to be trapped on the tracks (and we should assume that he was in no way at fault for being trapped there). Unless she stops the train, the trapped-​ahead man will be killed. If she stops the train, however, the rescue mission will be aborted and the five will die.96

The common intuition in this case is that Renee cannot justify allowing the trapped-​ahead man to be run over for the sake of saving the five. But it seems that the trapped-​ahead man has only a restricting claim. He certainly does not own the track on which he is trapped; he seems, rather, to have a claim like that of the sidetrack man, a claim not to be killed as a side effect of the agent pursuing her goal of saving the five. In truth, however, I think that what this case shows is that property claims have to be interpreted broadly to encompass property-​like claims as well as claims that are, strictly speaking, property claims. The case for saying that the trapped-​ahead man has a property-​like claim is that his temporal priority on the track gives him a prior claim to deny

 My argument with regard to this case is now settling into a position in between the one I originally took in Walen, 2014, pp. 456–​57, when I held that the trapped-​ahead man’s claim was non-​ restricting, and the one I took in Walen, 2016, pp. 244–​47, when I held that his claim was restricting. 96

90  |  The Mechanics Of claims And Permissible Killing in war

others the right to use it unless and until he can be freed from it. It is not a true property claim as he cannot do what one can normally do with prop­ erty: he has no claim to use the stretch of track on which he is stuck for his own purposes, nor may he sell it, nor may he refuse to vacate it if freed from it. But he does have a claim vis-​à-​vis Renee, namely that she may not count it as in her toolkit for getting where she wants to go. His claim on it is temporally prior to hers and thus is morally prior to hers. This reflects a deep presumption in our thoughts about property, that when it comes to unowned objects, first use provides a presumptive property-​like claim. Of course, the tracks are owned by someone or some organization. But as between the trapped-​ahead man and Renee, it functions like an unowned resource on which the trapped-​ahead man has a temporally prior claim.97 There are limits to this sort of property-​like claim. If the trapped-​ahead man were in some way to blame for his being trapped, he might forfeit his property-​like claim. And even if not to blame, he would need to have sufficiently clear temporal priority for his property-​like claim to vest. To see that, consider this case: Collision Course: The conditions are the same as in Rescue Ahead, except the trapped ahead man is replaced by someone (the “collision-​course man”) who is strapped onto a flimsy cart that is rolling down a sidetrack so that it will arrive at the spot where the sidetrack joins with the main track a second before the train with Renee would come to the same spot.98

Can it be that Renee should try to speed up just to make sure that she gets to the intersection first? Assume that if she did so, the collision-​course man would be killed in more or less the same way; the difference would be only that he would run into the train rather than having the train run into him. If it were clear that he would hit the train if Renee did not apply the brakes, the case would be intuitively quite different from Rescue Ahead. There would be much less normative pressure for her to stop the train. But it also seems implausible to say that she should speed up to ensure that he hits her, rather than that she hits him. Imagine that speeding up would introduce a 10 percent chance of derailing, injuring Renee and others on the train as well as preventing the rescue of the five. It seems a bad trade-​off  One might wonder why Renee does not have a prior claim to use of the track, assuming that she has a ticket to ride the train that is on it and that the trapped-​ahead man paid no such fee. The answer is that his blameless claim to need to withdraw it from use is a much stronger basis for claiming a property-​like right than a ticket. I return to how claims work when someone owns property in § 4.4. 98  I first discussed this case in Walen, 2016, pp. 245–​46. 97

A Theory Of rights  |  91

to introduce any such risk just to ensure that the train beats the cart to the merger spot. I  acknowledge that right-​of-​way notions have some moral weight, but they also seem beside the point in this case. The reason this is so, I suggest, is that if the collision-​course man had just a bare moment of temporal priority, that is insufficient to cause us to judge that his claim to priority over use of the track has vested. This is just one more example of the sort difficult decision points that inevitably must be faced to operate in the space of rights. (The first one concerned when luck vests.)99 But the fact that these decisions have to be made does not, I think, undermine the importance or applicability of the basic concepts. I turn now to the second problem case for the restricting claims principle. This case raises a similarly obscure issue concerning the directness of the link between a property claim and the interest it protects. I do not know how to spell out exactly what constitutes a sufficiently direct connection, but I can illustrate the issue with two cases in which it fails. Consider first: Trolley Switch Tool: In order to throw the switch and turn the trolley away from the five, Brenda needs to use a tool. She sees one lying nearby that she knows belongs to the sidetrack man. If she doesn’t use his tool, she can’t save the five, but if she uses it, she will kill him.100

The case seems, intuitively, more like Trolley Switch than Massive Man, but here Brenda uses the property of the owner and thereby brings about his death. I used to think that this case showed something special about second property, but consider this case: Massive Man Tool: In order to throw the switch, Bertha needs to push a massive man who is standing next to her into a switch. This will have three effects. First, it will turn the trolley away from the five. Second, it will cause the massive man a little pain. Third, it will dislodge a rock above the switch, which will then fall on and kill the massive man.

This case, too, seems intuitively more like Trolley Switch than Massive Man. But now Bertha is using the massive man himself, not some second property of his, as a means of saving the five. In both cases, I think that what explains the permissibility of acting is that the harm that is directly  See § 1.3.3.  I first discussed this in Walen, 2014, p. 455.

99

100

92  |  The Mechanics Of claims And Permissible Killing in war

connected to the use of another’s property (his tool or his body) is relatively minor. The harm that follows to that same person is a side effect of the use. The owner’s claim not to suffer the serious harm of death is a separate, restricting claim. Again, I do not know how to spell out exactly what makes the connection sufficiently direct, but it seems to be connected with whether it makes sense to think of the use as separable from the ultimate harm, such that one could imagine the use having a small effect on the owner and the ultimate harm befalling someone else, someone with nothing like an ownership claim in the object used. There will doubtless be hard, borderline cases. But the clear cases seem clear enough.

4.4. Refining the Restricting Claims Principle with Empowering Claims What I said in section 4.2 suffices, I think, to explain why claims not to be used as a means are substantially stronger than claims not to be harmed as a side effect, as well as claims to enjoy a benefit, all else equal. But the account is still in a fundamental way incomplete. The agent–​patient frontier is not limited to restricting claims seeking to push agents relative to their baseline freedom and agents pushing back primarily with negative agent-​ claims. There is another important way in which patient interests cross the frontier: they ground claims that push to empower agents to count as part of their toolkit what would normally be the exclusive property of others. I call these “empowering claims.”101 It is the conflict between property claims and empowering claims that fully sets the baseline freedom of an agent. It will help to explain the role of empowering claims, and to bring the whole picture into focus, if we think about cases in which the agent has to use someone else’s property as a means to save five. Consider, again, Brenda from Trolley Switch. I have been describing this case as though it is obvious that Brenda may throw the switch. But presumably the switch belongs to a trolley company and is the kind of thing that bystanders may not normally treat as theirs to use. Nevertheless, if using it is the only way to save lives, then the interests of those whose lives would be saved empower Brenda with a temporary right to throw the switch on their behalf. More precisely, their empowering claims outweigh the trolley company’s property claim to exclude Brenda from throwing the switch and put the switch, for this one purpose, in her toolkit. The empowering claims  I introduced empowering claims in Walen, 2016, pp. 229–​31. This section substantially develops the idea. 101

A Theory Of rights  |  93

successfully empower Brenda only because the trolley company’s interest in excluding bystanders from throwing its switches is quite weak—​ it serves no interest other than the bare interest in having exclusive use of its property; there is no other harm directly connected to the trolley company’s ownership interests that would be implicated by Brenda’s use of the switch. Compare that to the interest the five have in being saved. It is clear that any reasonable balance between property rights and empowering claims should favor the five. These same empowering claims would fail, however, to empower Bertha to topple the massive man onto the path of the trolley in order to save the five. If anything is going to be out of one agent’s toolkit, it is another agent’s body, unless the other agent offers his consent or has somehow forfeited his claim to control at least some set of uses of his body or his property claim directly protects only an insignificant interest and is outweighed by empowering claims that serve a very strong interest. We can assume that he has not consented to be used and has not forfeited his right not to be used. This leaves only the weighing question, and it clearly favors the massive man. The interest of the five is great, but his interest in denying them the use of his body is also great and directly connected to the denial. As it is a property claim, it is substantially stronger than their empowering claims. To better understand why the massive man’s property claim over his body should outweigh the empowering claims of the five, it helps to notice a deep symmetry in the theory: restricting claims push on agents in their agent mode, by pushing to restrict their freedom; empowering claims push on agents in their patient mode, by pushing to strip away the prop­ erty claim to exclude others from using their property. As already noted, property claims are the patient analogue of negative agent-​claims. They are both essential to giving people the freedom that is grounded in the autonomy principle. Both have limited priority over patient-​claims that reflect the welfare principle. Given these symmetries, empowering claims should register as having a strength on a par with restricting claims, all else equal, just as property claims should register as having a strength on a par with negative agent-​claims, all else equal. One might object that empowering claims should not be as weak as restricting claims, all else equal, because only one of the two factors that limit the strength of restricting claims applies to empowering claims. They both express a concern with welfare, and they both bring that concern to bear against a concern with autonomy. But only restricting claims also reflect the importance of competition between restricting claims. Insofar as different groups, with conflicting interests, have claims to empower agents

94  |  The Mechanics Of claims And Permissible Killing in war

to do different things, that in itself is no conflict. The agents in question might simply be empowered more than they otherwise would be. I think this objection overlooks the way in which competition matters for restricting claims:  it brings negative restricting claims down so that they do not excessively overpower positive restricting claims. There is no reason to assume that the competition also affects the strength of positive restricting claims. Their strength should simply reflect the normative force they exert against negative agent-​claims. Empowering claims operate in an analogous way, to reflect the interests of patients against the patient-​analog of negative agent-​claims, i.e., property claims. As such, it makes sense to treat empowering claims as having a strength akin to positive restricting claims that serve the same interests. We are now in a position to state more clearly the structures that constitute the restricting claims principle. The most important thing to note is that the principle operates in two steps. In step 1, an agent’s baseline freedom is set by the agent–​agent division, which sets her property claims, and by the influence of all relevant empowering claims. If some tool is in another’s toolkit, and the empowering claims in favor of an agent having use of it fail to overcome the other’s competing property claims, that is the end of the story; there is no role for restricting claims; she may not use the property in question. If, however, her baseline freedom includes the use of some tool, whether because it is her property or because empowering claims give her a very limited right to use it, we then come to step 2. In step 2, restricting claims push to restrict an agent’s freedom in different ways. The balance of those restricting claims pushes the agent either to do something or to refrain from doing something. Insofar as it pushes her to do something, she can rely on her strong negative agent-​claim not to have to serve the balance of restricting patient-​claims, and she will be obliged to do so only if having such an obligation does not substantially impinge on her ability to lead her own life. Insofar as the balance of patient-​ claims (including her own) pushes her not to do something, she may not do it except insofar as she is using her property in a way that causes no serious harm to others. It is at the second step that we set her freedom for stage 2, which takes into account special patient-​claims. It is important to be clear that empowering claims can succeed at step 1, but the corresponding restricting claims can fail at step 2. Recall the flipped version of Trolley Switch, in which the trolley is headed toward one person and he could be saved only by Brenda turning it onto another track where it would kill five. The empowering claim of the one would, at step 1, suffice to give Brenda the right, as against the trolley company’s claim that she keep

A Theory Of rights  |  95

her hands off its equipment, to throw the switch. But the restricting claims of the five would outweigh the restricting claim of the one with regard to that freedom. Thus, Brenda would be confronted with a balance of restricting claims that would prohibit her from turning the trolley in the final analysis. Having now fully described how the restricting claims principle works, I  present in flowchart form the full structure for analyzing whether an agent, A, has a right to perform act X, where X can be an act like turning a trolley or an omission like not turning the trolley.

Stage 1: Weigh the basic claims

Step 1: Does A have the means to X in her toolkit? (If she has a negative-agent claim to X, then she does.)

If no, do empowering claims temporarily put the means in her toolkit?

If yes, does the balance of patient-claims weigh against A Xing?

If yes, then does A have a negative or positive agent-claim?

If no, then go to Stage 2

If negative, then A may X unless her agent claim is weak and the balance of patient-claims pushes strongly to X.

If A’s negative agentclaim dominates, then go to Stage 2

If yes, then go to the “yes” box in the previous row

If positive, A may not X unless the harms to others are insubstantial. If they are, go to Stage 2

If the balance of patient-claims dominates, then A must X

Stage 2: Does anyone have a special claim that removes A’s liberty to X

96  |  The Mechanics Of claims And Permissible Killing in war

If no, then A may not X.

4.5. Two Final Clarifications Regarding the Restricting Claims Principle I want, now, to make two final clarificatory points about the restricting claims principle. The first clarifies the importance of focusing on justifications, and it does so in the context of risk. I have earlier emphasized that what matters, first and foremost, in the mechanics of claims is whether a justification for an act works. When considering what is justifiable in the sense relevant for rights, we need to appeal to the non-​esoteric facts.102 This leaves room for the possibility that the best anyone can say is that various acts come with various risks—​any appeal to the facts of what will happen if various things are done would be too esoteric to affect the right to act. To see this, consider the following hybrid of the Massive Man and Trolley Switch cases: Trolley Switch with Risk of Looping: Brenda can turn the trolley from the five onto the sidetrack man, and the odds are that the latter’s death will not be a means to saving the five. But there is a switch that randomly flips over many settings, one of which leads the tracks behind the sidetrack man to loop back to where the five are. If the switch happens to direct the trolley onto the loop, then what stops the trolley from hitting the five will be the sidetrack man’s body.

I have argued before, contrary to the standard intuition, that it is impermissible to turn a trolley onto a sidetrack man in a looping variation of Trolley Switch in which the trolley would be prevented from looping around and hitting the five only by hitting the man on the sidetrack.103 The reason is that Brenda is not simply turning the trolley away from the five; she could not justify doing so, given that it will kill the sidetrack man, were it not for the fact that she could save them by using him to stop it. But then his property claim blocks that justification.104  See Chapter 2, § 2.3.  See Walen, 2016, pp. 242–​44. 104  One might wonder, on this basis, what the restricting claims principle would say about a case in which Brenda could turn the trolley from five onto the sidetrack man but the sidetrack man protects 10 behind him. She could not justifiably turn the trolley from five onto 10, so it might seem that she uses him as a means of justifiably turning the trolley away from the five. But I think the cases are meaningfully different. The justification for turning the trolley away from the five is that it will save them; the sidetrack man is not a means to doing that. His presence does then save the 10, but she does not use him to achieve that end. All she does is rely on the fact that only one person will die. She may not act to ensure that he stays put, but I am assuming that his staying put is a given. See ibid., pp. 225–​26. 102 103

A Theory Of rights  |  97

This hybrid case, however, is different. To see the difference, suppose that the odds are 80 percent that the tracks will not loop and that the trolley would simply turn away from the five and onto the one, killing him as a side effect. How should we handle the remaining 20 percent probability that the sidetrack man will be used as a means of saving the five? If we took it as an evil with a value, then we might think that it outweighs the claims of the five. But that is to mistakenly treat wrongdoing as a special evil. Rather, we should treat that possibility as one in which the claims of the five are blocked. As a result, we should treat the case as one in which there is an 80 percent chance that the five’s claims go through and a 20 percent chance that they fail. This is like a case in which the likelihood of Brenda successfully saving the five is 80 percent and the chance that she will kill the one is 100 percent. If we suppose that an 80 percent probability of saving the five is sufficient to justify imposing on the sidetrack man a 100 percent certainty that he will die, then it is permissible for her to turn the trolley; and if he ends up serving as a means, that’s a merely foreseen accident, not part of the justification of the act. If we now suppose that the odds of saving the five in a non-​looping case are only 50 percent and that one cannot justify Brenda imposing a 100 percent chance of death on someone to achieve a 50 percent chance of saving five, then the case comes out differently. The point is: what matters is not whether he is in fact used as a means but whether that fact appears in the justification for the act. That justificatory path is blocked, but other justificatory paths that remain open can still carry weight, discounted by the odds of their working. Second, the ideas of restricting and empowering claims are crucial for thinking about how to take the general welfare into account without stepping outside of the space of rights. The best model here is the state as the agent charged with pursuing the general good. We should think of the state’s power to tax and take property for the general good as grounded in the empowering claims of the people whose interests it serves. I  do not want to overstate this point. I reject a Lockean notion of natural prop­ erty rights that the state disrupts whenever it taxes. Much of the property regime is set by the state in the first place and should be set to achieve distributive justice, which is not reducible to the sorts of deontological considerations picked up by the restricting claims principle. But the state’s power, with regard to legally vested property, to tax it and to take it with compensation should be understood to rest on the empowering claims of those whose interests are served. On the flip side of the ledger, restricting and property claims both come into play to explain the restrictions on state

98  |  The Mechanics Of claims And Permissible Killing in war

action that the state must observe—​as highlighted in Chapter 2, section 3. This way of thinking about the general welfare reinforces a point I made in the first section of this chapter, to the effect that while the content of the space of rights must reflect certain considerations other than the claims of individuals, we should nonetheless accept that all welfare that can register through competing claims on agents does register that way.

A Theory Of rights  |  99



4 Putting the Mechanics of Claims in Perspective

T

his chapter covers two basic themes. First, I argue that the mechanics of claims is substantively preferable to the infringement model. I  do that in two ways. First, I  consider the main reason to adopt the infringement model:  its account of the duty to compensate those who are “rightfully wronged.” I  argue that this model is inferior to the account of compensation that one can find in the mechanics of claims. Second, I review four other problems that I think are particularly telling against the infringement model. Then I turn to the second theme, acknowledging the limits of the mechanics of claims and the need for some notion of threshold deontology. But I  also explain how threshold deontology can complement the mechanics of claims.

1. A Fresh Approach to Compensation I believe that the main reason the infringement model has captured the imagination of most moral philosophers is that it is taken to provide a good account of the fact that sometimes it seems permissible for A to do X, but it also seems that P’s right that A not do X is still in play in the sense that it leaves a “moral residue,” including but not limited to a right to compensation if A does X.1 In this section, I argue, first, that the infringement model does not have an advantage over the mechanics of claims in terms

 This is the core of Judith Jarvis Thomson’s argument against the thesis that all rights are absolute. See Thomson, 1990, especially ­chapter 3. 1

of accounting for claims for compensation and, second, that it actually fares worse in accounting for claims for compensation.

1.1. The Infringement Model’s Non-​Unique Ability to Account for Compensation To examine the infringement model’s account of the right to compensation, I  focus on Judith Jarvis Thomson’s argument that when rights are infringed, the right-​holder is wronged and therefore has a right to compensation. (I return to the idea that we might want to say that right-​holders may permissibly be wronged in section 2.) Thomson’s argument has two parts: that we need the infringement model to make sense of cases in which compensation is owed and that we need the infringement model to make sense of cases in which compensation is not owed. I take these in turn. Thomson makes the positive part of her argument by discussing a case in which an agent has made two promises and it turns out that she cannot keep both: she promises a banana to C, then promises one to D, and then discovers that she has only one banana.2 Suppose that, all things considered, she ought to give her banana to C. Thomson thinks that she nonetheless should make amends to D. The moral residue of D’s claim on the agent implies that she ought to seek release from the promise, perhaps by negotiating a new way to satisfy D, or, failing that, she ought to compensate D. Thomson then asks, what could ground this residue if D simply did not have a right against the agent? The residue makes sense, she says, because the right is still there, even if it is justifiably infringed. This may work as an argument against specificationism, the view that rights are absolute when they apply but that their range of application has to be tailored to ensure that they do not mistakenly imply that agents have duties that they do not in fact have.3 But it has no force against the mechanics of claims because the latter does not suppose that patients like C and D either simply have a right or lack a right. Rather, they both have claims on the agent. If C’s claim is stronger, then C may have a right and D not. But D still has a claim. And that claim may give rise to secondary rights, such as the right to compensation. Thus, if the mechanics of claims is otherwise plausible, we do not need the infringement model to make sense of the right to compensation.

 Thomson, 1990, introduces this example on pp. 84–​85.  I discussed specificationism in Chapter 3, § 1.1. For an argument that specificationist attempts to get around this objection fail, see Liberto, 2014, pp. 179–​85. 2 3

Putting The Mechanics Of Claims In Perspective  |  101

Thomson’s negative argument, however, offers at least prima facie reason to think that the mechanics of claims cannot be used to account for claims for compensation; if it succeeded, then the pair of arguments would support the infringement model. The negative argument concerns cases in which patients have interests that are morally significant but that, when left unaddressed, ground no claim to compensation. For example, suppose that Brenda decides that she does not want to kill the sidetrack man for the sake of the five in Trolley Switch. Their heirs would have no claim for compensation against either her or the sidetrack man. Thomson accounts for these cases by arguing that people do not, in general, have rights to be saved. This enables her to explain why there is no residue if an agent does not save them.4 One need not accept Thomson’s libertarian view of the limits of positive rights to see her point. One can accept that people have some positive rights to aid, when having such a right would not be too costly to others or the agent who would have to act to provide the aid. For example, one could accept that Brenda would have a duty to save the five if there were no sidetrack man on the other track and if all she had to do was throw a switch. In that case, she would owe compensation to their heirs if she fails to save them. But if one agrees that no compensation is owed in the original case, one might agree with her claim that the infringement model provides a good account of why: they have no right to be saved in that circumstance. The question is whether the mechanics of claims has an equally plausible account of why the heirs of the five would not be owed compensation if Brenda does not save them but the heirs of the one would be owed compensation if Brenda kills him to save them. I argue here that by appealing to the idea of shifting luck, it does. Recall that the autonomy principle holds that we each have to own our own luck.5 The idea of owning one’s own luck implies that compensation is due in cases like Trolley Switch only if the bad luck that properly belongs to one set of patients is transferred to another set. The ones onto whom bad luck was transferred are owed a debt by the ones who were saved. In Trolley Switch the five start off with the bad luck of being threatened by a runaway trolley. If Brenda does not save them, she does not transfer their bad luck and their heirs are owed no compensation. If, however, she does

 Thomson, 1990, pp. 160–​63. Exceptions would include cases in which they have, effectively, a special claim on an agent to save them because, say, a promise was made or they are owed that sort of attention as compensation for earlier harms. 5  See Chapter 3, § 1.3.3. 4

102  |  The Mechanics Of claims And Permissible Killing in war

save them, then she transfers their bad luck onto the sidetrack man and his heirs now have a claim against them—​and perhaps against her6—​for compensation.7 One could argue that the sidetrack man has the bad luck of being in the way of the rescue of the five. But that kind of luck is symmetrical between the five and the sidetrack man. They have the bad luck of being threatened when someone else has a claim that they not be rescued. This kind of symmetrical bad luck is already picked up in the fact that they have competing restricting claims. But there is another kind of bad luck that only the five have: if nothing is done to save them, they will suffer harm. The transferring of luck is, of course, not the only way to ground claims for compensation. If an agent wrongs a patient, then the agent presumably owes the patient compensation. Moreover, a promisor may take on not just the duty to perform but the duty to seek release if performance is not possible and, failing that, the duty to compensate the promisee if she does not perform the promise made—​though that is a matter of special claims, not basic claims. At any rate, the point is that we can say all that we need to say about compensation and moral residues within the framework of the mechanics of claims. Thus, the infringement model cannot be supported as the only way to make sense of when people have and when they lack claims for compensation.

1.2. Why the Mechanics of Claims Is Better on Compensation It is not simply that the mechanics of claims provides an adequate alternative to the infringement model; the infringement model actually provides a poor account of claims for compensation. I show this by examining a clever criticism Patrick Tomlin raised of Victor Tadros’s duty theory of  The argument that she might owe the sidetrack man’s heirs compensation is that she was not duty-​ bound to shift the trolley onto him and, thus, has to take responsibility for the choice. The argument against is that she is responding to the balance of patient-​claims on her, and she should not face an extra disincentive from doing so. I remain agnostic about which is the stronger argument. 7  How much compensation they have a right to is another matter. Tort law, outside the domain of contributory negligence, tends to assume that compensation should always aim to make the victim whole. I think that is a mistake. Matters of compensation should not be framed as one of determining who should carry the burden of some harm, as if it was not someone’s burden already. They should reflect the degree of responsibility one party has for the welfare of another. The strength of that duty, which corresponds to the strength of the claim for compensation, should depend on a range of factors: Is anyone culpable and, if so, how much? Was one party using another as a means of helping herself avoid harm, or did she cause harm simply as a side effect? And how much responsibility should one party assume for the welfare of others she happens to harm, or who happen to be harmed for her sake, if she did nothing to bring about that harm? These matters are beyond the scope of this book. 6

Putting The Mechanics Of Claims In Perspective  |  103

punishment.8 My aim is not to examine Tadros’s theory; it is only to show how the infringement model runs into trouble with the claim for compensation.9 According to Tadros, the best justification for punishment is that those who commit crimes acquire a duty first to their victims and then to other potential victims of crime in the future. This gives them a duty to submit to punishment that would deter potential criminals.10 Tomlin’s criticism is based on noting that having committed a crime is not the only way to acquire a duty to submit to treatment that is, at least for the recipient of the treatment, just like punishment. He introduces the following variations on a case to bring this point to life. Minor Punishment: Sarah has committed a crime and is punished by being sentenced to jail for six months. Voluntary “Punishment”: Beth has committed no crime, but she volunteers to be “punished”—​locked up in jail for six months—​for the sake of some greater good. Given the magnitude of the good she would cause and the amount she would suffer, she had no duty to submit to this “punishment” before she promised to do so. But having volunteered to endure the “punishment,” she acquired a duty to submit to it. Involuntary “Punishment”: Ruth has likewise committed no crime, but the benefits of “punishing” her for six months would be tremendous, and her suffering would not be too great. She therefore has a duty to submit to it even though she has not volunteered to submit to it.11

Sarah, Beth, and Ruth all have a duty to endure the same harsh treatment; but it seems that the way we evaluate their treatment is quite different: Sarah’s  Tomlin, 2017. Full disclosure: I was an external reviewer on this paper and pushed Tomlin, when he was still an unknown author to me, to consider whether the mechanics of claims provides a superior way of handling the issue he addressed with the infringement model. He included reference to my way of handling claims but sensibly chose to try to focus on the more familiar infringement model. 9  The argument that follows is a stronger one than that offered by John Oberdiek, 2004. According to Oberdiek, the infringement model fails to properly account for compensation because a rights violator would normally owe greater compensation than a rights infringer, all else equal; but the model treats both the same (pp. 334–​35). An infringement theorist could respond that baseline compensation is owed for an infringement, while something like punitive compensation is also owed for a violation. 10  See generally Tadros, 2011. 11   These are my case names and descriptions. Tomlin introduces Sarah’s case in Tomlin, 2017, p. 240, Beth’s case on p. 244, and Ruth’s case on p. 232. Regarding Ruth’s case, if one thinks that six months in jail is too great a burden to bear for a great social good, reduce the burden until it seems plausible that an agent who has done nothing to acquire a special responsibility for that good could nonetheless be required to bear that burden. 8

104  |  The Mechanics Of claims And Permissible Killing in war

suffering is less of a bad thing than the suffering of Ruth or Beth. Tomlin tries to make sense of this within the infringement model as follows: Sarah has forfeited her right not to be punished, Beth has waived her right not to be “punished,” and Ruth has her right not to be “punished” infringed for the sake of the greater good. But his effort, as Tomlin acknowledges, is a strain. Here is why the infringement model looks odd in this context. First, it seems that Sarah and Beth are in the same position, that of having lost their right not to be punished. But while it is intuitive to say that Sarah has no right to compensation, Beth, intuitively, has at least some claim to compensation. If her right to compensation depends on her having a right that has been infringed, however, then it is not clear why Beth should have such a right or claim. Second, while the infringement model seems to explain why Ruth would have a claim for compensation, it is in the peculiar position of saying that she retains a right not to be “punished” even though she has a duty to submit to being “punished” in just that way. Tomlin tries to resolve these difficulties, and arguably he resolves the first problem without doing violence to the infringement model; but the second problem shows, I  think, that the infringement model does not actually provide a good account of claims for compensation. Starting with Beth’s claim for compensation in Voluntary “Punishment,” Tomlin suggests that we might account for it by saying that “waiving a right [at least sometimes] involves allowing others to infringe one’s right, rather than removing it altogether.”12 This suggestion, however, fails to cover certain plausible ways of making sense of Beth’s case. It seems to presuppose that Beth is volunteering to submit herself to others who will wrong her. But we need not imagine her case that way. She and her “punishers” might agree, essentially, to put on a show of punishment in order to avert some great harm to others. This show could involve real suffering on Beth’s part, but the suffering would be something she endures in full cooperation with those who inflict it on her. In that case, it seems simply wrong to say that they “infringe” her right in any meaningful way. They infringe her right not to be given harsh treatment no more than a dentist who fills a cavity infringes her patient’s right not to suffer. The difference between Beth’s case and that of a dental patient is that the dental patient suffers for his own sake, whereas Beth suffers for the sake of others. I  don’t mean to imply that Beth’s motivation is selfless;

 Tomlin, 2017, p. 250.

12

Putting The Mechanics Of Claims In Perspective  |  105

she might have volunteered with the expectation or even the promise that she would be compensated. But the point is that whatever her motivation, whatever bargaining went on, she is suffering for the sake of others. That, however, is exactly why she has a claim for compensation: she takes on someone else’s suffering. Even if she volunteered selflessly, with no expectation of compensation, such a significant sacrifice for others gives her some claim for compensation. She might judge that she has no standing to press that claim if there was no negotiation in advance. But she has the claim nonetheless,13 and that explains why anyone who knew that he was helped by her act would wrong her if he did not offer her some compensation. His failure would be wrong not because it expressed the vice of ingratitude but because she has a claim to be compensated for suffering on his behalf, a claim that only she can waive. And the key point is that this has nothing to do with infringing and everything to do with the redistribution of harm from others onto her. Those who embrace the infringement model can handle Beth’s case simply by saying that her claim for compensation comes from her choosing to take on someone else’s suffering. They do not have to hold that only rights infringements or violations ground claims for compensation. An infringement theorist can be a pluralist about compensation rather than embrace Tomlin’s suggestion regarding waivers amounting to permission to infringe. The more problematic case for the infringement model is Involuntary “Punishment.” Tomlin acknowledges that it is very “odd” to say that “Ruth has an enforceable duty to bear a given level of harm, and a right not to be harmed to the same level,”14 and he offers two possible ways to recover from this. First, he suggests that the right might be “held against a different group of people [than those] to whom the duty is owed.”15 Then, he suggests that we “more carefully specify the duty to be harmed [as] a duty to allow your right to be infringed.”16 Taken together, the idea seems to be that Ruth might owe the duty to submit to suffering to those who would benefit from it and that she would perform that duty by allowing her right to be infringed by those who inflict the suffering. But this “solution” faces two significant problems. First, if her “punishers” cooperate with her, all aiming to put on a show for the sake  This is the exception to the general connection between standing to demand respect for a right and having a right that I mentioned in Chapter 3, § 1.3.1. 14  Tomlin, 2017, p. 249). 15  Ibid. 16  Ibid., p. 250 (emphasis in original). 13

106  |  The Mechanics Of claims And Permissible Killing in war

of others, then Ruth’s case simply turns into Beth’s, and the notion of infringing her right would seem not to fit.17 Second, if they understand what they are doing as enforcing on Ruth the duty that she has, and it is Ruth who has the moral failing, failing to see that she should volunteer as Beth does, then they become “wrongdoers,” people who infringe her right because of her failing. That surely is a mistake. Surely her failing cannot turn them into wrongdoers. Granted, the element of cooperation is gone; but by hypothesis, her duty is an enforceable one, and that is all they are doing: enforcing her duty. This shows that we again have a case in which compensation is owed not because the person to whom it is owed is wronged but because she has taken on suffering for the sake of others. The problem for the infringement model is that it cannot so easily say that her claim for compensation is independent of the infringement of her right. For starters, at least if she does not volunteer, it seems odd to say that her right not to be “punished” is not infringed. Surely, she has a claim not to be “punished” for the sake of others. Even if it is permissible to force this on her, it is a forcing that runs counter to a pro tanto right, exactly the sort of case that the infringement model aims to explain. Therefore, it is hard to see how the infringement model could avoid explaining her claim for compensation by reference to this infringement. But again, there’s something wrong with the explanation the infringement model offers in this case. By hypothesis, it’s her enforceable duty to take on such “punishment.” Enforcing such a duty seems to involve no wrongdoing at all. But the infringement model accounts for a duty of compensation by saying that it is to make up for having suffered a wrong. Without the wrongdoing, there seems to be no wrong; and without the wrong, there seems to be no reason to offer compensation. A supporter of the infringement model could avoid the problem here by saying that, contrary to first appearances, Ruth’s rights are not infringed if she is forced to suffer “punishment.” Such a supporter of the infringement model could say that when a person is forced to endure what she has an enforceable duty to endure, then she is not wronged and her rights are not infringed. But it is worth pointing out the cost of this concession. It implies that there is a fundamental difference between cases in which a patient is justifiably harmed as a side effect of helping others (in which cases she is wronged) and cases in which she is justifiably harmed as a means of

 They might still be different in this way: Beth’s choice was supererogatory, whereas Ruth’s was, by hypothesis, not. Still, if they are cooperating, then they are presumably infringing her rights no more than if they cooperate with Beth. 17

Putting The Mechanics Of Claims In Perspective  |  107

helping others (in which cases she is not wronged). It is fundamentally unclear, however, why either should count as a wrong. In both cases, the harming is justified. The pressure to adopt the infringement model comes from the thought that in both cases compensation is owed. But if compensation is disconnected from being wronged in the means case, why insist that it is connected to being wronged in the side effect case? I conclude that the infringement model approaches compensation wrongly from the get-​go. The claim for compensation has nothing to do with being “wronged”; it has to do only with having absorbed someone else’s bad luck. The idea of an “infringed right” grounding the claim to compensation is just a misplaced, idle, fifth wheel.

2. Other Problems for the Infringement Model In this section, I review four other reasons why the infringement model should be rejected.

2.1. Intrinsic Oddness of the Infringement Model The idea of rightfully wronging someone, which sits at the core of the infringement model, seems unnecessarily paradoxical and obscure. It is now common parlance, and the paradoxical nature of rightfully wronging someone is no longer keenly felt; but it seems to me that it nonetheless is paradoxical. The fact that an act wrongs someone should be a basis for concluding that the act is wrong, that is, that it must not be performed. But that is not the conclusion offered by the infringement model. Admittedly, I too think that there are cases in which the good lost by respecting rights is so great that violating a right is, all things considered, the right thing to do. But I think such cases are rare and problematic. (I return to them in the last section of this chapter, when I discuss threshold deontology.) The problem with the infringement model is that it finds such cases all over, whenever one person has to be harmed for the sake of a greater number or competing rights claims cannot all be satisfied (as in Thomson’s banana case). I say that that confuses conflicting claims, a normal moral phenomenon, with the breakdown of the coherence of the space of rights, an extreme and rare moral phenomenon. In addition, I find that it obscures what rightfully wronging is supposed to mean. In a case of paradigmatic wronging—​rights violating according to the infringement model—​the person wronged has a complaint that the

108  |  The Mechanics Of claims And Permissible Killing in war

act is not only something he has a right that the agent not perform but also that it is morally unjustified, all things considered. But in a case of rightfully wronging, the person “wronged” should be able to acknowledge that the agent acted rightly. If we needed the notion of wronging to capture the notion that the person “wronged” has a claim for compensation, then I would see the point. But as I argued in the last section, we don’t need the notion of wronging to do that work. And without serving that role, I am not sure what else it is supposed to mean. Given these problems, I am left with the thought that it is more perspicuous and morally accurate to use the language of rightfully contravening claims—​recognizing that contravening claims sometimes gives rise to a secondary claim for compensation—​rather than the language of rightfully wronging. I understand that some find it jarring to say, in cases in which it seems permissible to kill one for the sake of others, that the one is not “wronged.” To make matters vivid, imagine a case in which people are killed as a side effect of a justified military attack. They may have been fully innocent, threats to no one, and in no way responsible for the fact that military equipment was at the relevant time in their vicinity. And yet, as the result of a voluntary act, they are burned, pierced with shrapnel, and, in some cases, torn to pieces. How can I say they are not wronged? While I  acknowledge the emotional appeal of this line of thought, I think it simply confuses the horror of certain instances of harming with wrongdoing. And I say that because infringement theorists cannot avoid coming to a similar conclusion. To see that, consider that according to the infringement model there is a fundamental difference between cases in which a person has forfeited his right not to be harmed (in which case he is not wronged) and cases in which a person retains his right not to be harmed (in which case, if the harming is justified, his right is infringed and he is wronged). Some infringement theorists, like Jeff McMahan, think that all an agent has to do to forfeit her right not to be killed is make a responsible choice which brings about a situation in which someone has to die. Recall Conscientious Driver, the case in which Connie is driving and is doing so carefully and non-​culpably but loses control of her car because of a freak accident and is now threatening to run over Peter, a pedestrian. I said when first discussing this case that it is at least plausible that Peter may kill Connie if doing so is necessary to protect himself from her car.18  See Chapter 1, § 1.

18

Putting The Mechanics Of Claims In Perspective  |  109

If McMahan is right about her forfeiting the right not to be killed, then despite the fact that her death could be just as horrific as the ones that often occur in war, he would not wrong her. If this is morally plausible, then there is nothing implausible about my claim that a justified killing of the sidetrack man in Trolley Switch involves no wronging. One might turn this argument around and say that it shows only that McMahan’s threshold for forfeiture is too low. But then one has to ask, How high should it be? Should we require some minimal level of culpability, such as negligence? What if we suppose that if only she had read her driver’s manual Connie would have realized that she had driven a thousand miles beyond the point where she should have had her brakes checked? But what if we also suppose that the thought that she might be overdue to have her brakes checked never crossed her mind? Can that level of culpability be sufficient to say that killing her is a completely different kind of thing? Can it be that it would be a horrifying but justifiable wrong to her if she had driven a thousand miles less when she had the accident, but now it’s just her bad luck? That seems an awfully steep normative cliff to dive off for such a tiny failure. Perhaps, then, we should require gross negligence or even recklessness, defined as awareness that she was imposing a substantial and unjustifiable risk on others. Criminal law normally requires at least gross negligence, if not recklessness, for criminal culpability.19 And I am inclined to require at least that much before finding rights forfeiture myself. But I say that because I think forfeiture is particularly relevant to eliminative killing only when the thought is that countless individuals can be killed for the sake of just one other.20 For that I want some real culpability. That’s not what is at issue, however, when thinking about forfeiture as a basis for justifying eliminative killing inside the infringement model. In that context, a high bar for forfeiture just raises other problems. For example, it means that Peter, who cannot appeal to a lesser-​evil justification, could not justifiably kill Connie in self-​defense even if she were negligent in her driving. But the idea that he may not defend himself even if she negligently caused the problem is an extreme position that few would take.21 Thus, the infringement model faces a dilemma: either it uses a threshold for forfeiture that is so high that acts of self-​defense that should be accepted are wrongly ruled  See Dressler, 2015, p. 132.  See Chapter 3, § 1.2.3. It is also relevant, I think, for the justifiability of using a threat to protect others. See Chapter 1, § 3.1. 21  It might be tempting to appeal to an agent-​center prerogative here to say that he may kill her. But I argued in Chapter 3, § 2.1.4, that such prerogatives should be rejected. 19 20

110  |  The Mechanics Of claims And Permissible Killing in war

impermissible or it uses a lower threshold for forfeiture, and then it too says that people who have done no grave wrong themselves may be killed without wronging them.

2.2. A Substantive Objection Regarding an Inappropriate Normative Gulf McMahan and others have appealed to criteria such as responsible choice or negligent choice to try to draw a line between those who may not be taken to have forfeited their right not to be killed and those who may.22 Their positions seem plausible enough if considering cases in which one person threatens only one other person. But the difference between forfeiture and retention of a right does not result merely in the drawing of a line; it opens up a substantial and implausibly wide normative gulf. If someone retains the right not to be killed, only a substantial consequentialist improvement or stronger competing rights would justify “wronging” that person. The flip side of that is that if someone has forfeited a right, then he not only has lost that protection but is now at a distinct rights-​based disadvantage compared to another who retains his right not to be killed and to act in self-​defense. Using the example Conscientious Driver to illustrate, if Connie has not forfeited her right not to be killed, then it would take some number of threatened people—​for the sake of concreteness, let us assume the number is three—​to justify eliminatively killing her. But if she has forfeited her right, then the infringement model seems to imply that it would be permissible to kill not only her but a few others as well who have likewise forfeited their rights if doing so were necessary to save Peter. It is in this sense that the infringement model opens up a gulf between two possibilities:  (a) when the threats have forfeited their rights, then it is permissible to eliminatively kill many of them to save one person and (b) when the threats have not forfeited their rights, then many people would have to be threatened before it is permissible to eliminatively kill one threat. The problem here is that the underlying normative difference is so slight that it seems that a gulf like that should not open up. Indeed, I will argue in Chapters 6 and 7 that there should be a relatively smooth function ranging over the following cases. At the high end is killing an innocent bystander as a side effect. Suppose, again, that it would take at least three people on the path straight ahead of the trolley in Trolley Switch to justify turning the  See Chapter 1, § 1.

22

Putting The Mechanics Of Claims In Perspective  |  111

trolley onto one innocent bystander. It is slightly easier, I argue, to justify killing an innocent threat; and as a result, it is at least plausible that one may eliminatively kill one innocent threat to save one innocent victim. It is more clearly permissible to eliminatively kill one person who responsibly chose to act in a way that happened to cause her to become a threat to an innocent victim. It would take even less to justify killing a negligent threat—​flipping the ratio: perhaps two or three negligent threats could justifiably be killed to save one innocent victim. At some point—​arguably when the threats are grossly negligent but maybe only when they are reckless or even grossly reckless—​forfeiture becomes plausible, and then any number of threats could be killed to save an innocent victim (though even then they would retain some residual claim not to be killed such that it would be impermissible to kill them if doing so were unnecessary to save the victim or if doing so would protect an innocent victim from only a minor harm). The relatively smooth change—​from favoring the innocent bystander over the innocent victim already threatened to rough equality between the innocent victim and the innocent threat to favoring the innocent victim more as the threat becomes more responsible for being a threat—​is easily accounted for if one uses the mechanics of claims. The mechanics of claims gives claims different weights in determining what is ultimately justifiable by looking not only to forfeiture or waiver but also to fair allocation of responsibility for a situation arising and causal role. It is much harder to account for such a relatively smooth function using the infringement model, which seems to toggle suddenly from having a right and requiring a substantially greater good to overcome it to having forfeited a right, in which case the presumption is reversed. In other words, the dramatic shift entailed by having or forfeiting a right in the infringement model does not fit the relative smoothness of the underlying moral reality. The mechanics of claims can much more naturally take that relative smoothness on board. There is one way that I  have seen in which the infringement model might be read to bridge that gulf. One might argue that even innocent threats are responsible for the threats they pose to others, simply in virtue of the threat being theirs; and one might use that responsibility to ground a duty to do something for the sake of those threatened. Assuming that duty is enforceable, the sacrifice that a threat would owe could then be extracted from her. The sacrifice that can permissibly be extracted from her—​assuming it is less than her life—​can then be subtracted from the harm caused to her if she is killed to come up with a morally discounted harm that is to be used in a lesser-​evil justification. This could enable us to

112  |  The Mechanics Of claims And Permissible Killing in war

justify killing one innocent threat to save one innocent victim as a matter of a hybrid rights-​and-​lesser-​evil justification. I discuss this possibility in Chapter 6, section 3, and argue that it fails. Given that it does, I conclude that the infringement model cannot bridge the gulf and that its reliance on forfeiture to explain why certain acts of self-​defense are permissible is substantively misleading.

2.3. Occam’s Razor and the Rights–​Consequentialism Hybrid The infringement model is wedded to the view that rights sometimes yield not only to more stringent rights but to considerations of the greater good or the imperative of avoiding a greater evil. This means that the infringement model is committed to using two fundamentally different sorts of normative frameworks and creating from them a single, all-​things-​considered moral conclusion. Pluralism has been quite popular in moral philosophy for some time now. But it remains, as far as I can tell, completely unclear how one is supposed to make sense of the combination of rights and lesser-​evil justifications. I have been stipulating, for the sake of argument, that it would take saving three or more lives to overcome the right not to be killed in cases like Trolley Switch. But I have yet to read any account from someone operating inside the infringement model of why the numbers should be anything like that. And it is hard to see how the account could be forthcoming. It looks like we have nothing other than raw intuition to guide us concerning how important the greater good must be before rights yield to it. By contrast, as I argued in Chapter 3, the mechanics of claims has some basis for making such claims. The importance of owning one’s luck means that negative claims should be stronger than positive claims, all else equal. But the significance of having a restricting claim means that the gap between negative and positive claims should be relatively small. Whether that means that it takes saving two lives or three to justify taking one in Trolley Switch, I cannot say. That kind of precision is not to be expected from moral theories. But we have at least some basis for saying that it should be in the range of a few lives. And that, I will argue in Chapters 6 and 7, provides us an anchor point for thinking about how to deal with threats. No such information is to be found in the infringement model. Admittedly, this point is not decisive. The mechanics of claims too has to confront normative choices that extend beyond its theoretical framework. As I argue in the next section, there comes a point—​and as to where it is I have only intuition to guide me—​at which the space of rights seems

Putting The Mechanics Of Claims In Perspective  |  113

to break down and we enter the land of threshold deontology, in which lives saved outweigh rights. So arguably I have no basis for complaining that the infringement model also has to rely only on intuition in deciding when rights give way. But I think it makes a difference that we get some meaningful guidance regarding less extreme cases from the mechanics of claims. Less of the overall moral picture seems arbitrary if we use the mechanics of claims. And I think it is fair to invoke Occam’s razor to say that if we can make our understanding of the moral landscape less arbitrary and more principled, that is a reason in favor of adopting the view that provides such enhanced coherence.

2.4. Other Forms of Explanatory Power in the Mechanics of Claims The explanatory edge of the mechanics of claims is not limited to having some basis for making comparisons between claims not to be killed and claims to be saved; it shows itself also in making sense of important phenomena like the restricting claims principle, which gives us insight into why the means principle should be accepted. The restricting claims principle presupposes that even patients who have positive claims that would not be counted as rights under the infringement model have claims that compete against negative claims not to be harmed. The mechanics of claims, by making room for concepts like empowering claims, also helps to make sense of how agents get the authority to use property that is not theirs to help others in emergency situations. This sort of structure is just not to be found in the infringement model. My point is not that the infringement model cannot accommodate differences between types of rights. An infringement theorist can distinguish rights not to be used as a means from rights not to be harmed as a side effect and can assign different rights different weights or stringencies. The problem I emphasize here is not the ability to make such distinctions; it is the ability to make sense of them. The restricting claims principle is premised on the idea that all interests that can sensibly be picked up by claims are picked up by claims, that the strength of a claim depends not only on the strength of the interest served but on the type of claim that picks it up, and that a structure for distinguishing types of claims and according them different weights can be grounded in the way that doing so allows the space of rights to respect the three fundamental principles. These ideas are at least partly alien to the infringement model as that model is constrained

114  |  The Mechanics Of claims And Permissible Killing in war

by the need to rein in the idea of rightfully wronging, to deny that the five in the trolley cases even have claims (rights) in play. If it took them to have rights to be saved, it would have to say that those rights are “infringed” in various cases, for example, Massive Man, in which the massive man’s more stringent right prevents Bertha from saving them. To avoid this kind of implausible implication, it registers their interests as something that comes in from outside the space of rights. But then it can do nothing but assume, with no clear basis, that people have particularly strong claims not to be used as a means without their consent. This failure to provide a framework for making sense of such ideas is another strike against the infringement model and in favor of the mechanics of claims.23

3. Threshold Deontology 3.1. A Positive Account of Threshold Deontology I take the space of rights to serve a singular function:  enabling people to navigate a normative space in which the three basic principles are harmonized. Saying that they are harmonized does not mean that there is no space for rightful competition, a point I take up in Chapter 8. It means only that competition itself has to be justified as a proper expression of the rights of all involved. Harmonization means that the status of all as free and equal citizens of the space of rights, citizens whose welfare matters, is respected. But now I  have to admit that harmonization is an aspiration, and it presumes that the world cooperates well enough. I have already mentioned one way in which the world might not cooperate well enough for morality to be fully coherent: one might face a deep conflict between the love and loyalty one feels for and to certain others and the demands of the balance of patient-​claims. For example, a parent might feel obliged to turn a trolley away from her child onto another, even though the other has a right not to have a trolley turned onto him for the sake of one person.24 This can be seen as a conflict arising between the agent’s perspective on the autonomy principle, which protects her right to pursue basic projects, which  I note here that similar points can be made about the principles I develop in Chapters 8 and 9, dealing with intervening agency. That account is thoroughly grounded in the mechanics of claims, and I can see no way of making sense of it outside of that framework. Insofar as it is an attractive account of the moral relevance of intervening agency and cannot be recast in terms of the infringement model, that provides another reason to reject the infringement model. 24  See Chapter 3, § 2.2. 23

Putting The Mechanics Of Claims In Perspective  |  115

may take on a moral life of their own (as it were), and the equality principle, which calls on her to respect the rights of others as she would want her own rights respected. Normally these principles can be harmonized. Normally a moral person would be willing even to die rather than violate the rights of others. But situations could arise in which morally compelling commitments cannot be given up, even for the sake of respecting rights. A similar conflict can arise if the demands of the welfare principle become irreconcilable with the demands of the autonomy principle. The latter implies that the space of rights cannot demand that an agent treat herself primarily or fundamentally as a tool for the welfare of others; she may be required to make small sacrifices for others, but there is a limit to the size of the sacrifice that can be demanded consistently with the premise that her life is fundamentally hers to live. Likewise, as property claims represent the patient analogue of this limit of the autonomy principle, there is a limit to how much can be done to use another as a tool if she is merely trying to live her own life, has done nothing to assume special duties to others, and her patient-​claims impose no “externalities” on the welfare of others. But, at the same time, there seems to be no limit to how much weight the welfare principle can accord the welfare of others as their numbers go up. The claims of two to be saved count for more than the claim of one to be saved, all else equal; the claims of 10 count more than the claims of two; and the claims of 100 count more than the claims of 10. I take no position here on how these claims add up: whether linearly or with a certain kind of discounting. I say only that they do add up, that there is no upper limit, and that there comes a point at which one cannot comfortably refuse to recognize that the moral reasons in favor of sacrificing one person outweigh his fundamental right to live his own life. I consider both of these cases—​in which it seems right to say that it is in some sense permissible, if not obligatory, to do something inconsistent with the respect presupposed by the space of rights—​cases of threshold deontology. That is, they are cases in which some sort of threshold is crossed such that it is no longer morally wrong to violate rights. I used to think that these sorts of cases presented the agent with a true dilemma: a situation in which no matter what she does, she acts in a way that is genuinely, all things considered, wrong.25 I have now come around to the view that in many of these cases it is not especially difficult for an agent to know what she has most reason to do. The defining feature of having crossed the

 Thomas Nagel, 1979b, p. 73, at least entertained this position.

25

116  |  The Mechanics Of claims And Permissible Killing in war

threshold is not that she faces a true dilemma but that the aspiration of the space of rights to harmonize competing claims while respecting the three foundational principles of rights breaks down. The tricky thing to state is why such cases are different from cases of permissible competition, such as competition in sports or business.26 The similarity is this:  in both permissible competition cases and threshold deontology cases, there is at least a sense in which one side is permitted to do things that the other side is permitted to resist. A  football player for one team who tries to run the ball down the field may be tackled by a player from the other team who is trying to stop him; a business owner who sets up shop somewhere may face competition from a competing business that aims to drive him out of business27; a parent may have sufficient reason to turn a trolley from her child onto another, but the other may use force to prevent her from doing so; a bystander may have sufficient reason to topple a massive man in front of a trolley if necessary to save a whole city, but the massive man may—​this is the tough case—​ use force to prevent him from doing so. But, in an important sense, the first two cases are not like the last two. The question is, How so? My answer is that normal competition is justified by the welfare principle but constrained by the overall operation of the space of rights. For example, sports can be good for health, entertainment, the sublimation of tribalism, and other purposes. But all sports have rules that limit what may be done to win, both as essential to the definition of the sport and to prevent excessive injury to the players.28 Competitive business practices are justified because of the way such competition incentivizes business to try harder to provide goods and services that will have general social value. But, again, there are limits on what business may do. They may sometimes aim to drive each other out of business (when doing so would not so undermine competition that it would harm the consumer) but not with violence, theft, or a variety of other means that would be counterproductive if all had an equal right to use them and unfair if only some could use them. In

 I discuss competition in more detail in Chapter 8.   The law on anticompetitive practices limits how a company may act while seeking to drive a competitor out of business. But not all practices that seek to drive away the competition are anticompetitive. As the Federal Trade Commission says, “A company violates the law only if it tries to maintain or acquire a monopoly through unreasonable methods” (see https://​www.ftc.gov/​enforcement/​anticompetitive-​practices). 28  Players may feel that the latter sort of limit is paternalistic, but it can also be justified as a matter of group rationality. Each competitor will feel pressure to do whatever is permitted to win (and sometimes what is not permitted), and it is therefore in the interest of each to have rules that prohibit practices that are excessively damaging to the players. 26 27

Putting The Mechanics Of Claims In Perspective  |  117

sum, the rules of competition, in whatever venue it is permitted, should be justifiable to all as both respectful and productive of the general good that all have reason to value. This justificatory structure is missing in threshold deontology. All sides may acknowledge that what the others do makes sense and is in a sense justified. It would be justified in the sense that an agent can say, “If I were in her shoes, I would do the same thing.” But the justification does not appeal to a general good that can be pursued in a way that respects the rights of all. Rather, the justification appeals only to what each would say is the balance of reasons for her. An agent claiming that justification should be able to see that the competition arises only because what makes most sense for her cannot be harmonized with what makes most sense for another.29 The breakdown of harmonization explains the germ of truth in the idea that threshold deontology cases present dilemmas. When competition is justified, it is right to compete within the rules of fair competition. When competition is just the residue of a failure of rights to harmonize different points of view, there is a kind of tragic dilemma from the point of view of morality. If we step into that point of view, then no matter what the outcome, we must say either that someone’s rights were respected or that some agents were prevented from doing what they had most reason to do, and not because of rightful competition. I can illustrate with the two cases on the table. Start with the parent version of a one-​on-​one variation of Trolley Switch. If the parent does not turn the trolley, she fails to do what she had most moral reason to do, namely save her child. If she does turn the trolley, however, then she fails to respect the right of the sidetrack man not to be killed. Turn now to the variation on Massive Man in which if he is not pushed a whole city is destroyed. If the agent does not topple the massive man, she fails to do what she has most moral reason to do, namely save the city. If she does topple the massive man, however, then the right thing to say is less clear. That is because it is not clear what right he has in this situation, one in which rights seem to break down. But at the very least, we can say that the core commitments of the autonomy principle—​that he has his own life to lead and that his body may not be used as communal property to benefit others who have no special claims on him when the cost to him is his life—​are not respected.

 This point is different from the one I made in Chapter 2, § 2, about evidence-​relative justification. When harmonization breaks down, each can say, “If I were in her shoes, I would do what she is doing, but standing where I stand, I may or must try to stop her.” And each can say that even if all of them know all the relevant facts. 29

118  |  The Mechanics Of claims And Permissible Killing in war

As one of my complaints against the infringement model is that it is unnecessarily paradoxical when it asserts that one might rightfully wrong another, I should say here why my view of threshold deontology is different. There are two differences worth flagging. First, the position taken by the infringement model about infringing rights is that it is fully permissible to do so. The agent who does so may owe compensation, but she has a right to do so, if not also a duty. By contrast, an agent who does what is morally best in a threshold deontology situation does not have a right to do so. Rather, the space of rights itself ceases to function properly, and the answer to the question What may I do? is no longer limited by the rights of others. In fact, the idea of what an agent has a right to do ceases to be well defined as the harmonization that rights presuppose falls apart. Second, this must be an extraordinary event on the mechanics of claims. It represents a failure of the space of rights to function as it should, and if such failures were common, we could not rely on rights as we do. But for the infringement model the prospect of rightfully infringing rights is a commonplace. It may not be desirable to find oneself in a position where one has reason to infringe the rights of another, but there is nothing particularly extraordinary about it. It is the very ordinariness of rightfully wronging others on the infringement model that shows that the model is unnecessarily paradoxical.30

3.2. Overcoming Problems with Threshold Deontology Using This Account I believe that this account of threshold deontology escapes three of the four “theoretical difficulties” for threshold deontology cited by Larry Alexander and Michael Moore.31 First, they believe that threshold deontology might problematically imply that “if we are one-​life-​at-​risk short of the threshold [then it is permissible] to pull one more person into danger who will then be saved, along with the others at risk.” But they say this only because they suppose that reaching the threshold somehow enables one to do what is straightforwardly better. On my view, however, even  It may seem that one virtue of the infringement model is that, like a willow, rights yield rather than break. But even that virtue may be illusory. See David Sussman (2018), who accepts the infringement model but thinks that the conflict in a case like Conscientious Driver is one in which the driver and the pedestrian are “essentially [in] a moral state of nature, where the only way for each to defend their basic rights is to violate the rights of the other” (p. 804). I reject this sort of descent into mutual combat in Chapter 8, § 4. 31  Alexander and Moore, 2016, § 4. All quotes in the text discussing their view come from the same paragraph. 30

Putting The Mechanics Of Claims In Perspective  |  119

if there were a discrete threshold of lives saved at which it became in some sense permissible to topple the massive man in front of a trolley, this is not a straightforward gain—​as if we were chafing at the shackles that rights impose on us and could finally, with one more life saved, break free. Rather, the threshold serves only to mark the point at which concern for the welfare of the many induces a breakdown in the space of rights such that a bystander has sufficient moral reason to use the massive man to save the many. There is no reason why one committed to respecting rights should prefer that to a situation in which it remains clear that Bertha may not kill the massive man to save the many. Second, Alexander and Moore worry that “there is some uncertainty about how one is to reason after the threshold has been reached: are we to calculate at the margin on straight consequentialist grounds, use an agent-​weighted mode of summing, or do something else?” In my view, this again reflects a mistaken idea that one breaks out of deontology into something like consequentialism. The better thought is that one knows one has crossed the threshold for deontology because one has a clear sense that the aspiration for harmonization of the three principles has broken down. If one were the parent who can save her child by turning the threat, one would reason something like this: “There are things I may not do to others to save my child. I may not push another under the bus to save my child. I may not save my child if doing so would entail killing many others as a side effect. But the priority of a single claim not to be killed as a side effect over a single claim to be saved is just not that profound. On the other hand, my duty to my child is very profound. I cannot sit by and let him die when the right on the other side is so thinly justified.” Meanwhile, the bystander on the other side can reason something like this: “I know that if I were in her shoes, I would try to save my child even at the expense of the life of an innocent bystander. Nonetheless, my claim not to be killed is stronger than her child’s claim to be saved, and I, therefore, have a right not to be killed. She, therefore, wrongs me, and I may attack her as necessary to save myself. Moreover, others who are not equally committed to preserving the life of her child may help me but not her.” There is no special problem with seeing how this reasoning goes. It simply goes into conflict. Admittedly, the situation in which the number threatened is so great that it becomes morally appropriate to topple the massive man in front of the trolley is more complicated. For the agent, it is clear: respect for the autonomy of the massive man gives way to the call of saving a great number. But, as I indicated in the prior section, it is less clear what to say for the massive man. Does he have a duty to throw himself in front of the

120  |  The Mechanics Of claims And Permissible Killing in war

trolley? Not necessarily; I think that it would be permissible for another to topple him in front before he would have a duty to topple himself. But even if he had such a duty, it would be inconsistent with his having his own life to lead and thus would be a moral duty without being a duty that finds expression in the space of rights. What about his fighting back? It is odd to say that he has a right to fight back as we’re saying that the space of rights breaks down. Would it be morally wrong for him to fight back? I think morality can offer no clear answer here. His moral interest in preserving his life is legitimate, but he should also respect the decision of the other to seek to use him at lethal cost to him. The idea of a right answer seems too tied up with rights to survive the breakdown of rights. He is morally adrift. But that does not mean that he faces a dilemma in which anything he does is morally wrong. Nor does it indicate that “there is some uncertainty about [whether he] is . . . to calculate at the margin on straight consequentialist grounds, use an agent-​weighted mode of summing, or do something else.” It is not that kind of uncertainty. It is more like the uncertainty an originalist about statutory and constitutional interpretation faces when a situation arises that could not have been anticipated when the text in question was written: the guidance just gives out, and one has to muddle along as best one can. Third, Alexander and Moore claim that “threshold deontology threatens to collapse into a kind of consequentialism.” But this is just not true. On my model, at least, consequentialism only seems to take over because people have focused on cases like that of toppling the massive man to save a city. But that’s not consequentialism taking over in any robust sense. It’s just the welfare principle dominating the autonomy principle such that harmonization no longer works. Nothing about that suggests that we are ultimately always really committed to maximizing welfare or any other consequentialist good. Finally, they worry about “giving a theoretically tenable account of the location of such a threshold. . . . Why is the threshold for torture of the innocent at one thousand lives, say, as opposed to nine hundred or two thousand?” But this worry is not unique to threshold deontology. It occurs wherever we have vague boundaries. A libertarian like Larry Alexander might object to my account on multiple grounds. First, he might object to the idea that normal deontology or rights can impose duties on some to help others.32 Second, he might object

 According to Alexander, 2000, p. 898, a person “generally has no affirmative duty to sacrifice himself to save others (by giving up his life, giving up a kidney, or even by making the slight effort and taking the slight risk of an easy rescue).” 32

Putting The Mechanics Of Claims In Perspective  |  121

that when such duties must be imposed for the sake of the greater good, that is a matter of threshold deontology.33 And third, he might therefore object that threshold deontology is not a matter of a rare breakdown of the harmonization of rights. But this libertarian view of rights is implausible. Even libertarians like Alexander seem willing to recognize that positive claims have some role to play in the space of rights. They recognize, that is, that it is permissible to do things like turn a trolley from five onto one, and thus they are committed to according weight to positive claims. If they are to deny that one ever has a positive duty to help or that property claims can be outweighed by empowering claims, they must see negative agent-​claims and property claims as essentially infinitely strong. But that is implausible in two ways. First, it makes it impossible to make sense of Brenda having a right to turn the trolley in a Trolley Switch case in which the switch is privately owned, except as an instance of threshold deontology with regard to the trolley company’s ownership rights.34 But the idea that this involves threshold deontology with regard to the trivial property interest of the trolley company, but not with regard to the profound welfare interest of the sidetrack man, is bizarre. Second, on a more theoretical level, this idea that negative agent-​claims and property claims are infinitely strong is out of balance. The autonomy principle ensures that one’s life is fundamentally or primarily one’s own to lead. But having to respond to the balance of patient-​claims when it’s not a big sacrifice does not challenge that (unless the sacrifice is likely to repeat in such a way as to consume the agent’s basic freedom). One might object that the true libertarian embraces absolutism about autonomy, holding that the individual’s life is entirely her own. But this thought is misleading, at least for libertarians with fairly standard moral commitments. At the end of the day, a libertarian like Alexander will come to fairly standard positions about when a person has a duty to help others and when a person may be used as a means to help others. The difference is that Alexander frames those duties and permissions as a matter of threshold deontology. But by doing so he makes threshold deontology a rather regular occurrence, and he loses the ability to make sense of the thought that something problematic and rare happens

 Ibid., p. 899.  If the trolley company owns the switch, then it may have a responsibility to allow the switch to be thrown. But if the switch owner is not responsible for the trolley going out of control, then that explanation for why Brenda should have access to the switch is not available. 33 34

122  |  The Mechanics Of claims And Permissible Killing in war

when we cross the threshold. Moreover, his threshold deontology has problems of its own—​the four I just finished discussing. Thus, I think it is misplaced. In sum, I think we must accept that there is such a thing as threshold deontology, but we should not conceive it as involving a point at which rights yield and it becomes straightforwardly permissible to act like a consequentialist. The threshold is the point at which the goal of harmonizing the different claims of the different citizens who populate the space of rights becomes unattainable. In my view, we have every reason to be optimistic that such breakdowns are relatively rare. They do not occur whenever the infringement model would say that it is permissible to infringe rights. They do not occur whenever it is permissible to use someone as a means. Nor do they occur simply because the conditions of war have arisen. Those sorts of events, including the events in war, can usually be handled within the mechanics of claims as perfectly normal cases in which the claims of all involved can be harmonized. For the rest of this book, then, I will assume that we do not confront the problem of threshold deontology.

4. Final Comment on the Infringement Model I cannot show that there is no way for supporters of the infringement model to adjust to handle the problems I  have raised here. If an infringement model theorist abandons the connection between infringement and compensation, expands the range of interests protected by rights, and abandons the idea of a separate and distinct set of consequentialist considerations, then her version of the infringement model and the mechanics of claims may become mere notational variations on each other.35 But I treat these connections and ideas as substantive commitments of the infringement model. And defined in terms of those commitments, it is a model that is misleading and distorting. Sadly, the problems that flow from using the infringement model infect most rights discourse. But they need not do so. The mechanics of claims provides a clear and superior alternative. Thus, I think it worth the effort to try to argue as strenuously as I can against the infringement model and in favor of the mechanics of claims.

 Note the parallel in this possible reform of the infringement model and the idea of consequentializing deontology. See, e.g., Dreier, 2011. One can make a theory fit, but only by abandoning what was distinctive about it, which is fundamentally not enlightening. 35

Putting The Mechanics Of Claims In Perspective  |  123



5 Avoiding a Misplaced Emphasis on Intentions

T

he mechanics of claims is focused on the causal roles patients play and the causal means available to agents.1 Other things matter for claim strength as well, such as responsibility for choices. That sort of responsibility shows up especially clearly in the notions of forfeiture and waiver. I will also explore, in the next two chapters, how it shows up in weakened claims when forfeiture and waiver are not in play. For the sake of the discussion in this chapter, however, the relevant point is that the mechanics of claims makes the fundamental distinction between property claims and restricting claims by reference to the causal role patients play in potential justifications for acts. This focus on causal roles contrasts with the focus many people put on agent intentions. Indeed, the norm in just war theory is to appeal to intentions, not causal roles. This chapter is meant to defuse the objection that I give intentions short shrift. I start the chapter with a discussion of Seth Lazar’s use of intentions in his effort to shore up the idea that it is harder to justify targeting noncombatants than it is to justify targeting combatants in war (the principle of distinction2). I focus on Lazar because his argument and the problems I want to raise for it provide a case study regarding why one should not put weight on intentions in the way that many scholars do. Moreover, it enables me to start making a bridge back from the general discussion of the last two chapters to the more specific issue of the permissible use of 1 2

 Sections 1, 2, and 4 of this chapter largely reproduce the material in Walen, 2017.  For a discussion of the principle of distinction, see Chapter 1, § 2.

defensive force. But I also add to the discussion of Lazar a more general argument in defense of my giving intentions relatively little significance in accounting for the permissible use of defensive force. I  conclude the chapter by applying the lesson of these arguments to Lazar’s argument that eliminatively killing noncombatants is generally impermissible and to the use of intentions in just war theory generally.

1. Introduction to Lazar’s Argument Lazar is concerned that just war theory has taken a bad turn, one that undermines the protection noncombatants are normally—​and he thinks properly—​thought to enjoy. In response, he argues in favor of a principle related to the principle of distinction, one that he dubs moral distinction: “In war, with rare exceptions, killing noncombatants is worse than killing combatants.”3 This is not merely an evaluative claim about what is “worse”; it is offered as an action-​guiding claim, concerning what is more readily justified. One argument Lazar offers in defense of moral distinction is based on the claim that “intentionally killing civilians is more opportunistic than intentionally killing soldiers. . . . Each kind of killing involves mixed kinds of agency. But the mixture in anti-​civilian violence is more opportunistic than it is for counterforce attacks.”4 Lazar thinks this premise can support moral distinction because he also thinks (a) that opportunistic killing (killing that aims to exploit the target’s presence) is harder to justify, all else equal, than eliminative killing and (b) that the greater the prevalence of the former in the mixed motives of an agent, the harder it is to justify the action in question. I argue that Lazar mistakenly attributes moral significance to the idea of “mixed kinds of agency.” My primary reason is that intentions are generally only of secondary importance when it comes to determining the permissibility of actions in the context of life-​or-​death choices or choices involving serious bodily harms.

2. The Limited Significance of Mixed Motives I start with the thought that an agent often acts for a combination of reasons. I  assume that if some of those reasons are morally suspect or  Lazar, 2015, p. 2.  Ibid., p. 72 (italics in the original).

3 4

Avoiding A Misplaced Emphasis On Intentions  |  125

worse, that can increase the extent to which her action is blameworthy. But whether she acts impermissibly is a different matter. I argue here that if her motives are mixed and if she limits herself to performing only morally permissible acts, at least on the occasion under consideration, then she acts permissibly. Indeed, I argue that even if her motives were purely illicit, the problem with her action would not be the sort of problem we care about when we worry about wrongful eliminative killing. An otherwise permissible killing carried out for entirely illicit reasons is nothing like murder. Before proceeding, I want to quickly clear up an ambiguity. The notion of mixed agency refers to the reasons for action in play in an agent’s action. But the notion of a reason can be understood as the objective reasons that might be offered to justify an act or as the subjective reasons for which an agent acts. Clearly, it is the subjective reasons that are relevant for Lazar. If his argument turned on the objective reasons that could be offered to justify an act, it would obviously fail. It would fail because one cannot undermine the justifiability of an act by proposing bad reasons for it. Any permissible act could be defended in any number of wrongheaded ways. That simply cannot make the act itself impermissible.5 Wrongheaded reasons can conceivably be morally relevant only if they are the reasons for action motivating a particular agent. To argue that mixed reasons are not relevant to permissibility, even if interpreted as subjective reasons, I  introduce three different kinds of bystanders at a trolley switch. Each is considering whether to divert a trolley that cannot be stopped from a track where it will kill five to a track where it will kill the sidetrack man. They all see that there are five people on the track straight ahead and only the sidetrack man on the other track, and they all know—​assuming that this is both true and knowable—​that saving the five makes the act of turning the trolley morally permissible. Moreover, they all care about that fact; none would turn the trolley if the five were not there. In addition, they all know—​same assumptions—​that the “act”—​ the omission understood without reference to the reasons for which it is chosen—​of allowing the trolley to hit the five, at least for an agent who has not waived her negative agent-​claim not to kill, is

 Suppose the reasons against killing in some particular case just barely tip the scales against killing if one looks at it simply as a case of eliminative killing. Might it not be that adding an opportunistic justification might tip the balance? That seems unlikely, but suppose that in a particular case it works like that. Then offering that reason would not compromise the justifiability of the act; it would establish it. 5

126  |  The Mechanics Of claims And Permissible Killing in war

permissible. All three turn the trolley. They differ, however, in the reasons for which they act. Benevolent Bystander: Betty regrets that her act will kill the sidetrack man, but she is more concerned to save the five than not to kill him. Mixed-​Motive Bystander: Mitzi unreasonably hates the sidetrack man and is glad to have a chance to kill him, but she also sees that it is important to save the five. She acts on both reasons. Moreover, she believes that she would act like Betty if she did not hate the sidetrack man. Malicious Bystander: Malia unreasonably hates the sidetrack man and is glad to have a chance to kill him. She would not turn the trolley to save the five; her only reason for turning it is that it would kill the sidetrack man.

Betty deserves praise. Mitzi deserves some praise and some criticism. And Malia deserves more criticism, and arguably no praise, for doing the right thing for a bad reason. Before continuing, it is important to be clear that I envision the hatred that Mitzi and Malia feel for the sidetrack man provides each with a bad reason to want to kill him. Imagine that each hates the sidetrack man because he got a job she wanted and she is consumed with jealousy. I raise this point because not all malice is unreasonable. Suppose the sidetrack man raped her, got away with it, and teases her about it whenever he sees her. That, on its own, would not suffice to justify killing him, but it would arguably be a reason to be willing to take on the burden of killing him when the agent would otherwise not be willing to kill to save five. If her malice were reasonable in that way, then he could not reasonably object to her killing him for that reason, not when killing him is otherwise justifiable. It might not be admirable if she acts on reasonable malice, but doing so would not be a clear moral failure. But again, I assume that the malice each feels is of the unreasonable variety. The question I want to address now is, Do the actions of Betty, Mitzi, and Malia differ in terms of permissibility? There are two sorts of ways to think about the relevance to permissibility of illicit reasons for action: (1) illicit reasons could be seen as intrinsically wrongmaking or (2)  illicit reasons could be seen as objectionable from the point of view of those who have claims on the agent. The first framing may make sense within a perfectionist morality, but it should be rejected by a fundamentally liberal morality. While a liberal morality must be able to make sense of the thought that agents have moral reasons to strive to be better people, and

Avoiding A Misplaced Emphasis On Intentions  |  127

thus have moral reasons to focus on the reasons for action that they choose to put into action, it should treat permissibility as a notion reflecting the constraints that respect for others—​expressed in terms of norms of rights and justice—​imposes on agents.6 That is to say that a liberal morality should embrace patient-​focused, not agent-​focused, limits on the permissible exercise of liberty.7 The alternative would be to confuse paternalistic reasons, which aim at protecting an agent’s moral purity, with reasons for making an act impermissible.8 To avoid that, a liberal morality should accept that the reasons on which agents act are relevant to permissibility only insofar as they are objectionable from the point of view of those who have claims on the agent. Contrary to many critics of the relevance of intentions to permissibility, I think it is not hard to see the appeal in saying that the sidetrack man has a claim not to be killed for unreasonable malicious reasons.9 It shows him disrespect, and our status as beings whose welfare matters implies that it is perfectly reasonable for him to object to someone acting toward him on the basis of a disrespectful reason or motive. Thus, it is at least worth considering whether Mitzi and Malia act impermissibly. Some have objected that the idea that someone like Malia, much less Mitzi, acts impermissibly is “absurd” because it implies that she must simply let the five die.10 But this absurd implication follows only if she could avoid acting on her malicious reason for action only by not saving the five. If, instead, she can choose to perform the act for a better reason, then it is not absurd to hold that that is what she must do. I believe that she can choose to perform the act of turning the trolley for a better reason.11 I say this because I think we can choose to commit ourselves across a range of possible cases, just as we can choose to pursue  To be clear, my point is not as narrow as the harm principle. As I argued in Chapter 3, § 2, an agent might lack the right to perform an action for reasons of justice that cannot be reduced to claims against her. In addition, some actions are impermissible, even if they cause no harm, because they are attempts to wrong others. 7  See Kamm, 1992b, p. 382 (note: I use the term agent-​focused differently from Kamm). See also Scanlon, 1998, p. 153. While I differ with Scanlon—​because I think that contractualism cannot be properly spelled out without developing a prior substantive theory of rights—​I agree that permissibility reflects the “principles for the general regulation of behavior that no one could reasonably reject as a basis for informed, unforced general agreement.” 8  I credit Scheffler, 1982, pp. 96–​97, with first making this point about paternalism. 9  Warren Quinn, for example, thinks that people who have pro tanto rights not to be harmed have stronger pro tanto rights not to be harmed when the other deliberately involves “them in something in order to further his purpose precisely by way of their being so involved.” Quinn, 1993, p. 184. See also Tadros, 2011, c­ hapter 7. 10  See Thomson, 1999, p. 515. 11  See Kolodny, 2011, pp. 104–​105. See also Tadros, 2011, pp. 145–​46. 6

128  |  The Mechanics Of claims And Permissible Killing in war

one of two or more actual options. That sort of counterfactual commitment can give practical meaning to the thought that an agent is acting for one set of reasons and not for another. In Malicious Bystander, I think Malia can make a commitment to act for the morally relevant reasons, which, by hypothesis, she recognizes, and not to act for her malicious reason. The commitment to act for the morally relevant reasons rather than malice would affect her choices by causing her either not to turn the trolley onto the sidetrack man or to turn it to save the five even if she were to discover that the one on the other track was not someone she hated and wanted dead. Arguably, she must reject the first possibility because she is willing to kill for an illicit reason. This effectively forfeits her claim that she should not have to bear the burden of killing for the sake of the five. Given the balance of patient-​claims on her, she would then have an obligation to save the five even if she were to discover that the one on the other track is not someone she hates and wants dead. There is no reason this obligation should be any more obscure to her or any more difficult to conform to than other moral obligations. We thus can answer the absurdity challenge. Nevertheless, there are two ways in which this result shows less about the relevance of acting on illicit reasons than one might think. First, it is unclear how to extend this reasoning to Mitzi, the agent from Mixed-​ Motive Bystander. She could try to convince herself that hatred of the one does not provide her with any reason to kill him. If we assume that she has a sound moral understanding, then she should recognize that her malice is not a good reason for action. Nonetheless, it is much less clear how she can assure herself that she is not acting on it. By hypothesis, she would turn the trolley even if she did not hate the sidetrack man. She may intellectually reject her malice toward the sidetrack man as a basis for action, but she may feel the pull of acting on it all the same. And it is not clear how she can simply choose to rid herself of that pull. She can focus on the second-​order reasons not to take malice to give her a reason, but that may not suffice to rid her of malice. And if she is honest with herself (and us), she may have to admit that her malice gives her some extra motivational interest in doing what she already takes herself to have sufficient moral reason to do, namely to turn the trolley. She may be blameworthy for having this sort of hatred in her and for letting it play a motivational role in her action,12 but given (a) the moral desirability of her turning the

 See Smith, 2005.

12

Avoiding A Misplaced Emphasis On Intentions  |  129

trolley and (b) the difficulty of her purifying her reasons for action, it does not seem reasonable to say that she violates any right of the sidetrack man if she acts with malice as one of her motivating reasons for action. To say that would be to embrace the absurd result that morality should prefer that five innocents die than that one innocent be killed—​an innocent who can justifiably be killed for the sake of the five—​if he is going to be killed by someone acting in any way on an illicit reason for action. Second, even if we assume, again, for the sake of argument, that it is impermissible to kill someone solely out of malice, as Malia does, that still leaves unanswered an important question:  What kind of a wrong is that? Is it like murder? Is it a wrong like killing someone when there is no justification for the act itself? One need only briefly consider two cases to see that it is a much lesser wrong.13 Contrast Malia, who seeks to exploit the opportunity to kill her enemy when the act of doing so is justifiable, with an unconstrained malicious bystander who turns a trolley onto a sidetrack man even though doing so saves no one. The latter commits murder; the former commits, at most, a much lesser dignity-​based  wrong. This discussion of Malicious Bystander shows, I think, that acting on an objectionable reason is at most a weak wrongmaking factor. It may make actions like Malia’s impermissible, but it would not make them the kind of serious wrongs that we care about when we think of murder. The action the malicious bystander performs may be a wrongful killing, but it is not a wrongful killing. And in cases like Mixed-​Motive Bystander, it seems to have no impact on the permissibility of her actions at all.

3. Objection: Respect as a Subjective State One might object that a malicious killing is actually worse than a benign yet wrongful killing and that this shows that intentions have a more significant role than I have just argued. To see this, compare Malia with the agent in Massive Man, Bertha, who, we can assume, is just as benevolent as Betty, the agent in Benevolent Bystander. Bertha is wrong if she thinks she is permitted to use the massive man to save the five. But at least she is motivated only by the pull of their claims for aid. By contrast, Malia is motivated by nothing of moral value. Many will want to say that

 I first made this argument in Walen, 2016, pp. 218–​19.

13

130  |  The Mechanics Of claims And Permissible Killing in war

a malicious trolley switcher commits a greater wrong than a benevolent toppler of a massive man. I think the intuition that the malicious trolley switcher commits a greater wrong than the benevolent massive-​man toppler is mistaken. It rests on confusing the degree of culpability for a wrong with the gravity of the wrong itself. No matter how culpable a person’s reasons for action, the magnitude of the wrong she can commit is limited by the wrongness of the act she commits or aims to commit. No matter how malicious a petty thief, her crime is not as grave as a wrongful killing.14 And no matter how benevolent the wrongful killer, her crime is graver than that of a petty thief. Arguably the worst dignity-​based wrongs are worse than petty theft. Consider a dentist who knocks his patients unconscious in order to do dental work on them and who then gropes the breasts and genitals of women he finds attractive. A victim of his groping behavior may not be physically harmed; she may not even be psychologically harmed if she does not find out about what he did. But surely he commits a serious crime. Nevertheless, I remain convinced that a single incident of such a dignity-​ based harm is a less serious wrong than taking the life of someone who has a right not to be killed, even with the best of (misguided) intentions. Matthew Liao objects that the wrong committed by Malia in Malicious Bystander is not a mere dignity-​based harm; it is “murder.”15 His argument is that acting on a malicious intention can make an act impermissible, and the kind of impermissible act we are considering is a wrongful killing. Moreover, he means it in the sense of a wrongful killing. That is, he thinks that since the act of turning the trolley is only permissible and not obligatory, the fact that the bystander acts wrongly if she acts with the aim of killing implies that she ought not to do it: she ought not to throw the switch. If she does throw the switch, she kills wrongly, and that is murder. The problems with this argument are twofold. First, as I pointed out in the previous section, there is an alternative way for the bystander not to act on a malicious intention, and that is to perform the same act but for a different reason: if not out of concern for the five, then at least out of a sense of duty. Second, the idea that turning the trolley is simply a permissible option overlooks the structure underlying that fact: the balance of patient-​ claims calls for the agent to turn it, but the agent has (unless she waives  This example shows that we have to be somewhat careful with the notion of an act. An act like theft inherently involves an intention: the “further intention,” as it is called in the criminal law, not to return the property. But we can still distinguish different reasons for committing theft and take those to be the different action variations. 15  Liao, 2012, p. 716. 14

Avoiding A Misplaced Emphasis On Intentions  |  131

or forfeits it) a negative agent-​claim not to have to turn it.16 Looking at just the balance of patient-​claims, it calls on her to turn the trolley. In that context, it is implausible to say that the action is a wrong like murder. The sidetrack man has no right not to have the trolley turned onto him. He has, at most, a right not to be acted against for a disrespectful reason. Again, that wrong is merely a dignity-​based wrong; it is wrong because it is objectionably offensive, not because it wrongfully deprives him of his life.17 Liao also makes the argument that what is fundamentally wrong in cases of acting on disrespectful intentions is acting on a particular problematic attitude:  disrespect of the person.18 If that is right, then disrespectful intentions have to be the basis for our judgments that wrongful actions are wrongful. Taking this point for all it’s worth, the thought is that I was mistaken to suggest that we should start by considering the permissibility of acts, stripped of the particular intentions with which they are performed. Rather, we should take as our starting point Warren Quinn’s thought: “Each person is to be treated, so far as possible, as existing only for purposes that he can share.”19 And in parsing that imperative, we should recognize that the notion of “treatment” concerns the kinds of intentions each forms with regard to others. There are many problems with this position. First and foremost, as I emphasized when introducing the restricting claims principle, this focus on intentions wrongly puts the focus of deontology on the mind of the agent, rather than the claims of the patient.20 Unpacking that just a bit, disrespect is a problem because the patient deserves respect:  on that all sides agree. But what does that respect mean in practice? Does it translate directly into a requirement that the agent show respect for the person in her attitudes? That would require an agent to act on good will but would not do

 I am not saying that Liao overlooks this structure. He shows he is aware of it: ibid., pp. 715–​16. “Within certain limits, so the thought goes, agents should not be required to cause harm to others, especially if they have deep personal commitments against doing so.” But he doesn’t recognize its significance for this issue. 16

 What if the agent did not see that there were five on the other track and thought of herself as an unconstrained malicious bystander? She has then committed a sort of attempted murder. In a purely fact-​relative sense, she performs a good act. Someone else could even permissibly shout encouragement if he knows that it is permissible to turn the trolley. See Walen, 2004. But her intention was to murder, and thus her action is culpable like an attempted murder. 18  Liao, 2012, pp. 723–​24. Liao acknowledges that he is drawing, in this argument, on the earlier work of Quinn, 1993. Lazar, 2018, p. 337, rests his response to me on a similar thought: “I think that the reasons for which we act reflect our attitudes to those whom our actions affect, and that those attitudes are morally fundamental.” 19  Quinn, 1993, p. 192. 20  Chapter 3, § 4.1. 17

132  |  The Mechanics Of claims And Permissible Killing in war

much to explain or ground the particular rights of patients. The mechanics of claims assumes that we can do a lot to work out what rights patients have even before worrying about the attitudes of agents. These rights provide, as it were, cash value for respect: agents must not violate the rights of patients. To violate the rights of others is to fail to show them the respect they are owed as fellow citizens in the space of rights. This is the core notion of respect that a Kantian should accept, not some immediate turn into the intentions of agents. To repeat, I  do not mean to disparage the claims patients have on the intentions and actions of agents. To the contrary, I recognize a large number of ways in which intentions are relevant to permissibility. Clearly, patients have claims not to be manipulated into consenting to interactions on the basis of misrepresentation of the other’s intentions.21 I  have also argued that it is sometimes impermissible for an agent to act on intentions that include impermissible acts within their scope (i.e., to execute any step of a plan that requires, or if certain circumstances arise, might require, the agent to perform an impermissible act).22 In addition, it is sometimes impermissible to act on certain reasons, like racist reasons, that contribute to widespread social problems that make it difficult for others to have a fair chance to lead a good life. And I have already indicated that I think that patients have claims not to be treated offensively.23 Nonetheless, the relevance of intentions is, for two reasons, limited and secondary. First, the claims patients have with regard to such intentions supplement the basic set of rights that can be marked even without reference to the intentions of agents. Second, the reasons intentions are sometimes relevant to the permissibility of an action turn on the agent orienting herself to doing things that can be understood as unjustifiable, or at least problematic, prior to saying anything about the agent’s intention. The second problem with the argument that deontology is fundamentally anchored in the imperative not to act on disrespectful intentions is that it not only fails to understand the nature of patients’ claims but also fails to come to grips with the claims of agents. Agents, in a liberal society, enjoy   Scanlon, 2008, p.  40, offers this kind of example to explain the derivative significance of intentions—​derivative of the claim not to be deceived into giving one’s consent. 22  See Walen, 2006, and Walen, 2013. Attempted wrongdoings are acute wrongs of that sort. 23  I also now agree with Victor Tadros that a case in which A contributes to causing a harm to V but does so knowing that she cannot stop others from causing the harm to V and knowing that by contributing she will actually make things slightly better for V is a case in which intention matters for permissibility. If she were to perform the same act intending to harm V, she would act culpably and therefore impermissibly. See Tadros, 2011, pp. 159–​60, and Tadros 2018, pp. 422–​23 (Poisoned Pipe case). I was less willing to see this case in its own terms in Walen, 2013. 21

Avoiding A Misplaced Emphasis On Intentions  |  133

the basic freedom to lead their own lives. That freedom has to be balanced against patient-​claims not to be harmed. But when the latter claims are relatively insignificant, agents should enjoy their basic freedom untrammeled. This came up once before, in Chapter 3, section 2, where I argued that “we need to recognize a basic freedom to use one’s property in ways that cause only minor annoyance or offense to others.”24 Here I raise the related idea that agents have a claim to act on illicit reasons as long as they constrain themselves to performing only those acts that do not impose serious costs on patients who have a right not to bear those costs. Again, I  am disparaging neither the thought that people have claims not to suffer offensive behavior nor the thought that the reasons on which agents act can make their actions offensive. But suffering from offensive action is a relatively low-​level harm in itself. Nor can it be made into a great harm by being linked to some great harm. The great harm of dying at the hands of a malicious agent is, normally, dying25; the fact that it is at the hands of a malicious agent adds, literally, insult to injury; but the insult should not be confused with the core injury. Thus, the claim not to suffer offense does relatively little to constrain agents and at most gives rise to minor wrongs when agents act on offensive reasons. In sum, making the disrespectful intentions on which a person acts the basic wrongmaking feature of her action is fundamentally backward. It is only a secondary wrongmaking feature.

4. Application of the Preceding Argument to Just War Theory 4.1. Application to Lazar’s Argument Let us now take these points and reconsider Lazar’s claim that “[I]‌ntentionally killing civilians is more opportunistic than intentionally

 See also my point about the right to do wrong and the liberal idea that there is a distinction to be drawn between whether an act is virtuous or vicious and whether it is permissible or not. Chapter 3, § 1.3.1. 25  I say “normally” because one can imagine cases in which an agent would not mind dying but her whole sense of purpose is tied up in not being humiliated. And one can imagine that if she is killed by a malicious agent, she would be humiliated. (Conversely, one can imagine cases in which a person would find death by an enemy noble, whereas death by a deflected trolley would be humiliating.) But even if one is dealing with such a case, the objective weight of the claim not to be humiliated is less than the objective weight of the claim not to be killed. This is clear if one thinks about the kind of force that would be proportionate to stop a humiliation versus a wrongful killing. 24

134  |  The Mechanics Of claims And Permissible Killing in war

killing soldiers. . . . Each kind of killing involves mixed kinds of agency. But the mixture in anti-​civilian violence is more opportunistic than it is for counterforce attacks.”26 Suppose that a given combatant (or her commander) sees that there is an eliminative reason to kill a set of noncombatants on the other side. And suppose that, all things considered, such a killing would be a permissible act. Would acting based on some mixture of opportunistic reasons that seek to take advantage of the person’s presence turn the act into an impermissible one? Not if the legitimate reason also plays a guiding role. In such a case, a mixture of justifications in the minds of the actors would be just as irrelevant to the permissibility of their actions as it was in Mixed-​Motive Bystander. What if the combatant (or her commander) acts primarily on opportunistic reasons? Would that affect the permissibility of her actions? Maybe. But if she limits herself to acts she knows are permissible, she would then be guilty of only a relatively small wrong, nothing on a par with wrongfully killing people who, regardless of the intention of the agent, may not be killed. This is not the sort of wrong on which to hang much of interest for just war theory. It is too minor. In his reply to an earlier version of this chapter, Lazar adds this argument: [S]‌uppose that you can choose between two courses of action to achieve the same result. Both involve mixed modes of agency; one is more eliminative than opportunistic; the other is more opportunistic than eliminative. Which should you choose? Clearly the more eliminative one. This tells in favour of killing combatants rather than noncombatants.27

My response is, Sure, all else equal, you should choose the more eliminative action; it is less offensive. But suppose the one with the more eliminative mix is 10 percent less likely to be effective or would cost 10 percent more to carry out. And suppose that both are otherwise fully justifiable. Should the mix of intentions outweigh that? I say, clearly not.28  Lazar, 2015, p. 60.  Lazar, 2018, pp. 327–​28. 28   Lazar also speculates in his reply piece that commanders and the combatants who do the killing would have different intentions with regard to different targets of killing. With regard to noncombatants, he writes:  “Their deaths can count as eliminative only at the strategic level. For the people carrying out the killing, the primary motivation is likely to be merely murderous, or else opportunistic.” Ibid. I don’t see this as a significant new argument, but I also have a contrary suspicion. I suspect that many combatants kill out of anger or hatred, whether they are killing enemy combatants or enemy civilians; and I suspect that the rest are equally likely to kill simply because they are following orders, with the general belief that doing so is relevant to trying to win the war. 26 27

Avoiding A Misplaced Emphasis On Intentions  |  135

4.2. Lessons for Just War Theory About the Relevance of Intentions I conclude this chapter by pointing out, as I  did in Chapter  1, that the implications of this argument for just war theory are significant. Almost all just war theorists frame the principle of distinction with reference to the intentions of the agents acting. But they should not do so. They should read the distinction in terms of possible justifications for different acts. There is no need to invoke intentions as the restricting claims principle provides an intention-​free way of making sense of why it is much harder to justify harmfully using another as a means, unless he has waived or forfeited his claim not to be so used, than it is to justify causing another to suffer harm as a side effect of pursuing some goal. And as we have just seen, even if illicit reasons for action have a wrongmaking power, their application in the context of just war theory is most likely to give rise to a dignity-​based wrong. That is the wrong kind of wrong for just war theory to focus on. I think this mistaken reliance on intentions in just war theory is mostly not a “deep” problem. I think it mostly reflects a vestige of the history of moral thought, in which the doctrine of double effect29 was dominant in explaining the principle of distinction. It should be fairly easy for most theorists to recast their thoughts in terms of the restricting claims principle. The deeper point that I want to emphasize is that what really matters for the deontological grounding of just war theory is the causal roles of the patients whose claims are in the balance. The restricting claims principle explains why the causal role of being a means by which an agent can achieve some good, as opposed to being someone who would merely be affected by what the agent has a baseline freedom to do, affects the strength of the patient’s claims. But the role of being a threat is different from either of those roles, and it has its own moral significance. As noted in Chapter 1, it is actually easier to justify causing harm to a threat who lacks the right to threaten than it is to justify causing harm to one with a claim not to be harmed as a side effect of achieving other ends. This view is quite different from that which results from framing the distinction in terms of intentions because if one frames the act of eliminative killing in terms of intentions, it will be a form of intentional killing and therefore harder to justify than killing as a side effect.

 This doctrine holds, roughly, that it is harder to justify intentionally causing a harm (or “evil”) than causing it as a side (or double) effect. 29

136  |  The Mechanics Of claims And Permissible Killing in war



6 Innocent Threats Versus Innocent Bystanders

T

his chapter is the first part of the two-​part argument that there is no fundamental, principled reason why noncombatants who help unjust combatants to wage war may not be eliminatively killed. It sets the moral foundation by arguing that it is easier to justify eliminatively killing innocent threats than it is to justify the side effect killing of innocent bystanders; Chapter 7 brings the argument to fruition by moving from that premise to the conclusion regarding noncombatants who help unjust combatants to wage war. I set a foundation by explaining why it is hard to justify killing an innocent bystander to save an innocent victim. I then argue that the choice whether to kill an innocent threat or allow an innocent victim to die is a closer call. I argue, in particular, that we cannot reason from the premise that a threat owes a victim some sacrifice to the conclusion that his harm can be discounted in the way suggested by those who would invoke a lesser-​evil justification. I argue, instead, that there are both agent and patient effects of being a threat, effects that make it easier to justify killing a threat, even an innocent threat, than an innocent bystander. Finally, I respond to objections.

1. Introduction Is it permissible to kill an innocent threat to save the life of an innocent victim who would otherwise be killed by him? As noted in Chapter 1, this has been the subject of a long-​running debate in moral philosophy. Some

say no; some say yes. I say it’s a close call. But I also say that it is clear that innocent threats have weaker claims not to be killed than innocent bystanders and that it is therefore easier to justify eliminatively killing innocent threats than it is to justify killing innocent bystanders as a side effect of saving others. This may seem like a morally marginal issue. Innocent threats are incredibly rare. Why should anyone care? There are two reasons. First, substantively, establishing that innocent threats have weaker claims not to be killed than innocent bystanders forms the basis for the argument pursued in Chapter 7 regarding the permissibility of eliminatively killing noncombatants who help unjust combatants to kill. Second, this topic provides an interesting case study regarding the way rights work and thus lies at the theoretical core of this book. In particular, the deontological discussion of the permissibility of eliminatively killing innocent threats has been framed, to date, by one of two assumptions. According to one assumption, intentions matter and make it harder to justify intentionally killing an innocent threat than foreseeably killing an innocent bystander. I explained why I reject this assumption in the previous chapter. According to the other assumption—​one framed by the infringement model—​we have to appeal to the idea of a lesser-​evil justification to make sense of the justifiability of killing innocent threats. Some think such a justification is available only if many more innocent victims would be saved than innocent threats killed. Others appeal to the idea that threats owe victims a special duty of care, and that their interests can therefore be discounted, to argue that a lesser-​evil justification can be found for killing an innocent threat to save an innocent victim. I argue that both approaches to a lesser-​evil justification are misguided and, thus, continue my argument against the infringement model.

2. Innocent Victims and Innocent Bystanders The first step in arguing that innocent threats have weaker claims not to be killed than innocent bystanders is to get a sense for the strength of the claims of innocent bystanders. I do that by contrasting the claims of innocent bystanders with the claims of innocent victims. An innocent victim is someone who is threatened by another and who has done nothing to bring that threat on himself or to forfeit his claim not to be killed. An innocent bystander faces no threat at the start of any scenario, though he might become threatened by some agent’s effort to save an innocent

138  |  The Mechanics Of claims And Permissible Killing in war

victim. After arguing that an innocent bystander’s claim not to be killed is stronger than an innocent victim’s claim to be saved, I will reintroduce an innocent threat to see how his claim would stack up against that of an innocent bystander. I made the argument in Chapter 3 that an innocent bystander’s claim is stronger than that of an innocent victim, all else equal, because it is the innocent victim’s bad luck to be threatened.1 He may not simply shift his bad luck onto an innocent bystander. Nor may anyone else shift his bad luck onto an innocent bystander on his behalf. The innocent victim has a positive restricting claim to be saved, and such claims are weaker than negative restricting claims, all else equal. This is a direct implication of the autonomy principle and the way it requires people to own their luck. If the number of innocent victims with claims to be rescued is sufficiently great compared to the number of innocent bystanders with claims not to be harmed, then it would be presumptively permissible to shift the bad luck of the victims onto the bystanders. Likewise, if the harm an innocent victim faces is significantly greater than the harm an innocent bystander would be caused if the victim were saved, then it would be presumptively permissible to shift the bad luck of the victim onto the bystander. But in those cases in which luck shifting is permissible, those who are harmed would acquire a claim for compensation against those who benefited from the shift.2 This idea of shifting bad luck conjures the image of shifting a threat, as in Trolley Switch. But it is meant to apply more generally. I  think of it as applying equally if the agent has to do something like blow up the threat to save an innocent victim, foreseeing that such an act will cause harm to a bystander. Many, however, think that those cases are not equivalent. They think that it is substantially easier to justify redirecting an existing threat, as in Trolley Switch, than it is to justify introducing a new threat that will cause harm as a side effect of saving innocent victims.3 I take this opportunity to argue that they are mistaken. There are too many people who have taken such a view to try to engage them all, but I will engage the one whose views seem to have had the most influence: Frances Kamm.

 See Chapter 3, § 1.3.3.  See Chapter 4, § 4.1. 3  See Draper, 2016, p. 139 (citing others who agree: Philippa Foot, Gregory Kavka, Samuel Rickless, and Frances Kamm). See also McMahan, 2014b, pp. 131–​32. 1 2

Innocent Threats Versus Innocent Bystanders  |  139

Kamm offers what she calls the principle of permissible harm to explain when it is permissible to cause harm to some to save others.4 In a recent “rough” statement of the principle, she describes it this way: Actions are permissible if greater good or a component of it (or means having these as a noncausal flip side) leads to lesser harm even directly. Actions are impermissible if mere means that produce greater good (like [an explosion that blows up a threat]) cause lesser harm at least directly, and actions are impermissible if mere means cause lesser harms (such as toppling people in front of a trolley) that are mere means to producing greater goods.5

Kamm’s account presupposes familiarity with a number of technical terms; I endeavor here to explain the two that are most in need of explaining. By “a component of greater good,” Kamm means to pick up the idea that what is strictly caused in turning a trolley in Trolley Switch is that the five live a moment longer than they otherwise would have. If they would live only for that moment, then no greater good would have been brought about. But that moment is part of the causal chain that leads to further moments of life for the five and is a component of the total good that their future lives represent. The “noncausal flip side” reflects the idea that turning the trolley from the five does not cause them to continue living the way that it causes the death of the sidetrack man. It is, rather, a logical or “noncausal” consequence of their not being hit that they continue to live. With those terms clarified, we can see that what her principle says is that relative location in the causal stream of greater good (or its noncausal flip side) and lesser harm is fundamental to permissibility. It is permissible to turn a trolley away from five, even though it will hit and directly kill the sidetrack man, because turning the trolley away from the five comes first. It causally leads to the death of the sidetrack man. But the five being saved is tied (by the noncausal flip side relation) to the act of turning the trolley and, thus, is causally upstream from the lesser harm. That makes it permissible. By contrast, if Brenda had to toss a bomb at the trolley (assume it’s empty, so no one on board will die) to prevent it from hitting

 Kamm, 2007, pp. 24–​26; Kamm 2015, pp. 66–​67. Kamm, 2007, pp. 138–​69, also offered a more nuanced theory, the doctrine of productive purity; but I think it suffices to discuss the principle of permissible harm, which is much easier to summarize. 5  Kamm, 2015, p. 66. 4

140  |  The Mechanics Of claims And Permissible Killing in war

the five, foreseeing that the bomb would also kill a bystander, that would not be permissible. It would fail the first of the impermissibility prongs as it would directly cause the lesser harm (the death of the bystander) but would not bring about a greater good, or the noncausal flip side of it, at an earlier point along the causal path. Rather, the explosion would have two effects that are causally parallel: blowing up the trolley and blowing up the bystander. As the good effect (or its noncausal flip side) is not causally upstream from the lesser evil, it is impermissible to proceed. I have two fundamental complaints against this account: (1) though it is designed to provide an account of a range of intuitions, it actually seems to me to be quite counterintuitive; (2) it puts tremendous weight on the location of events (or their noncausal flip sides) in the causal stream without answering the basic question, Why should mere causal facts matter? I take these in turn. Thomas Hurka convincingly explores why the principle has counterintuitive results. His discussion is extensive,6 but I think it suffices to focus on the first example he gives. He considers two ways a bomb dropped on a munitions factory might kill people near the factory as a side effect: If they’ll be killed by a flying piece of the bomb, the bombing is morally impermissible. That’s because the good our act can achieve is the destruction of the factory, the bomb’s exploding is a causal means to that, and causing even a lesser evil by means to a greater good is forbidden. But if the civilians will be killed by a flying piece of factory, the bombing is permitted. The factory’s exploding isn’t a causal means to the destruction of the factory; it’s effectively the same thing. . . . So the principle that allows the diverting of the trolley also allows bombing in which civilians are killed by flying pieces of factory but doesn’t allow bombings in which the same civilians are killed by flying pieces of bomb.7

Hurka finds “this implication very hard to accept.” As he later fleshes the point out, The problem in the bombing cases is that the principle attaches moral significance to something that doesn’t have it—​namely the difference between one flying object and another. The reason why the principle isn’t appealing

 Hurka, 2015, pp. 137–​46.  Ibid., p. 138.

6 7

Innocent Threats Versus Innocent Bystanders  |  141

is that it makes . . . an arbitrary distinction between flying objects that are means to the same good end.8

This seems to me exactly right. And it applies just as well to trolley cases as to war cases. No one can reasonably defend a moral distinction between turning the trolley and causing all of it to hit and kill a bystander and blowing up a trolley and causing part of it to hit and kill a bystander. But there can also be no meaningful moral difference between blowing up the trolley and causing part of it to kill a bystander and blowing up a trolley and causing part of the bomb to kill a bystander.9 Now if the theory behind the principle of permissible harm were particularly well grounded in deeper moral principles, then perhaps these counterintuitive implications should be accepted. But it is not. The best Kamm has to offer in defense of the principle of permissible harm is the following: I believe that on the causal structures distinguished by the [relevant causal principle] supervenes a moral distinction between substituting one person for another and subordinating one person to another. If this distinction is an important part of respect for persons, then the necessity (given our act) of certain causal structures will be consistent with respect for persons, and the necessity of others will not be.10

I agree with her that we need an account of why important moral differences supervene on differences in causal roles. And she has some suggestive things to say about why using as a means is subordinating and why certain other cases of impermissible acts also might look like subordination. But in the end, she has no systematic account to offer of why subordination is particularly hard to justify or why it should be linked to her causal principle. Nothing she says comes close to the account offered in Chapter 3 of the restricting claims principle. In the end, I conclude that there is no theoretical reason to embrace her principle of permissible harm. And thus,

 Ibid., p. 146.  Kamm seems to be moved by Foot’s case involving creating a gas to save some patients in a hospital knowing it will kill one. See Kamm, 2015, pp. 222–​24. Kamm agrees with Foot that this is impermissible, and she takes the implication to be that it is not always permissible to cause a lesser harm as a side effect. I think the lesson she draws is overly general, however. I have long thought that this case draws on intuitions that are specially relevant to the medical setting. The other patients may have a claim not to be harmed much like that of the man on the track in Rescue Ahead. 10  Kamm, 2007, p. 165. She repeats the basic idea in Kamm, 2015, p. 69. 8 9

142  |  The Mechanics Of claims And Permissible Killing in war

I conclude that there is no reason to accept its implication that introducing a new threat (rather than diverting an existing threat) and thereby causing a harm as a side effect is impermissible in very much the same way as using another as a means. In sum, we should assume that the negative restricting claims of innocent bystanders are stronger, though not much stronger, than the positive restricting claims of innocent victims. It doesn’t matter how those innocent bystanders with negative restricting claims might be killed—​whether by diverting a threat or creating a new threat that takes out the old one. What matters is only that negative restricting claims are somewhat stronger than positive restricting claims. For the sake of having a concrete number to use in the arguments that follow, I will continue to assume that negative claims are roughly three times stronger than the positive ones, all else equal.

3. Innocent Threats and Innocent Victims 3.1. Stating the Problem Turning now to the claims of innocent threats, one might reason as follows: it is intuitively permissible for an innocent victim to kill an innocent threat in self-​defense. If we reject the idea that people have some sort of agent-​centered prerogative to favor themselves and kill on their own behalf,11 then if an innocent victim may kill an innocent threat in self-​ defense, it must also be permissible for a third party to kill an innocent threat in defense of an innocent victim. But it is impermissible for both innocent victims and third parties to kill an innocent bystander to save an innocent victim. Therefore, the claims of innocent threats not to be killed must be weaker than the claims of innocent bystanders. The problem with this argument is relying on the first intuition. Though it is a widely shared intuition, it is, I think, an unreliable one.12 It could easily be that the intuition is an overextension of the idea that innocent people can normally defend themselves against threats who, normally, are to some extent culpable. Or it might reflect the confused idea of an agent-​ centered prerogative. Moreover, there are good moral theorists who conclude that the intuition does not stand up to scrutiny.13 Indeed, I no longer

 See Chapter 3, § 2.1.4.  For more on why it might be unreliable, see Grabczynska and Ferzan, 2009, p. 249. For more on the general unreliability of intuitions, see Chapter 2, § 1. See also Cushman, 2016. 13  See Chapter 1, § 1. 11 12

Innocent Threats Versus Innocent Bystanders  |  143

have a clear sense of whether it is permissible to kill an innocent threat to save an innocent victim. I therefore conclude that we need to put that intuition aside and look carefully at the theoretical principles involved. The ground for the claim that innocent bystanders have stronger claims not to be harmed than innocent victims have to be saved is the idea that innocent victims have to own their luck. So the natural place to start is to ask who, as between an innocent threat and an innocent victim, initially has the bad luck. Unfortunately, it is not clear. To see this, consider the following: Falling Man: The falling man has been tossed by a wild animal, through no fault of his own, down a well, where he will crush and kill Vicky (and survive himself) unless he or she does something to save her life. Vicky cannot get out of the way, but she has a pole that she can position to impale the falling man and save herself. Vicky is not the only one who could move the pole: the falling man has a whip that he could use to position the pole so that it would impale him and save Vicky. If Vicky were not present, the falling man would not die; thus, his life does not depend on his using her as a means of saving him.14

Who suffers from bad luck here? On the one hand, one can say that it is the bad luck of Vicky to be threatened by the falling man.15 If that were the end of the story, then we should conclude that neither she nor any third party may shift her bad luck onto the falling man and that the falling man is not obliged to shift the pole so as to save her. On the other hand, one can say that it is the bad luck of the falling man to be an innocent threat. He should not want to be an innocent threat, and even though it is not his fault, having become one, he has the moral burden of suffering harm rather than passing his burden on to Vicky.16 How are we to decide?

3.2. The Hybrid Argument: Its Insights and Its Failures Victor Tadros argues that even if it is not clear that an innocent threat has the moral duty to sacrifice his life to avoid killing an innocent victim, he has a duty to absorb, if he can, some substantial costs in order not to be  This is a variation on a case that Tadros, 2011, p. 249, calls Fat Man’s Remote. The origin of the case goes back at least to Nozick, 1974, p. 34. 15  This is McMahan’s position in his reply to Shlomit Wallerstein in McMahan, 2009b, p. 404. 16  I defended this second position in Walen, 1997, pp. 1100–​1103. I now think the argument less convincing than I once did. 14

144  |  The Mechanics Of claims And Permissible Killing in war

a threat. This allows us to discount the harm to him by the amount he already owes to Vicky. And that, Tadros thinks, would make killing him justifiable as a lesser evil.17 This is a hybrid rights–​lesser-​evil argument, taking advantage of both the rights and the consequentialist elements of the infringement model. It is distinctive, however, in that it takes the unusual step of arguing for an infringement after first discounting the strength of a right by appeal to a duty. I think Tadros’s first step is correct, but his second step—​the one that appeals to a hybrid of rights theory and consequentialism—​fails. Nonetheless, the first step is informative and helps point us toward a better second step. 3.2.1. Insights

The first step in Tadros’s argument appeals to this thought: “I must bear a greater cost to avert threats to others that would come about through my body than I must bear to avert threats to others that are independent of me.”18 One might object that one must bear a greater cost only if one has done something to be responsible for what one’s body does. But the objector’s principle seems to presuppose that we have to own our luck only when it flows somehow from a free choice. That, however, is inconsistent with the position I have taken on owned luck. The innocent victim may have made no responsible choice that caused him to be the initial victim, rather than the bystander, yet it is the victim’s bad luck to be the victim. Indeed, it is also the bystander’s bad luck to present a moral obstacle to the rescue of the victim; that is what explains why his claim not to be harmed is restricting and is as weak as it is.19 Accepting that one has to own one’s luck undermines the objector’s position, but it does not tell directly in favor of Tadros’s position. What tells in favor of Tadros’s position is that we should want not to be innocent threats; we should strive with some urgency not to be innocent threats.

 Tadros, 2011, pp. 252–​56. These ideas have also been taken up by Frowe, 2014, p. 69. Others use the same sort of hybrid reasoning in slightly different ways. See McMahan, 2014a, pp. 438–​40; Bazargan, 2014. McMahan’s argument is actually importantly different as it applies to cases in which a person is liable to one harm and would have to suffer a larger harm for a problem to be solved. That argument would not apply to innocent threats as they have not done anything to become liable to any harm. 18  Tadros, 2011, p. 255. Similar ideas were expressed by Kamm, 1992a, pp. 47–​48. 19  See Chapter 3, § 4.1. Note, the argument in the text suggests that an innocent bystander whose claim not to be harmed poses a moral obstacle to the rescue of an innocent victim would have a stronger duty to help the innocent victim than an innocent bystander whose claims did not pose such a moral obstacle. I accept that implication. 17

Innocent Threats Versus Innocent Bystanders  |  145

I am not making the self-​interested and circular point that we have reason to strive not to be innocent threats because we have reason to want not to be in a position where we can be justifiably harmed or killed. I  am making the point that agents have a moral reason to prevent their bodies from harming others. We have a special relationship with our bodies; we are embodied; when our bodies do well, we benefit; when our bodies are harmed, we suffer. We claim the right to enjoy at least some of the good things that might happen to or flow from our bodies (health, pleasure, physical power, social power); likewise, we must accept a special responsibility to ensure that our bodies do not inflict unjustifiable harm on others.20 This implies that it is bad luck to become an innocent threat; it is something that one has a special burden to try to prevent and to try to undo. This special burden should give an innocent threat more reason to sacrifice his welfare to save an innocent victim from the threat that he poses than an innocent bystander would have to save an innocent victim from a threat that something or someone else poses. And even if an innocent threat can do nothing to make that sacrifice, his conditional duty to make it negates his claim not to have to suffer an equivalent harm if doing so would save the innocent victim. Thus, an innocent threat may lack a right not to suffer a certain level of harm for the sake of an innocent victim, even though an innocent bystander would have a right not to suffer that much harm for the sake of an innocent victim. 3.2.2. Failures

The second step in Tadros’s argument appeals to the notion of a lesser evil. It goes like this: if the falling man has a duty to accept some harm for the sake of Vicky (e.g., a broken foot that he would suffer if he steered himself away from her), then the total harm that the falling man would suffer if killed for her sake can be “discounted” by the harm that he would be obliged to absorb.21 If we weigh the discounted harm he would suffer—​ loss of life minus a broken foot—​against the harm Vicky would suffer—​ loss of life—​we see that Vicky would suffer the greater morally relevant harm. This matters because, as Tadros put it, “as soon as other things are somewhat unequal, I am permitted to transfer my bad luck to you.”22

 This argument echoes one of Tadros’s arguments in favor of his claim, the one quoted at the start of the previous paragraph in the text. Tadros, 2011, pp. 253–​55. 21  Ibid., p. 251. 22  Ibid. 20

146  |  The Mechanics Of claims And Permissible Killing in war

There are three problems with this argument. First, it cannot literally be true that “as soon as other things are somewhat unequal, I am permitted to transfer my bad luck to you.” It would not be true, for example, that an innocent victim on a trolley track could turn the trolley onto a sidetrack man if his death would be slightly quicker and less painful than hers. There must be some threshold difference that makes it permissible to turn the trolley from one onto another. The threshold would reflect the difference in strength, all else equal, between a negative restricting claim and a competing positive restricting claim. Second, this argument encourages a kind of exaggerated view of what an innocent threat would owe an innocent victim. Suppose one thinks it at least plausible that it is permissible to kill an innocent threat to save an innocent victim. And suppose that one thinks that killing is justified on a lesser-​evil basis only when the ratio of good to harm is at least three to one. Then this argument would imply that the innocent threat must owe two-​thirds of the value of his life to the innocent victim. It is hard to say just what that would be, but it is surely well beyond a broken foot; presumably, it would be something more like being paralyzed from the neck down. This seems far-​fetched; this is, after all, someone who has done nothing at all to be responsible for the threat. This sort of sacrifice seems greater than can be supported by Tadros’s argument based on responsibility for one’s body. It seems more a supererogatory act than an enforceable duty. Last but not least, there is a big hole in the argument, and the appeal to a lesser evil covers that hole up without filling it in.23 The problem is that it is not obvious why the fact that the falling man owes some sacrifice to Vicky implies that his claim not to be killed is any weaker than an innocent bystander’s. To see this, change the nature of the debt to a substantial monetary debt—​thereby removing any possible conflation with the significance of being a threat—​and change the case to one in which an agent must choose which of two innocent bystanders to divert a threat onto, a diversion necessary to save five. More concretely, suppose the agent has the choice of diverting the threat onto Dennis, who owes each of the five a large sum of money, or onto Norman, who owes them nothing. The fact that Dennis owes them money does not in any obvious way imply that he should be the one onto whom she turns the trolley. His claim not to be

 I am indebted to Adil Haque for pressing this point at the Ethics, Law, and Armed Conflict workshop in Florence in 2016. Uwe Steinhoff also makes a similar point. See Steinhoff, 2018. 23

Innocent Threats Versus Innocent Bystanders  |  147

killed seems unaffected by his financial debts. Instead, it seems that she confronts two equal claims not to be killed and that she should do something like flip a coin to decide who to kill. One might respond to the argument just made by saying that collecting financial debts and killing are very different sorts of acts, involving very different claims in the patient. That is, one might say that it is more reasonable to think that if A owes B a physical sacrifice, then it is less of a leap to say that A’s claim not to be killed is also thereby weakened. Those two losses seem to be more connected, perhaps by a fundamental interest in physical integrity. That connection is lost if we switch to monetary debts and physical damage. But this reply does not withstand scrutiny. It is still mysterious why owing someone a relatively small physical injury (if necessary to save the person) implies that one has less of a right not to suffer a much more serious physical injury (if necessary). To see this, imagine that Dennis promised an agent that if she needed help doing something really important (saving lives would do), he would help out with all his might, up to the point at which he might suffer a broken foot (or an equivalent physical harm). Norman has made her no such promise. Now she has to decide whether to turn the trolley onto Dennis or Norman. It seems to me that she may not simply turn the trolley onto Dennis because he has waived his right to suffer a much lesser harm than death if necessary to help. She must again treat their claims not to be killed as equally strong.24 The problem with Tadros’s argument was trying to go from a debt that the falling man owes to Vicky to some reason why it is more justifiable than it otherwise would be to do something that is not about collecting the debt but rather goes well beyond it. The appeal to a lesser-​evil framework at this juncture is the problem. We cannot simply tack on consequentialist reasoning, even when it is sophisticated enough to take into account the notion of morally discounted harms, and hope to retain a clear sense of how competing claims work. We need to start over and ask why else an innocent threat’s claim not to be killed might be affected by his status as a threat.

 One might again object that waiver is not the same as forfeiture. But it is unclear how one could defend that difference if the waiver comes with a debt that is otherwise functionally just like the debt that Tadros has in mind. Steinhoff is again very helpful on this point. Ibid. 24

148  |  The Mechanics Of claims And Permissible Killing in war

3.3. The Dual Significance of Being a Threat The better view is that an innocent threat’s debt to his innocent victim and the weakness of the threat’s claim not to be killed have a common cause; one does not ground the other. The common cause is the obligation to take on some of the burden of being a threat, something one should want not to be. One way to take on that burden is as an agent with a duty to do something to help Vicky, the innocent victim. That duty can give rise to a related loss of a patient-​claim not to be forced to sacrifice what one would have a duty to sacrifice if one had the chance to act as an agent.25 The other way to take on that burden is directly as a patient. Patients do not do anything—​that is the essence of what it is to be a patient—​but the strength of their claims should be adjusted to fairly reflect the impact of their claims on others. This idea that the strength of patient-​claims has to reflect the way those claims would impact others is the essence of the restricting claims principle. A similar thought applies here: fairness calls for treating the claims of innocent threats as weaker than they would otherwise be (if they were the claims of mere innocent bystanders with restricting claims) because otherwise they would take on no responsibility, qua patients, for the fact that their bodies are threatening others. To see the significance of that, assume that a number of innocent victims are threatened by an innocent threat so that it would clearly be justifiable to cause harm to an innocent bystander as a side effect of saving them and then add that an agent could instead harm the innocent threat. If the threat’s claim not to be harmed were not weaker than the bystander’s, their luck would essentially be pooled. That would shift the threat’s bad luck onto the bystander. That is what is ruled out by the autonomy principle, with its implication that we each carry the burdens and benefits of our own luck first and foremost. At the end of the day, I  am not suggesting that the claims of innocent threats have to be so weak that one could justifiably kill one innocent threat to save one innocent victim. That depends on who has the greater burden of bad luck. As noted, it is unclear which of those two has the burden of the bad luck: the victim or the threat. They both have bad luck in a way. But this uncertainty explains why reasonable people can disagree

 This is not to say that this debt could be extracted only by using the person as a means of saving the patient. An equivalent harm could also justifiably be caused as a side effect. See Kamm, 2007, pp. 169–​73 (discussing what she calls the principle of secondary permissibility). 25

Innocent Threats Versus Innocent Bystanders  |  149

about whether killing an innocent threat to save an innocent victim is permissible. It is a feature of the view, not a bug. What I do think should be clear, at the end of the day, is that the claims of an innocent threat are weaker than those of an innocent bystander who has no special responsibility for the threat to the innocent victim.

3.4. A Final Review of the Hybrid Argument and the Infringement Model With this alternative account in view, we can see the mistake in the hybrid argument more clearly. It was a kind of category error, conflating patient-​claims that reflect an agent’s duties with patient-​claims that reflect the burdens of bad luck when thinking about competing restricting claims. The hybrid model simply assumed that having weaker property claims (not to be used) would transfer into having weaker restricting claims (not to be killed). But this was not a valid assumption. Rather, the right way to understand the situation is that, in this particular case (as opposed to the other debt cases we have considered), the same bad luck that grounds the weakened property claim also grounds a weakened restricting claim. Moreover, once we see that, we no longer need to invoke the consequentialist side of the hybrid argument at all. All the moral work can be done, and is more plausibly done, within the mechanics of claims. We can also see now how this category mistake led to the second problem with the hybrid argument.26 That was the objection that the hybrid argument works only if either the threshold for a lesser-​evil justification is implausibly low or an innocent threat owes an innocent victim a sacrifice that is implausibly close to the sacrifice of his life. The hybrid view is driven to this dilemma because it assumes that the innocent threat’s claim must be the claim not to be used as a means as that is the claim that corresponds to an agent duty. If we recognize instead that the innocent threat’s claim is a restricting claim, not a property claim, then it is not so implausible that it could be weaker than the innocent victim’s claim to be saved. Finally, we have now completed the argument against the infringement model. In Chapter 4, section 2.2, I noted that the infringement model seems unable to account for the fact that the claims of innocent threats and innocent victims seem to be roughly equal. Instead, it seems to imply that there is a normative gulf that arises between those cases in which threats have forfeited their right not to be killed (in those cases a great many  See § 3.2.2.

26

150  |  The Mechanics Of claims And Permissible Killing in war

threats may be killed to save an innocent victim) and those cases in which they have not forfeited their right not to be killed (in those cases, a number of victims would have to be threatened for it to be permissible to infringe a threat’s right not to be killed). I also noted that there was one way the infringement model might seek to bridge that gulf: the hybrid model just discussed. As it fails, I conclude that the infringement model has no way to bridge that gulf and is thus substantively implausible.

4. Objections and Replies 4.1. The Nature of the Claims of Innocent Threats The conclusion to the previous argument is that it is easier to justify the eliminative killing of innocent threats than the side effect killing of innocent bystanders. But one might object that this argument presupposes that the competing claims in Falling Man are relevantly like the competing claims of an innocent victim and an innocent bystander—​restricting claims. One might doubt that. After all, an agent would eliminatively kill a threat, and that seems to involve a kind of targeting not present in killing a bystander. Thus, it is at least fair to raise the question, Is an innocent threat’s claim a kind of restricting claim, such that this framework can be brought to bear on his claim as well? If not, if the claims of innocent threats are more like property claims, then even if they are slightly weakened by the duty to make a small sacrifice (one much less than sacrificing their own lives), they may still outweigh both the claims of innocent victims to be saved and the claims of innocent bystanders not to be killed. A good way to see that an innocent threat’s claim not to be killed is indeed a type of restricting claim is to recognize that an innocent threat is like a passenger, helplessly riding along on a threat that happens to be his body. His case is effectively the logical limit of a series of cases that start with him riding along on a much larger threat to which he contributes only a marginal amount of mass and then shrinking the other mass down to his clothes. In a case in which an innocent threat contributes only a marginal amount of the total threatening mass, he is fundamentally just a patient along for the ride, and his claim is clearly fundamentally like that of an innocent bystander with a negative restricting claim. But there is no reason why his claim should undergo a profound qualitative change as his mass becomes a larger and larger portion of the total mass that threatens the innocent victim. His claims should change in a way: as he becomes a larger

Innocent Threats Versus Innocent Bystanders  |  151

portion of the threat, he should have more of an agent duty to do something about the threat and, as a result, a correspondingly weaker property claim not to be harmed. Likewise, as he becomes a larger portion of the threat, his restricting claim not to be killed should progressively weaken relative to that of an innocent bystander. But even so, his claim not to be eliminatively killed is essentially like an innocent bystander’s restricting claim not to be killed, though weaker insofar as he also has to own the bad luck of being a threat.27 It makes absolutely no sense for his restricting claim to transition, at some point, as the percentage of the total mass that is his goes up, into a property claim.

4.2. A Temporal Objection One might also object that my argument with regard to the nature of an innocent threat’s claim not to be eliminatively killed has implausible implications for the permissibility of using past innocent threats as a means of rectifying damage that they have caused. Jeff McMahan makes the argument this way: It is hard to see how this mere difference in timing—​the difference between having caused a present threat and causing a present threat—​could by itself make a decisive moral difference, making it permissible to kill one who is the present cause of a present threat . . . when it is clearly impermissible to kill one who is the past cause of a present threat.28

McMahan is clearly right about one premise: it would be impermissible to lethally use the fallen man (past tense of the falling man) as a means simply because he had inflicted harm on Vicky. Imagine that he had fallen on her and crushed her ribs so that her heart has been badly damaged and she will soon die if she does not receive a heart transplant. Furthermore, imagine that his heart could be used to save her life and that there is no other heart available to save her life. The fallen man should not have to sacrifice his heart for Vicky. But there is nothing paradoxical about taking this position while also holding that he may be killed to prevent him from killing Vicky—​a   This argument echoes the core of the argument that Michael Otsuka originally used to argue that it was impermissible to kill an innocent threat in defense of an innocent victim. See Otsuka, 1994, p. 85. 28  McMahan, 2002, p. 406. Others seem to find this argument convincing. See Draper, 2016, p. 72n5; Haque, 2017, p. 67. 27

152  |  The Mechanics Of claims And Permissible Killing in war

position that I think might be true. Rather, it simply reflects the different justifications one would be using for killing him and the different claims that he would have in each situation. The falling man can be eliminatively killed. His claim not to be eliminatively killed is fundamentally like the restricting claim of an innocent bystander, only somewhat weaker. The fallen man can no longer be eliminatively killed. He can only be killed as a means of saving Vicky. His claim not to be used as a means is a property claim. Insofar as he has a duty to sacrifice himself for Vicky’s sake—​a duty grounded in the harm he has non-​culpably inflicted on her—​his claim not to be used is correspondingly weakened. But if we assume that he does not have a duty to sacrifice himself for her, then it is reasonable to assume that he may not to be used as a means of saving her if so using him would cost him his life. In the end, McMahan’s argument, by focusing on the temporal dimension, loses track of what really matters: the nature of the claim at issue. That often changes from one time to another, but there is nothing paradoxical about that. It simply reflects changes in the causal paths that are open to the agent.

Innocent Threats Versus Innocent Bystanders  |  153



7 From Innocent Threats

to Noncombatants Innocently Aiding Unjust Combatants

T

his chapter is the second part of the two-​part argument that there is no fundamental, principled reason why noncombatants who help unjust combatants to wage war may not be eliminatively killed. Building on the prior chapter’s argument that it is easier to justify eliminatively killing an innocent threat than an innocent bystander, I argue that it should be at least as easy—​if conditions of necessity and proportionality can be met—​to justify killing noncombatants who help unjust combatants to wage war, even if they act non-​culpably.

1. Overview of the Terms and the Argument I proceed in four steps. First, I consider non-​culpable aggressors, people who intentionally threaten harm but who are non-​culpable in failing to recognize that they have no right to do what they do. Then I consider a category of agents that I call innocent threateners. Agents in this category do not intentionally use force against innocent victims; they may, in fact, have no idea that they pose a threat. Nonetheless, they act in ways that threaten harm to others.1 I argue that it is easier to justify eliminatively killing innocent aggressors than innocent threats, our baseline case from the previous chapter. I then argue that the claims not to be killed of innocent threateners fall in between those of these first two types of agents.  Some would include a separate category for innocent aggressors—​defined as people who are acting but not as responsible agents (e.g., someone who is temporarily insane or a small child). See, e.g., Otsuka, 1994, p. 91. I think they raise no new issues and would treat them like innocent threats. 1

Both non-​culpable aggressors and innocent threateners can threaten others directly or through the intervening agency of others.2 I consider the difference that is made by the intervening agency of others and conclude that indirectness makes a difference only when those who threaten indirectly face the alternative of bearing some cost to preempt the misbehavior of others. This is not the case when dealing with noncombatants who aid or abet unjust combatants but who do so non-​culpably because they are non-​ culpably ignorant either of the fact that they are aiding unjust combatants or of the fact that the combatants they aid are engaged in unjust actions. Finally, in step four, I consider the difference made when an individual’s acts do not, by themselves, threaten harm even though her acts, in combination with the acts of many others, do threaten harm. I conclude that the fact that one agent is acting with—​though not necessarily in joint intention with—​others does not make it fundamentally harder to justify the eliminative killing of indirect innocent threateners; it only raises the cost in a way that is relevant to necessity and proportionality. In sum, it is actually easier to justify the eliminative killing of noncombatants who aid unjust combatants than it is to justify their killing as a matter of collateral damage. That does not mean that it is justifiable, all things considered. But contrary to the view of most current writers in just war theory, I  conclude that questions of their liability to being eliminatively killed through acts that forfeit their claim not to be eliminatively killed do not present an important limit on the justification of their being eliminatively killed.

2. Non-​Culpable Aggressors It should be easier to justify the eliminative killing of non-​ culpable aggressors than innocent threats. This is because non-​culpable aggressors are more responsible for their acts than innocent threats are for the motions of their bodies. Non-​culpable aggressors have to own the risk that they will cause harm without having the right to do so. I make this point not to suggest that they have forfeited their right not to be killed if killing them is necessary in order to save innocent victims but to establish that their

 Innocent threats, who threaten only through the involuntary motions of their bodies, could conceivably do the same, by, for example, giving a potential victim a reason to fire a weapon that would then endanger third parties. Such incidents would be exceedingly rare. More importantly, I think the moral treatment of indirect threatening developed in this chapter can be straightforwardly extended to such cases. 2

From Innocent Threats To Innocently Aiding Unjust Combatants  |  155

claims not to be killed should be weaker than those of innocent threats because they have to take even more responsibility for what they do. To illustrate the point, recall Hologram, in which Polly projects a holographic gun onto Howard’s hand, Diane takes the gun to be real and reaches for her own gun to shoot Howard in self-​defense, and Howard sees what is going on and pulls his gun to shoot Diane in self-​defense. Diane’s mistake, when she takes Howard to be aggressing against her, is reasonable; and her aggression against him is non-​culpable, given what she reasonably, though mistakenly, believes. She is, therefore, a non-​culpable aggressor. But given the facts, she threatens to kill Howard without having any right to do so. Is Howard permitted to kill her in self-​defense? If Diane’s claim not to be killed were as a strong as that of an innocent threat, the answer would be unclear. But the falling man is not even acting, and yet he has to take some ownership of what his body is doing. Surely Diane has more reason to take ownership of her choice, even though it reasonably seemed to her the right choice to make. If she has to take more ownership of the luck of her choice, then it would seem that she should own the lion’s share of the bad luck, as between her and Howard. Yes, he still has the bad luck to be threatened. But recall that claims not to be killed are not inherently stronger than claims to be saved; rather, they are normally stronger because it is normally the case that the bad luck in a choice situation belongs first and foremost to the person who needs to be saved.3 In this case, however, the bad luck arguably belongs first and foremost to Diane, the agent who has the bad luck to have made a choice that, as a matter of fact, she had no right to make. Thus, her claim not to be killed should be weaker than Howard’s claim to be saved (by himself or another). What I have just said is not offered as a knock-​down argument in favor of the permissibility of eliminatively killing a non-​culpable aggressor to save a single innocent victim. It would not be ridiculous to hold the position that it is primarily the bad luck of the innocent victim to be threatened and that the normative force of posing a threat in the form of non-​culpable aggression is insufficient to permit the innocent victim or a third party to shift the innocent victim’s bad luck onto the non-​culpable aggressor. But the fact that we are normatively more connected with and responsible for our choices than for the passive motions of our bodies suggests that it should be at least somewhat easier to justify the eliminative killing of a non-​culpable aggressor than of an innocent threat.

3

 See Chapter 3, § 1.3.3.

156  |  The Mechanics Of claims And Permissible Killing in war

3. Direct Innocent Threateners Turning now to direct innocent threateners, they are a sort of hybrid between innocent threats and non-​culpable aggressors. To see this, consider the following. Crosswalk: Wendy is about to press a button to get a walk sign at a cross walk, but something has gone wrong, and though Wendy has no forewarning of this fact, the button will send out a signal that will cause an autonomous vehicle’s navigation system to go awry, thereby causing the vehicle to hit and kill an innocent victim.4

Wendy is like a non-​culpable aggressor in that she makes a nonculpable choice to act and, as a result, if she is not stopped, some innocent victim will die. She is also, however, like an innocent threat in that she does not aggress. I believe her claim not to be eliminatively killed is therefore of intermediate strength: weaker than that of an innocent threat but stronger than a non-​culpable aggressor’s. One might seek to minimize the significance of Wendy’s act, and thereby assimilate her case to that of an innocent threat, by pointing out that she has no reason to think that she might harm someone at all. By contrast, Diane is trying to harm Howard, and she should know that there is some risk that she is mistaken in thinking she has the right to do so. But we should inquire into why that difference might be significant. It is not because Diane necessarily knows that she risks unjustly attacking Howard. She might fail to recognize that there is any risk that he is not a culpable aggressor. Thus, she might be no more aware that there is a risk that she threatens a harm without a right to do so than Wendy. The essential difference between their cases is that Diane knows she will, if not stopped, harm Howard; Wendy does not know that she will harm anyone. It is not clear, however, why this should make it significantly easier to justify the eliminative killing of Diane than Wendy. Jonathan Quong attempts to explain why it is qualitatively easier to justify the eliminative killing of innocent aggressors like Diane than innocent threateners like Wendy by appealing to the idea of a justified practice in which “each person’s claims or interests have been fairly taken

 This is my updated version of a case that goes back at least to Judith Jarvis Thomson’s, Day’s End. See Thomson, 1990, p. 229. 4

From Innocent Threats To Innocently Aiding Unjust Combatants  |  157

into account.”5 He illustrates his notion of a justified practice by considering Conscientious Driver. Quong reasons that as a society we think the benefits of driving outweigh the risks, and he thus concludes, contra Jeff McMahan, that someone who conscientiously drives and nonetheless finds herself threatening another should not be held liable to being eliminatively killed. If a conscientious driver may not be eliminatively killed, despite being a threat to the life of an innocent victim, then clearly Wendy, whose practice of pushing the button at a crosswalk is even less threatening and equally embraced by society, may not be killed. The contrast, according to Quong, is with Diane, who knows, or ought to know, that she is taking a “gamble with the moral rights of others” in the sense that she “know[s]‌(or ought to know)” that what she does “would be wrong but for the assumption that [he] lack[s] important rights.”6 This is superficially appealing, but it is triply implausible when examined carefully. First, it has implausible implications in particular cases. As Michael Otsuka points out, Quong’s view implies that a driver should be treated no differently from a helpless child left riding in the car after the driver has somehow been thrown from it. But they are different. The child is essentially, given how little her mass contributes to the threat, just an innocent bystander aboard the car and may not be killed to save an innocent victim. The driver seems to be in a different position; it is easier to justify eliminatively killing her because of her choice to act in a way that she knows (or should have known) endangers others. Second, Quong wrongly assumes that an act that is justified at one level of description is justified at all levels of description.7 Yes, drivers who drive conscientiously are acting as they have a right to act qua drivers. But that does not mean that they have a right to do what they do on each occasion when they drive conscientiously. The fact that a driver does not intend to harm an innocent victim explains why her act is an “accident.” But the fact that she now threatens an innocent victim means that she made some choice that, from a fact-​relative point of view,8 she had no right to make.

 Quong, 2012, p. 66.  Ibid. 7  The notion of levels of description is different from the point I made in Chapter 1, that an act that is justifiable in some way is justifiable. For an act to be justifiable in some way, it has to be justifiable in its full specificity as an act. If some extra fact would show that it is unjustifiable, then that act is unjustifiable. 8  See Chapter 2, § 2.2 5 6

158  |  The Mechanics Of claims And Permissible Killing in war

Third, it is not so clear why society should be less supportive of the practice of defending oneself when one has good reason to believe, as Diane does, that doing so is necessary. Not only is doing so not blameworthy, overall, but the practice is likely to save innocent lives. In that way, it might be even more justifiable than conscientious driving. There are two reasons Quong’s account goes so wrong. First, he is looking to explain and defend the right of self-​defense against certain kinds of threats by looking to explain why they might be held to have forfeited their right not to be killed. Second, he uses a notion of forfeiture that is implausible. Insofar as he says that someone like Diane should be held to have forfeited her right not to be killed, his position is too quick to find forfeiture. And insofar as he says that a driver may not be eliminatively killed, his position is too limiting. The solution, again, is to use a high threshold of culpability for forfeiture and look for reasons other than forfeiture to explain why self-​defense is justifiable against non-​culpable aggressors and innocent threateners. In all of these cases, there is good reason to say that the agent has to take ownership of the bad luck of having chosen to do something that threatens another when she had no right to threaten in that factual context. Arguably, the cases fall on a spectrum: Diane, choosing to aggress, is most responsible for her choice if it proves to be wrong; a conscientious driver, choosing to impose risks, is somewhat less responsible; Wendy, who acts but has no idea she imposes risks, is less responsible still. But even Wendy, who merely chooses to act, should have to own the bad luck of being a threat more than an innocent threat. And thus, we should accept that it is easier to justify eliminatively killing a direct innocent threatener than doing the same to an innocent threat, which, in turn, is easier to justify than killing an innocent bystander as a side effect.

4. Indirectly Threatening Harm The discussion so far implies that eliminatively killing those who directly threaten harm, without having a right to do so, is permissible when the ratio of those killed to those saved is at least close to even. But to see how this conclusion applies to noncombatants who aid unjust combatants in killing innocent victims, we need to turn to considering the difference made by being an indirect innocent threatener. Many think that the intervening agency of another can make it harder to justify the eliminative killing of indirect innocent threateners compared to the eliminative killing

From Innocent Threats To Innocently Aiding Unjust Combatants  |  159

of otherwise similarly placed direct innocent threateners. But I will argue this is true only in a limited range of cases that do not include most noncombatant innocent threateners.9 This is a complex argument, however, so I break this section into three subsections.

4.1. An Initial Argument That Indirectness Does Not Matter To start our exploration of the relevance of indirectness, consider a case in which the indirect agent acts culpably: Taunting: Tanya knows that her colleague Brett is morbidly embarrassed by his receding hairline, and she believes that if teased about it he will fly into a murderous rage. She also knows that he keeps a loaded gun in his desk drawer. Nevertheless, she is about to tease him about it. If she does, he will pull out his gun and start to shoot. Anticipating his reaction, she will duck behind a filing cabinet. She will survive unscathed, but he will shoot and kill three innocent victims in the crowded office where they work.

David Rodin, from whose work I  take this case, plausibly claims that Tanya “do[es] not have the same responsibility as the gunman.”10 Nevertheless, if Angie, a third party, could foresee that three innocents would die if Tanya were allowed to tease Brett and if Angie knew she could not kill Brett in time to save the innocent victims, it would seem that she would be permitted to eliminatively kill Tanya before she taunts Brett. Now, consider a variation on the case. Change two features of the case. First, substitute a mechanism for Brett. Suppose that if Tanya presses a button, a gun would pop out of a desk and start firing, killing three innocent people. That change by itself turns Tanya from an indirect to a direct threat to the three innocent victims. Now add a second change. Suppose that Tanya is an innocent direct threat to the three innocent victims. She wants to turn on a light and reasonably thinks the button is a light switch. In that case, her situation is very much like Wendy’s. If I  was right in section 3 that it is easier to justify eliminatively killing an innocent threatener like Wendy than an innocent bystander and if an agent may kill an innocent bystander to save three or more innocent victims, then it should   For a more general argument that the wrongful agency of others, whether intervening or not, cannot make it permissible for an agent to do what she could not permissibly do if the choice had arisen in some other way, see Tadros, 2016b. Tadros’s nearly complete rejection of the relevance of intervening agency goes too far, as I argue in Chapter 9. But I think much of his argument succeeds. 10  Rodin, 2014, p. 83. 9

160  |  The Mechanics Of claims And Permissible Killing in war

follow that it is permissible for Angie to kill Tanya to prevent her from pressing the button that will kill three innocent victims. Now, let us put Brett back in but keep the culpability essentially out. Suppose Tanya did not know that taunting Brett would cause him to go into a murderous rage; she thought he would just make a nasty face at her.11 But suppose that Angie knew that he would go into a murderous rage. Angie sees Tanya warming up to the taunt and knows that the only way to stop her is to eliminatively kill her. Admittedly, it’s harder to stom­ ach shooting her in this case than in the previous case. But it’s worth going slowly to be clear what that difference represents. Perhaps the thought is that Brett carries some of the blame. But if Brett cannot be shot to save the innocent victims, then that is as irrelevant in this case as in the original case. The relevant question is, Why should the innocent victims have to die if Brett would play a role in causing their death, even though they may be saved if a mechanism would take his place in the causal order? Why should his responsibility as a free agent—​assuming that he is not so blinded by rage as not to count as a free agent—​imply that the innocent victims have to die? Once we are clear that that is the issue, keeping in mind that in both cases the only way Angie could save them is by shooting Tanya, it should seem quite peculiar and implausible to say that if another agent, rather than a mechanism, is proximately connected to their death, then Tanya may not be killed and the others may not be saved. In other words, once we are clear about the issue, Tanya’s being an indirect innocent threatener seems to make her eliminative killing no less justifiable than if she were a direct innocent threatener.

4.2. Objection and Reply One might object that there are other cases in which it seems that intervening agency makes a difference; and thus, unless there is some way to distinguish those cases from the non-​culpable version of Taunting, we have reason to doubt what I just said about that case. To see the problem, consider the following pair of cases.

 If Tanya taunts Brett, even if she has no idea that he might “go ballistic,” she is more culpable than if she innocently presses a button like Wendy. One shouldn’t taunt others. She is nevertheless still essentially innocent for purposes of my discussion; that sort of teasing is not the kind of culpable act that could cause one to forfeit one’s claim not to be killed—​unlike culpable Tanya. Thus, liability is removed from the discussion, and we are left with a non-​liable, indirect innocent threatener. 11

From Innocent Threats To Innocently Aiding Unjust Combatants  |  161

Coercion I: Alvin threatens to kill Barbie unless Barbie kills Ken with Barbie’s long-​range rifle. Alvin cannot kill Ken himself or by means of anyone else except Barbie. Neither Alvin nor Barbie has a right to kill Ken. Coercion II: Alvin threatens to kill Carla unless Carla gives him her long-​ range rifle so that he can use it to kill Ken. Alvin cannot kill Ken without Carla’s rifle. Neither Alvin nor Carla has a right to kill Ken.12

Assume that Amelie cannot kill Alvin to save Ken, but she can kill Barbie or Carla and thereby save Ken. Kai Draper claims that the eliminative killing of Barbie seems justified in Coercion I but that the eliminative killing of Carla does not seem justified in Coercion II.13 I agree with his claim about Coercion I. Barbie would kill Ken as a means of saving herself from Alvin. She may not do that. Her attempt to do so would be culpable, and she may be eliminatively killed to protect Ken from her culpable, wrongful action.14 Coercion II is less clear. Certainly, there seems to be some difference between the cases. The questions are, What is the relevant difference? And does the difference make it impermissible for Amelie to kill Carla to save Ken? One might try to distinguish Carla from Barbie by pointing out that Carla would not kill Ken to save herself; she merely enables someone else to do so. But is that difference capable of holding much moral significance? No. Carla still has to choose to endanger Ken to save herself. Imagine that Alvin was working with someone else, an expert sharpshooter who needed a rifle. Suppose Alvin said to Carla, “Give my friend here your rifle, or I’ll shoot you with my pistol.” Clearly that is no different from demanding that Carla give him (Alvin) her rifle. But now how is that different from a range of other mechanisms he might demand she employ to kill Ken: “Release the lion!” “Unleash the floodwaters!” “Shoot him!”? Why is “Give my friend your rifle!” meaningfully different? I can see no reason it would be. No matter the means chosen, Carla would be choosing to endanger Ken as a means of saving herself. And the fact that she would be merely enabling another agent does not seem morally relevant either.15  This formulation of the cases closely tracks that given by Haque, 2017, p. 75. The cases originate, however, with Draper, 2016, p. 207. 13  Draper, 2016, p. 207. 14  I return to this sort of case and explain more carefully why Barbie may not kill in a case like this in Chapter 8. 15  See also Tadros, 2018, pp. 413–​415, where he persuasively argues (a) that it is hard to distinguish enabling from causing in the first place and (b) that even if the distinction can be drawn, it does not affect the justifiability of eliminative killing. 12

162  |  The Mechanics Of claims And Permissible Killing in war

If there is some relevant difference, it would seem to be that another agent is the direct cause. Adil Haque might be read to embrace this last thought. He says that “the duty not to enable others to kill is less stringent than the duty not to kill.”16 This sentence invokes both the notion of enabling and the notion of intervening agency. But there’s reason to think that he thinks the latter is what is doing the moral work: he argues that Carla would not kill Ken directly; it is Alvin who would kill Ken directly. Carla’s “relationship with the unjust threat to V is mediated through Ken’s intervening agency.”17 The question is, Is there something about the fact that Alvin’s agency would be directly responsible for Ken’s death that should allow Carla to play the role of enabler? The most obvious thought is that Carla’s choice to give Alvin her rifle does not simply make it the case that Ken will die; Alvin still has to choose to shoot Ken, and he could very well decide not to. But the reasonable response to this is to say that Carla should treat Alvin as a more or less reliable mechanism. We do this with each other all of the time. We give mail to letter carriers, confident that they will send it where it needs to go; we drive through green lights, confident that drivers who could hit us will instead stop at the red light; we hand money over to store clerks, confident they will give us appropriate change, etc. So, suppose that Carla is equally confident (and rightly so) that if she gives Alvin her rifle, he will shoot Ken. Even though she does not share his ends and is in no sense working “with” him in some sort of conspiracy to shoot Ken, it seems to me that she cannot plead that his free choice cuts off her responsibility for Ken’s death. Yes, his free will explains why he is responsible, but responsibility is not in short supply. She too can be responsible. And if she could be shot to save Ken if she were loading a mechanism that would kill Ken if she is not stopped, then we need something better than Alvin’s free will to explain why she should be allowed to load him up to kill Ken. In sum, so far, while Coercion II does seem, intuitively, different from Coercion I, the intuition still seems ungrounded. I  offer, therefore, one last hypothesis to explain it:  squeamishness. We are more repulsed by the thought of directly killing. That makes us think that someone who directly kills has acted worse than someone who enables another to kill. But while squeamishness has its place in moral thinking—​as a basis for

 Haque, 2017, p. 75.  Ibid. For a similar, if more extreme, view, see Kamm, 2010, p. 715. Tadros, 2016b, p. 122, lists others who take a similar view, including David Rodin, Jeff McMahan, and Helen Frowe. 16 17

From Innocent Threats To Innocently Aiding Unjust Combatants  |  163

an agent’s claim not to have to do something she finds morally repulsive, even if doing it would respond to the balance of patient-​claims on her—​it would be misplaced to rely on it here where the question is whether the patient’s (Ken’s) claim not to be killed is strong enough to prohibit Barbie or Carla from saving herself from Alvin. If it is that strong, then Barbie’s squeamishness is irrelevant. If it isn’t, then Barbie’s squeamishness would at most give her a reason not to save herself; it would not make saving herself impermissible. In neither case can it explain why Barbie would act impermissibly while Carla would act permissibly if either helped Alvin achieve his goal of killing Ken. I conclude that we are left where I  said we were after the prior subsection: intervening agency does not make eliminatively killing an indirect threat harder to justify than eliminatively killing a direct threat, all else equal.

4.3. A Way That Indirectness May Matter There is one way in which intervening agency can make it the case that an agent has a right to do something, even though it results in another agent wrongly threatening a victim. Agents have a strong claim not to have to sacrifice themselves for the sake of potential victims. If A says to B, “Allow me to hurt you, or I will hurt others even worse,” then A is effectively demanding that B sacrifice herself to save those whom A will otherwise harm even worse. B has a strong agent-​claim not to have to make such a sacrifice and may therefore have a right to fight off A. In doing so, she is effectively refusing to be sacrificed for their sake. The harm to A’s victims is something B has a right to bring about, through A’s agency, because B’s bringing it about is just the flip side of her refusal to sacrifice herself for their sake.18 Importantly, this is not something that arises only in the context of intervening agency. It is fundamentally an extension of the claim to be free to duck (i.e., to refuse to be a threat absorber) and thereby to pass a threat on to others. The thing one ducks can be an intervening agent, an animal that is not a moral agent, or an inanimate object. Thus, while this is a way that intervening agency can matter, it is not a way that intervening agency uniquely matters. Moreover, this kind of intervening agency is not at stake in the version of Taunting with the minimally culpable Tanya or in Coercion II, with  I take up this idea in some depth in Chapter 9.

18

164  |  The Mechanics Of claims And Permissible Killing in war

Carla. Minimally culpable Tanya is not even aware of the harm that Brett might cause, and it is not as if he said to her, “Allow me to do X to you, or I will shoot others.” Carla’s case comes closer to fitting the pattern but still does not quite fit it. Alvin does say to Carla that he will do something to harm her if she does not help him. But he does not demand that she allow him to shoot her as the price of his not shooting Ken. He demands that she help him shoot Ken or he will shoot her. She does not have to suffer being shot by Alvin as a means of protecting Ken. She can protect Ken directly by not arming Alvin. She would get shot as a consequence of not helping Alvin, but that is no different from Barbie’s case in Coercion I. Thus, again, this model for intervening agency cannot help establish the difference that Haque and Draper claim exists. Instead, I  conclude that intervening agency generally makes no difference to the justifiability of eliminative killing.

5. Threatening Only in Numbers The argument so far has shown that even non-​culpable noncombatants who aid unjust combatants may, in some cases, justifiably be eliminatively killed like direct innocent threateners. But one might think that it is only in the rarest cases that noncombatants ever make a meaningful difference, as individuals, to the ability of unjust combatants to kill innocent victims. Perhaps the most brilliant physicists and engineers can help to build weapons that enable unjust combatants to kill more effectively than would otherwise be possible, and thus eliminative killing them would effectively eliminate a threat. Perhaps there are others who likewise have irreplaceable, or at least difficult-​to-​replace, skills that meaningfully help unjust combatants: the best-​connected financiers, the most rabble-​rousing populists, etc. But the typical worker in a munitions factory, oil refinery, or steel mill, if she is killed, will almost certainly be replaced by another. Thus, killing such a worker is pointless; it would not actually eliminate the threat posed by unjust combatants. It therefore cannot be justified as an “eliminative” killing. It is rather a superfluous killing. The obvious rejoinder to this objection is that if country A were fighting a just war against country B and it killed enough workers in munitions plants, steel mills, oil refineries, etc., of B, that would have a tendency to eliminate the capacity of these industries to produce the weapons and the means of using weapons that B’s unjust combatants rely on. In that case, the question of justifiability would turn on whether A would have to kill a

From Innocent Threats To Innocently Aiding Unjust Combatants  |  165

disproportionate number of people to undermine the ability of B’s unjust combatants to wage war. If not, then we have so far seen no reason why it should be banned. There are two responses to this obvious rejoinder that I now want to address. First, one might object that such eliminative killing would, as a matter of fact, lead to terrorism. Here is why. Suppose that A is willing and able to kill 10,000 noncombatants in B to substantially undermine B’s war effort, and suppose that doing so is baseline proportionate. Then consider Frances Kamm’s principle of secondary permissibility. According to that principle, it is sometimes permissible to perform an act “that would not have been permitted if it were the only act one could perform, when this alternative act is the one, of all those that could reasonably be done, that minimizes the harm that would permissibly have been done to the same person.”19 Generalizing, it should be equally permissible to perform an act that minimizes harm when doing so is a Pareto superior alternative for those affected, harming some less and no one more. Thus, if A could kill only 100 noncombatants in B and terrorize the other 9,900 workers into staying away from their jobs, that would be permissible too. And given that it would save so many lives, it would seem to be required by the principle of necessity. Thus, we seem to slip from eliminatively killing a large number to terrorism. I want to suggest that this is not a bug but at least a potential feature of the theory. Killing the 100 should not be confused with terrorism. Terrorism is justified by appealing to the good effects of harming some and deterring others as a result.20 The justification using the principle of secondary permissibility is superficially the same but fundamentally different. It presupposes that A is first willing and able to perform a permissible act of eliminatively killing a larger group, and it then relies on the terrorist effect only insofar as that enables the agent to save some of those who she would otherwise kill without killing anyone else. Needless to say, this sort of appeal to the principle of secondary permissibility is rife with practical danger. It is easy to see how a country that lacked either the will or the means to eliminatively kill a large group of others might decide simply to engage in terrorism instead. It is also easy to see how a country might cheat on the analysis and reason like this: we would have to kill a disproportionately large number of noncombatants (10,000) to significantly degrade the enemy’s war-​making abilities, and  Kamm, 2007, p. 170.  See Chapter 1, § 3.1.

19 20

166  |  The Mechanics Of claims And Permissible Killing in war

we cannot justify eliminatively killing that many noncombatants; but we could achieve the same effect by killing many fewer noncombatants (100) and relying on the terrorist effect; it would be permissible to eliminatively kill 100 people to significantly degrade the enemy’s war-​making abilities; so it is permissible to engage in terrorism as long as we kill no more than the 100 needed to significantly degrade the enemy’s war-​making abilities. This form of reasoning is not licensed by the principle of secondary permissibility; it is a terrorist’s cheat. But it would doubtless be a tempting cheat. That temptation is the sort of thing the law of war would have to reasonably assess; if it were too strong, some legal measures should be adopted to help keep the temptation at bay. But for the sake of a first-​order moral argument, the principle of secondary permissibility shows how it might not be a problem to engage in relatively small amounts of terroristic killing as a substitute for larger amounts of eliminative killing—​as long as the eliminative killing was a live, permissible option and those killed by a terrorist attack would otherwise have died in the eliminative killing attack.21 The second objection that I want to address was raised by Haque, who argues that we cannot justify eliminatively killing many people just because of the magnitude of the harm they would cause in the aggregate. He claims that “we lose our rights by our own actions, not by the independent actions of others.”22 He would allow that one could justify eliminatively killing noncombatants who exercise collective agency, jointly aiming to do something such that destroying or disrupting their group would actually suppress the threat they unjustly facilitate. But in other cases his view is: “Since contributing civilians neither jointly pose unjust threats nor jointly enable unjust threats they are not morally liable to defensive killing or to non-​defensive eliminative killing.”23 Haque is correct, I think, about liability to harm; but he is mistaken in his conclusion about the permissibility of eliminatively killing those who threaten only in the aggregate. To see this, consider the following case: Exploding the Well: Ten people are independently visiting a stream that feeds into a well. Each is non-​culpably ignorant of the fact that she has a chemical on her shoes that, when mixed in sufficient quantity with a chemical in the  On first-​order morality and why the law of war might properly deviate from it, see Chapter 2, § 3.2.   Haque, 2017, p.  77. Fabre, 2009, p.  61, makes a superficially similar sort of claim, that each person’s causal contribution must reach a certain threshold. Frowe, 2014, pp.  175–​177, offers a convincing reply. 23  Haque, 2017, p. 78. 21 22

From Innocent Threats To Innocently Aiding Unjust Combatants  |  167

well, will cause the well to explode, killing the 20 people who live near it. If at least eight of these 10 people carrying the chemical on their shoes get to the stream, the well will explode. The only way to prevent eight people from getting to the stream is to kill at least three of the 10.

Note that in this case, no one person does anything that would kill any victims if she alone were acting. It takes eight of them walking to the stream to kill the victims. Moreover, they are not acting in coordination; they have no shared plan. Can the facts about their independent agency really make it impermissible to kill three to save 20? I think not. Suppose they were engaged in the collective action of visiting the stream. Why would that matter? Their collective agency is unrelated to any culpability. If three of them could be killed to prevent eight from reaching the stream if they were collectively acting to do that, then three can be killed in the given case, even though they are not collectively trying to do anything together. Collective action is irrelevant to the permissibility of killing them. What Haque was thinking, I  suggest, is that it would be permissible to eliminatively kill them only if they had forfeited their right not to be killed. If forfeiture requires a certain level of culpability—​as I think and as I assume Haque thinks—​then merely contributing to some bad effect collectively caused is insufficient for an agent to forfeit her right not to be killed. An agent would have to have joined in culpable joint agency—​ think conspiracy—​to have forfeited her right not to be killed. His point is that noncombatants who collectively help unjust combatants normally are not engaged in culpable joint agency to help unjust combatants, and thus they cannot justifiably be targeted for the help they provide unjust combatants. But once one recognizes that rights forfeiture is not necessary for justifiable eliminative killing and that merely being an indirect threat is morally sufficient to reduce one’s claim not to be killed, then one can see that Haque’s argument misses the mark. It is the person’s causal role that fundamentally matters, and causal roles can be established through mere aggregation. They do not depend on joint or collective agency.

6. Conclusion The steps in the argument so far lead to the conclusion that we should consider the eliminative killing of noncombatants who support the acts of unjust combatants easier to justify than the side effect (collateral

168  |  The Mechanics Of claims And Permissible Killing in war

damage) killing of noncombatants. This is because of the following chain of reasoning:  It is easier to justify eliminatively killing innocent threats than it is to justify killing innocent bystanders. It is easier still to justify eliminatively killing innocent threateners than it is to justify eliminatively killing innocent threats. Being an indirect innocent threatener rather than a direct one makes little to no difference.24 And finally, the fact that unjust combatants would often have to be killed en masse to effectively eliminate the threat they pose does not by itself make eliminatively killing them harder to justify, except insofar as it makes it less likely that such killing will be proportional to the good done. To be clear, this is a radical suggestion. It does not simply mean that there is a second justification for killing factory workers who happen to be in a factory if and when a country’s military bombs the factory; it means that it may be justifiable to kill factory workers in their beds or out getting a coffee, just as one might target unjust combatants themselves.25 I do not mean to imply that all or even many noncombatants who aid or abet unjust combatants may be targeted for eliminative killing. In the conclusion to this book, Chapter 10, I discuss a range of reasons why it might be, all things considered, morally impermissible to eliminatively kill noncombatants in most situations. But the point of this and the previous chapter is that there is no fundamental and generally applicable reason not to do so. I close out this chapter with two final thoughts. First, I  want to emphasize that nothing I  have said here would justify punitively targeting noncombatants for what they have done to help unjust combatants. The justifiability of punishment is a separate matter from the justifiability of eliminative killing, which concerns the targeting of ongoing threats. Nor would anything I have said here justify using erstwhile enablers of threats as a means of undoing the threats they have enabled, at least not if the harm to them is substantial. Using as a means, when the harm is substantial, is justifiable only if the person so used has waived or forfeited his right not to be so used. And most noncombatants do nothing so culpable as to make them liable to being used at great cost to them. Second, my focus on the moral significance of being a threat, without worrying about liability to harm, helps to avoid a persistent worry in the just war literature, namely that voting for a government that predictably

 It might make a difference if one had a claim of non-​sacrifice. See § 4.3 and Chapter 8. But I assume that noncombatants lack that claim. 25  See Chapter 1, note 43. 24

From Innocent Threats To Innocently Aiding Unjust Combatants  |  169

would start an unjust war makes noncombatants liable to be killed. Both McMahan and Frowe agree that this is the wrong result, and both have something to say about why their view does not imply it; but both of their views run into problems. McMahan says voters are not liable because that sort of contribution is too trivial to ground liability,26 but Frowe convincingly argues that he relies too much on the size of a person’s causal contribution.27 Frowe uses a slightly higher standard for liability—​she requires at least negligence, rather than mere responsible action. She recognizes that voting for a warmongering government could count as negligent and thus could make noncombatants liable to eliminative killing. But she tries to minimize the implausible implications of that by appeal to the same sort of factors I  endorse in Chapter  10, factors that limit the degree to which it is morally justifiable to target noncombatants who contribute to an unjust threat. Her use of liability suggests, however, that these same noncombatants have forfeited their claim not to be killed rather broadly. Indeed, she accepts the implication that noncombatants “will be liable to harms that are a means of averting threats even if those means are not the most efficient way of averting those threats.”28 This is an even more radical thesis than mine, as it makes terrorist attacks on noncombatants as easy to justify as eliminative killings of noncombatant threats—​an implausible result. One point of this chapter is to say that we can avoid this sort of problem by using the mechanics of claims. We can then leave the threshold for forfeiture high enough that voting for a government that predictably wages an unjust war does not make all who voted for it liable to be killed. I could accept that those who voted for it because it would engage in warmongering should be held liable to be killed, but those who voted for it for other reasons, not believing or at least not hoping that it would wage an unjust war, would not be liable to be killed, even if they were negligent in their assessment of what the government would do. That is to say that I can assume that most people who voted for a government that predictably would start and did start an unjust war are not thereby made liable to be killed. And yet I can handle those cases in which it seems permissible to engage in eliminative killing by recognizing that the strength of a patient’s claim not to be targeted for eliminative killing depends on whether he imposes on others, directly or indirectly, a threat that he has no right to impose.

 McMahan, 2009a, p. 225. See also Fabre, 2009, p. 61.  Frowe, 2014, pp. 175–​181. See also Tadros, 2018. 28  Frowe, 2014, p. 179. 26 27

170  |  The Mechanics Of claims And Permissible Killing in war



8 Negative Agent-​Claims and the Agent–​Patient Inference

T

his chapter’s goal is twofold: to refine the thought, argued for in Chapter 3, that an agent may be partial to herself, when serious harms to others are at stake, only insofar as she is exercising a negative agent-​claim, and to describe the conditions under which having a right to act comes with a claim-​right against others that they not interfere.1 These are connected points because they are both necessary to address a commonly held view, to wit, that agents often have the right to resist another agent who is rightfully engaged in self-​ defense or warfare. The argument proceeds in four parts. First, I argue that the idea of negative agent-​claims should be construed broadly to include claims to act when the reason not to act is that the agent confronts the claims of others that she allow herself to be sacrificed for the sake of others. Second, I discuss when one can make the agent–​patient inference, moving from an agent’s right to act to her having a patient-​right against others that they not interfere with her acting. In this context, I argue that justified competition negates the agent–​patient inference but that competition is justified only in a limited range of situations. Third, I argue that the key to the agent–​patient inference, other than there being no justified competition, is that the agent’s act is justified; it does not require the agent herself to have a stake in the act. Finally, I bring this to bear on just war theory.

 Large sections of this chapter and Chapter 9 are published separately in Walen, 2018.

1

1. From Ducking to Actively Causing Harm The idea that agents have strong negative agent-​claims but no positive ones (when confronting patient-​claims not to suffer significant harms)2 may seem to imply that they could never justifiably do anything that would cause significant harm to another unless the balance of patient-​claims (including their own patient-​claims) called for doing it. But that is too quick. The space of rights cannot require an agent to make substantial sacrifices for the sake of others (unless they have special claims on her), but more generally it cannot require her to serve, at substantial cost to her, as a tool for their welfare. If she needs to act to avoid serving as a tool for their welfare, she has a negative agent-​claim to perform that act. When such agent-​ claims ground a right, I call it the right to resist being sacrificed for others, or the right of non-​sacrifice for short. This extension of negative agent-​ claims reflects the core of the autonomy principle as much as the paradigmatic agent-​claim not to have to act. But to be properly limited, it can justify an agent causing harm to others only insofar as she is contravening claims that she serve as a means of protecting them. I argue for the right of non-​sacrifice first in the context of a case in which the agent acts and thereby allows others to suffer harm. I then argue that the same principles apply when an agent acts and thereby causes others to suffer harm.

1.1. Allowing Harm by Ducking Let us start with an illustration. Cart Avoidance: David is on a crowded sidewalk, and he sees a heavy cart rolling down a steep hill at him. He can tell that if he stays where he is, he will be knocked down and seriously injured by the cart. He can also tell that if he jumps out of harm’s way, two people behind him, strangers to him, will be knocked down and seriously injured by the cart. Moreover, he can tell that if he takes the blow, they will be unharmed. Paula sees that he might jump out of harm’s way and considers blocking him from doing so.

If one thinks that David is not required to interpose himself in front of the cart, it is hard to see how one can deny that he may effectively duck by

2

 See Chapter 3, § 2.1.

172  |  The Mechanics Of claims And Permissible Killing in war

jumping out of the way. But note, if he jumps out of harm’s way, he does something that makes the two behind him worse off. This is not a case of him merely allowing them to suffer harm. He allows them to suffer by removing what protects them: him. If he were removing some object that they had a right to rely on—​say a metal barrier installed by the city—​then he would be impermissibly harming them. What makes his act permissible and what makes us want to call it an “allowing” is that by removing himself he is exercising his right not to sacrifice himself for them.3 If David may jump out of harm’s way, then Paula may not interfere with his act for the purpose of saving the two behind him. She cannot justify blocking him because that justification relies on using him as a means of saving them. One could imagine a case in which she had some other reason to impede his getting out of harm’s way: perhaps to get out of his way, she would have to jump in front of a truck and sacrifice herself. She would have a right to refuse to do that for his sake. Likewise, if her getting out of his way would cause her to push others off a cliff, causing them a greater harm, then she has a justification for refusing to do so. What she cannot do is justify blocking him from getting out of the cart’s path for the sake of the two who would thereby be saved. This is an example of the agent–​patient inference in action: his right to act gives rise to a claim of non-​interference. I also believe she has reason to support his choice.4 That reason is not an expression of the agent–​patient inference as it is not about responding to his right to act. Nevertheless, it is grounded in the same core principles. This can be made clear if we recall the principles developed in Chapter 3, section 4.  The key is getting the various claims of the two, David and Paula, framed in the right way. One might think that they should be assessed like this: David and the two all have positive restricting claims to be saved from the cart. Two positive restricting claims beat one positive restricting claim, all else equal, so the balance of patient-​claims on Paula calls for her to allow him to be hit. But this misunderstands David’s role and the nature of his claims on Paula. To properly understand the nature of David’s claims on Paula, we have to look at the justifications Paula might offer for her choice. Insofar as she is  For a similar analysis, see Alexander, 2016, pp. 261–​262, and Ferzan, 2017.  There are some who argue in a similar case—​one in which one person is in the path of a trolley that would go on to hit and kill five if he is removed—​that the agent should allow one to serve as a means of saving a number of others. Mogensen, 2016, argues for this point explicitly; Segev, 2016, is less certain but thinks that there is no good account of why an agent should help the one. See also Kagan, 1989, p. 164. 3 4

Negative Agent-claims And The Agent–patient Inference  |  173

making a claim not to have to expend her resources (her time and energy) on David’s behalf, his claim is a positive restricting claim that pushes to restrict her freedom not to work on his behalf, and it comes up against her negative agent-​claim. But if it would be easy for her to save him—​suppose all she would have to do is alert him to the danger by crying “look out”—​then her negative agent-​claim loses to his restricting claim that protects his life. The other justification she can offer for not crying “look out” is that she needs him to stay put for the other two to be saved. The problem with this justification is that it presupposes that she is permitted to use him as a tool for saving others, but he is not hers to use that way. Imagine the exchange they might have if they could communicate. She might say, “I think it better that you be hit and seriously injured than that two others suffer that fate.” David could rightly reply, “But you’re using me as a means to save them!” Paula’s best response would be, “Well, I’m not actively using you; I’m merely allowing you to be hit so that they can be saved.” This last move fails to take into account that she would still be justifying her choice by appeal to his utility as a means of stopping the cart and protecting the others. That justification requires him to serve as her tool for saving them, and he has a property claim not to serve as her tool in that way. Her last answer is tempting if one thinks that the distinction between doing and allowing is fundamental. If it were, then one could defend her choice not to warn him by saying that she merely allows him to die and that doing so is justified because it leads to less harm. But the distinction between doing and allowing is not fundamental; it is an ambiguous amalgam of two other distinctions. It has an agent-​centered aspect, given by the distinction between negative and positive agent-​claims; and it has a patient-​centered aspect, given by the distinction between negative and positive patient-​claims. Those need to be pulled apart. Once pulled apart, we can see that her negative agent-​claim is outweighed by his positive restricting claim. And we can see that his patient-​claims are not limited to a positive restricting claim to be helped; they also include a property claim not to be used as a tool for saving the two. That latter claim blocks the claims of the two from ever reaching her; as a result, she lacks the baseline freedom to help them. She might seek to rely on the claims of the two to gain the freedom to use him to save them. In that justificatory context, their claims would function as empowering claims.5 But just as the claims of the five in Massive

 See Chapter 3, § 4.4. Alexander and Ferzan, 2018, pp. 117–​19, rightly take me to task for my view, in Walen, 2014, that in a case like this the two have negative restricting claims that compete with David’s positive property claim. 5

174  |  The Mechanics Of claims And Permissible Killing in war

Man cannot put the massive man in Bertha’s toolkit, so in Cart Avoidance the empowering claims of the two cannot put David in Paula’s toolkit. His property claim over his body defeats their empowering claims which push to put his body in Paula’s toolkit. Again, since she lacks the baseline freedom to use him as a tool for saving them, their restricting claims get no purchase on her at all. Since she lacks the baseline freedom to use him to save the two, the justification that he could be used to save the two fails. And since neither justification works, she has a pro tanto duty to David to save him.6 The primary temptation to think otherwise, it seems to me, reflects the idea that David has an agent-​centered prerogative to favor himself over the two, while Paula has a different point of view and should be neutral between him and the two behind him. But that temptation should be resisted. As I argued in Chapter 3, section 2.1.4, the idea of an agent-​ centered prerogative should be rejected. Here we see another reason why. It sets David in an implausibly antagonistic relationship with the others in the case. It presents the case as if he is entitled to favor himself, while the two are entitled to favor themselves, and Paula has reason to choose to favor the two because they outnumber him. But if this would justify Paula leaving him to get hit, even if saving him would be no personal sacrifice for her, it’s unclear why she may not take their side more actively. Why not block him from saving himself? If he is merely acting on an agent-​centered prerogative to save himself, then, in terms of impartial considerations, he would be cutting against the grain. Clearly, this is wrong with regard to blocking him. I  can see no reason why it is not equally misguided with regard to the choice to let him suffer for their sake.7 It may also help to explain just what the right of non-​sacrifice means if I add a fact to Cart Avoidance. Suppose that if David were to get out of harm’s way, the two behind him would protect three behind them from an equally grave harm. Can Paula now justify allowing David to be hit by the cart so as to prevent the two from serving as a means of protecting the three? It might seem like my account would say yes as it is worse to have two serve as a means than to have one serve as a means. But the answer   Segev, 2016, p.  2058, writes, in response to my view, as developed in earlier works, that “it is doubtful if we should consider not saving a person . . . as using him in the relevant sense.” But he offers no argument for this position. I suspect it is based on the unfounded assumption that using requires acting. But it is not a matter of acting or not; it is a matter of the causal role of another. It is a question of whether he serves as a means by which she can achieve her justifying end or not. 7  I say more about the issue of moral competition in § 2. 6

Negative Agent-claims And The Agent–patient Inference  |  175

is no. What matters is not that some people happen to serve as a means of saving others. What matters is only what has to be appealed to in order to justify Paula’s choice. The reason Paula must take the balance of patient-​claims on her to call for her to help David is that she cannot justify treating David as a means of saving the two. Their claims on her to be saved are failed empowering claims. The fact that they would save three may be a kind of moral bonus if they are hit, but it plays no justificatory role in choosing to save David. The only thing that is relevant to that choice is that David’s claim not to be a means of saving them defeats their empowering claims.8 One might wonder how to square this judgment with the earlier point I made, namely that Paula could justify her choice not to get out of David’s way by reference to the fact that if she did so, she would push two off a cliff. One might ask, Doesn’t his claim not to be used as a means outweigh their claims not to be pushed off a cliff just as it outweighs the claims of the two below him not to be hit by a cart, at least as long as the injuries caused by being hit by the cart are comparable to those that the two would suffer if pushed off a cliff? But this question frames things too much as a consequentialist would, as though his claim not to be used as a means of saving two floats free of the justificatory context in which it arises and simply has weight on Paula. That is not how the mechanics of claims works. Claims take their nature from the role they play in a potential justification for an agent’s act. Paula might try to justify blocking David for the sake of the two behind him, but that justification involves treating him as a means of saving them. It fails. Paula might also try to justify blocking David from moving out of harm’s way because making space for him would cause harm that is at least as grave to two others. In that justification, his relevant claim is a claim to be saved (“make room for me”), and that restricting claim loses to the two restricting claims not to suffer an equivalent harm—​both sets of claims are restricting because Paula has the baseline freedom to move or not move and is normatively

 An interesting problem arises if we iterate on the problem. Suppose that behind the three were four, and so on, until the last row contained a million people. And suppose that at each step it is possible to save those in the front row, exposing those behind them. If Paula had to follow the reasoning described in the text at each step (putting aside her agent-​claim not to have to perform nearly a million acts) but could not save the last row of people, then in the choice to save David she would, in the end, allow a million people to die. No step involves doing the wrong thing, so it is hard to see why she should stop anywhere along the way. But all the steps, taken together, would be wrong. This is surely an instance in which threshold deontology kicks in (see Chapter 4, § 3), and the right thing to say, I suggest, is that she should let David die to save, in some extended sense, a million. I am grateful to Chris Frugé for raising this problem. 8

176  |  The Mechanics Of claims And Permissible Killing in war

restricted only because of the effects that would flow from her moving or not moving. Therefore, that second justification for blocking him can succeed where the other one does not.

1.2. Causing Harm That Is Like Ducking Having argued that an agent may do something that allows harm to befall others if doing so exercises her right of non-​sacrifice, I turn now to arguing that the right of non-​sacrifice does not depend on allowing harm to occur. I do so by describing a case in which an agent can interpose a shield that would deflect a harm from one onto two. Interposing a shield changes the causal relationship between the act and the death of others. Rather than merely allowing the two to die, it actively causes the two to die. Doing so would be impermissible if the threat were merely turned onto two who did not depend on the one being there to absorb the threat—​two negative restricting claims not to be harmed are stronger than one positive restricting claim to avoid the same harm.9 But one can imagine that the two are the very people who depend on the one being used. Then it would seem permissible. Consider the following: Shielding the Massive Man: Mike, a massive man, has been toppled off a bridge into the path of a boulder rolling down a ravine towards two people below. If hit, he will be crushed, but he will prevent the boulder from crushing the two below. He cannot get out of the way, but he can position a shield in front of himself, thereby causing the boulder to bounce off to the side, where it would then continue down the ravine and hit the two below.

It is clear that Mike may position the shield to save himself from the boulder. Again, he is not merely deflecting it onto two others; he is preventing them from relying on him to absorb the danger. It so happens that the thing he does to prevent them from relying on him causes the danger to hit them after having taken a slightly different route. But he has a right to do it all  For a classic deflection case from the legal literature, see Scott v. Shepherd, 96 Eng. Rep. 525 (K.B. 1773), the “squib case.” In that case, a live small explosive device, a squib, was tossed into a crowded market. It was picked up and tossed away twice, and after the second toss it exploded and took out plaintiff Scott’s eye. None of the intermediaries were held liable because they were held to be acting under “compulsive necessity.” I agree that this should generally be excusing—​the exception being for people who are trained to handle situations like this. As regards justification, I agree with judge Blackstone, who dissented in the case. According to him, those who tossed the squib were not justified in doing so. They “had . . . a right to protect themselves by removing the squib, but should have taken care to do it in such a manner as not to endamage others” (p. 527). 9

Negative Agent-claims And The Agent–patient Inference  |  177

the same. Moreover, just as he may save himself, so he may be saved. A third party—​call her Sue—​who had a shield may give it to him despite the fact that doing so will kill two others. In fact, she should give it to him as long as doing so would not impose a grave cost on her. Sue may not justify refraining from giving him the shield by reference to the fact that if he is hit, two others will live. Their claim on her goes via his being used as a means to help them, and she may not take herself to have the use of him for their sake.

2. From Agent-​Rights to Patient-​Rights I have just discussed two cases (with some variations) in which the fact that an agent has a right to do something implies that he has a corresponding patient-​right of non-​interference (at least not if the interference is justified as a form of interference), as well as a claim for help. But it is time now to address head on the question of when the agent–​patient inference is valid. It is not always valid, and the claim that it is valid in any particular instance reflects a substantive moral conclusion about competition and its role in that kind of case.

2.1. The Problem with Moving from Agent-​Rights to Patient-​Rights The need to examine the agent–​patient inference with some care is established by the fact that the Hohfeldian category of a privilege or lib­ erty rights does not imply anything about duties on the part of others.10 Consider three examples:11 A might have the right to try to run a football down the field, and B might have the right to tackle him; C might have the right to set up a business, and D might have the right to set up a competing business aiming to drive C out of business; E might have the right to give a speech, and F might have the right to organize a protest to drown her out. We need an account, then, to explain why these cases should be different from what I said about Cart Avoidance and Shielding the Massive Man. Without an account of why those two cases should be different from the

 See Hohfeld, 1913. The point is emphasized in Moore, 2018. Moore takes this to be a problem for Hohfeldian rights. I think, to the contrary, that it is a virtue of Hohfeldian rights that it leaves this issue open for resolution on substantive grounds. 11  The following is a more extensive treatment of an issue raised in Chapter 4, § 3.1. 10

178  |  The Mechanics Of claims And Permissible Killing in war

cases of sports, business competition, and free speech, one might think that my argument in the previous section missed something important. To accentuate the need for an account of why these cases should be different, consider this case: Trolley Turners: Brenda can save five victims only by throwing a switch and turning a trolley from the path of the five onto another track, where it would then hit a sidetrack man. The sidetrack man also has a switch, and it would do the same thing: turn the trolley from the five onto him.

I take it that the sidetrack man has a right not to turn the trolley onto himself. But I also take it that Brenda may turn it onto him.12 This seems to be a counterexample to the agent–​patient inference. Here the sidetrack man is not even doing anything; he is merely choosing not to do something so as not to sacrifice himself for others, and yet Brenda seems permitted not only not to help him but to work to undermine that very right by killing him in just the way he is choosing not to be killed. My response to this case has two parts. The first part appeals to the notion of responsiveness; the second part deals with the justifiability of competition.

2.2. Responsiveness Responsiveness shows why Trolley Turners is not actually a counterexample to the agent–​patient inference. The explanation of the sidetrack man’s right not to turn the trolley onto himself is that he has a strong negative agent-​ claim not to have to sacrifice himself for the sake of others. But that claim does not govern the choice of someone else with a switch. His claim with regard to what another agent does with her own switch is only a negative, restricting patient-​claim not to be hit by the trolley. That claim is plausibly outweighed by the competing positive, restricting patient-​claims of the five to be saved. Responsiveness is the dog that does not bark in this case. Brenda’s choice to turn the trolley—​should she make that choice—​would not, in the relevant sense, be in response to his exercising his right not to turn it onto

  Thomson, 2008, relied in part on this tension to change her long-​held position and argue that someone in Brenda’s position may not turn the trolley onto the sidetrack man. David Wasserman and I wrote our 2012 article largely to respond to the mistakes we believe drove Thomson to her new position. The current argument brings out new arguments not covered in that paper. 12

Negative Agent-claims And The Agent–patient Inference  |  179

himself. It is responsive in a weak sense: if she saw that he had already turned the trolley onto himself, she would have no reason to throw her own switch; but if she believed he might not throw his switch, then she would have reason to throw hers. Still, her throwing her switch would not be a response to his inaction. A response to his inaction, in the strong sense that I have in mind, would be an effort to make him act. But that is not what she is trying to do. The notion of responsiveness comes into greater relief if we contrast her act of turning the trolley with his possible act of shooting her in self-​ defense. Imagine, that is, that he not only chooses not to throw his own switch but is bent on making sure that she does not throw her switch; he will kill her, if he can, to prevent her from turning the trolley onto him. His shooting her would be a response to her threatening to turn the trolley onto him. Her throwing her own switch is not responsive to his inaction in the same way. To further reinforce the point about responsiveness, let us return to the possibility briefly mentioned two paragraphs back. Imagine a case in which Brenda would have the ability to save five only by causing the sidetrack man to act, even though he has a right not to act. Suppose that Brenda lacked a switch herself but could shoot him with a dart containing a serum that would have the same effect on him as hypnotizing him: it would cause him to do what she tells him to do. Suppose that she shoots him with the dart and then tells him to turn the trolley onto himself. In that case, she would be responding to his right not to act and making him act nonetheless. His right not to be used as a means in this context establishes that she may not do that. That right is the responsive reflection of his own agent-​ right not to sacrifice himself for the five. In other words, it is an example of the agent–​patient inference. His right that she not respond to his choice not to save the others by responsively interfering with it (by forcing him to act) does not, however, give him a right that she not act in ways that will effectively negate his choice. The justifications for the two acts are completely different.13

 An interesting side question arises if she could turn the trolley onto him without using him as a means of doing so, but it would be better for him if he were forced to turn it onto himself—​suppose that he would then be injured but not killed. Frances Kamm’s principle of secondary permissibility suggests that in that instance Brenda would be permitted to make him turn the trolley onto himself. See Kamm, 2007, p. 170. See also the earlier discussion of this principle in Chapter 7, § 5. 13

180  |  The Mechanics Of claims And Permissible Killing in war

2.3. Justified Competition Let us turn now to the role of justified competition. As noted, A  might have the right to try to run a football down the field, and B might have the right to tackle him; C might have the right to set up a business, and D might have the right to set up a competing business aiming to drive her out of business; E might have the right to give a speech, and F might have the right to organize a protest to drown her out. In each case, the second agent has a right to take action directly responsive to what the first agent does, even though the first agent has a right to do it. The question is, Is there a principled way to describe when these cases arise and when such competition is not justified? I think the answer is yes. Moreover, I think that if we examine the reasons why competition is sometimes good or at least justifiable, we will see that they do not extend to allowing competition in the two cases that matter for eliminative killing: the right of non-​sacrifice and the right to respond to the balance of patient-​claims. Responsive competition is permitted, I will argue, only when the justification for the type of activity in question presupposes that it has a kind of value that can be achieved only when competition is permitted. The type of activity will then control the question of permissibility. To see this, let us look at our examples of responsive competition. In sports, competition is often central to the nature of the activity, and the activity itself can be good for health, entertainment, the sublimation of tribalism, and other purposes. These purposes seem to outweigh the reasons to limit or eliminate competitive sports, such as the claim not to be pressured into playing a sport that one does not want to play. Such pressure may be a side effect of permitting the practice to exist; but it should be possible to keep it from being unduly coercive, and the costs to those who want to engage in sports of banning them are, I presume, higher. Framing this in terms of the mechanics of claims, we can model the point by thinking about the claims on relevant governmental bodies, such as legislatures and regulatory bodies charged with supervising particular sports in particular jurisdictions. I assume these bodies have the authority (i.e., the baseline freedom) either to establish the legality of particular sports (presumably as long as certain regulations are adhered to) or to ban them. That makes the competing claims on those bodies restricting. I don’t mean to suggest that these bodies must engage this question for sports to be permissible; the long existence of many sports may provide a presumption in favor of their permissibility. My point is that the possible engagement of legislative bodies provides a concrete way of thinking about how

Negative Agent-claims And The Agent–patient Inference  |  181

to model, within the mechanics of claims, the competing interests that justify permitting or banning a type of activity in a particular society. Even if the legislative bodies have not addressed such issues in a particular jurisdiction, one can ask how they should legislate and get a sense for whether a particular type of activity should be held to be permissible or not. And it seems to me that the balance of claims weighs in favor of accepting the existence of, and regulating, most competitive sports.14 In business, economics tells us that competition leads to better and/​ or cheaper products and services for consumers. And, of course, certain entrepreneurial types will want to be free to set up businesses and to compete to thrive in the marketplace. Again, framing the permissibility of the practice of competitive business in terms of claims on regulatory bodies, the idea is that consumers’ claims to enjoy the fruits of a competitive marketplace and entrepreneurs’ claims to enjoy the fruits of competing in the marketplace, taken together, outweigh any claims that businesspeople might have to engage in business without competition. One might object that business competition need not involve acts that are directly responsive to a competitor; an entrepreneur can try to succeed in business simply by, for example, providing high-​quality goods and services for an attractive price and making that fact known to potential customers. If she succeeds, her competitors may suffer financial losses, but that result need not be an essential part of the justification for her economic activity. At the same time, it is hard to argue that all acts that aim to bankrupt a business owner’s competitors unethical and unjustifiable. Bankrupting her competitors may enable an entrepreneur to expand her business and make a greater profit, a perfectly appropriate goal in the capitalist marketplace. Certain ways of trying to drive her competitors out of business are and should be impermissible, and monopolistic practices should generally be discouraged as bad for the economy. But there is nothing per se morally inappropriate about making business decisions that aim to make more profit by driving out some of the competition.15 Turning to the context of free speech, there are different morally plausible ways to balance the relevant considerations, such as the autonomy interests of speakers, the hearers’ interests in getting information, and the interests all have in sorting truth from falsity, in the production of art, in

 This is not to say that all sports as they now are practiced in particular states are justifiable. For example, American football is arguably unjustifiable given the pressure on children to play and take the risk of concussions that endanger their long-​term mental health. 15  See Chapter 4, note 25, for antitrust limits on this right. 14

182  |  The Mechanics Of claims And Permissible Killing in war

the preservation of civility, and in the prevention of harms that sometimes follow from speech. Rules for free speech generally support speaking in civil ways with those with whom one disagrees. In that sense, they may permit competition over ideas but not competition to shout down another speaker. But in certain settings, such as political rallies, it is at least arguably reasonable for them to permit more raucous modes of interaction, ones that tolerate some speakers seeking to shout down others. In all three contexts—​sports, business, and speech—​and in others as well, where competition is either encouraged or at least tolerated, one agent can justify performing certain acts in ways that are directly responsive to other agents taking rightful action because of the social value—​ which can be understood in terms of claims on the relevant regulatory authorities—​that the activity provides.16 And in those cases, it would undermine the general smooth functioning of the activity to be overly critical of particular instances of the activity. Sure, there should be exceptions when considerations from outside of sports, business, or speech make it morally impermissible to directly interfere with another. For example, in sports leagues for young people and even many for adult amateurs, the importance of preventing a competitor from being demoralized can make it morally impermissible to continue to compete hard and “run up the score.” But in general, if the activity is justified, then competing within it is justified. Note, however, that the range of responsive actions that are permissible is always limited. It is impermissible for an agent to shoot her opponent in any of these contexts. The permissible responsive actions have to fit generalizable rules for competition that further the justifying goals of the activity, and the activity itself has to be justified. Note too that responsive competition cannot be justified by straightforward appeal to the autonomy principle and the idea that we should all be at liberty to lead our own lives. Competition is not like the freedom to do what one wants with one’s own property as long as one causes no serious harm to others. Competition is a distinct permission to responsively engage another agent, trying to prevent her from doing what she has a right to do. That requires a special justification, one that draws on the benefits that are available only if competition is permitted and that is sensitive to the harms that competition can cause. Outside of the limited contexts in

 Once a competive practice is established as justified, an agent’s claim to be free to perform competitive acts should count as a strong negative agent-​claim as against competing restricting claims of competitors. Insofar as the practice can rightly be revised, however, agents may be stripped of these liberty rights. 16

Negative Agent-claims And The Agent–patient Inference  |  183

which competition is permissible, an agent may not act in a way that is directly responsive to another’s permissible action, aiming to interfere with it. The default position for agents acting in a space of rights, a normative space designed to harmonize people living together, has to be support for others exercising their rights, except insofar as an agent has a negative agent-​claim not to have to sacrifice her own ends for the ends of others.

2.4. The Bear and the Plank One might object to my representation of the limited nature of justifiable competition on the ground that it does not fit certain other cases in which it seems that people may compete with each other even though the effect is that the loser will die. Consider this famous example: Running from the Bear: Two campers, Alexandra and Bryce, are confronted with a ravenous bear. As Alexandra grabs her running shoes, Bryce points out that no one can outrun a bear. “I don’t have to outrun him,” she replies. “I only have to outrun you.”17

I share the common intuition on this case, namely that Alexandra is permitted to compete with Bryce to see who can get away from the bear the fastest, even though the loser will die. The question is, How can this be squared with the limits of competition just sketched? No special justification for competition seems to apply in this case. Before trying to explain this case, it is important to note that the restrictions on permissible competition are still quite tight. As the authors who first brought the case into the literature point out, Alexandra may not tie Bryce’s ankles, or knock her “unconscious and throw” her “to the bear.”18 How can we explain why Alexandra may compete with Bryce over who can get away from the bear and yet not let her use whatever means are available to her? I think the answer is that the justification for Alexandra and Bryce trying to outrun each other is not that each is permitted to try to make the other serve as the means by which she escapes. Rather, it is that each is permitted to try not to be the means by which the other lives. In other words, rather than being permitted to responsively compete to use each other (which

 I take this, with only small changes, from Boorse and Sorensen, 1988, p. 115. The example has an earlier history as a joke. 18  Ibid., p. 116. 17

184  |  The Mechanics Of claims And Permissible Killing in war

would presumably license more direct attacks), each is permitted to do her best, without attacking the other, to exercise her right of non-​sacrifice. She may in fact be relying on the other to serve as her means of escape, but she need not justify her act that way. She can justify it simply as an effort to escape and an effort not to be used as the means of saving the other. One might wonder why Alexandra and Bryce should not be required to find some fair method of deciding who should be allowed to run away and who should have to stay behind. As Victor Tadros says, “we have good reason to object to the idea that those who have more power are permitted to use that power against those who have less.”19 Why then allow the fast to use her speed against the slow? Why not require them to flip a coin? Sure, there might be a practical reason not to do so. Flipping a coin might result in a Pareto inferior solution—​say because the bear would be more likely to catch both of them while they sort out who can run away. That would be a good reason not to flip a coin. But if we assume that there is time to do a coin flip without significantly reducing the odds that faster one would get away and without significantly increasing the odds that the slower one would die, then fairness seems to call for it. To see the force of this objection, consider this case: Plank: Two shipwrecked people find themselves on a plank. No one is aware of any reason why one has any priority over the other. The plank does not belong to one of them, nor did one get there, as far as anyone can tell, ahead of the other. However, the plank can only keep one of them out of shark-​infested waters; if both try to hang on, both will be eaten.

Tadros asks of this case, “Is each of [them] permitted to try to throw the other one off?”20 His answer is no. “[They] must find some fair procedure, such as tossing a coin, to determine who is to survive and [they] must submit to the outcome of that procedure.”21 If Tadros is right with regard to Plank—​and I think he is—​then why allow a foot race in Running from the Bear? Why is it permissible to try to outrun a slower person but not to wrestle a plank away from a weaker person? If a fair procedure is called for in one, why not in the other? Or if competition is permissible in one, why is it not permissible in the other?

 Tadros, 2011, p. 208.  Ibid., p. 203. 21  Ibid. 19 20

Negative Agent-claims And The Agent–patient Inference  |  185

The answer is that the cases are deeply different. The two on the plank have equal claim to enjoy the benefit of it, and when people have roughly equal conflicting claims over an indivisible good, the ultimate right to enjoy it should be established with some fair mechanism.22 Fighting, even just trying to wrest the plank from the other’s grasp, is not a fair mechanism. It is a mechanism which favors the strong over the weak. It may become necessary and justifiable to struggle if the other will not use and abide by a fair means of deciding who gets to stay on the plank, for then the other is culpable and has forfeited her equal claim to the plank. But if the other is willing to use and abide by fair means of determining who can stay on the plank, then both are obliged to use and abide by them.23 By contrast, in Running from the Bear, each is simply refusing to sacrifice her own interests for the sake of the other. That is to say that each is exercising a negative agent-​claim not to sacrifice herself for the other. The demand that the faster person submit to a coin flip to decide who gets to run away and who has to stay behind and confront the bear is effectively the demand that the faster person accept a 50 percent chance of having to serve as the means by which the slower person could escape the bear. That is more of a sacrifice than anyone has a duty to make for another who does not have special claims on her. It might seem that the fast one owes a debt of gratitude to the slow one because, but for the slow one, she could not escape the bear. But even if she owes some such debt, it is not enough of a debt to ground a duty to sacrifice herself or even to accept a 50 percent risk of having to sacrifice herself. To see that, change the case so that the bear simply came upon Bryce and was ready to kill her. Had it not been the case that Bryce was there, the bear would have come upon Alexandra and killed her. As it is, however, Alexandra is not threatened. Note, she is equally in Bryce’s debt in this variation. Now imagine that she could throw himself at the bear and save Bryce. Even recognizing that, but for Bryce’s presence, Alexandra would be facing death, it would be a supererogatory gesture for her to throw herself at the bear. It is Bryce’s bad luck that the bear went for her first, and she has to own that luck. Likewise, in the original case, it is Bryce’s bad luck that Alexandra can get away faster. Bryce has no right to  See Broome, 1984. I made this point in Chapter 3, § 1.3.3.  I think the same is true of racing to the plank. If both see it and could get to it in time, there is no reason to let racing determine who gets it. It is different if one gets there first, without racing to deprive the other of the resource, and then starts to rely on it. At that point, her luck vests and her claim shifts to a negative claim not to be deprived of the resource. See Chapter 3, §§ 1.3.3 and 4.3, for discussions of vesting luck. 22 23

186  |  The Mechanics Of claims And Permissible Killing in war

demand that Alexandra at least offer her a 50 percent chance of being the one who gets away. In the end, the puzzle presented by Running from the Bear reflects the fact that it looks like more of a competition case than it is. There is a competitive element because the faster one has the ability to get away and the effort to get away faster works just like a competitive race. But the faster one may not responsively act to ensure that the slower one plays the role of threat absorber. Each may merely seek to avoid playing that role by running away. If the slower one happens to lose, that is simply her bad luck.

3. Justified Acts and the Agent–​Patient Inference In this section I want to refine what is presupposed by the agent–​patient inference. In the previous section, I argued that the case has to be one in which responsive competition is not justified. In this section I will argue that the other key is that the agent’s act is justified; she does not have to have any interest in the act herself. To make my argument, I will start by revisiting Trolley Switch. Many think that if the sidetrack man had a gun, he would be permitted to engage in self-​defense against Brenda, shooting her before she turned the trolley onto him. I argue here that this is false. I make the point first with regard to a good Brenda. I then discuss the difference that results if Brenda is culpable. Then I introduce third parties into the mix.

3.1. Good Agents and the Agent–​Patient Inference Brenda’s right to turn the trolley is based on the balance of patient-​claims on her, which I assume favors saving the five. The question is whether the sidetrack man may responsively oppose her act. Let us examine the possible justifications for him doing so one at a time. First, consider the balance of patient-​claims on him. Does it favor his shooting her? Clearly not. In this scenario, he is the agent, and there are seven patients: himself (he will be killed if he does not shoot her), the five, and Brenda. How do the patient-​claims compete? Against him shooting her we have her negative claim not to be shot and five negative claims not to have their rescue interfered with.24 Weighed against that, we have  We should distinguish claims not to be killed from claims not to have a rescue interfered with. The latter are arguably weaker, though no weaker than claims to be saved. As far as I know, this distinction was first highlighted in Hanser, 1999. 24

Negative Agent-claims And The Agent–patient Inference  |  187

only his positive claim to be saved from her action—​his claim on her is negative, but his claim for himself, taking her choice to act as a given, is positive. His claim clearly loses in that balance. Second, consider whether Brenda would have forfeited her patient-​ claim not to be killed by trying to kill him. She has done nothing wrong or culpable. Indeed, for taking on the burden of responding to the balance of patient-​claims on her, she deserves praise.25 Third, might the sidetrack man have an agent-​claim that gives him a right to proceed nonetheless? No, his claim is merely the positive agent-​ claim to do something harmful to others for his own sake, a claim that has no weight against negative claims not to suffer significant harm. Finally, might this be a situation in which competition is called for? It is hard to see why. I can see no plausible social good that legislators should recognize that would explain why he should be permitted to compete with her in this case. The closest thing to a plausible candidate is the thought that struggling to preserve one’s life has value on its own.26 But that value should be limited to struggling against those who have no right to act. I can see no reason to grant it more general value than that. Rather, granting it value more broadly than that is just another way of resuscitating the notion of an agent-​centered prerogative. And I  have said enough in Chapter 3 about why I reject that idea.

3.2. Culpable Agents and the Agent–​Patient Inference Now consider whether it makes a difference to what Brenda and the sidetrack man may do if Brenda is a culpable agent. Imagine, then, that Brenda has culpably brought about the situation in which the five need to be saved.27 Suppose that she recklessly damaged the trolley’s brakes and now sees that the five will be killed if she does not direct the trolley onto the sidetrack man. Her turning the trolley would still be justified. That is because, regardless of whether she was reckless, the claims of the five on her outweigh the claim of the sidetrack man. It is not that the claims on her are completely unchanged. The five would now have a special claim on

 See Chapter 4, § 2, where I argued for a high threshold for forfeiture.  See Statman, 2008, arguing for a right of self-​defense that would allow a defender to inflict harm on an attacker even if doing so would not prevent the attack. My argument in the text does not address the correctness of Statman’s suggestion, at least with regard to unjust attackers. It does imply that it should be limited so as not to apply to just attackers. 27  I believe the argument in the text would work just as well if her culpability lay in the reasons for which she acted, rather than in the acts she performed. 25 26

188  |  The Mechanics Of claims And Permissible Killing in war

her that she do what she can to save them. Though she would have had the liberty not to save them had she been non-​culpable—​a liberty she would have because of her negative agent-​claim not to be a killer for their sake—​ she would presumably have lost that liberty in virtue of her culpable act. But the competition between the basic patient-​claims of the five and the one should remain unaffected. The special claims of the five on Brenda do not change the weight of their restricting claims on her.28 The question is, Might the sidetrack man now have a right to stop her in self-​defense? Here is one reason why one might think that he does: Brenda’s culpability causes her to be indebted to the five. If she turns the trolley, as she is obliged to do to satisfy that debt, she now takes on a debt to the sidetrack man. If she could sacrifice herself to save him, she should do that. We are supposing that she cannot sacrifice herself to save him, but he can engage in self-​defense to save himself. Since by doing that he is only claiming what she owes him, he must have the right to engage in self-​defense. The problem with this argument is that it focuses too narrowly on Brenda and the sidetrack man, leaving the five out of the picture. If the sidetrack man shoots Brenda in a way that would save him, that would, per force, interfere with the rescue of the five. Thus, the question has to be asked, What are their claims on him? The answer is that they remain negative claims not to have their rescue interfered with. These claims still outweigh his positive claim to be saved. The fact that Brenda’s claim not to be killed drops out does not change the overall balance in favor of letting the sidetrack man kill her. The important thing to note about this variation is that Brenda’s personal interest in turning the trolley can drop out—​she might decide to turn it out of duty rather than out of a personal desire to save the five—​and yet she would retain her right to act and “her” related right not to be responsively attacked.29

3.3. Third Parties and the Agent–​Patient Inference Let us turn now to the case of a third party, call her Thelma, who could help or interfere with the sidetrack man or culpable Brenda. The first thing to say is that she could not permissibly shoot the sidetrack man simply to

 See Chapter 3, § 2.2.  I use scare quotes because the right is not fully hers and not hers at all if she has culpably caused the problem. I argue that it is not fully her right in the next section. 28 29

Negative Agent-claims And The Agent–patient Inference  |  189

protect Brenda. That is, she may not shoot him for Brenda’s sake. By her culpable act, Brenda has forfeited her own claim not to be shot. If the sidetrack man or another could kill her in a way that would save both the five and the sidetrack man (perhaps if she’s shot, she will fall in the path of the trolley and stop it), then that would be permissible. But Thelma, like the sidetrack man, may not shoot Brenda to save him if doing so would prevent her from saving the five. What, then, is culpable Brenda to Thelma if Thelma decides to act for the balance of patient-​claims on her? She is not just a person in her own right, with her own claims (some of which she has forfeited); she is also a kind of tool for Thelma. Given that Brenda will do what Thelma would want done, Thelma has reason to protect Brenda for the sake of the five. She could even permissibly shoot the sidetrack man to stop him from shooting Brenda. This point picks up on the point with which I  ended the last subsection: Brenda’s right not to be killed is not fully grounded in her own interest. Indeed, it is not fully her right; it is partly the right of the five whose interests ground her agent-​right, and if she is culpable, it is fully their right.30 Since culpable Brenda matters, in the balance of claims, not as an agent but as a tool, we need to ask, What if the sidetrack man could defend himself just by attacking the switching mechanism by which an agent like Brenda or Thelma would turn the trolley onto him? One might be tempted to say that surely he may do that.31 That is, one might be tempted to say that an essential part of the reason he may not attack good Brenda is that he would be inflicting a very serious harm, death, on an agent who has done nothing to forfeit her claim not to be killed. If we replace good Brenda with a mechanism set up by Thelma, then the conflict is between the sidetrack man and Thelma, who would suffer only the frustration of her plan to save the five. His life, one might want to say, should win over her claim not to have her plans frustrated. But this leaves out the claims of the five just as surely as it would if culpable Brenda were the mechanism. When it comes to determining what may be done, the agent–​patient inference need not reflect the presence

 Patient-​rights are grounded in the patients’ own interests. See Chapter 3, § 2.1. An agent need not have her own interest to have a right; agent-​rights can be identified as belonging to an agent because she is the one who occupies the position where competing claims get weighed against each other. Given that the right not to be responsively attacked concerns the agent as a patient and that culpable Brenda has forfeited her right not to be attacked, it is not really her right; it is the right of the five, whose interests ground her agent-​right. 31  I raised this point in Chapter 1, § 3.2. See McMahan, 2014b, p. 110. 30

190  |  The Mechanics Of claims And Permissible Killing in war

of an agent’s right. If her interests are not the ground of her right, if her agent-​right is grounded in the balance of patient-​claims on her, then it is the balance of patient-​claims that demands that other agents not interfere with the agent. And that would bar interference even with a mechanism set up to pursue that balance of claims. In sum, an agent may have a right or a duty to act even if she has no agent-​claims. And another may not responsively undermine an agent doing what she has a right or duty to do except insofar as competition is independently justified. If competition is not justified, then the fact that an agent has a right to act implies that every other person’s interest has already been fully taken into account. Another agent who can address the problem that justifies her act in a better way may have a right to interfere with her as part of a generally better solution to the problem.32 But no one may respond by simply seeking to prevent her from successfully acting.

4. Application to Just War Theory I conclude this chapter by extending the discussion of the agent–​patient inference and the limits of competition to the context of self-​defense against justified combatants. I start by noting that the issues in the discussion of the sidetrack man’s claim to engage in self-​defense against Brenda have the structure of a classic case from just war theory: a bomber pilot is engaged in a justified attack as a side effect of which bystanders will die, and some of those bystanders contemplate killing the pilot in self-​defense. Following on the point with which I ended the previous section, I will argue that, contrary to the common view, it is impermissible for people threatened by the justified acts of just combatants to eliminatively kill those combatants in self-​ defense. Before doing so, I will engage two reasons why one might take the contrary position. First, there is the thought that even if the bomber pilot has a right to bomb where she does, she threatens to infringe the rights of those bystanders whom she will kill if she is not stopped. Frances Kamm embraces this thought when she writes that the sidetrack man would  Recall Preventing Murder, Chapter 2, § 2.3. If Mark decided not to shoot Victor, he could justifiably interfere with Andrea’s plans to shoot him, as long as he does not cause her too much harm (measured taking into account the fact that his plans up to that point have forfeited his claim not to be harmed). 32

Negative Agent-claims And The Agent–patient Inference  |  191

have the right to redirect the trolley back to the five or to those “who had permissibly redirected the trolley” toward him.33 She thinks that this “may be explained, in part, by his having a general right not to be harmed that would be infringed when the trolley is redirected towards him.”34 Thus, she and others think that it is permissible for those bystanders to attack the pilot to try to prevent their rights from being infringed. In response, I want to say that this presents another substantive reason to resist the infringement model: it leads people to believe that justified acts can be resisted because they “wrong” someone. But, as I argued in Chapter 4, section 2, we should resist the idea that it makes sense (outside of cases of threshold deontology) to rightfully wrong anyone. We should stick with the mechanics of claims, according to which the heirs of those who are justifiably killed for the sake of others have a claim for compensation; but that does not imply that they were wronged, and it does not imply that the agents who do the justified killing may be attacked. Jeff McMahan helpfully offers another reason why combatants might have to suffer more for the sake of the innocent noncombatants they threaten: professional combatants, as part of their profession, accept certain risks to protect bystanders.35 If they had a strong duty of that nature, then arguably justified bombers may be killed if killing them could somehow protect innocent bystanders—​as long as doing so would not undermine the effectiveness of their mission. But it is hard to see how such situations would arise. Moreover, as McMahan writes, “it is doubtful whether the special duties of professional combatants include a duty to sacrifice their lives to avoid killing innocent bystanders.”36 Assuming his doubt is correct, then McMahan’s point would support neither using just combatants as a means of protecting bystanders nor killing just combatants as a side effect of protecting bystanders.37

 Kamm, 2007, p. 167. See also Hosein, 2014.  Kamm, 2007, p. 167. The other part of the explanation for Kamm seems to be a revanchist idea that the sidetrack man is presumptively allowed to “resist[] an improvement in the world and return[] things to the way they were for the five.” It is unclear to me what speaks in favor of this idea. It is true that the five might owe the sidetrack man or his heirs compensation if someone shifts bad luck onto him. But that does not imply that he is in any way free to set things back. 35  McMahan, 2014b, pp. 134–​135. See also Richardson, 2012, for the more general thought that acts that connect one person to another might give rise to special duties of care. 36  McMahan, 2014b, p. 135. 37   See my discussion of the disconnect between small obligations and large harms in Chapter  6, § 3.3.2. 33 34

192  |  The Mechanics Of claims And Permissible Killing in war

This leads me, at last, to making the leap from Trolley Switch to just war theory. Nothing I have said here should fail to make that leap. If one side in an armed conflict is justified in using force to protect itself or others against unjust aggression, then no one is justified in using force on the other side in response. Admittedly, this extension is hugely controversial. War is thought by many to be an example of justified competition. According to the law of armed conflict, combatants may permissibly target combatants on the other side, and those who are targeted may take responsive action aiming to prevent these attacks. But according to reductive individualism, war is not a special activity with its own moral rules; rather, it must respect the rights of individuals just as every other activity must do.38 Moreover, one cannot justify war as a socially useful competitive activity like sports, business, or speech.39 War is, rather, a horror to be avoided if possible. It should be engaged in only when necessary to protect large groups of individuals from suffering serious injustices at the hands of one or more large, coordinated groups of others whose unjust activities are a threat to them. On this view, just combatants are like Brenda: they may not be targeted for doing what they have both a right and a morally sufficient reason to do. It may be that the targeting of just combatants must be legally permitted because the law cannot find adequate institutional ways of regulating it. Quite often, when conflicts arise between states or between states and nonstate actors, there is no impartial body with the power to enforce its judgment that can adjudicate who is in the right and who in the wrong. Lacking such a body, each side has no choice but to judge of its own case as best it can. Moreover, I accept that the law has some moral relevance. There are reasons to accept legitimate law, even if it is unjust.40 This implies that combatants for an unjust side have some moral reasons to do as they are legally directed to do, even if their orders are unjust. But I have also argued that the duty to respect legitimate law has limited weight when contrasted with the duty not to do or tolerate substantive injustice.41 The substantive injustice of attacking just combatants is clear and profound. Thus, it is my view that individuals of good conscience should still recognize that it is

 See Chapter 2, § 3.  According to a chivalrous notion of war, the statement in the text is mistaken; war is a competitive forum for expressing martial virtue. If this view was ever sufficient to justify war—​and I reject the claim that it was—​it certainly cannot justify modern warfare, no matter how much modern armies draw on chivalrous attitudes to raise armies and motivate combatants to fight. 40  See Walen, 2001; Renzo, 2018. 41  See, again, Walen, 2001; Renzo, 2018. 38 39

Negative Agent-claims And The Agent–patient Inference  |  193

impermissible to respond to the permissible acts of just combatants with attempts to interfere with their activities. Combatants may have a duty to obey orders (it may be what they ought to do—​see Chapter 2, section 2.2) if it is unclear whether the enemy combatants are engaged in a just or an unjust cause, but if it is clear that their cause is just, then they may not be attacked.

194  |  The Mechanics Of claims And Permissible Killing in war



9 Intervening Agency and the Right of Non-​Sacrifice

T

his chapter argues for the extension of an idea from the previous chapter, that of a right of non-​sacrifice. I argue that this right exists in certain cases of intervening agency: when the intervening agent presents the primary agent with a choice either to submit to suffering a harm or to resist, knowing that the intervening agent will then harm others. I argue that the effect of an intervening agent presenting a primary agent with this choice is to create for the primary agent a situation in which she must either allow herself to be harmed for the sake of others or save herself but thereby bring it about that the others will be harmed. An agent has a claim of non-​sacrifice when presented with such a choice.1 The argument proceeds in four parts. First, I explain why such a case is puzzling; second, I  dismiss two unsuccessful attempts to resolve the puzzle; third, I  explain why intervening agency sometimes has the effect of changing the nature of the causal structure in which an agent acts, though noting as well that intervening agency is not unique in having that effect; and finally, I explore how these ideas might be relevant to just war theory and eliminative killing in that context.

1. Introduction to the Puzzle The last chapter highlighted two cases, Cart Avoidance and Shielding the Massive Man, in which an agent can permissibly act to save himself from  Much of the argument in this chapter is similar to arguments deployed in Ferzan, 2017. This is not a surprise as we discussed these ideas together over a number of years. 1

threats, despite the fact that his act would lead to more harm overall. The explanation appealed to the right of non-​sacrifice. In both cases, those who would be harmed as a result of the agent’s self-​protective acts would need the agent to serve, at significant cost to him, as a means of protecting them. Their claim that he suffer significant harm for them is substantially weaker than his claim to be free to act to avoid suffering such harm for their sake. I now want to introduce another case that intuitively has the same structure, though it also has an important difference: the presence of an intervening agent. The important challenge of the case is explaining why it should be interpreted as an exercise of the right of non-​sacrifice, rather than as a case of impermissibly harming others to save oneself. The question, in other words, is why this case should be thought of as a species of ducking harm, rather than a species of deflecting harm. Consider, then, the following: Bomb: Barry attacks Lisa. Lisa knows that if (and only if) she fends him off, he will detonate a bomb that will kill two people. She also knows that if she doesn’t fend him off, he will cut off one of her legs.2

Intuitively, Lisa is permitted to fend off Barry’s attack. Intuitively, that reflects the idea that she is not required to absorb such a significant harm from him to protect the other two. And intuitively, the fact that he is an intervening agent plays some role in explaining why she is permitted to fend him off. But it is important to be clear about just how problematic these intuitions are. The first challenge is to explain why her act should not count as something akin to a mere deflection of a threat from herself onto others. Recall what I  said in the last chapter, when introducing Shielding the Massive Man: “two negative restricting claims not to be harmed are stronger than one positive restricting claim to avoid the same harm.”3 The dominance of the negative claims is even stronger here, given that she would be inflicting a greater harm on others than she is avoiding herself. The challenge is to explain why fending off Barry is not an instance of deflecting and augmenting a threat. One might be tempted to say that the reason Bomb is morally different from a mere deflection case is that Barry’s free choice intervenes and cuts off Lisa’s responsibility for the welfare of the two whom he will kill if she 2 3

 This case is based on one presented by Frowe, 2014, p. 135.  See Chapter 8, p. 177.

196  |  The Mechanics Of claims And Permissible Killing in war

fends him off. There is a germ of truth in this, but we must separate the wheat from the chaff. It is not as simple as saying if another agent’s will is interposed between what one agent does and harms coming to others, then the first agent is not morally responsible for the harm that comes to the others. Recall the discussion from Chapter 7, section 4, in which I argued that even if the first agent lacks the mens rea to be an accomplice of the second, the first can still be viewed as a threat who may be eliminatively killed if necessary to save victims of the second. Thus, the second challenge is to explain why Lisa should not be viewed as simply a threat to the other two, one who may be eliminatively killed if necessary to save them. My answer is that Lisa can appeal to the right of non-​sacrifice and that Barry’s intervening agency affects the nature of her act so that she can appeal to the right of non-​sacrifice rather than see herself as engaged in a deflection of harm from herself onto others. But before defending that position, I want to argue that two other attempts to make sense of Bomb do not work. The first seeks to reject the intuition by distinguishing Bomb from Cart Avoidance and Shielding the Massive Man. The second seeks to capture the intuition by appeal to the supposed importance of rights violations.

2. Two Unsuccessful Approaches to the Puzzle 2.1. Misplaced Use of Counterfactual Baselines One way to avoid the puzzle presented by Bomb is to say that Lisa lacks the right to act that David in Cart Avoidance and Mike in Shielding the Massive Man have. And one reason to say that is that the counterfactual significance of her presence is morally relevantly different from theirs. If David were not in the way of the cart, the two would be hit by it nonetheless; if Mike were not in the way of the boulder, the two would be killed by it nonetheless. But if Lisa were not present for Barry to mutilate, there is no reason to think that Barry would kill the two nonetheless. Indeed, we can assume that he threatens them only as a means of pressuring her into submitting to him; if she were not present, he would have no reason to threaten them. In this sense, it seems that Lisa’s presence is uniquely problematic for the other two. And that suggests that what Lisa seeks to do, in fending off Barry, is not so much to prevent herself from being used as a means of keeping them safe as it is to deflect a threat from her onto them. Since her patient-​claim to have her leg saved (by herself or

Intervening Agency And The Right Of Non-sacrifice  |  197

another) is weaker than their patient-​claims not to be killed and since positive agent-​claims have no weight against claims not to suffer serious harm, she should have no right to act. This is an important objection, but it turns on the notion of a relevant counterfactual baseline. As I argued twice before in this book, it is a mistake to appeal to counterfactuals involving a person not being present.4 If we look to determine what kind of claim a patient has by imagining what would be the case if he were not present, counterfactuals can be developed such that one would get certain cases wrong, implying that the impermissible is permissible and vice versa. Similar problems arise when asking what would happen if an agent were not present. The claim that an agent like Barry would not be motivated to blow up a bomb killing others if Lisa simply disappeared is only one of a variety of possible counterfactuals, and there is no basis for picking the relevant one. Moreover, the same could be said for Mike and David. And yet what they—​all three—​may do to save themselves does not plausibly depend on the counterfactuals. To show this, I will start with Lisa’s case. How are we to imagine that she “simply disappears”? If we imagine it like a magic trick, who knows what Barry might do. Maybe he would be enraged and blow up even more people. But suppose that what we mean by her not being present is that something causes her not to be present when and where he otherwise would have encountered her. There are still many ways to imagine alternative possibilities that are all equally close to the “real” world of the case. To see how this might play out, consider the following backstory, explaining why Barry would be so angry if she fought him off. Lisa and Barry met a month before. They had two dates, and at the end of the second she thanked him for his time but told him that she was not interested in seeing him again. He continued to call her, however, and she neither picked up nor returned his calls. Her refusal to return his calls made him angrier and angrier. On the day of the event in question, he decided that she “deserved” to lose something of great value too. She was a competitive runner, and he thought that chopping off one of her legs would be fitting “punishment” for all the pain she had caused him. With that backstory in mind, consider these possibilities: 1. He finds her, corners her, and lets her know that if she fends him off, he will set off a bomb that kills two innocent bystanders.  See Chapter 2, § 4 (discussing the nature of eliminative killing), and Chapter 3, § 4.1 (discussing the restricting claims principle). See also Walen, 2016. 4

198  |  The Mechanics Of claims And Permissible Killing in war

2. She sees him coming and runs away before he reaches her; though still angry, he does not harm anyone. 3. She dies just as he comes to her; he is very saddened and gives to charity in her name. 4. She leaves shortly before he arrives; he looks around and finds another woman, who he mistakenly believes is her, kissing another man; he becomes so enraged that he kills both of them, then realizes his mistake and commits a very public suicide, killing himself and 10 others. The counterfactual that suggests that she may not fend him off is number 2; relative to that as the baseline, fending him off causes harm to others. But I can see no reason why it is any closer to the first possibility than any of the others. That is, it is not clear why we should say that the baseline world from which to compare the effect of her fending him off is that alternate possible world, rather than the one in which 12 innocents die or the one in which he gives to charity in her name. It seems that the only meaningful comparison, the only one relevant to her choice, concerns what would happen in the world in which she finds herself if she either does or does not fend him off. As already noted, a similar point can be made for Cart Avoidance and Shielding the Massive Man. I’ll content myself with demonstrating it with Cart Avoidance. The claim, meant to drive a contrast with Bomb, was that if David were not present, the others would be hit by the cart, so his ducking changes nothing. But suppose that one way he would not be present in a maximally close possible world is that he was turned aside by seeing a friend just before walking to that spot. And suppose that if this had happened, then the cart would not have been rolling down the hill. Why? Suppose it was launched at him by nefarious Ned. Ned had no good reason for aiming the cart at him. In fact, Ned had mistaken David for someone else, whom Ned hates. But there it is: if David had not been present, the other two would not be endangered. Adding in this backstory about Ned’s misdirected malice does not give David any extra obligation to stay put and absorb the cart’s blow. It is not his fault that Ned aimed at him; the mere fact that someone aimed it at him cannot mean that he is required to sit still and protect the others who would be harmed if he ducked. What matters, assuming that he has no special responsibility to absorb the blow, is only his causal role: he can be the means of saving others at his own expense, or he can move his body out of harm’s way and allow the others to be hit. He has a right to act to save himself from serving as their

Intervening Agency And The Right Of Non-sacrifice  |  199

shield regardless of the counterfactual. I say the same thing must be true for Lisa. Either she allows Barry to chop her leg off and thereby prevents him from killing others, or she fends him off and brings it about that two others die.5 If she has a right not to serve as the means of protecting the others, then she has a right to fend him off.

2.2. A Misplaced Appeal to the Importance of Rights Violations Helen Frowe offers an alternative account of why Lisa may fend off Barry, one that appeals to both a general discount for harms brought about via intervening agency and the importance of rights violations. I  have said enough about why I reject the idea that there is a significant general discount for harms brought about via intervening agency,6 but I’ll grant Frowe’s appeal to it for the sake of argument, to highlight the problem with her appeal to the importance of rights violations. Frowe supposes that Lisa can discount her responsibility for the harm that Barry would cause if she fends him off “by 50 per cent.”7 She is not insisting that this is the proper discount rate; she is only putting a plausible number on it to show that something else seems to be going on—​for even supposing that the discount rate for intervening agency is 50 percent, it seems that Lisa would still not be entitled to fend off Barry. The reason is that an act that kills 50 percent of two people—​in her example it is five people—​brings about disproportionate harm if it only saves the agent her leg. Frowe’s solution, her way to explain why Lisa may nonetheless fend off Barry, is to claim that the balance tips further in favor of Lisa because there is “a further cost” to her, a rights violation.8 To understand this supposed cost, it helps to have in mind the contrasting example Frowe is working with: River: Two people are on a raft in dangerous waters and will soon drown if not rescued. Rachel can pull the life raft to shore, but if she does, she will suffer the loss of her leg.9

 I use the phrase “brings it about” rather than “causes” to suggest that her bringing it about might not be, in some metaphysically strict sense, causal. I say this to accommodate the thought that Barry is acting as a free agent and the fact that the law regards such intervening free choice as breaking the causal chain: novus actus interveniens. My challenge to the view that intervening agency generally breaks the chain of responsibility need not take issue with the idea that it breaks the causal chain. 6  See Chapter 7, § 4. 7  Frowe, 2014, p. 130. 8  Ibid., p. 136. 9  This is a lightly modified version of Frowe’s case. Ibid., p. 133. 5

200  |  The Mechanics Of claims And Permissible Killing in war

Rachel and Lisa would both lose their legs if they choose to act to ensure the well-​being of the other two, but Rachel would not have any of her rights violated in the process. Lisa, however, faces the choice between fending Barry off or suffering both a lost leg and a serious rights violation at his hands. Of course, that is not the only difference between them. To save her leg, Lisa would act to bring about, indirectly, the death of two, whereas Rachel would merely allow two to die. But according to Frowe, the “additional cost” of suffering a rights violation “could counteract [Lisa’s] greater responsibility for the harms in Bomb compared to [Rachel’s] responsibility for the harms in River.”10 Thus, according to Frowe, with both factors in the balance we can see why Lisa “is required to bear roughly the same amount of cost to prevent mediated harms to others as [she] would be to rescue others from harm.”11 Just as Rachel is not required to sacrifice her leg for the others, so Lisa is permitted to save her leg despite the harm it will indirectly cause to others. This is a bad account of Bomb. One problem is that there are threatened rights violations on both sides of the balance in Bomb. Barry threatens to violate Lisa’s right not to have her leg cut off, but he also threatens to violate the rights of two people not to be blown up. If Lisa gets to appeal to the fact that she not only would lose her leg—​a fact that she shares with Rachel in River—​but would be wronged in the process, then the two get to appeal to the fact that they not only will die—​a fact they share with the two in River—​but will be wrongfully killed. If there are rights violations on both sides, it would seem that they cancel out, leaving Frowe with no account of how to overcome the fact that Lisa in Bomb would kill (though indirectly) two, while Rachel in River would merely allow two to die.12 Even if this problem did not sink Frowe’s account, there is another level at which it is implausible. It requires us to suppose that a harm that comes with a rights violation is much worse than a harm that comes without a rights violation. Recall, even if we grant a 50 percent discount for intervening agency, Lisa would still effectively be killing one person to save her leg. If we think that it takes at least a three-​to-​one ratio of saving to harming to justify harming some to save others, the fact that she would be wronged by Barry has to register as being a harm on a par with three deaths, despite the fact that the physical harm is only a lost leg.

 Ibid., p. 136.  Ibid. 12  In fact, Frowe recognizes that rights violations could come into play on both sides of a case like this. Ibid. I am at a loss as to why she did not see that this posed a problem for her account of the case. 10 11

Intervening Agency And The Right Of Non-sacrifice  |  201

But even if one thought that being wronged was some sort of multiplier of the harm one suffers, it is quite far-​fetched to say that it can raise the harm of losing a leg to the level of three deaths. Of course, Frowe might think that the intervening agency discount is higher, say 90 percent. But we could just put a few more victims on the other end, and it would still create a problem. Or Frowe might think that Lisa also benefits from an agent-​centered prerogative, but I said enough about why that is a bad idea in Chapter 3. In the end, we must simply reject Frowe’s account of Bomb. The reason to give it this much attention is that it reflects a deep and common misunderstanding about rights.13 It treats rights violations as a bad thing that sometimes happens to people. It treats rights violations as akin to bodily damage, an extra harm that we have to put in the proportionality balance. But that is a profound misunderstanding of how rights work. It is not much worse to be murdered than to be killed by a wild animal or any other natural force. Yes, our reactive emotions are engaged more by a murder, but that is because we—​in our saner moments—​blame murderers, not animals. It is not because the claim to be protected from murder is much different from the claim to be saved from death by animal. It is a kind of category mistake to think that the normative force of rights is carried by how bad it is to infringe or violate them.14 The point of rights is not to capture a special kind of value; it is to mark what is permissible. An agent must respect the rights of others not because doing so produces more value than not doing so but because she has no right to violate the rights of others. If she operates in the space of rights—​claiming them, trading them, acting on them, etc.—​then her reason to respect rights is that rights provide her with a normative framework for acting. We can go deeper and appreciate why the space of rights is important. As I indicated in Chapter 3, the reason is that the space of rights enables us to respect each other as free and equal beings whose welfare matters. It enables us to relate to each other in a way that is deeply valuable for humans, who are social creatures and who are deeply committed to notions of justice, fairness, and rights. Living outside the space  Tadros, 2016b, pp. 110–​15, offers a similar rebuttal to the argument that an agent has greater latitude to allow harm to innocent victims if another has wrongly chosen not to do his fair share and thus has threatened to wrongfully impose an extra burden on the agent. 14  I touched on this point before, in Chapter 3, § 1.2.3 (citing Scanlon, 1998, p. 83). I should add that my view here was, in a way, anticipated by Tadros, 2011, p. 201: “We have reason to protect people from being harmed. Whether harm also amounts to a rights violation is relatively unimportant, at least when the harms with which we are concerned are very significant, such as loss of life.” We differ because he thinks rights are fundamentally unimportant; I think they are unimportant only insofar as one is fundamentally concerned with welfare. 13

202  |  The Mechanics Of claims And Permissible Killing in war

of rights is an option only for gods, beasts, and sociopaths.15 If one is a normal human, committed to living as a citizen in the space of rights, one’s reason to respect the rights of others is not the value of doing so; it is the nature of rights itself that they provide such reasons.

3. Intervening Agency and Causal Roles I return now to the argument that Lisa may fend off Barry as an exercise of her right of non-​sacrifice. And I pick up with the question of why we should think that she is protecting herself from being used to save them from Barry rather than deflecting the threat Barry poses from her onto them. In other words, I return to the question of the causal role she plays. To get started, consider the following case, in which the agent, unlike Lisa, clearly is not exercising her right of non-​sacrifice: Grenade and Bystanders: Bobby is coming toward Laura. She knows that if he catches her, he will cut off one of her legs. She also knows that she cannot outrun him or fight him off if he gets to her. The only thing she can do to save herself is throw a grenade at him. But to defend herself with the grenade, she would have to throw it where it would not only kill Bobby but also kill two bystanders.

Perhaps we would excuse Laura if she used the grenade; nonetheless, she may not use it. She may not prefer her leg to the lives of two bystanders. But is there really a difference between Lisa and Laura in terms of their causal roles? Laura would kill two directly, whereas Lisa would kill two by bringing it about that Barry will kill two. The question is, Why should that make their causal roles different in the sense that Lisa can appeal to an agent-​claim to protect herself from being sacrificed and Laura cannot? To see why their causal roles are different, and why that makes a difference, note that Laura cannot claim that her submitting to Bobby is the means by which she refrains from hitting the two bystanders with a grenade. She can achieve that simply by not throwing the grenade. The unfortunate result of her not hitting Bobby and the bystanders with the grenade is that Bobby will cut off her leg. But that would happen as a result of her refusing to kill the bystanders. It is not a means by which she ensures that

 To paraphrase Aristotle’s Politics, 1984, book 1, ­chapter 2, line 1253a27 (on why man needs to live in a political community). 15

Intervening Agency And The Right Of Non-sacrifice  |  203

they are not killed. By contrast, Lisa can save the bystanders from Barry’s bomb only by submitting to Barry. For both Lisa and Laura, the reason to make the decision that would lead to losing a leg is to avoid playing a causal role in the deaths of two bystanders. But for Lisa the question is whether to submit to Barry as a means of keeping him from harming them. By contrast, for Laura the question is whether to refuse to harm others as a side effect of defending herself. One might still think that what is really doing the moral work is that Barry, but not Bobby, serves as an intervening agent. But if one thinks this because one thinks that intervening agency is giving Lisa the right to discount the harm that she indirectly causes the two bystanders in her case—​a discount based on the idea that he is the one primarily responsible for their harm—​then one is misunderstanding the relevance of intervening agency. Intervening agency is relevant if and only if it helps establish that someone is indeed being used as a means of preventing others from suffering harm. It does not provide a general “discount” for harm caused indirectly to others. I have argued this point before in terms of the thought that agents should treat others who might do harm as a result of what the agents themselves do as mechanisms they might unleash. I argued that it makes no sense to say that the claims of the victims of intervening mechanisms have to be discounted if the intervening mechanisms are free agents but not otherwise.16 Here I want to introduce an argument that comes at the supposed relevance of intervening agency from the other side. It aims to show that even without intervening agency, we can get cases that seem to work just like Bomb, as long as it is clear that the agent is exercising the right of non-​ sacrifice. Consider, then, the following: Hungry Bear: A hungry bear has come upon Edith and is hoping to kill and eat her. Edith has bear spray and can fight the bear off. If she uses the spray, however, the bear will turn aside and find two other hikers who are not far away and will kill and eat them.

Edith’s fighting off the bear is much like Lisa’s fighting off Barry. The bear wants to inflict harm on Edith just as Barry wants to inflict harm on Lisa. The bear will also harm others unless it can harm (and eat) Edith, just as Barry will harm others unless he can harm Lisa. And just as Lisa may

 See Chapter 8, §4.

16

204  |  The Mechanics Of claims And Permissible Killing in war

fight off Barry, so may Edith fight off the bear. Both may do this because they have a right of non-​sacrifice. But the hungry bear is not a free agent. It has some agent-​like features (wanting, acting), but it lacks all capacity to reflect on its wants and make reasoned choices. It is a brute. It cannot reasonably be punished because it cannot be responsible for what it does. Therefore, it cannot serve as an intervening agent. This shows that the structure of Bomb does not depend simply on the relevance of intervening agency. It depends primarily on the agent-​claim not to be sacrificed for others. But now to see how intervening agency is relevant to Bomb—​to see the kind of role it has to play to turn a case from a deflection case into a ducking case—​consider another bear case that also seems like Bomb, only in this case the moral restrictions seem more like Grenade and Bystanders: Momma Bear: Holly is out hiking and stumbles upon a bear cub in the woods. The bear cub lets out a squeal of fear and the momma bear comes running angrily at Holly. Holly can fight the bear off using her bear spray. But if she does, the enraged bear will then turn off down the hill where it will maul and kill two other hikers.

Holly’s use of bear spray seems like a deflection of a threat from herself onto others. Deflections that would harm others as much as or more than oneself are not permitted.17 The others have negative patient-​claims not to be harmed that outweigh Holly’s positive patient-​claim to be saved, and her agent-​claim is the positive claim to save herself, a claim that has no weight against their claims not to be harmed. That is why this case is like Grenade and Bystanders:  Holly, like Laura, threatens impermissibly to harm others to save herself. These cases raise two questions. The first question is, Why should we think that Momma Bear is different from Hungry Bear?18 The answer is that the hungry bear wants to eat someone, and Edith has a negative agent-​ claim not to have to allow herself to serve as the means by which others are  Doug Husak posed this question in personal communication: What if there were a sign at the entrance to the park saying, “Beware of bears. Do not enter without bear spray”? Then it seems that Holly could permissibly deflect the bear onto others because they should be able to deflect it from themselves as well. In this instance, if they cannot defend themselves, it is not her fault; it is their problem because they assumed the risk of entering the park without bear spray. This is a quite gen­ eral point: all of the cases in this book that involve deflecting harm assume that the vulnerability of others is not their responsibility. 18  I acknowledge that the intuition that they are different may not be strong. But the theoretical account of the difference, I think, is robust; and, to me, that matters more. See Chapter 2, § 1. 17

Intervening Agency And The Right Of Non-sacrifice  |  205

protected from becoming bear food.19 The momma bear, however, would not harm Holly as a means of satisfying some general hunger that also threatens the other hikers. The momma bear was going for Holly because she believed that Holly threatened her cub. If Holly refused to defend herself, she could not argue that the reason is that she was allowing herself to be used to save the others; her reason would be simply to avoid deflecting a harm onto the others. The second question is, Why is Momma Bear different from Bomb? They are similar in that in both the force that threatens Holly or Lisa has it in for her and would go on to harm others only if the primary target fights the threat off. The difference is that the agent in Bomb, Barry, makes the connection between Lisa and the other two. His agency makes it the case that she would have to sacrifice herself to him to save them from him. He does not have to explicitly draw the link; he does not have to communicate to her: “If you do not allow me to cut off your leg, I will blow up a bomb and kill two others.” But if it is clear to her that this is how he has framed the situation, and therefore her choice, then his agency sets up the link. He sets up a causal structure, running through his choice, such that if she does not submit to him, then he will harm others. Her submitting to him is, in that causal structure, a means of preventing him from harming others. His agency makes it the case that her claim is a claim not to be the means by which they are protected from him. Without that particular sort of agency—​making a conditional link—​Lisa’s choice would turn into Holly’s. One might object that the conditional link is present for both Lisa and Holly. If Lisa resists Barry, he will kill two others; if Holly resists the momma bear, she (the bear) will kill two others. But this objection overlooks the difference in the background causal structure. Lisa faces a choice in which Barry has already put two others in danger unless she submits to him.20 Holly, by contrast, faces a choice in which the momma bear has not already put others in harm’s way; they are in harm’s way only insofar as Holly deflects the momma bear onto them. This difference between exercising the right of non-​sacrifice and a deflecting may seem

 A similar point might arise with alarm systems that deter robbers from entering the property with the system. Those might seem to be aimed merely at deflecting harm onto others. But I think they are more plausibly framed as devices with which one chooses not to absorb the threat of robbery—​on the assumption that robbers operate like hungry bears—​that otherwise would target others. More generally, one does not have to leave oneself a soft target. 20  It does not have to be determinate which two people it is he will kill. He might decide to hold a lottery to pick two. The point is that there are two who will die if she fends him off. 19

206  |  The Mechanics Of claims And Permissible Killing in war

subtle, but it is present in all the cases we have examined so far. Shielding the Massive Man, for example, is an exercise of the right of non-​sacrifice and not a deflection because the two would be saved only by Mike’s serving as the means to absorb a threat that would, but for his absorbing it, harm them. This notion that others are already in danger and would suffer harm but for another agent absorbing the threat is the essence of a case in which an agent can exercise the right of non-​sacrifice. It is present in Bomb and Hungry Bear; it is absent in Momma Bear and other deflecting cases. The way that Barry establishes that Lisa has to submit to being harmed to save the others reflects the power of his free will. Unlike a bear, Barry has the capacity to choose to delink her harm and the harm to others. He could simply change his plan. He could decide not to attack her in the first place. Or he could decide that if she fights him off, he will simply go off, tend to his wounded ego, and leave everyone else in peace. It is his choice to create a situation in which she has to submit to him for the others to be safe. But we have to be careful to make sense of why his having free will matters. The real moral magic here is not that his will simply happens to be placed between her choice and harm to others; it is that he requires her to bend to his will, to refrain from fighting him off, if she is to save the others from him. In further support of this, consider a case in which Barry acts more like an animal than a person with a plan. Imagine that he has a bomb strapped to himself and is coming to attack Lisa. She can defend herself with pepper spray and cause him to go staggering away, where the bomb will go off in a crowd. If she does not use the spray on him, however, she will be unable to fight him off; he will come up and grab her, and then the bomb will go off, killing her and him but no bystanders. In this case, I think the only fair thing to say is that his agency does not matter as it does in Bomb. It is as if he simply tossed a live bomb at her. She can either toss it away toward others or refrain from doing so. She must refrain from doing so as the alternative is to deflect a threat onto others.21 What makes Bomb different is, again, that Barry has set up a causal structure that already threatens others, one in which she has to submit to him to save them.22 The lesson of these cases is this: the reason intervening agency is sometimes morally relevant is that it can transform a deflection into a using.  See Chapter 8, note 9 (on the squib case).  Frowe’s central case in her discussion of mediated harm is Angry Rape, a case in which a rapist attacks a potential victim and the victim knows that if she fends him off, he will go and rape two other women. But the reason he will rape two other women is not that he would be following through on a plan to use that threat to pressure her into giving in; he would do so because “he will be so angry” at having been fought off. Frowe, 2014, p. 129. That case, I am afraid, has to be analyzed as a deflection case. 21 22

Intervening Agency And The Right Of Non-sacrifice  |  207

It can do so because we attribute to intervening agents the ability to establish a kind of conditional link that, in essence, uses others as hostages and requires an agent to choose whether to sacrifice herself to save them. Agents, however, have a strong claim not to have to sacrifice themselves for others, and they have an equally strong claim to be free to prevent themselves from being sacrificed for others. Thus, this sort of intervening agency permits an agent to resist another, even if the other will then go harm third parties.23 Moreover, it is important to be clear that this is a robust right. Just as third parties may not interfere with David in Cart Avoidance and Mike in Shielding the Massive Man saving themselves and have a reason to help them, despite the fact that doing so will lead to more deaths, so a third party may not interfere with Lisa fending off Barry and has a reason, in fact, to help her do so. Her case is absolutely on a par with their cases. Finally, to wrap up this section, I want to return one last time to the idea that intervening agency provides agents some sort of significant, general discount. That idea should be dismissed. I have given many reasons why. I now return to a case from Chapter 7 to offer one last reason: Coercion II: Alvin threatens to kill Carla unless Carla gives him her long-​ range rifle so that he can use it to kill Ken. Alvin cannot kill Ken without Carla’s rifle. Neither Alvin nor Carla has a right to kill Ken.

Alvin’s agency intervenes between Carla and his other potential victims. If intervening agency worked via discounting and if discounting were substantially responsible for explaining why Lisa may fight off Barry, then Carla would be permitted to help Alvin—​or at least it would be substantially more likely that she would be. But I  believe that Carla may not help Alvin. This is not because Carla is somehow complicit with Alvin. She does not share his goal of shooting Ken. She would help him only to avoid being shot herself. The real reason Carla may not assist Alvin is that his intervening agency does not free her of responsibility for the harm she would help him to cause, and she has no claim of non-​sacrifice.  This analysis accounts for the intuition in a prominent case in just war theory: a surgeon is unjustly attacking A, but he will go on to save many lives if and only if A does not defend herself. (The case seems to originate in Hanser, 1999.) A’s submitting to the attack is necessary, in this case, to ensure a benefit, rather than to prevent a harm. But that seems to make no difference to the basic analysis. A may defend herself. 23

208  |  The Mechanics Of claims And Permissible Killing in war

Why does Carla have no claim of non-​sacrifice? Alvin is trying to use her as a means of harming the others. Is that not a matter of being used which she has a right to resist? It is, but it does not give her a right of non-​ sacrifice. If she refuses to be used as a means, the others are not harmed—​ quite the reverse of Bomb. Rather, if she refuses to be used as a means, she will then suffer harm. She can protect them simply by not helping him; she does not have to suffer in order to protect them. Putting it in terms of the causal chain, Lisa has to accept harm as a means of preventing Barry from harming others; Carla’s harm would be the result of not helping Alvin harm another, but it would not be the means of preventing him from doing so. It, like the harm Laura would suffer in Grenade and Bystanders, is simply a result of her choosing not to endanger others. And again, if one looked to discounting to explain the case, then any discount that would substantially explain Bomb would get Coercion II wrong. In sum, I  think the idea that agents enjoy a substantial discount for intervening agency was always a confused notion, conflating the fact that an intervening agent’s will could turn a deflecting case into a using case with the thought that the intervening agent’s will somehow generally makes it easier for agents to justify bringing it about that intervening agents cause harm to innocent victims. This is not to deny that Alvin deserves more blame than Carla if he shoots Ken. But blame and permissibility are not zero-​sum games. We can blame both, say that one is more blameworthy than the other, and say that both act impermissibly.

4. Application of the Claim of Non-​Sacrifice The question I want to end this chapter with is whether this account of the claim of non-​sacrifice is likely to be relevant to eliminative killing in the context of just war theory. There is a kind of case in which it may come up, one that resembles Bomb. Coercive Invasion: A group of combatants from X conveys a message to the people of Y that they intend to invade, and that if the people of Y resist the invasion, then the combatants will kill a large number of others (including members of Y who are not involved in the resistance).

In such a case, the members of Y have a claim of non-​sacrifice to resist, but that does not mean that they have a right to resist. Assuming that

Intervening Agency And The Right Of Non-sacrifice  |  209

the invading force is unjustly invading, the right to resist would depend on three things: (1) how severe a loss or harm they would suffer if they allowed the invaders to take control; (2) how likely it is that their resistance would be successful, and how likely it is that it would do nothing more than stoke their pride for a short time, while also leading to the death of others; and (3) how many others would suffer as they would not otherwise suffer if members of Y resisted. Unpacking those conditions, starting with the first, suppose that life in Y would go on as normal and that the only change is that a local autocrat would be replaced by someone who is probably no worse, and maybe even more inclined, after a short transition period, to promote the rule of law, respect for human rights, and democracy. In that case, the loss the members of Y would suffer would be essentially one of dignity:  the indignity of being overrun by an outside power. Arguably, that is no basis for taking action that would lead to the death of innocents, even if the deaths would be the result of the invading force acting unjustly and demanding that the locals accept their invasion without resistance. Comparing this to Bomb, it would be like Lisa fending off Barry, knowing he will then kill two others, even though all he wants to do is slap her face. Negative agent-​claims, property claims, and claim of non-​sacrifice are not infinitely strong; if the interest they protect is small and the competing claims are strong, the balance can go against those claims. On the other hand, if the harm of being taken over is that a legitimate, representative, democratic government that respects human rights is going to be replaced by some stooge from another country who will rule unjustly and treat the community as a resource to be plundered, then resistance may well be justified, even if innocents will die who otherwise would not.24 As for the second condition, success matters. If the invaders are going to set up an unjust, rapacious regime whether one resists or not, then the only good one can get from resisting is again one of dignity. That may suffice if the question is whether one can inflict harm on a culpable aggressor.25 But if serious harm will also befall innocent third parties, then we are back to the reasons for not being permitted to resist discussed in condition 1.  If there is a chance of success, however, then the good of resisting might suffice to justify doing so. How high the odds must be is

 This position is meant to contrast with the position taken in Rodin, 2014. Rodin is highly skeptical of the right to resist for reasons less than the protection of life itself. For a helpful response to Rodin, see Frowe, 2014, § 5.4. 25  See Statman, 2008. 24

210  |  The Mechanics Of claims And Permissible Killing in war

hard to say.26 I will content myself with saying that they must reflect the relevant proportionality considerations, which themselves are weighted, though not absolutely weighted, in favor of the claim of non-​sacrifice. As for the third condition, if the invaders, if met with resistance, would make only token, symbolic strikes at others, then it would be relatively easy to justify resistance. If their strikes would kill hundreds or thousands of innocents, however, then resistance would be harder to justify. In that case, it might be that the harm of failing to resist would have to be on the scale of widespread injustice, mayhem, or killings for the right of non-​ sacrifice to kick in.27 And, of course, it would matter how much overlap there is between those who would suffer retaliatory strikes from the invaders and those who would suffer under the rule of the invaders. The more those groups overlap, the less retaliatory strikes would inflict some extra harm; the less they overlap, the clearer the reason not to resist for the sake of those who would be harmed. If all three conditions are met and the right of non-​sacrifice applies, then no one—​not members of the community that would be attacked by the unjust combatants, not those acting on their behalf—​would be permitted to interfere with those members of the community who rightfully resist the invasion of the unjust combatants.

 See Chapter 3, § 4.5, and the discussion of Trolley Switch with Risk of Looping for the idea of discounting claims when it is unclear whether they will apply. 27  If it does and the harm threatened to others is great enough, that might put us into the space of threshold deontology. But I say only “might.” It might also be the case that the balance still tips clearly in favor of the right of non-​sacrifice. 26

Intervening Agency And The Right Of Non-sacrifice  |  211

10 Conclusions

A

s the book itself has operated on two levels, the conclusion too will operate at two levels. First, I offer some concluding thoughts on the permissibility of eliminative killing, focused on the eliminative killing of noncombatants who provide aid to unjust combatants. Then I summarize the important conclusions about rights theory.

1. Conclusions About Eliminative Killing I have argued for and elaborated on what is meant by the thesis stated in Chapter 1: permissible eliminative killing.1 I described that thesis with six conditions. The first was essentially a success condition, that eliminative killing has to at least reduce the magnitude of the harm victims would suffer or the risk of harm that the victims face. The next three conditions articulated what I  mean by the traditional necessity and proportionality conditions. The fifth condition articulated the agent–​patient inference, pointing out that eliminative killing could not be justified if the agent who poses the threat has a right to pose it. And the sixth condition was a catch-​all for other moral reasons that might affect the permissibility of eliminative killing. We can now look back on that thesis and draw out some concrete implications for just war theory. The most significant general point is this:  noncombatants who help unjust combatants kill unjustly have at most contingent rights against being eliminatively killed. In practice, this

1

 Chapter 1, § 2.

means that a wide range of noncombatants might permissibly be targeted for eliminative killing: people who design or manufacture weapons, who manufacture parts for weapons, who provide other equipment needed by combatants such as reconnaissance drones, who do linguistic analysis of information intercepted from the enemy, or who raise money to provide resources for combatants. If the activities of these people are helping an unjust aggressor to unjustly kill people and are thereby increasing the number of people wrongfully killed, then there is no fundamental moral prohibition on killing them. They may be targeted wherever they are, whatever they are doing, as long as doing so is necessary to reduce the threat posed by an unjust enemy, is proportional to the good that will be caused, and is not for some other reason morally prohibited. The prior general point does not imply that all or even many noncombatants who aid or abet unjust combatants may justifiably be eliminatively killed. I identify four reasons why it may not be justifiable to eliminatively kill noncombatants, all things considered. First, the limits imposed by necessity and proportionality are not trivial limits. It may generally be disproportionate to eliminatively kill noncombatants because the number of noncombatants a country would have to kill to limit their ability to help unjust combatants to wage an unjust war would be too great, given the good thereby achieved. Moreover, one cannot justify killing a few as a matter of eliminative killing if one is relying on having killed a few to dissuade others; that justification relies on the justification of terrorism, a separate issue.2 There may also be better alternative ways to combat an unjust aggressor. Indeed, targeting noncombatants may be worse than unnecessary; it may, in fact, be counterproductive.3 One complication that cannot be avoided is the culpability of noncombatants. If many are highly culpable for supporting the war, then the proportionality calculus changes. They may have forfeited their claim to be treated as indirect innocent threateners; they may instead be indirect culpable threateners, whose claims not to be killed are weak. This could be because of their political activities, such as voting for an aggressive government because it is aggressive, vocally supporting aggressive policies, or suppressing dissenting voices. Or it could be because they realize or

 See Chapter 1, § 3, and Chapter 7, § 5.  See, for example, A. C. Grayling, 2006, pp. 256–​259 (arguing that area bombing was no more effective than precision bombing in reducing the ability of the Germans to produce the guns, planes, and other tools of war during World War II). 2 3

Conclusions  |  213

should realize that they are helping unjust combatants and that they should stop doing so. But we should be very wary of assuming that noncombatants are culpable in a way that meaningfully forfeits their claims not to be killed. Culpability for moral choices presupposes that one could reasonably be expected to know better than to do what one does.4 It is reasonable to expect people to adhere to the norms of the community in which they are raised but not so reasonable to expect people to know that the community’s norms are indefensible. Not many people in the time of Socrates thought that slavery was an injustice, and it would be bizarre to hold all who endorsed the institution culpable for that mistake.5 Likewise, in most belligerent countries, the population is generally whipped up into a nationalist fervor before the war is launched, and most believe that the patriotic and right thing to do is to support the troops. Of course, sometimes the idea that a war is unjust is readily available and well supported in a country that wages it. Many Americans opposed the invasion of Iraq under George W.  Bush. Those who were gung-​ho for that war were arguably in some sense culpable. But in truth, it is much more reasonable to say that the Bush administration, which cherry-​picked facts or outright lied about the basis for going to war, bears all the culpability. The average citizen who believed that Saddam Hussein had weapons of mass destruction or who thought that we would be welcomed as liberators should probably be forgiven her well-​meaning mistake. Perhaps Hitler’s “willing executioners”6 crossed the line not only in the human rights violations they supported domestically but in the aggression they supported internationally. Perhaps they forfeited their claim to be treated as innocent threateners. But it is a judgment we should be reluctant to make. And, of course, even if some in a population are culpable and forfeit their status as innocent threateners, others are not—​a point I take up in my fourth point.7

 See Doug Husak, 2016.  This is not to say that one should not make negative judgments about the moral quality of people who did things like own slaves in communities where that was common. But forming judgments about how admirable someone like Plato was is different from saying that he acted culpably in owning slaves. Thomas Jefferson is a harder case as the moral reasons not to own slaves were not so foreign to him. 6   I  take the term from Daniel Goldhagen, 1997; it refers to the Germans who knew about and supported the killing of Jews and others in concentration camps. 7  Unjust combatants may also, often, have no idea that they are acting unjustly and thus not forfeit their right not to be eliminatively killed as well. See Seth Lazar, 2009 and 2015. But the real problem for them is terroristic killing. See Chapter 2, § 3.2. It also occurs to me, as I review page proofs, that the defense of this practice need not appeal to forfeiture, but that is for future work. 4 5

214  |  The Mechanics Of claims And Permissible Killing in war

Second, some noncombatants have a right to do what they do that contributes to the acts of unjust aggressors, and they benefit from the agent–​patient inference. Some may have the right to act based on the right of non-​sacrifice. Recall the case of Coercive Invasion. Suppose that the invaders from X said that they would attack country Z if members of Y resisted the invasion. Members of Z might be tempted to respond to the citizens of Y as a means of stopping the unjust attack from X. But they may not do so. They must restrict their response to X. Another basis for the right to act in ways that contribute to the acts of unjust combatants appeals to the fact that what people do often has a double effect: while it might help unjust combatants, it might also simultaneously help others and do so to such a degree that it is justified overall. Consider, for example, farmers who raise food that feeds both unjust combatants and a number of innocent noncombatants. Were they not to raise food, it would harm both. If the good of feeding the general population outweighs the bad of feeding unjust combatants and if there is no reasonable way for farmers to control who gets to eat their food, then their acts of growing food are justified overall. It would be unjust to eliminatively kill them to stop them from aiding unjust combatants. They benefit from the agent–​patient inference. Third, it will often be difficult to distinguish noncombatants who can justifiably be eliminatively killed, given these first two points, from those who cannot. Thus, another reason not to proceed is that one will inadvertently kill noncombatants who may not justifiably be targeted. How strong is their claim not to be killed? They would be killed as a side effect of pursuing the justified eliminative killing of threats. Thus, their claims are essentially negative restricting claims not to be killed as collateral damage from the policy. But their claims not to be so killed are stronger than the claims of indirect innocent threateners, raising the overall justificatory burden. Fourth, there is the moral relevance of the law, which may have to recognize extra reasons not to target noncombatants. One central reason to ban the targeting of noncombatants is that any exceptions to a general ban on targeting noncombatants may have a tendency to be misused; it may not be possible to effectively adjudicate and enforce morally refined limits. If true, this might provide all countries with a reason to adopt or retain laws that ban the targeting of noncombatants more generally.8 And for  As noted before, Chapter 2, § 3.2, Victor Tadros, unpublished manuscript, gives a host of arguments to doubt that this and related arguments are convincing. 8

Conclusions  |  215

individuals operating inside of a particular country’s legal regime, the fact that the regime legitimately bans targeting noncombatants may provide strong, if defeasible, moral reason not to do so.9 All that said, there may be instances in which noncombatants threaten to provide significant aid to unjust combatants, in which it is not too hard to tell who provides such aid, and in which it is clear that the best way to prevent the aid from putting a great many innocent victims at great risk is to target certain noncombatants for eliminative killing. My central practical point in this book is that the moral permissibility of eliminative killing in such cases does not depend, pace McMahan, Frowe, and others, on their being liable (through rights forfeiture) to defensive killing.

2. Conclusions About Rights Theory The standard model for rights is the infringement model. It is based on the idea that rights provide only a limited set of inputs into final determinations of who may do what to whom. Rights, on that model, have to be weighed against other rights and, more importantly, have to be complemented with consequentialist considerations. The idea that we have rights that may permissibly be infringed is a natural one, ubiquitous in natural-​language rights discourse. But if one wants to understand who may do what to whom, one is better off jettisoning this model and adopting, instead, the mechanics of claims. The mechanics of claims is a systematic approach to determining who may do what to whom. It is based on the idea that normal humans need to be able to relate to each other inside a shared normative space, the space of rights. This space is not merely some conglomeration of rights that are grounded in the will of agents, the interests of rights-​holders, or any such narrow basis. It is a systematically constructed normative space that gives expression to its three foundational principles: the autonomy principle, the equality principle, and the welfare principle. Rights emerge, in this space, from the balance of patient-​claims on agents, taking into account the agent’s claims as well. Claims in this space are much like rights as conceived by those who use the infringement model, but they cover all the interests that can sensibly be covered, rather than some subset of interests; and when they are contravened, nothing follows about a person being wronged. It may be that the person whose 9

 See Walen, 2001.

216  |  The Mechanics Of claims And Permissible Killing in war

claims were contravened has a claim for compensation, or some other related claims, but that is a separate matter. Just war theory provides an important set of issues that help highlight the need to use the mechanics of claims rather than the infringement model. Many of those who use the latter are drawn to overreliance on the idea of liability to explain why it seems unproblematic to kill threats, but their doing so gives rise to problems in accounting for the right range of forfeited rights. If they use a high bar for forfeiture, they have trouble explaining how a just war can be fought; but if they use a low bar for forfeiture, they seem to allow too much harm to be done. Nor is there any sweet spot in the middle. The use of forfeiture opens up too big a normative gulf between those threats who have forfeited their right not to be targeted and those more innocent threats who have not. In addition, those who use the infringement model lack the tools for making proper sense of a range of phenomena, from the means principle to when compensation is owed to the relevance of intervening agency—​ phenomena that are better explained using the mechanics of claims. It is my hope that the account of the mechanics of claims offered here is clear enough and appealing enough for others to explore its utility further—​and to seek to refine it further, as well.

Conclusions  |  217

TABLE OF CASES (WITH PAGE OF FIRST USE)

Bayonet, p. 39 Benevolent Bystander, p. 127 Bomb, p. 196 Cart Avoidance, p. 172 Coercion I, p. 161 Coercion II, p. 162 Coercive Invasion, p. 210 Collision Course, p. 91 Conscientious Driver, p. 3 Crosswalk, p. 157 Drug Owner, p. 88 Exploding the Well, p. 167 Falling Man, p. 144 Future Tech Prevention, p. 27 Grenade and Bystanders, p. 203 Hologram, p. 4 Hungry Bear, pp. 204-​05 Involuntary “Punishment”, p. 104 Malicious Bystander, p. 127 Massive Man, p. 85 Massive Man Tool, p. 92 Mine Shafts, p. 24 Minor Punishment, p. 104 Mixed-​Motive Bystander, p. 127 Momma Bear, p. 205 Murderer, p. 59 Plank, p. 185 Preventing Murder, p. 26 Rescue Ahead, p. 90 Risky Vaccine, p. 28

219

River, p. 201 Running from the Bear, p. 184 Shielding the Massive Man, p. 177 Sniper, p. 40 Spooky Prevention, p. 27 Startled Prevention, p. 27 Taunting, p. 160 Trolley Switch, p. 51 Trolley Switch Tool, p. 92 Trolley Switch with Risk of Looping, p. 97 Trolley Turners, p. 179 Voluntary “Punishment”, p. 104

220  |  Table of Cases

GLOSSARY OF TERMS

Action:  an act performed for a particular reason. Agent:  a person viewed as someone with choices to make. Agent–​agent division:  the dividing of the world’s useful material along property lines. Agent–​patient frontier:  the normative space in which patient-​claims come up against agent-​claims. Agent–​patient inference:  if an agent has a right to do X, then others may not justify acting in ways that interfere with her doing X by appeal to the fact that what they do interferes with her doing X, unless the conditions justifying competition apply. Autonomy principle:  we all have our own lives to lead; the space of rights must give us substantial freedom to do so, while also holding us accountable for our choices. Baseline freedom:  the freedom an agent has to act based on the property in her toolkit, augmented by any empowering claims that may put items in her toolkit for limited purposes. This freedom can be restricted by restricting claims. Baseline proportionality:  one condition for an act being permissible is that the claims on an agent that call for her to do it outweigh the claims on her not to do it. This sort of proportionality considers only the balance of patient-​claims, and looks only at the option of acting versus not acting. Basic patient-​claims:  claims of patients on agents that presuppose no special relationship between patient and agent. Claim:  a pro tanto right. Claim of non-​sacrifice:  the claim to be allowed to act to avoid serving as a means of protecting others from harm. Comparative proportionality:  one condition for an act being permissible is that it not unnecessarily cause harm. Given that the Pareto condition in necessity rarely applies, the question is whether, given a choice between two options, the agent can justify choosing one unsupported by the balance of patient-​claims because of her own agent-​ right to take an option that is costlier to her off the table. Contravening a claim:  acting contrary to the normative force of a claim. Eliminative killing:  killing a threat when the justification for doing so is that it eliminates or reduces the threat posed to potential victims.

221

Empowering claims: claims that push to put property that belongs to one person (artificial or real) in the toolkit of another for the limited purpose of achieving a particular goal. Equality principle:  we are all fundamentally equal as members or “citizens” of the space of rights. Infringement model: the conception of rights according to which rights have to be weighed against competing rights and competing consequentialist considerations and according to which rights do not necessarily control what may be done. If a right is not outweighed, it must be respected; failure to do so violates the right and wrongs the person. If a right is outweighed, it is permissible (and perhaps mandatory) to disrespect it. Doing so “infringes” the right, wrongs the right-​holder, and gives rise to a claim for compensation but is permissible nonetheless. Innocent aggressor: a person who threatens to use harmful force against another, knowing that the force would be harmful but non-​culpably ignorant of the fact that she lacks the right to do so. Innocent threat:  a person who is a threat to another because of the involuntary motions of her body. Innocent threatener: a person who is a threat to another because of some act she performs when she is non-​culpably ignorant of the fact that her act would endanger another. Liability to harm X:  forfeiture of the primary claim to avoid suffering X, leaving a weak residual claim not to suffer X unless suffering X is unavoidable if a proportionate good is to be achieved, where the proportionate good can be much less significant to the person with a claim to enjoy it than X is to the person liable to suffer it. Mechanics of claims:  the conception of rights according to which, at step one, pro tanto rights emerge from the balance of basic patient-​claims on agents, modified by whatever agent-​claims the agent can bring to bear, and at step two, pro tanto agent-​rights may be further limited by special claims on the agent. Necessity: if an act that serves the interests of some but harms others can serve the interests of the former at less cost to the latter, then it is unnecessary and impermissible. This Pareto condition is hard to meet, and when not met, the issue turns into one of comparative proportionality. Negative agent-​claims:  claims an agent has not to have to respond to the balance of patient-​claims on her when those claims call on her to act. These are normally quite strong and include the agent’s interest in benefiting from not performing the act. Patient:  a person viewed as having welfare interests that are affected by the choices of agents, including her own choices insofar as they might harm her. Patient-​claim:  the claim a patient has to avoid being harmed by an agent or to receive a benefit from an agent. An agent’s interest in benefiting from her own act registers as a patient-​claim. Positive agent-​claims:  claims an agent has to do what she wants to do, despite the balance of patient-​claims pushing her not to do so. These have no weight if the patient-​ claims pushing to restrict the agent serve to protect against significant harm. The agent’s interest in performing the act registers as a patient-​claim. Property claims:  claims a patient has over some item that puts it in his toolbox and outside the toolbox of others and thereby limits their basic freedom.

222  |  Glossary of Terms

Restricting claims:  claims that push to restrict an agent relative to her baseline freedom and thereby also push to make others with competing restricting claims normatively worse off. Restricting claims principle:  restricting claims are substantially weaker than property claims, all else equal. Space of rights:  the normative space grounded in the autonomy, equality, and welfare principles and structured by the mechanics of claims. Special patient-​claims:  patient-​claims on an agent that presuppose some special relationship between patient and agent (e.g., promisee to promisor). These can restrict the liberty agents otherwise would have but cannot augment it. Toolkit:  the property that an agent may take herself to have the baseline freedom to use. Welfare principle:  our welfare matters, and that value aggregates and affects the rights we have.

Glossary of Terms  |  223

BIBLIOGRAPHY

Alexander, Larry. “Deontology at the Threshold.” San Diego Law Review 37 (2000): 893–​912. —​ —​ —​ . “The Means Principle.” In Legal, Moral, and Metaphysical Truths:  The Philosophy of Michael S.  Moore, edited by Kimberly Kessler Ferzan and Stephen Morse, 251–​64. New York: Cambridge University Press, 2016. Alexander, Larry, and Kimberly Kessler Ferzan. Crime and Culpability:  A Theory of Criminal Law. With Stephen J. Morse. New York: Cambridge University Press, 2009. Alexander, Larry, and Kimberly Kessler Ferzan. Reflections on Crime and Culpability: Problems and Puzzles. New York: Cambridge University Press, 2018. Alexander, Larry, and Michael Moore. “Deontological Ethics,” in The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. Stanford University, 2016. http://​plato.stanford.edu/​archives/​win2016/​entries/​ethics-​deontological/​. Anscombe, G. E. M. “Who Is Wronged?” Oxford Review 5 (1967): 16–​17. Aristotle. The Politics. Translated by Lord Carnes. Chicago:  University of Chicago Press, 1984. Bazargan, Saba. “Killing Minimally Responsible Threats.” Ethics 125 (2014): 114–​36. Benbaji, Yitzhak. “A Defense of the Traditional War Convention.” Ethics 118 (2008): 464–​95. Boorse, Christopher, and Roy Sorensen. “Ducking Harm.” Journal of Philosophy 85 (1988): 115–​34. Broome, John. “Uncertainty and Fairness.” Economic Journal 94 (1984): 624–​32. Calabresi, Guido, and A. Douglas Melamed. “Property Rules, Liability Rules, and Inalienability:  One View of the Cathedral.” Harvard Law Review 85 (1972): 1089–​1128. Cushman, Fiery. “The Psychological Origins of the Doctrine of Double Effect.” Criminal Law and Philosophy 10 (2016): 763–​76. Draper, Kai. War and Individual Rights. New York: Oxford University Press, 2016. Dreier, Jamie. “In Defense of Consequentializing.” Oxford Studies in Normative Ethics 1 (2011): 97–​118.

225

Dressler, Joshua. Understanding Criminal Law. 7th ed. San Francisco:  Matthew Bender, 2015. Duff, R. A. Punishment, Communication, and Community. New York: Oxford University Press, 2001. Dworkin, Ronald. Sovereign Virtue: The Theory and Practice of Equality. Cambridge, MA: Harvard University Press, 2000. Fabre, Cécile. “Guns, Food, and Liability to Attack in War.” Ethics 120 (2009): 36–​63. —​—​—​. Cosmopolitan War. New York: Oxford University Press, 2012. Feinberg, Joel. Rights, Justice, and the Bounds of Liberty. Princeton, NJ:  Princeton University Press, 1980. Ferzan, Kimberly Kessler. “Forfeiture and Self-​Defense.” In The Ethics of Self-​Defense, edited by Christian Coons and Michael Webber, 233–​ 53. New  York:  Oxford University Press, 2016. —​—​—​. “Omissions, Acts, and the Duty to Rescue.” In The Ethics and Law of Omissions, edited by Dana Kay Nelkin and Samuel C. Rickless, 217–​34. New  York:  Oxford University Press, 2017. Foot, Philippa. “The Problem of Abortion and the Doctrine of the Double Effect.” In Virtues and Vices, 19–​32. Berkeley:  University of California Press, 1978. First published 1967. Frowe, Helen. Defensive Killing. New York: Oxford University Press, 2014. Gilbert, Margaret. Rights and Demands:  A Foundational Inquiry. Oxford:  Oxford University Press, 2018. Goldhagen, Daniel J. Hitler’s Willing Executioners:  Ordinary Germans and the Holocaust. New York: Vintage Books, 1997. Grabczynska, Arlette, and Kimberly Kessler Ferzan. “Justifying ‘Killing in Self-​ Defence.’” Journal of Criminal Law and Criminology 99 (2009): 235–​53. Grayling, A. C. Among the Dead Cities:  The History and Moral Legacy of the WWII Bombing of Civilians in Germany and Japan. New York: Walker, 2006. Hanser, Matthew. “Killing, Letting Die and Preventing People from Being Saved.” Utilitas 11 (1999): 277–​95. Haque, Adil. Law and Morality at War. New York: Oxford University Press, 2017. Harman, Elizabeth. “Morally Permissible Moral Mistakes.” Ethics 126 (2016): 366–​93. Hohfeld, Wesley. “Some Fundamental Legal Conceptions as Applied in Judicial Reasoning.” Yale Law Journal 23 (1913): 16–​59. Hosein, Adam. “Are Justified Aggressors a Threat to the Rights Theory of Self-​Defense?” In How We Fight: Ethics in War, edited by Helen Frowe and Gerald Lang, 87–​103. New York: Oxford University Press, 2014. Hurka, Thomas. “Trolleys and Permissible Harm.” In The Trolley Problem Mysteries, by F. M. Kamm, edited by Eric Rakowski, 135–​50. New York: Oxford University Press, 2015. Husak, Doug. Ignorance of Law: A Philosophical Inquiry. New York: Oxford University Press, 2016. Kagan, Shelly. The Limits of Morality. New York: Oxford University Press, 1989. Kamm, Frances. Creation and Abortion. New York: Oxford University Press, 1992a. —​—​—​. “Non-​Consequentialism, the Person as an End-​in-​Itself, and the Significance of Status.” Philosophy and Public Affairs 21 (1992b): 354–​89.

226  | Bibliography

—​—​—​. Intricate Ethics. New York: Oxford University Press, 2007. —​—​—​. “Substitution, Subordination and Responsibility: A Reply to Scanlon, McMahan and Rosen.” Philosophy and Phenomenological Research 80 (2010): 702–​22. —​—​—​. The Trolley Problem Mysteries. New York: Oxford University Press, 2015. Kant, Immanuel. Grounding for the Metaphysics of Morals. Translated by James W. Ellington. 1785. Reprint, Indianapolis, IN: Hackett Publishing, 1993. —​—​—​. The Metaphysics of Morals. Translated by Mary Gregor. 1797. Reprint, New York: Cambridge University Press, 1991. Kleingeld, Pauline. “The Principle of Autonomy in Kant’s Moral Theory:  Its Rise and Fall.” In Kant on Persons and Agency, edited by Eric Watkins, 61–​ 80. Cambridge: Cambridge University Press, 2018. Kolodny, Niko. “Scanlon’s Investigation:  The Relevance of Intent to Permissibility.” Analytic Philosophy 52 (2011): 100–​23. Lazar, Seth. “Responsibility, Risk, and Killing in Self-​ Defense.” Ethics 119 (2009): 699–​728. —​—​—​. “Necessity in Self-​Defense and War.” Philosophy and Public Affairs 40 (2012): 3–​44. —​—​—​. Sparing Civilians. New York: Oxford University Press, 2015. —​—​—​. “Method in the Morality of War.” In The Oxford Handbook of Ethics of War, edited by Seth Lazar and Helen Frowe. Oxford University Press, 2016. doi:10.1093/​ oxfordhb/​9780199943418.013.25. —​—​—​. “Deontological Decision Theory and Agent-​Centered Options.” Ethics 127 (2017): 579–​609. —​—​—​. “Strengthening Moral Distinction.” Law and Philosophy 37 (2018): 327–​49. Liao, S. Matthew. “Intentions and Moral Permissibility: The Case of Acting Permissibly with Bad Intentions.” Law and Philosophy 31 (2012): 703–​24. Liberto, Hallie. “The Moral Specification of Rights:  A Restricted Account.” Law and Philosophy 33 (2014): 175–​206. Lippert-​Rasmussen, Kasper. “Justice and Bad Luck.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. Stanford University, 2018. https://​plato.stanford.edu/​archives/​sum2018/​entries/​justice-​bad-​luck/​. Locke, John. Second Treatise of Government. 1690. Reprint, Indianapolis, IN: Hackett Publishing, 1980. McMahan, Jeff. “Self-​Defense and the Problem of the Innocent Attacker.” Ethics 104 (1994): 252–​90. —​—​—​. The Ethics of Killing:  Problems at the Margins of Life. New  York:  Oxford University Press, 2002. —​—— ​ ​. “The Morality of War and the Law of War.” In Just and Unjust Warriors, edited by David Rodin and Henry Shue, 19–​43. New  York:  Oxford University Press, 2008. —​—​—​. Killing in War. New York: Oxford University Press, 2009a. —​—​—​. “Self-​Defense Against Morally Innocent Threats” and “Reply to Commentators.” In Criminal Law Conversations, edited by Paul H. Robinson, Kimberly Ferzan, and Stephen Garvey, 385–​94, 404–​406. New York: Oxford University Press, 2009b. —​—​—​. “Proportionate Defense.” Journal of Transnational Law and Policy 23 (2013–​14): 1–​36.

Bibliography  |  227

—​—​—​. “Proportionality and Just Cause:  A Comment on Kamm.” Journal of Moral Philosophy 11 (2014a): 428–​53. —​—​—​. “Self-​Defense Against Justified Threateners.” In How We Fight: Ethics in War, edited by Helen Frowe and Gerald Lang, 104–​37. New  York:  Oxford University Press, 2014b. Melzer, Nils. Interpretive Guidance on the Notion of Direct Participation in Hostilities Under International Humanitarian Law. Geneva:  International Committee of the Red Cross, 2009. https://​www.icrc.org/​eng/​assets/​files/​other/​icrc-​002-​0990.pdf. Mill, John Stuart. On Liberty. 1859. Reprint, New York: W. W. Norton, 1975. Mogensen, Andreas. “Should We Prevent Optimific Wrongs?” Utilitas 28 (2016): 215–​26. Moore, Michael. “Steinhoff, Hohfeld, and Self-​Defense.” San Diego Law Review 55 (2018): 315-​38. Morris, Herbert. “Persons and Punishment.” Monist 52 (1968): 475–​501. Nagel, Thomas. “Moral Luck.” In Mortal Questions, 24–​38. New  York:  Cambridge University Press, 1979a. —​ —​ —​ . “War and Massacre.” In Mortal Questions, 53–​ 74. New  York:  Cambridge University Press, 1979b. —​ —​ —​ . “Personal Rights and Public Space.” Philosophy and Public Affairs 24 (1995): 83–​107. Nozick, Robert. Anarchy, State, and Utopia. New York: Basic Books, 1974. Oberdiek, John. “Lost in Moral Space: On the Infringing/​Violating Distinction and Its Place in the Theory of Rights.” Law and Philosophy 23 (2004): 325–​46. —​—​—​. “Specifying Rights out of Necessity.” Oxford Journal of Legal Studies 28 (2008): 127–​46. Otsuka, Michael. “Killing the Innocent in Self-​Defense.” Philosophy and Public Affairs 23 (1994): 74–​94. —​ —​ —​ . “The Moral Responsibility Account of Liability to Defensive Killing.” In The Ethics of Self-​Defense, edited by Christian Coons and Michael Weber, 51–​68. New York: Oxford University Press, 2016. Øverland, Gerhard. “Moral Obstacles: An Alternative to the Doctrine of Double Effect.” Ethics 124 (2014): 481–​506. Parfit, Derek. “Innumerate Ethics.” Philosophy and Public Affairs 7 (1978): 285–​301. —​—​—​. Reasons and Persons. New York: Oxford University Press, 1986. —​—​—​. On What Matters. Vol. 1. New York: Oxford University Press, 2011. Pettit, Philip. “Responsibility Incorporated.” Ethics 117 (2007): 171–​201. Quinn, Warren. “Actions, Intentions, and Consequences: The Doctrine of Double Effect.” In Morality and Action, 175–​93. New York: Cambridge University Press, 1993. First published 1989. Quong, Jonathan. “Killing in Self-​Defense.” Ethics 119 (2009): 507–​37. —​—​—​. “Liability to Defensive Harm.” Philosophy and Public Affairs 40 (2012): 45–​77. Ramakrishnan, Ketan. “Treating People as Tools.” Philosophy and Public Affairs 44 (2016): 133–​65. Rawls, John. A Theory of Justice. New York: Oxford University Press, 1971. —​—​—​. Political Liberalism. New York: Columbia University Press, 1993. Raz, Joseph. The Morality of Freedom. New York: Oxford University Press, 1986. Regan, Donald. Utilitarianism and Cooperation. New York: Oxford University Press,  1982.

228  | Bibliography

Reichenberg, Gregory. “Just War and Regular War: Competing Paradigms.” In Just and Unjust Warriors, edited by David Rodin and Henry Shue, 193–​213. New York: Oxford University Press, 2008. Renzo, Massimo. “Political Authority and Unjust Wars.” Philosophy and Phenomenological Research (2018) doi:10.1111/​phpr.12487. Richardson, Henry. “Moral Entanglements: Ad Hoc Intimacies and Ancillary Duties of Care.” Journal of Moral Philosophy 9 (2012): 376–​409. Rodin, David. War and Self-​Defense. New York: Oxford University Press, 2002. —​—​—​. “The Myth of National Self-​Defense.” In The Morality of Defensive War, edited by Cécile Fabre and Seth Lazar, 69–​89. New York: Oxford University Press, 2014. Ross, W. D. The Right and the Good. 1930. Reprint, Indianapolis, IN:  Hackett Publishing 1988. Saunders, Ben. “A Defense of Weighted Lotteries in Life Saving Cases.” Ethical Theory and Moral Practice 12 (2009): 279–​90. Scanlon, T. M. What We Owe to Each Other. Cambridge, MA:  Harvard University Press, 1998. —​—​—​. Moral Dimensions: Permissibility, Meaning, Blame. Cambridge, MA: Harvard University Press, 2008. Scheffler, Samuel. The Rejection of Consequentialism:  A Philosophical Investigation of the Considerations Underlying Rival Moral Conceptions. New  York:  Oxford University Press, 1982. Segev, Re’em. “Should We Prevent Deontological Wrongdoing?” Philosophical Studies 173 (2016): 2049–​68. Sellars, Wilfrid. Empiricism and the Philosophy of Mind. Cambridge, MA:  Harvard University Press, 1997. Sen, Amartya. “Rights and Agency.” In Consequentialism and Its Critics, edited by Samuel Scheffler, 187–​223. New York: Oxford University Press, 1988. First published 1982. Setiya, Kieran. “Love and the Value of Life.” Philosophical Review 123 (2014): 251–​80. Shafer-​ Landau, Russ. “Specifying Absolute Rights.” Arizona Law Review 37 (1995): 209–​25. Shiffrin, Seana. “Harm and Its Moral Significance.” Legal Theory 18 (2012): 357–​98. Simons, Kenneth. “Self-​Defense, Necessity and the Duty to Compensate, in Law and Morality.” San Diego Law Review 55 (2018): 357-​79. Smith, Angela. “Responsibility for Attitudes:  Activity and Passivity in Mental Life.” Ethics 115 (2005): 236–​71. Stanczyk, Lucas. “Productive Justice.” Philosophy and Public Affairs 40 (2012): 144–​64. Statman, Daniel. “On the Success Condition for Legitimate Self-​Defense.” Ethics 118 (2008): 659–​86. Steinhoff, Uwe. “Rights, Liability, and the Moral Equality of Combatants.” Journal of Ethics 12 (2012): 339–​66. —​—​—​. “Against a ‘Combined Liability-​Lesser-​Evil Justification.’” Philosophia (2018). https://​doi.org/​10.1007/​s11406-​018-​9992-​7. —​—​—​. Just War Theory:  Self-​Defense, Necessity, and the Ethics of Armed Conflict. New York: Oxford University Press (forthcoming). Sussman, David. “Is Agent-​Regret Rational?” Ethics 128 (2018): 788–​808. Tadros, Victor. The Ends of Harm. New York: Oxford University Press, 2011.

Bibliography  |  229

—​—​—​. “Causation, Culpability, and Liability.” In The Ethics of Self-​Defense, edited by Christian Coons and Michael Webber, 110–​30. New  York:  Oxford University Press, 2016a. —​—​—​. “Permissibility in a World of Wrongdoing.” Philosophy and Public Affairs 44 (2016b): 101–​32. —​—​—​. “Causal Contributions and Liability.” Ethics 128 (2018): 402–​31. —​—​—​. “To Do, To Die, To Reason Why; Why Law and Morality Should Converge.” Unpublished manuscript (last modified Microsoft Word file dated Dec. 6, 2017). Taurek, John. “Should the Numbers Count.” Philosophy and Public Affairs 6 (1977): 293–​316. Temkin, Larry. Rethinking the Good: Moral Ideals and the Nature of Practical Reasoning. New York: Oxford University Press, 2012. Thomson, Judith Jarvis. “Killing, Letting Die, and the Trolley Problem.” Monist 59 (1976): 204–​17. —​—​—​. The Realm of Rights. Cambridge, MA: Harvard University Press, 1990. —​—​—​. “Self-​Defense.” Philosophy and Public Affairs 20 (1991): 283–​310. —​—​—​. “Physician Assisted Suicide: Two Arguments.” Ethics 109 (1999): 497–​518. —​—​—​. “Turning the Trolley.” Philosophy and Public Affairs 36 (2008): 359–​74. Tomlin, Patrick. “Innocence Lost:  A Problem for Punishment as Duty.” Law and Philosophy 36 (2017): 225–​54. Waldron, Jeremy. “A Right to do Wrong.” In Liberal Rights. 63–​87. New York: Cambridge University Press, 1993. First published 1981. Walen, Alec. “Doing, Allowing, and Disabling: Some Principles Governing Deontological Restrictions.” Philosophical Studies 80 (1995): 183–​215. —​ —​ —​ . “Consensual Sex Without Assuming the Risk of Carrying an Unwanted Fetus: Another Foundation for the Right to an Abortion.” Brooklyn Law Review 63 (1997): 1051–​1140. —​—​—​. “Reasonable Illegal Force:  Justice and Legitimacy in a Pluralistic, Liberal Society.” Ethics 111 (2001): 344–​73. —​—​—​. “Permissibly Encouraging the Impermissible.” Journal of Value Inquiry 38 (2004): 341–​54. —​ —​ —​ . “The Doctrine of Illicit Intentions.” Philosophy and Public Affairs 34 (2006): 39–​67. —​—​—​. “Crime, Culpability and Moral Luck:  Comment on Alexander, Ferzan and Morse.” Law and Philosophy 29 (2010): 373–​84. —​—​—​. “Wrongdoing Without Motives: Why Tadros Is Wrong About Wrongdoing and Motivation.” Law and Philosophy 32 (2013): 217–​40. —​—​—​. “Transcending the Means Principle.” Law and Philosophy 33 (2014): 427–​64. —​ —​ —​ . “The Restricting Claims Principle Revisited.” Law and Philosophy 35 (2016): 211–​47. —​—​—​. “Eliminative Killing and the Targeting of Noncombatants: Comments on Seth Lazar’s Sparing Civilians.” Law and Philosophy 37 (2017): 313–​25. —​—​—​. “A Right to Cause Harm.” San Diego Law Review 55 (2018): 381-​417. —​ —​ —​ . “Risk and Limited Aggregation.” Unpublished manuscript (last modified Microsoft Word file dated Nov. 26, 2018).

230  | Bibliography

Walen, Alec, and David Wasserman. “Agents, Impartiality, and the Priority of Claims over Duties: Diagnosing Why Thomson Still Gets the Trolley Problem Wrong by Appeal to the ‘Mechanics of Claims.’” Journal of Moral Philosophy 9 (2012): 545–​71. Walzer, Michael. Just and Unjust Wars. New York: Basic Books, 1992. Wedgwood, Ralph. “Defending Double Effect.” Ratio 24 (2011): 384–​401. Wellman, Christopher. “The Rights Forfeiture Theory of Punishment.” Ethics 122 (2012): 371–​93. Wenar, Leif. “The Nature of Claim-​Rights.” Ethics 123 (2013): 202–​29. Woodward, Fiona. “The New Problem of Numbers in Morality.” Ethical Theory and Moral Practice 17 (2014): 631–​41. Woodward, James. “The Non-​Identity Problem.” Ethics 96 (1986): 804–​31.

Bibliography  |  231

INDEX

acts action vs., 9, 37–​38, 43–​44 reasons and, 128–​30 agency causal roles and, 203–​9 indirect agency, 17, 155, 159–​65 intervening agency, 19–​20, 115n23, 155, 159–​60, 160n9, 163–​65, 195–​97, 200n5, 200, 201–​2, 203–​9,  217 joint agency, 29n11, 168 mixed agency, 126–​30 opportunistic agency, 38n30, 38, 39, 41, 125, 126n5, 134–​35, 135n28 See also causal roles; eliminative killing agent-​agent division, 87–​88, 89–​90, 95 See also restricting claims principle agent-​centered prerogatives, 6–​7, 73–​76, 110n21, 143–​44, 175, 188 agent-​claims use of term, 57n39, 65 anti-​paternalism and,  69–​70 being a threat and, 149 culpability and, 188–​89 harm and, 172–​78 negative agent-​claims, 66, 68–​71, 73, 75–​76, 89, 93, 94, 95, 122, 131–​32, 164, 171–​72, 173–​74, 179, 183n16, 183–​84, 186, 204–​5, 210

non-​sacrifice and, 169n24, 172, 175–​76, 177, 181, 184–​85, 195–​97, 203–​11,  215 positive agent-​claims, 68–​69, 71–​76, 174, 188, 205 welfare interest and, 70 agent-​neutral value, 56–​57, 58n40, 58, 59 agent-​patient frontier, 87–​88, 93 See also restricting claims principle agent-​patient inference, 13, 171, 172–​73, 178–​80, 187–​93, 212, 215 agent-​relative topology, 58–​59, 59n42 aggregation, 50n20, 50–​53, 52n27, 59, 168 aggressors culpable aggressors, 2, 34–​35, 52–​53, 65–​66,  210–​11 innocent aggressor, 6, 7, 154n1, 157–​58 non-​culpable aggressors, 154–​56, 157, 159 unjust aggressor, 34–​35, 212–​13 See also combatants; threateners; threats Alexander, Larry, 119–​23 alienation, 7n28, 52–​53 allowing. See causation authority authority of individuals, 33, 114 reductive individualism and, 30–​33 social contracts and, 32–​33 state or governmental authority, 30–​33, 36–​37,  181–​82

233

autonomous capacity, 48–​50, 54, 204–​5,  207 autonomy principle use of term, 47, 48–​49 anti-​paternalism,  69–​70 competition and, 183–​84 conflict with other principles, 115–​16,  121 freedom to lead one’s own life and, 74, 118, 122 freedom not to serve as a tool and, 69, 75–​76, 89, 118 harmonization with other principles and, 79, 89, 94, 115–​16, 118 Kantian autonomy, 48 negative agent claims and, 69, 75–​76, 89, 118, 172 owned luck and, 60–​65, 102–​3 partiality to self, 67 prioritization over the welfare principle, 89, 94, 118 property claims and, 88–​89 respect for individuals, 52 restricting claims principle and, 89–​90 topology of rights and, 58–​59 value vs, 47 See also mechanics of claims; restricting claims principle; space of rights baseline freedom, 84–​90, 93–​95, 136, 174–​75, 176–​77,  181–​82 See also restricting claims principle baseline proportionality, 10–​11, 12, 13, 15n46, 166 benefits, use of term, 54–​55 benevolence, 127, 130–​31 bystanders. See innocent bystanders causal roles counterfactuals and, 86 intervening agency and, 203–​9 just war theory and, 20, 136 justification and, 9, 41–​42, 142–​43 means or tool, 82–​83, 136, 175n6, 199–​200

234  | Index

mechanics of claims and, 112, 124 noncausal flip side, 140–​41 using, 41 See also agency; innocent bystander; innocent victim; side effect; threat causation allowing (or passive causation), 52, 90, 126–​27,  172–​77 doing (or active causation) vs allowing, 52,  172–​78 ducking harm, 172–​78 enabling,  162–​63 intention and, 9, 38 redirection or deflection of threats, 64,  139–​43 See also agency; eliminative killing chance. See luck children,  48–​50 civilians use of term, 14 noncombatants vs., 14 permissible eliminative killing and,  14–​17 principle of distinction and, 15n45, 15–​16,  17–​18 See also noncombatants claims use of term as pro tanto rights, 8, 57n39 aggregation of, 50n20, 50–​53, 59, 60, 168 alienation of claims, 7n28, 52–​53 basic claims, 65–​67, 73, 75, 77–​78 contravened,  80–​81 empowering claims, 67, 83n88, 93–​96, 98–​99, 114, 122, 174–​76 justificatory role for setting the nature of claims, 176–​77 property claims, 67–​68, 87–​99, 116, 122, 124, 150–​53, 174–​75, 183n16, 210 special claims, 11n37, 65, 67–​68, 74–​75, 77–​79, 95, 102n4, 103, 172, 188–​89 See also agent-​claims; infringement model; innocent bystanders; mechanics of claims; patient-​claims; restricting claims principle

combatants use of term, 14–​63, 15n45 terrorism and, 13–​14 unjust combatants, 6, 6n22, 17, 34–​35, 138, 154–​55, 159–​63, 165–​70,  212–​16 comparative proportionality, 10–​11n34, 11, 12 compensation duty to compensate, 100–​8 infringement model and the duty to compensate, 79–​80, 101–​2, 103–​8,  147–​49 tort duty to compensate, 80n80, 103n7 transferred luck and, 102–​3, 108 competition exception to agent-​patient inference, 171, 178–​79,  181–​84 just war theory and, 193 justified competition, 181–​87 responsiveness and, 179–​80 consequentialism geometry of, 58, 58n40 mistaken appeal to, 36, 53, 58, 59n42, 81,  176–​77 reductive individualism and, 36 rights-​consequentialism hybrid, 113–​14, 144–​45,  146–​48 sophisticated conception of, 81, 148 threshold deontology and, 120, 121 constructivism, 48n13 contractualism, 46–​47, 51–​53, 128n7 counterfactuals agency and, 29, 128–​29 causal roles and, 39–​42, 128–​29 intentions and, 39n33, 39–​42, 128–​29 misused, 39n35, 39–​42, 83n88, 86, 197–​200 restricting claims principle and, 84–​85 culpability agent-​patient inference and, 188–​89 culpable threats, 3, 6, 35–​36, 143–​44 eliminative killing and, 213–​16 forfeiture and the threshold for forfeiture and, 7, 110–​11, 159, 168, 170, 213–​14, 217

gravity of a wrong in contrast to, 131 minimally culpable agents, 161n11,  164–​65 negligence and, 63, 110–​11, 169–​70 non-​culpable noncombatants, 165,  213–​14 recklessness and, 63, 110–​12, 188–​89 democracy, 32, 210 deontology use of term, 36, 46–​47, 51 intentions and, 126–​34 paradox of, 59n42 See also means principle; restricting claims principle; threshold deontology dignitary good, harm, loss or wrong, 130–​32, 136,  210–​11 discounting and intervening agency,  208–​9 distinction. See principle of distinction distributive justice, 64, 98–​99 doctrine of double effect, 136n29, 136 Draper, Kai, 162, 164–​65 ducking harm. See causation duty duty of justice not tied to rights, 56–​57 duty of professional combatants, 192 duty to obey the law, 13, 36–​37, 44n2, 56, 57, 193–​94 patient-​rights and, 66 social contract and, 31–​34 See also compensation; mechanics of claims; rights eliminative killing use of term, 8–​9, 10 causal structure of, 37–​42 culpability and, 213–​14 directness of threat reduction and, 14 intention and, 9, 125–​30, 136 necessity condition, 11–​12 proportionality and, 10–​11, 12, 213 terrorism, contrast with, 13n41, 13–​14, 14n43, 166–​67, 213 See also permissible eliminative killing; side effect

Index  |  235

equality principle use of term, 47, 49 animals and, 49–​50 conflict with other principles, 115–​16 harmonization with other principles and, 79, 115–​16 relevance to law, 72–​73 relevance to mechanics of claims, 67, 75–​76,  216 value vs, 47, 49 See also mechanics of claims; restricting claims principle; space of rights excuse. See justification fairness, 56, 61–​62, 85, 149, 185 Ferzan, Kimberley Kessler, 76n77, 174n5, 195n1 forfeiture compensation and, 109–​10 culpability and, 110–​11, 159, 168,  213–​14 just war theory and, 217 liability and, 4–​5, 17–​18, 30, 33–​34, 35–​36, 80n80, 155, 157–​58, 161n11, 167, 169–​70, 216, 217 limits of, 65–​66 mechanics of claims and, 80–​81 normative gulf and, 111–​13, 150–​51,  217 responsible choice and, 3–​4, 8, 109–​10,  111 threshold for, 7, 110–​11, 159, 168, 170, 213–​14,  217 See also alienation; waiver Frowe, Helen, 17–​18, 169–​70, 200–​2 Haque, Adil, 163, 164–​65, 167–​68 harm benefits vs. harm, 11, 54–​55 ducking harm, 172–​78 morally discounted harm and, 112–​13,  146–​49 See also causation; proportionality harmonization, 78–​79, 89, 94, 115–​23, 118n29,  183–​84

236  | Index

heart lottery, 60–​62, 152 Hohfeld, Wesley, 49–​50, 57n39, 69, 77–​78,  178–​79 Hurka, Thomas,  141–​42 imminence, 12, 33 infringement model use of term, 19, 79–​82 compensation and, 101–​3, 104–​8, 123 consequentialism and, 123 fit with common rights discourse, 80 hybrid rights–​lesser-​evil arguments and, 144–​45,  150–​51 incompleteness of, 45, 79–​80, 81,  216–​17 intervening agency and, 217 just war theory and, 217 lesser evil and, 138, 144–​45, 146–​48,  150 means principle and, 217 mechanics of claims vs., 19, 79–​82, 100–​8, 123, 192, 217 normative gulf and, 111–​13, 150–​51,  217 Occam’s razor and, 113–​14 rightfully infringing and wronging, 45, 108–​11, 119, 192 specificationism and, 45n6, 45, 101 threshold deontology and, 119 tort law and, 80n80 innocent bystanders claims of, 111–​13 innocent threats and, 2, 137–​38, 143 innocent victims and, 138–​43 intentions and, 125–​30 luck and, 139, 144 restricting claims and, 139, 143 right not to be killed, 2, 6 right to act in self-​defense, 18–​19,  187–​94 innocent threateners use of term, 154 claims of, 157–​58, 159 direct innocent threateners, 155,  157–​59 forfeiting innocent status, 213–​14

indirect innocent threateners, 155, 159–​69,  213–​14 threatening in numbers, 165–​68 innocent threats use of term, 2 claims of, 111–​13, 149–​50, 151–​53 forfeiture of rights and, 2–​8 innocent bystanders and, 2, 137–​38,  143 innocent victims and, 143–​48 innocent victims use of term, 3 claims of, 111–​13 culpable aggressors and, 52–​53 innocent bystanders and, 138–​43 innocent threats and, 143–​48 intention Benevolence,  127–​31 causal structure and, 9, 38 counterfactuals and, 39n33, 39–​42,  128–​29 deontological relevance of, 127–​30, 132–​34, 136, 138 disrespectful intention, 128, 130–​34 doctrine of double effect, 136n29, 136 eliminative killing and, 9, 125–​30, 136 illicit intentions and reasons, 125–​26, 127–​30, 133–​34,  136 just war theory and, 124–​25, 135, 136 malice,  127–​34 mixed motive, 125–​30 offensive intentions, 131–​34 principle of distinction and, 124–​25,  136 restricting claims principle and, 132–​33 scope of, 133 secondary significance of, 125, 133 intuitions, 21–​22, 51, 80n80, 84, 90, 97, 113–​14, 131, 141, 143–​44, 163–​64, 184, 196, 197, 208n23 just war theory agent-​patient inference and, 13,  191–​93 causal role and, 20, 136 competition and, 193

deontological grounding of, 136 forfeiture and, 217 infringement model and, 217 intention and, 124–​25, 135, 136 intervening agency and, 19–​20 killing civilians or noncombatants and, 14–​15, 17–​18,  212–​15 mechanics of claims and, 217 non-​sacrifice and,  209–​11 objections to reductive individualism and,  34–​37 principle of distinction and, 124–​25,  136 reductive individualism and, 29–​30 restricting claims principle and, 136 restriction on killing just combatants and, 18–​19,  191–​94 rights discourse and, 6, 8, 217 self-​defense and, 1–​2, 6, 8, 29–​30, 36,  191–​94 justice. See distributive justice; retributive justice justification acts and actions, 9, 37–​38, 43–​44, 97 causal roles and, 9, 41–​42, 142–​43 claims, the nature of, 176–​77 evidence-​relative justification and, 12,  24–​26 excuse vs, 6, 10, 27, 73, 73n70, 203 fact-​relative justification and, 12,  24–​29 hybrid rights–​lesser-​evil arguments, 144–​45, 146–​48,  150 lesser-​evil justification and, 3, 6, 81, 110–​11, 112–​13, 137, 138, 140–​41, 144–​45, 146–​48,  150 restricting claims principle and, 97–​98 risk and, 97–​98 threshold deontology and, 115–​23 Kamm, Frances, 139–​40, 142–​43, 166,  191–​92 Kant, Immanuel, 44, 44n3, 48n16, 48 killing. See eliminative killing; opportunistic killing; side effect; terrorism

Index  |  237

law criminal, 12, 32–​33, 63, 131, 131n14, 135 humanitarian (law of armed conflict), 1n2, 1, 12, 15n47, 15–​17, 18–​19, 166–​67,  193 moral relevance of, 13, 36–​37, 44n2, 56, 57, 72–​73, 193–​94, 215–​16 tort law, 80n80, 103n7 Lazar, Seth, 5–​6, 35–​36, 39–​40, 81, 124–​25, 126, 134–​35, 135n28 lesser evil infringement model and, 138, 144–​45, 146–​48,  150 justification and, 3, 6, 81, 110–​11, 112–​13, 137, 138, 140–​41, 144–​45, 146–​48,  150 liability. See forfeiture Liao, Matthew, 131–​32 Liberto, Hallie, 45n6 Locke, John, 33, 98–​99 love and loyalty, demands of, 78, 115–​16 luck choice vs. brute luck, 61–​64 compensation and, 102–​3, 108 owned luck, 60–​65, 92, 102–​3, 113, 144, 145–​46, 149–​50, 151–​52, 156, 159, 186n23, 186–​87 shifting or transferring luck, 102–​3, 139, 146, 149, 156, 192n34 McMahan, Jeff, 2–​5, 17–​19, 18n60, 30n15, 30, 37, 109–​10, 111, 145n17, 152–​53, 157–​58, 169–​70, 192, 216 means principle, 43, 48, 82–​83, 83n88, 114, 217 See also restricting claims principle mechanics of claims use of term, 65, 66–​67, 83n88 balancing or weighing claims, 65,  66–​68 capaciousness of, 81 causal role and, 112, 124 compensation and, 82, 103–​8 consequentialism and, 81

238  | Index

explanatory power and, 80, 103–​8, 112, 114–​15, 170,  216–​17 framework for restricting claims principle, 83 functional account of, 65–​79, 176–​77 infringement model vs., 19, 79–​82, 100–​8, 123, 192, 217 intentions and, 124, 132–​33 just war theory and, 217 partiality/​impartiality,  66–​68 rights-​consequentialism hybrid and, 113–​14,  150 stages, 65–​66, 95 threshold deontology and, 119, 123 See also agent claims; agent-​ centered prerogatives; claims; means principle; restricting claims principle Mill, John Stuart, 71–​72 Moore, Michael, 119–​21, 178n10 moral residue. See compensation motives. See intentions Nagel, Thomas, 43n1, 46–​47, 62–​63 necessity use of term, 11–​12 eliminative killing and, 14–​15, 52–​53, 81, 109–​10, 111–​12, 120, 155–​56, 159, 193, 196–​97, 212–​13 fact-​relative justification and, 26–​29 killing as a side effect and, 147–​48 imminence and, 12 proportionality and, 10–​11n34, 12 reductive individualism and, 35–​36 using as a means and, 117, 148, 166 noncombatants aid to unjust noncombatants and, 6n22, 17, 34–​35, 138, 154–​55, 159–​63, 165–​70,  212–​16 civilians vs., 14 eliminative killing and, 6, 15n44, 17–​19, 37,  212–​16 intention and, 124–​25, 134–​35, 135n28 killing as a side effect, 215 terrorism and, 13–​14

non-​sacrifice agent-​claims and, 169n24, 172, 175–​76, 177, 181, 184–​85, 195–​97, 203–​11,  215 just war theory and, 209–​11 numbers counting. See aggregation obstacles, moral, 8–​9, 9n31, 83n88, 145 opportunistic killing, 38n30, 38, 41, 125, 126n5, 134–​35, 135n28 other-​defense, 1, 10 Otsuka, Michael, 2–​3, 3n6, 4–​5, 38, 152n27, 158 pacifism, 6, 34–​35 paternalism, 69–​70,  127–​28 patient-​claims use of term, 8 agent-​patient frontier, 87–​88, 93 aggregation and, 50n20 balance of, 13, 49, 51–​53, 57–​58, 65, 67, 68–​71, 73–​78, 83, 88–​89, 95–​96, 103n6, 115–​16, 122, 128–​29, 131–​32, 136, 172, 173, 176, 181, 187–​88, 189–​91,  216–​17 basic claims, 65–​67, 73, 75, 77–​78 being a threat and, 149–​50 innocent threats and, 111–​13, 149–​50,  151–​53 intention and, 132–​34 negative patient-​claims, use of term, 58 positive patient-​claims, use of term, 58 special patient-​claims, 11n37, 65, 67–​68, 74–​75, 77–​79, 95, 102n4, 103, 172, 188–​89 threshold deontology and, 115–​23 welfare principle and, 70, 81, 89, 98–​99 See also claims; forfeiture; property; restricting claims principle; waiver permissible eliminative killing use of term, 9–​13 comparison with law of armed conflict on eliminative killing, 13–​18 restrictive implications of, 18–​19

terrorism and, 13n41, 13–​14, 14n43, 166–​67,  213 See also eliminative killing principle of distinction use of term, 15n45, 15 intentions and, 124–​25, 136 permissible eliminative killing and,  15–​18 principle of permissible harm, 140n4,  140–​43 principle of secondary permissibility, 149n25, 166–​67, 180n13 pro tanto duties, 17, 55 pro tanto rights use of term, 8, 57n39 restricted by special patient-​claims, 67n58, 71n65 infringement model and, 107 intentions and, 128n9 See also claims property Basic freedom to use, 71–​73 first vs. second, 67n61, 67, 76, 87–​88, 92 Lockean property, 98–​99 property baseline and, 87–​90 property claims, 67–​68, 87–​99, 116, 122, 124, 150–​53, 174–​75, 183n16, 210 property-​like claims, 68n63, 77, 90–​92,  91n97 toolkit and, 87–​88, 90–​91, 93–​95,  174–​75 See also claims; restricting claims principle proportionality baseline proportionality, 10–​11, 12, 13, 15n46, 166 comparative proportionality, 10–​11n34, 11, 12 eliminative killing and, 10–​11, 12, 213 necessity and, 10–​11n34, 12 reductive individualism and, 35–​36 Public Committee Against Torture in Israel vs. Israel, 15–​16n51,  16–​17 punishment, 31–​33, 32n18, 56n38, 56, 103–​8,  169

Index  |  239

Quinn, Warren, 128n9, 132 Quong, Jonathan, 157–​59 Rawls, John, 21, 43n1, 48n13, 48n15 reductive individualism use of term, 29–​30 authority and, 30–​33 consequentialism and, 36 just war theory and, 34–​37 necessity and proportionality and,  35–​36 political self-​determination and, 36 social contract and, 31–​33 state authority and, 31–​34 reflective equilibrium, 21–​44 respect for persons, 43–​44, 52, 78, 132–​33,  142 for rights, as cash value for respect for persons,  132–​33 as a subjective state, 130–​34 responsible choice, 3–​4, 8, 109–​10, 111, 124, 145 restricting claims principle use of term, 83n88, 83–​99 agent-​agent division, 87–​88, 89–​90,  95 agent-​patient frontier, 87, 88, 93 autonomy principle and, 89–​90 baseline freedom and, 84–​90, 93–​95, 136, 174–​75, 176–​77,  181–​82 doctrine of double effect and, 136 empowering claims and, 67, 83n88, 93–​96, 98–​99, 114, 122, 174–​76 fairness and, 85 intentions and, 132–​33 just war theory and, 136 justification and, 97–​98 means principle and, 43, 82–​83, 114, 217 mechanics of claims and, 82–​83, 83n88 negative externalities and, 85–​86, 88, 116 property baseline and, 87–​90 same ballpark strength/​rivals and, 83–​84, 85–​86,  89–​90

240  | Index

side effect and, 83–​86, 90–​92, 114–​15, 139, 143 welfare principle and, 98–​99 retributive justice, 56 rights use of term, 43–​44 animals, rights of, 49–​50 categorical duty to respect, 46n10, 53 common discourse, 80 constructivism, 48n13 grounding in three fundamental principles, 47–​48, 79 hybrid rights–​lesser-​evil arguments, 113–​14, 144–​45,  146–​48 indeterminate without the state and culture, 31–​32, 57, 72–​73 ineliminable, not mere intermediate conclusions, 46 interest theory, 44–​45 kind-​desire theory, 44n4 legal rights, 57, 72–​73, 98–​99 normative significance not a matter of value, 53, 200–​3 relational topological structure of rights, 58–​59, 59n42 specificationism, 45n6, 45, 101 as subset of possible wrongs, 55–​57 topology of rights, 58–​59 violations, 6, 79–​80, 104n9, 106, 197,  200–​3 will theory, 44–​45 See also claims; infringement model; mechanics of claims; space of rights risk assumption of risk, 5, 62–​63, 121n32, 182n14, 186–​87, 192, 205n17 handling conditions of risk, 28,  119–​20 imposition of, 3–​5, 74–​76, 91–​92, 97–​98, 110–​11, 155–​56,  157–​59 justification and, 97–​98 reducing risk, 9, 10, 212, 216 risk pools, 61 war and conditions of high risk, 6 Rodin, David, 12, 160, 210n24

Scanlon, T. M., 46–​47, 51–​53, 82–​83, 128n7, 133n21 self-​defense against culpable aggressors and threats, 2–​3, 6,  110–​11 against innocent threats, 2–​3, 6,  143–​44 against responsible, non-​culpable threats, 3–​5, 110–​11, 156, 159 against justified threats, 180, 187–​89 ineffective, 188n26 just war theory and, 1–​2, 6, 8, 29–​30, 36,  191–​94 mutual aggression and, 78–​79, 171 other-​defense vs., 1, 10 recent history of theory of, 2–​6 See also forfeiture; permissible eliminative killing Sellars, Wilfrid, 46 space of rights use of term, 46n10, 46, 53 breakdown in, 79, 108, 113–​14, 115–​23 citizens of, 49–​50, 67, 132–​33, 202–​3 comprehensiveness, 45n6, 45, 53–​54, 55–​56, 98–​99,  216–​17 foundational principles, 47, 78–​79, 89–​90,  202–​3 harmonization of three grounding principles and, 78–​79, 183–​84,  216–​17 indeterminate without the state, 31–​32, 57 infringement model and, 114–​15 Kantian conception, 44 topology of, 46–​47, 58–​59 specificationism, 45n6, 45, 101 state, the authority of, 30–​33, 36–​37, 181–​82 social contracts and, 32–​33 Statman, David, 188n26 Steinhoff. Uwe, 18n60, 27n9, 74n74, 147–​48n24 superfluous killing, 165 Tadros, Victor, 17n55, 37, 79n79, 103–​4, 133n23, 144–​48, 160n9, 162n15, 185, 202n14

Taurek, John, 51n21, 77n78 terrorism, 13n41, 13–​14, 14n43, 35–​36, 166–​67,  213 Thomson, Judith Jarvis, 2, 7–​8, 51n25, 100n1, 101–​2, 108, 179n12 threats use of term, 8–​9 culpable threats, 3, 6, 35–​36, 143–​44 indirect threats, 16, 17, 155, 159–​65 justified threats, 180, 187–​89 patient-​claims and,  149–​50 redirection or deflection of threats, 64,  139–​43 See also innocent threateners; innocent threats threshold deontology, 115–​23, 176n8, 211n27 Tomlin, Patrick, 103–​6, 104n8 Toolkit. See property torture, 9n29, 121 United Nations Charter, 1n2 utilitarianism. See consequentialism victim. See innocent victims violation. See rights waiver use of term, 7n28 mechanics of claims and, 80–​81 See also forfeiture Wasserman, David, 73 welfare principle use of term, 47 aggregation and, 51–​52 benefits and, 54–​55 comprehensively registering welfare tied directly to a claim, 53–​54 competition and, 117–​18, 181–​83 conflict with other principles, 89, 94, 116, 118, 121 duty to aid, 58, 62, 64 empowering claims and, 98–​99 expressing a value, 50–​55 harmonization with other principles, 75–​76, 79, 115, 216

Index  |  241

welfare principle (cont.) nature of claims through which welfare is registered, 52–​53 paternalism and, 70 restricting claims and, 88–​89, 98–​99 utilitarianism and, 51–​52

242  | Index

See also agent-​claims; patient-​claims; mechanics of claims; restricting claims principle; space of rights Wenar, Leif, 44n4 wronging, 79–​81, 108–​11, 114–​15, 119 See also infringement model