Operational Images: From the Visual to the Invisual 1517912105, 9781517912109

Table of contents :
Cover Page
Title Page
Copyright Page
Contents
Preface
Introduction: Between Light and Data
Chapter 1: Operations of Operations
Chapter 2: What Is Not an Image?: On AI, Data, and Invisuality
Chapter 3: The Measurement-Image: From Photogrammetry to Planetary Surface
Chapter 4: Operational Aesthetic
Chapter 5: The Post-lenticular City: Light into Data
Conclusion
Acknowledgments
Notes
Index
About the Author
Color Plates

Citation preview

OPERATIONAL IMAGES

This page intentionally left blank

OPERATIONAL IMAGES From the Visual to the Invisual

JUSSI PARIKKA

University of Minnesota Press Minneapolis London

The University of Minnesota Press gratefully acknowledges support for the open-­access edition of this book from the Academy of Performing Arts in Prague and the Operational Images and Visual Culture project, funded by Czech Science Foundation’s EXPRO grant 19-­26865X. Portions of chapter 4 are adapted from “From Planetary Depth to Surface Measure, or How to Read the Future from an Image,” in Deep Mediation: Thinking Space in Cinema and Digital Cultures, ed. Karen Redrobe and Jeff Scheible (Minneapolis: University of Minnesota Press, 2021). Chapter 5 was originally published as “On Seeing Where There’s Nothing to See: Practices of Light beyond Photography,” in Photography Off the Scale: Technologies and Theories of the Mass Image, ed. Tomáš Dvořák and Jussi Parikka (Edinburgh: Edinburgh University Press, 2021). Copyright 2023 by Jussi Parikka Operational Images: From the Visual to the Invisual is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0): https://creativecommons.org/licenses/by-nc-nd/4.0/. Published by the University of Minnesota Press 111 Third Avenue South, Suite 290 Minneapolis, MN 55401–­2520 http://www.upress.umn.edu ISBN 978-­1-­5179-­1 210-­9 (hc) ISBN 978-­1-­5179-­1 211-­6 (pb) A Cataloging-­in-­P ublication record for this book is available from the Library of Congress. Printed in the United States of America on acid-­f ree paper The University of Minnesota is an equal-­opportunity educator and employer. 32 31 30 29 28 27 26 25 24 23

10 9 8 7 6 5 4 3 2 1

CONTENTS

PREFACE

Operational Images, All the Way Down

INTRODUCTION

Between Light and Data

1

ONE

Operations of Operations

31

TWO

What Is Not an Image?

THREE

FOUR

FIVE

CONCLUSION

On AI, Data, and Invisuality

vii

57

The Measurement-­Image

From Photogrammetry to Planetary Surface

95

Operational Aesthetic

Cinema for Territorial Management

139

The Post-­lenticular City Light into Data

177

A Soft Montage of Operations

211

ACKNOWLEDGMENTS

221

NOTES

223

INDEX

269

This page intentionally left blank

PREFACE

Operational Images, All the Way Down This book focuses on operational images, a concept that first emerged in Harun Farocki’s audiovisual work (films and video installations) and writings. The often-­voiced but almost as often insufficiently developed definition is simple, and it is already somewhat evident in the term itself: there are some images that primarily operate; they are not necessarily representational or pictorial. Operational images trouble what an image is, as far as it shifts from representational to nonrepresentational, from the primacy of human perception of bodies, movement, and things to measurement, pattern analysis, navigation, and more. They change the scales and terms of reference. Such images have an epistemic force, while they also are involved in an intervention in the world, whether directly or indirectly. These sorts of images fix epistemic details, but while they hold them in place, they also are mobile in ways that become an essential part of their institutional activity. These points are explored in the pages that follow. Operational images appear most often in discussions of machine vision and automation of perception and observation, where images are part of broader systems of analysis, identification, tracking, and destruction. The term operational image seemed to speak directly to the period from, roughly, the first Gulf War in 1990 to the current phase of military and security management of territories with drones as part of an ecology of observation, analysis, and surveillance. Even a brief survey of Farocki’s films and existing literature reveals that this period of thirty years of militarized airspace and technological systems that incorporated much of the already existing logic of targeting would be the most straightforward reference vii

viii  Preface

point of the term. However, this is not to say that this book is primarily about war or airspace, or surveillance or drones—­even if such are central parts of the theme that admittedly ranges much further in the past than just the context of smart weapons and machine-­ readable landscapes, territories, and objects (including making people into objects). It was also not the sole interest in Farocki’s work, especially if we consider how, for him, operational images also emerged in architectural modeling, traffic control systems, construction of affective environments such as malls, and other examples that have also paved the way toward current topics of research in AI culture. Furthermore, this book is also not merely about a particular class of images that emerge in high-­tech contexts, even if, again admittedly, it is about visual culture and its transformation into invisual data culture. I am not seeking an aesthetics or politics of digital images. There are plenty of books on these topics already, and many of them are consulted and discussed in this one, too. This book is about a needed shift in emphasis to the operational image. From a discussion of the image, its particular formats, and even its functions, I move to a different, albeit closely related, question that concerns operations, a key term that ties to infrastructures, logistics, and all manner of actions that function to sustain, mobilize, analyze, and synthesize the thing we have grown to call “images.” I am after the coupling of perception and action, of images that control, regulate, and amplify how bodies operate.1 Both perception and action are terms defined in the course of this book: perception is not tied to any natural experience, optical capacity, or even one technique, and the action in question is not necessarily only the here and now of a human or other animal body but defined by more tricky causalities—­including a relation to the temporality of futures, future-­pasts, recursions, and more. In other words, operations do not act only on the present but across a broader timescale, including a temporalization of space (e.g., in forecasting systems). Allow me to offer a guiding rhetorical figure that should be considered throughout this book as part of its conceptual scaffolding.2 Operationalism is already an established term in scientific methodological literature,3 but also consider the rather famous and often-­ used parable of the turtle at the base of the universe. The narrative

Preface  ix

comes and goes in different forms, this one written down in Carl Sagan’s Broca’s Brain: Some ancient Asian cosmological views are close to the idea of an infinite regression of causes, as exemplified in the following apocryphal story: A Western traveler encountering an Oriental philosopher asks him to describe the nature of the world: “It is a great ball resting on the flat back of the world turtle.” “Ah yes, but what does the world turtle stand on?” “On the back of a still larger turtle.” “Yes, but what does he stand on?” “A very perceptive question. But it’s no use, mister; it’s turtles all the way down.”4

Instead of turtles, the history of photography, technical images, and visual media could be told in the same key: instruments built upon instruments, upon infrastructures, upon practices, upon techniques, upon further instruments and infrastructures, and so on. Operations built on operations that include elements that are material and semiotic, forms of knowing, and forms of mattering. As Geoffrey Winthrop-­Young points out, such turtled ontology is also shared by the cultural techniques approach in media theory, concerned with “operational sequences involving actors, things and practices that, coming together, give rise to established cultural practices.”5 Could the same then be said about the operational sequences involving images? That there is a fundamental relational ontology at play, one that could be described by way of scaffolding operations? At the very least, we can confidently concur with Michelle Henning when she writes that “the Victorian idea of the photograph as an image that makes itself no longer holds sway.”6 Other notions of images and diagrams, observation and operation start to take precedence as do other examples of complex apparatuses of sensing and inscription. An example will help. Joseph von Fraunhofer’s 1810s invention of the spectrometer— a later backbone of remote sensing and the analysis of materials through their spectral signature—­represents one case from an alternative history of visual technologies. To verify and measure the dark lines existing in the solar spectrum—­later named Fraunhofer lines—­the spectrometer apparatus was an assemblage of multiple

x  Preface

elements, camera-­like technologies (six lamps and a shutter), a prism, and a modified theodolite: “an ordnance surveying instrument originally designed to measure angles for the production of maps.”7 As Susan Schuppli writes: “His work would eventually come to be used by scientists to determine the chemical composition of a remote object—­our sun, some 149.6 million kilometers away—­not through direct testing but by treating it as an image, one whose chromatic variance could be translated into the complex language of chemistry.”8 Treating the world as an image beyond its pictorial qualities is one key instance where operations of epistemic value take place—­but there are multiple operations already in place in order to produce such images. Fraunhofer’s laboratory setup can be described as an essential part of the history of science of discovery that concerns the qualities of the solar spectrum. However, to add, as an artificial setup of said spectrum, it is also a microcosmos of light—­even of interplanetary light. It is a simulation of the behavior of solar light, too, which then introduces the possibility of using this standardized view of the dark lines on the continuity of the spectrum to scale up and down across light and materials. All sorts of material instruments and scalar operations are piled upon each other, explicated and implied. As photography (and yet not entirely just photography), Fraunhofer’s apparatus was a camera of sorts—­one built like turtles on top of each other. I would go so far as to say that Fraunhofer’s apparatus represents one central instance of the history of operational imaging vis-­à-­vis remote sensing9 and that what was captured was not a picture in the traditional sense but a microcosmos (or a condensation of its visual characteristics) through a detailed concrete arrangement that facilitated this diagram of the abstraction to emerge, and it demanded a precise need for accuracy in the placement of angles and distances, a measurement-­operation in its own right.10 The recursive world of turtles and their relations—­or camera-­ like instruments in laboratories and solar light—­is a helpful figure of thought. At the same time, a more conceptual reference point for such operative chains of support is likely to be found in the theoretical work of cultural techniques and operative ontologies. But more on that later. To answer the question that likely has emerged, What is this book

Preface  xi

about? It is not about turtles, but operations all the way down: images that exist only because of other operations—­and the operations that help us to understand the transformation of images into data, from visual to other forms and formats of registering the world beyond representation. One could justifiably argue this point about cinema already, as Thomas Elsaesser speculates about Farocki’s interest in simulations as not a replacement of reality but a chain of synthesis—­“the downward spiral of simulation of a simulation of a simulation”11—­that pertains to cinema and postcinema. In our case, we instead place emphasis on the operational transformation of the link between visuality, photography, spectral analysis, and data—­a transformation that did not happen with digital culture, but rather circa 1900, as the story that follows will show you.

This page intentionally left blank

INTRODUCTION

Between Light and Data Capturing Light Around 1889, Harvard College expanded its influence far outside Cambridge, Massachusetts. Having joined the College Observatory (first as a student, later as professor of astronomy), Solon Irving Bailey was sent much farther south, to Arequipa in Peru, to establish a new field station. This operation was to switch the hemisphere and find a spot elevated enough for ideal observation of the light traveling from distant celestial objects. Astronomic photography had a long history already by the 1890s, but this need for a new observatory emphasized the additional demand for what we would now call scientific infrastructure. After New Year’s Day in 1889, a boat trip from San Francisco took Bailey and his family to their destination identified earlier, “attracted by reports of the clear sky and slight rainfall on the high plateau of Peru, where also the whole southern sky is visible.”1 While the rhetorical emphasis on the clean, crisp observation place puts all of the weather conditions easily outside of history and into the physical sphere important for astronomy as a science of observation of laws (out there) and not things (here), during the difficult trip to find the perfect spot, Bailey observed and (in passing) noted the colonial legacy of the region: “I should place the population of the valley near Chosica in the days of the Incas at six thousand. Today there are perhaps five hundred. This well illustrates how Peru has changed since she fell into the hands of the Spanish conquerors.”2 Such awareness in his thoughts and diary did not, however, prevent the expedition from (re)naming the place they came to in a softer but still imperial manner: Mount Harvard. The eponymous name was entirely in tune with the aims of Edward 1

2  Introduction

Pickering, the long-­standing and renowned director of the Harvard College Observatory, to establish posts in the north and the south, “so the entire sky would be available for Harvard’s research.”3 Besides a number of adventurous anecdotes from that trip, the relation with a media technological context is especially interesting. Two themes concerning light intersected during the years Bailey spent in Peru, both of which were essential to the scientific work, while producing an aesthetic quality to the geographical placement. The sunlit high-­a ltitude plains—­causing occasional mountain sickness for the party looking for a suitable observation spot—­provided ideal landscapes while the photometric (measurement of the brightness of light) and photographic techniques provided technologies for the capture of slowly shifting objects in the night sky. Not that such exact spots of observation were known in advance; some of Bailey’s memoirs from the trip read as a persistent search for locations where measurements can be made, leading him to echo earlier advice about the exploratory spirit: “Of the clearness and steadiness of the atmosphere in these different places, there is no certain knowledge, and your only way is to investigate it for yourselves.”4 The investigation aimed to take pictures to send back to the college in Cambridge. Besides telescopes, the comparative analysis of photographic evidence became a key technique that needed a reliable data supply. It was, in some way, a case of what Michelle Henning has called “the unfettered image”: fixed as image, but migratory and journeying as an object.5 Here, what migrated were the comparative observations of the vast space outside the planetary sphere. As per the Harvard Observatory aim, to be able to observe the night sky from both the Northern and Southern Hemispheres gave a particular advantage to astronomers. Moreover, with the help of the photographic media, Southern data was relatively easily transported back to Cambridge for comparative, computational analysis. In Pickering’s words, “For many purposes the photographs take the place of the stars themselves, and discoveries are verified and errors corrected by daylight with a magnifying glass instead of at night with a telescope.”6 Bailey’s field station’s photographs were sent north. This part of the logistical story has become more well-­k nown in recent years, particularly the (female) computer pioneers of data analysis and

Introduction  3

astronomy, including Annie Jump Cannon and her work on star classifications7 and Henrietta Swan Leavitt, among others. Leavitt, later awarded the title “Curator of Astronomical Photographs” (held earlier by Williamina Fleming), left lasting contributions to the field (even if here the focus is only on parts that relate to the media technological operations that serve as infrastructure and instruments of astronomy as a science). Leavitt’s research impacted astronomy by demonstrating important traits about the periodicity of brightness, an essential element in measuring distances across the vastness of outer space. In addition, the Peruvian night sky had been photographed and recorded on glass plates that Leavitt stacked on top of each other for comparative data analysis and to produce insights into the shifts of moving stars, which in our case illuminates a key theme: early in its first official century, photography was already a measurement device that not only took pictures of people and things but offered a way to analyze the world, including the extraterrestrial. As such, the point about technical images and measurement has already been articulated; for example, Kelley Wilder gives a good overview of some of the practices of astronomical imaging before and after the Harvard period in question and opens up important points more generally too. Besides photographs where “the ability to measure appears to be a useful but unintended byproduct,”8 there were various intentional practices, mostly scientific, where this cultural technique was central. In astronomy, this included the Venus transit plates of 1874 and institutionalized work such as Carte de Ciel of the 1880s, “one of the most influential photographic observation projects in astronomy.”9 Beyond astronomy, Raman spectroscopy and photogrammetry were “methods that bent photographic observation to mathematization,”10 with surveying as a technique that was, as Wilder outlines, “heavily dependent on the idea of measurable photographs.”11 Here, the commentary on measurement serves to illuminate the expanded scope of operational images to be followed throughout this book. In aptly contrasting ways, the title of Leavitt’s 1908 paper, “1777 Variables in the Magellanic Clouds,” rings poetic, while the opening sentence nails the argument about images as infrastructures of analysis and comparison in a pithy, informative fashion: “In the

4  Introduction

spring of 1904, a comparison of two photographs of the Small Magellanic Cloud, taken with the 24-­inch Bruce Telescope, led to the discovery of a number of faint variable stars.”12 Where Bailey had engaged with the landscapes of Peru, its altitudes and terrains, the shipment to Cambridge provided the other side of this landscape; in Leavitt’s reading, the Magellanic Clouds—­or, more precisely, their photographic recording—­provided a dynamic, periodic landscape of light to be interpreted. Leavitt writes about light that she has been observing on those records: “The variables appear to fall into three or four distinct groups. The majority of the light curves have a striking resemblance, in form, to those of cluster variables. As a rule, they are faint during the greater part of the time, the maxima being very brief, while the increase of light usually does not occupy more than one-sixth to one-­tenth of the entire period.”13 Surely, Leavitt and others would have cursed Tesla’s Spacelink satellite program that hinders the subtle balance and periodicity of the sky with its mass flooding of the orbit. However, around the 1890s and 1900s, the night sky was still stable and observable through the gridded transparency of the glass plates that opened up possibilities of comparative analysis. While the sky had been pictured, read, observed, interpreted, and calculated for millennia, as John Durham Peters argues in his media theoretical insight to astronomical star-­gazing, the scientific analysis of movement and light became particularly interesting toward the fin de siècle.14 The employment of both media of visual technologies (photography and spectral analysis) and the possibilities to harness the planet’s spherical shape—­Northern and Southern Hemispheres into a binocular view of sorts—­a s part of the astronomic observation unit from Peru to Massachusetts provided the backbone for broader infrastructures of knowledge. The intersections of media and the sciences (in this case, astronomy) have impacted the transformation of photography as it became “digital” and how it has been part of data analysis and planetary infrastructure. And, as will be argued in chapter 3, even the shape of the planet measured in geodesic triangulation can be considered part of the story of the extended planetary image. As already mentioned, this link to scientific uses of photography, including in astronomy, should not be particularly surprising con-

Introduction  5

FIGURE 1. A flyspanker and a photographic plate likely used and annotated by Leavitt for the research on the Small Magellanic Cloud. As hand­held reference for standard magnitude of stars, flyspankers were made from existing astronomical plates of stars with different values of visible brightness. They were sorts of “model image” instruments. Courtesy of Harvard College Observatory, Photographic Glass Plate Collection.

sidering that perhaps the most famous words in the early history of photography (or, more specifically, the daguerreotype) were given by an astronomer, François Arago, in his 1839 address. This talk was given to convince the French Academies of Art and Science about the benefits of the new technique, which was why the talk aimed to make sure it was seen as a scientific one and therefore included specific attention paid to the various uses of measurement: beyond people or things, landscapes or scenes, this was a medium to measure photometrically the brightness of transmitted light and thus

6  Introduction

also provide an insight into what lies beyond this particular planet and how that can be easily recorded on a plate. Thus, the instrument became a central part of an experimental apparatus that unfolded a whole visualization process in developing an image.15 As pointed out by Wilder, the nineteenth-­century history of photography was filled with astronomical works and interests: William de Wivelselie Abney, E. E. Barnard, William Crookes, L. J. M. Daguerre, John Draper, Paul and Prosper Henry, Jules Janssen, Hermann Krone, Adolphe Neyt, Warren de la Rue, Lewis Morris Rutherfurd, Hermann Wilhelm Vogel, and John Adams Whipple are among a list of practitioners relevant to both sides of the technical expertise. Wilder argues that “much of their work revolved around adapting emulsions and photographic instruments to astronomical observation, and they produced everything from spectra of starlight, to photometric readings, to iconic images of the heavens.”16 While much of the focus in earlier research has been on the apparatuses and their relation to both histories of technology and, in some cases, scientific discourses of validity and reliability,17 adding an emphasis on Leavitt opens a particularly interesting avenue of consideration not only for the history of photography but also for the theoretical topic at hand, operational images. Through Leavitt and the work of the numerous female scientists at the observatory,18 questions of photographic plates as instrumental infrastructures for operations of measure become underlined. However, discussing Leavitt gives us insight into the work of remote sensing and images that become operationalized for data analysis, in addition to dispelling the often overly male-­inventor and scientist-­focused narratives (including those concerning operational images). This applies to many of the other contributions too: the photographic plate collection becomes the scaffolding for advanced work in spectral classification and cataloging (Annie Jump Cannon) and remote sensing (calculation of “chemical composition of stars”19 in Cecilia Payne-­Gaposchkin’s work), alongside the Leavitt Law as an example of operational images, too: “a tool to calculate distance in space with the use of Cepheid variable fluctuations.”20 Leavitt’s and others’ contribution thus is not so much on photography as a particular technology or genre of images but as the basis for an analysis of data and an extraction of features that become signifi-

Introduction  7

cant for questions of remote sensing. Such work becomes part of the lineage of operational images that are less interesting as images on their own but rather as part of a broader infrastructure of skills, labor, techniques, and technologies, and how institutions21 assemble images for their particular needs and uses. Less interesting as pictures, even if armed with a poetic allusion of deep space, but useful for the trained scientific eye, such photographs become an infrastructure for analysis. Leavitt’s comparative analysis identifies light and subsequently tables the sources of light as varieties of data points in the Magellanic Clouds. Besides its straightforward operational use in astronomy, it leads to interesting considerations concerning the role of the image. Far from the usual focus of art history, such images are one element in the extended role played by computation and data. And not digital in the sense we refer to them as now, but as part of the institutional uses where photographic images are essential in being able to scale and inscribe the distant movement into the flat (paper and glass) world of pictures and observation tables.22 In short, the photographic functions primarily as detection instead of depiction; subsequently, detections are also turned into calculations. Photography—­or the photographic—­condenses massive scales of measurement of light.23 As such, the intersection of earth, sky, light, shadow, and astronomy, and images through which to calculate are among the contrasting currents that define our interest in the actions of images. Leavitt and her fellow computers are part of astronomy as well as an alternative genealogy of visual practice that finds a resonating counterpoint in more recent times: infrastructures of visuality, images tabulated as data, data that becomes a diagram, and image models that enable ideas of remote sensing (spectra and periodicity of light as signals of distance, material composition, movement) to emerge as a key paradigm of visual—­and invisual—­culture. Bailey’s travels to the Southern Hemisphere that ended up in Arequipa to manage the Harvard field station can be named as a starting point for this story about the long chain of operations of images ending up as diagrams, graphs, and tables. This is not to say that diagrams weren’t already part of the Enlightenment period repertoire of managing data and images: as Bender and Marrinan note, the practices of diagrams in the modern sense emerged in the pages of Diderot

8  Introduction

and D’Alembert’s Encyclopedia of the late eighteenth ­century, “A diagram is a proliferation of manifestly selective packets of dissimilar data correlated in an explicitly process-­oriented array that has some of the attributes of a representation but is situated in the world like an object.”24 Here, in their words, image-­diagrams are more Pollock than Rembrandt, which in our case could be reconfigured to: more like the scattered light beams of stars in the night sky (Pollock) than pictorial, representational images (Rembrandt). Although, as becomes obvious later in the book, the two are not always so separated. Science studies likely have a lot to say about this combination of assemblages of diagrams and their material contexts, as instruments facilitating the collecting of data but also as the transforming, translating, and rather literally transporting of these epistemological observations from the field to the laboratory. Henning’s notion of the “unfettered image”—­a lready mentioned earlier—­is discussed in the analysis of scientific images and data too. Bruno Latour’s rather influential take on techniques of soil sampling in the Amazon exemplifies this through the specialist instrument of the pedocomparator: “The pedocomparator has made the forest-­ savanna transition into a laboratory phenomenon almost as two-­ dimensional as a diagram, as readily observed as a map, as easily reshuffled as a pack of cards, as simply transported as a suitcase.”25 This book is not concerned with such instruments but instead with the variety of objects and techniques that briefly featured in the ministory of astronomy and photographs. Nonetheless, a soil scientist’s instrument is a useful parallel for us because the methodological transformation will be recounted in different ways and alternating variations throughout this book: the photograph is only one stop in the long chain of routes and operations of images, and the variety of techniques are interlinked from instruments to diagrams, and from three-­dimensional materials or landscapes to flattened images, diagrams, and the formalized registers of analysis. One can run images through similar filters as is evident in the STS type approach offered by Latour. To continue for a short while with the argument mediating abstractions and objects: “[the pedocomparator] is lighter than the forest, yet heavier than the paper;

Introduction  9

it is less corruptible than the vibrant earth, but more corruptible than geometry; it is more mobile than the savanna, but less mobile than the diagram that I could send by phone if Boa Vista had a fax machine.”26 Importantly, diagrams enter the picture, too: With the diagram, in contrast, the forest-­savanna transition becomes paper, assimilable by every article in the world, and transportable to every text. The geometric form of the diagram renders it compatible with all the geometric transformations that have ever been recorded since centers of calculation have existed. What we lose in matter through successive reductions of the soil, we regain a hundredfold in the branching off to other forms that such reductions—­w ritten, calculated, and archival—­make possible.27

The useful takeaway here is that we should look at the operative transformations of the image as it pertains to the multiple materials and their abstractions, and abstractions coming back to transform the materials. What STS does with mapping the route of material objects of science, we can try to do with images as they work in institutions. This will be a key part of what this book establishes as the operational in the operational image. In our example of the astronomical something shifted when light was observed, recorded, and sent across a geographical distance—­ perhaps a lot of things transformed from the planes of observation and appreciation of high-altitude air and clear skies to the photographic exposure and the plates that became one chain along the way before Leavitt, Cannon, and others interpret and synthesize (compute) these into the pithy but important results that ensued (see Figure 2): “A straight line can readily be drawn among each of the two series of points corresponding to maxima and minima, thus showing that there is a simple relation between the brightness of the variables and their periods. The logarithm of the period increases by about 0.48 for each increase of one magnitude in brightness.”28 Where is the image here in this description, in such tables, curves, and diagrams? On the photographic plates or their analysis, the moments of exposure in the Peruvian landscapes, in the logistics of

10  Introduction

FIGURE 2. Examples (Table 1 and Figures 1 and 2) from Leavitt and Pickering’s 1912 article “Periods of 25 Variable Stars in the Small Magellanic Cloud.”

transporting those images to Cambridge, in the trained analysis of composite photographs, or somewhere else along the way of transforming light? Or, perhaps, the notion of operational ties together a multitude of such material events, sites, and their abstractions and assembles them into a useful notion of the operational image that, as a term, itself is invented much later and for a different purpose, but might itself become useful to speak of infrastructure, logistics, and images that transform from visual to invisual, from ways of seeing to ways of calculating.

Introduction  11

The Operational Image This book opened with the excursion into the Peruvian light that turned into the Harvard-­based astronomy of calculation, comparison, analysis, the write-up of scientific papers, and some pictures that got published. However, most of the thousands and thousands of technical images were never seen but by few pairs of eyes. But the book itself, as should be obvious by now, is on operational images.29 The term by the renowned German filmmaker, artist, and writer Harun Farocki (1944–­2014) appeared in the early 2000s in his video installation trilogy Eye/Machine I-­III (2001–­3) that investigates autonomous weapon systems, machine vision in industrial and other applications, and the broader move from representations to the primacy of operations. Farocki’s film installation series presents this switch as a particular kind of an image that emerges in those institutional practices although also articulating it through the various histories and spaces that condition both the emergence of such images and their industrial base: these include military test facilities, archives, laboratories, and factories. The institutional line of references is familiar in many of Farocki’s films that have investigated how contemporary images are intimately tied with the modern forms of industrial production, detaching from a history of images only as visual culture, to histories of chemistry, violence, labor, exploitation, and data. Already in Images of the World and the Inscription of War (1989), Farocki mapped a similar terrain of investigations of how to read landscapes, aerial imagery, targeting systems, and also other forms of modeling, simulation, and aesthetic techniques as they operate in the world in the fundamentally material sense. The film will also be featured later in this book—­even if my aim is not to offer an analysis of Farocki or his works as such.30 Nonetheless, the work with images about images sets a scene and opens up an artistic, epistemic, and research-­focused agenda. Eye/ Machine III (2003) is one such example where operational images are articulated across a set of cases: factory scenes of data and measurement to infrared aircraft detection systems, laser scanning of built structures, and engineering to robotic navigation systems that sense the space around them. Images produced in those situations

12  Introduction

are drawn from machine-­v ision systems of perception, embodied and embedded in autonomous or remote systems, and working through an artificial environmental relation where the image is a crucial part of movement and guidance.31 Operational images are, in Farocki’s words, “pictures that are part of an operation,”32 implying the primacy of action and function instead of a picture to be seen and interpreted for meanings. Perception is tightly coupled with action, immediate or delayed. This coupling systematically operationalizes terrains and targets. Hence guidance systems, movement, tracking, measurement, and precision are some of the terms that take precedence in such images that are often, across a visual history, bluntly put, “inconsequential.”33 It is also a condensation of an aesthetic program that relates to what images are seen, which ones are archived, and which of the multitude of images are merely used and erased: Images that appear so inconsequential that they are not stored—­the tapes are erased and are used again. Generally the images are stored and archived only in exceptional cases, but exceptional cases one is sure to encounter. Such images challenge the artist who is interested in a meaning that is not authorial and intentional, an artist interested in a sort of beauty that is not calculated. The US military command has surpassed us all in the art of showing something that comes close to the “unconscious visible.”34

The operational aesthetic returns in chapter 3, but before that, a word or two about where we are heading: while military contexts of machine vision have taken most of the space of commentary and attention when it comes to Farocki’s notion and its articulation in moving images and photography, it is clear that the breadth of examples tells a larger story than merely about genealogies of military vision systems. This is not to dismiss such a key trait. Farocki’s examples from a 1942 instructional film showing the operations of a V1 guided missile to the 1990s military systems that became a key topic for art and media theory from Jean Baudrillard to Paul Virilio are persistently apt in the context of contemporary drone warfare and in the media archaeology of military vision.35 Even Farocki him-

Introduction  13

self reads “the US military command” as part of a new aesthetic operationality of visibility. Furthermore, this concerns not only perception and sensing that are turned into images but also operations. The history of the centrality of “operations” can be traced to Operational (or Operations) Research (OR) in the U.S. and British military since the 1930s and especially in the war years of the 1940s: quantifiable analysis of military operations for purposes of optimization, something that then developed into the Cold War continuation of “speculative fabrications of systems analysis”36 such as the RAND corporation in the United States. These are institutional-­level “machine learning systems” that aim to formalize, train, and model based on available quantitative data. Learning itself becomes a formalized operation. For OR pioneers and controversial practitioners such as Herman Kahn, the successes of operations research in World War II proved the greater effectiveness of mathematics over time-­honored tactics. Systems analysis was unquestionably superior, in his view, despite the common belief that “experience” has been a better guide than “theory” in this kind of work.37

Of course, an opposition of theory vs. experience was a bit of a simplification considering how the pioneer of Operational Research Patrick Blackett (later Baron Blackett, and later also featured in Thomas Pynchon’s Gravity’s Rainbow) had defined the pillars of OR as based on “observation, experiment and reasoning.”38 A broader understanding of the scientific method had been rolled out and integrated into how space, strategy, tactics (including the evaluation of success of tactics), and logistics were to unfold based on data. Nonetheless, to keep with Kahn’s exaggeration in spirit and style, perhaps OR did more for “theory” than French 1960s structuralism and poststructuralism. Perhaps not, and it is definitely not the sort of theory we usually practice or want to practice in the humanities, but one point was made clear: experience is secondary, formalizable design and planning are primary. To program the battlefield, you program people first, while later on you have programmable machines such as the ones that populate examples of operational

14  Introduction

images as we know them now. Besides naming them as computers, they are more specifically, for example, platforms, as will be argued in chapter 2. To deal with large-­scale systems, logistics, and abstractions, one had to fine-­tune a different mindset: “In decisions regarding weapons systems development such as choosing between long-­range bombers with big fuel tanks or short-­range bombers with refueling capacities, ‘no one can . . . answer by instinct, by feeling his pulse, by drawing on experience,’” as RAND economist Charlie Hitch put it.39 In short, the centrality of complex calculations (e.g., logistics), the massive amount of data to be processed, decisions to be taken, and the multiple scales of abstraction were not commensurable with the cognitive capacities of humans in the traditional sense of even trained officers. The necessity to be able to rationalize, theorize, model, and potentially automate decision-­making in the context of complexity persisted from the war to the postwar period—­for example, in management theory, making it a part of systems thinking where any decision was part of a meshwork of other decisions, by other actors, in a recursive loop.40 Cultural techniques of quantification connected to modelling was one particular route offered in this history of what “operations” came to mean on and off the battlefield. Numbers count landscapes and what moves through them; they count routes and their optimal relations; they count possibilities and potentials; and numbers are the backbone for both images and industrialization. Data is not infallible and simply “objective,” as critical data studies have shown over and again,41 but it can be effective whether it is correct or not. Rolling out data-­driven decisions, systems, and operations is also an intervention in landscapes, social relations, values (financial and other), and more. While leaving a longer discussion of logistics and operational images for later, it shows already much of the relevant context that elaborates the implicit conditions of emergence of what Farocki coins in his terms and in his audiovisual work by way of the “soft montage”42 of archive and inconsequential images. One peculiar context of such images would then be the over 70-­year history of military-­driven operations research and subsequent management theory and some 150-­year history of photographic-­driven data analysis. In some ways, this all condenses into “an industrialisation of vision”43—­or even “indus-

Introduction  15

trialisation of thought,”44 as Farocki himself coined his interest in cinema and perception, directly echoing Virilio’s work on the “veritable market in synthetic perception.”45 The contemporary versions of that coinage relate to questions of artificial intelligence, machine vision, but also the genealogy of the concept of operations as it pertains to images, institutions, spaces, and nonhuman visuality. AI will be featured particularly in chapter 2. The industrialization of vision has been often coined as part of the industrialization of destruction, a theme that connects Farocki to the 1980s (and later) theorization of war and visuality.46 Technical processes of abstraction, images that are primarily for targeting and destruction feature as part of a genealogy of rationalized violence that human bodies are subjected to. As such, Farocki’s scene for operational images could be seen as a crystallization of much critical theory, thematically visible in the focus on the Holocaust, the Vietnam War (napalm in Farocki’s Inextinguishable Fire, 1969), the Gulf War, and the prison-­surveillance-­capitalist complex that features in many of his works.47 But this book argues that there needs to be nuance in how this concept of the operational image is read and used, avoiding the temptation of packing all sorts of abstractions—­ and abstract images of technical and calculational use—­into one camp of a kind of Enlightenment gone awry, a stream of violence and extraction that is merely about military power in the restricted sense of warfare. This is not to ignore the operational violence of capitalism (chapter 1 and throughout) or the colonial uses and functions of measurement and their neocolonial forms (chapters 3 and 4); but to take a position against abstraction on principle would be a mistake, leading us to insufficiently nuanced readings about technical images. We have plenty of those already, and in the context of environmental imaging, remote sensing, AI and platform culture, and many other crucial topics, we cannot anymore afford to miss the more detailed high-­res insights. In other words, one additional step proposed in this book includes a shift from military operations to the other, closely aligned use of force that defines the current landscape of operations: “Operations Other Than War.” This is not a nonmilitary form of power, but one that builds on particular logistical capacities and systemic, technological potentials of power primed for the contemporary

16  Introduction

planetary situation, from environmental issues to humanitarian assistance to enforcement of exclusion zones to handling pandemics. In some ways it relates to the twentieth-­century lineage of operations research, but it also becomes a way to tap into the contemporary moment of logistical wiring of bodies and territories. In Rosi Braidotti and Matthew Fuller’s words, the “conflict is played out, triggered, and modulated through means that include finance, smuggling, culture, drugs, media and fabrication, technologies, resources, psychological operations, networks, international law, ecologies, economic aid, and urban terror. War becomes postdisciplinary, multi­scalar, creative, and highly mediatic and technological, deploying specialized multiskilled teams and techniques.”48 In other words, war and conflict become part of the extended repertoire of media techniques of confusion, doubt, and misinformation, often paired with the deployment of “ruses, proxies, ambiguous agency, hyperbole, the operationalization of ‘mistakes’ and unattributable forces.”49 Hence, we can ask, What format of operational images speak to this state of war and violence? We might not (always) be at war, but we are (always) mobilized and operationalized. This could also be referred to as the perceptual and operational fine-­tuning of the “nonbattle,” a term first introduced by Virilio and developed by Brian Massumi. Operations and actions are embedded in a broader field of intensities and potentials, possibilities, and the modeling of futures. “In the nonbattle, the relation between action and waiting has been inverted. Waiting no longer stretches between actions. Action breaks into waiting.”50 The operational is nested here in the significance of knowing how soft power can work effectively. Massumi continues: “Soft power is how you act militarily in waiting, when you are not yet tangibly acting. . . . In the condition of nonbattle, when you have nothing on which to act tangibly, there is still one thing you can do: act on that condition. Act to change the conditions in which you wait.”51 Operations that act on the conditions of existence and conditions of further operations sound like a version of Virilio that Massumi restages and like a proposal that could come from the direction of Foucault’s analysis of architectures and diagrams. Diagrams feature in this book as one reference; one could also consider images as tableaux of information52 (in reference to Gilles Deleuze’s terms)

Introduction  17

that cut across traditional scales of experience, space, and meaning, and rearrange those like the abstract images rearrange the current technological cities (see chapter 5 for a discussion of the urban technologies of light and data). Indeed, the Farocki in question here is somewhat less the critic of Enlightenment reason (in the lineage of Theodor Adorno and Max Horkheimer) than a media archaeologist mapping what is visible, what is sayable, and importantly, what is countable. This line of arguments pulls Farocki from a thematic analysis of modern rational images to a method of mapping archaeologies and genealogies of images as they become working material for critical thought. This material, though, is thoroughly conditioned by a recursive loop between industrial production and images in and out of war. Raymond Bellour calls this, rather aptly with a Foucault-­inspired undertone, the photo-­diagram—­another phrasing of the methodological positions at play in operational images. Bellour’s note on Farocki’s material is fruitful for our purposes: “The photographs as well as the actual film recordings are equally ordered pieces of evidence of a reasoned assessment of the nature of the visible as defined on the basis of the very invisibilities that form it, leading to so many machinic and asubjective regulations, normativities and constraints.”53 Operational images have been discussed in cinema studies by, for example, Volker Pantenburg, Thomas Elsaesser, Pasi Väliaho, and Erika Balsom, and in contemporary art discourse by Trevor Paglen, Hito Steyerl, and Lawrence Lek, among others. The Harun Farocki Institute in Berlin navigates among cinema, art, and discursive work as a “platform for researching [Farocki’s] visual and discursive practice and supporting new projects that engage with the past, present, and the future of image cultures.”54 Farocki’s name stands at the intersection of multiple genealogies, practices, and concepts that are not reducible to a story of an auteur. I do not claim that previous writing about him has done this either; indeed, already Elsaesser identified many of Farocki’s works as “contributions to media archaeology, as well as an essential part of the prehistory of digital images”55 where questions of interface, simulation, and, indeed, operation become central hinges for an appreciation of particular kinds of genealogies of which the digital is only one technical term. As Elsaesser puts it, “These changes we tend to associate with the

18  Introduction

digital turn, but operational images just remind us that moving as well as still images have many histories, not all of which pass through the cinema or belong to art history. Digital images may merely have made these parallel histories more palpably present, but operational images, as Farocki clearly saw, have always been part of the visual culture that surrounds us.”56 Two intersecting, closely related points sum up this argument: On the one hand, it can be seen as a term that speaks to techniques of measurement, analysis, and synthesis through techniques of images but in particular institutional situations and uses. Operational images organize the world, but they also organize our sense and skills in terms of how we are trained to approach such images, from the photogrammetric mapping of landscapes to pattern recognition, astronomy datasets to Mars Rover imaging practices.57 On the other hand, the term relates to practices (and labor) of testing, administering, and planning also reflected in the sites of filming (in addition to what was mentioned above) where Farocki himself worked. These range from schools to offices to management ­training centers, and army field exercises, to paraphrase Elsaesser. To also quote his summary: “To operational images correspond operating instructions for life.”58 As instructions of life, the term implies the broader use of the term “algorithmic” as the training of bodies, setting institutional routines, and rehearsing automation in ways that tie machines to laboring human bodies. Imaging practices become operational in how they tie bodies into collective routines.59 What characterizes Farocki’s films as an “education image” (to refer to Antje Ehmann and Kodwo Eshun) is exactly this particular quality to attend to such material spaces and signs and images that define learning (“work desks, typewriters, books, diagrams, and equations that constitute the scenographies of learning”) as well as “scenes that dramatize narratives of learning.”60 But after learning becomes about machine learning and training refers to the training set, we also have to adjust the scope of these cultural techniques. The work of labeling images in practices of supervised machine learning is one scene of the training of both neural networks and the people involved in sustaining those networks.61 The discourse of the photographic, but also “education” and work, become thus restaged in ways that are not merely the factory or the earlier use

Introduction  19

of the industrial scene but globally distributed across platforms of logistics such as Amazon Mechanical Turk.62 Not that one image replaces the other, but the educational image, navigational image, instructive image, and operational image take place at moments and sites of transition, exchange, and transformation. The electronic switch—­a nd its relation to the circuit and circuit board, the techniques of control and optimization—­defines the way both twentieth-­and twenty-­first-­century operations and (technical) images63 become the historical site of connection even if not all I say in this book is reducible to such technological ur-­scenes. In other words, societal operations are part of the broader framework of discussion of this particular aspect of visual culture, even if it at first seems in exact contrast to Farocki’s own somewhat fragmented words about operational images: “Images without a social goal, not for edification, not for reflection.”64 Indeed, while Pantenburg picks up this phrase from Farocki, it is not meant that the topic is devoid of politics in relation to a variety of societal institutions. While the images are not interesting to look at as images, they have multiple chains of operations through which they are linked to a long line of institutional, epistemological, and other uses that trigger a different aesthetics that speaks to questions of what is now, perhaps, called the nonhuman image65 and the nonrepresentational image as they circulate across institutional sites and uses, from education to training and technicians to algorithmics of the everyday. As such, this book toes a fine line between an emphasis on the notion of operationality—­while not wanting to ignore that it is the image, here, that ties the discussion to being a contribution to visual and screen studies, photography, and moving image—­and the transformation of those in relation to questions of data and even artificial intelligence (machine learning). Before outlining some key terms, let me offer this interim summary of the claims and arguments of this book: First, operations and operationality are key concepts for contemporary visual and media theory while moving beyond the visual, the visible, and the lens-­based. The term “invisual” (introduced below) will become a significant addition in this context. Second, the operational image is irreducible to being merely about digital images, big data, or artificial intelligence (machine/deep learning). Those recent developments are not ignored

20  Introduction

but placed into a historical dialogue with questions of data, sensing, and spatial uses of images. Third, the term is “transdisciplinary,” and as such, it links discussions in media theory, art studies, digital aesthetics, architecture, and critical infrastructure with visual culture studies. In this book, I am also interested in the context of how concepts bind together different disciplines. Concepts, too, operate.

The Media Theory of Invisuality Operational images refer to those practices and infrastructures of images that are not necessarily particularly interesting to watch or see. Nonetheless, they organize much of what contemporary culture fundamentally looks at, including figures, graphs, diagrams, trajectories, models, plans, simulations, control screens, spreadsheets, and so on. Images might not necessarily want anything, but they do lots and get lots done.66 This is not necessarily an emptying of the image from desire and subjectivity but just demonstrating that those two terms always work as collectives and assemblages, as infrastructures and operations. Thus a media theoretical insight into this situation implies that we need to be aware of specific technological contexts of visuality and its transformation of some two hundred years of technical images (photography, photogrammetry, and more), even if in this book we go back also to the early 1700s and the geodesic measurement and production of the planetary image (see chapter 3). As Hoel and Lindseth put it: Operational approaches provide promising possibilities for rethinking images for at least three reasons: first, they offer dynamic approaches that analyze phenomena into doings and happenings rather than into things and static entities; second, they offer relational approaches that conceive identity in terms of open-­ended processes of becoming; and third, by so doing, they allow us to ascribe agency to images, and crucially, to conceive agency as distributed across interconnected assemblages of people, practices, and mediating artifacts.67

The above is a useful, underlying methodological guideline that should be kept in mind throughout this book. Furthermore, it fea-

Introduction  21

tures the various projects and examples included in these pages. This dynamic notion of the image that speaks of them as temporal, relational, and agential is particularly visible in recent audiovisual and written work that picks up a similar approach. Trevor Paglen’s homage to Farocki and operational images related it to an informational, even data-­driven framing. As a version of the image as information-­tableau, Paglen’s artistic-­practice-­ informed writing identifies invisible images and machine vision as the regime of operations that define the transformation of visual culture—­images that are “made by machines for other machines” but also such “invisible images [that] are actively watching us, poking and prodding, guiding our movements, inflicting pain and inducing pleasure.”68 Images have agency, but that agency is distributed across vast operational spheres, which are also geographically dispersed. While we could argue that all images, including the visible ones, are known to do exactly that—­cause bodily reactions and affects from pain to pleasure and much more—­it is the odd but necessary view to a data-­driven understanding of images that suggests this methodological angle as to how the operations bring such instructions of life into existence. As “doings and happenings,” operational images could actually turn out to be less about images than about situational awareness and (automated) decision-­making that concerns many of the contemporary data situations, such as dashboards.69 Operational images are in Paglen’s narrative both logistically crucial (automated vision and sorting systems that read car plates, shipping containers, faces, etc.) and quantitively plenty: billions and billions of images not merely taken but uploaded and processed on platforms such as Facebook. “DeepMask,” TensorFlow (Google), and other platform-­integrated forms of operationalizing digital images as they “identify people, places, objects, locations, emotions, gestures, faces, genders, economic statuses, relationships” are included in Paglen’s useful entry list of ways of seeing as an intervention in the world. AI systems are part of the scene that help to realize the essential shift: “The machine-­machine landscape is not one of representations so much as activations and operations.”70 Following on from this pithy summary, I aim to discuss in more detail how this landscape—­which sometimes is literally a landscape as a territorial formation in urban and nonurban situations—­operates through

22  Introduction

different layers of abstractions that are essential to how the logistics of the image and logistics of planetary space coalesce.71 Indeed, it is thus no surprise that one of the key topics to focus on is the logistical image as it facilitates the global flows of goods and people and how it turns things across the scale into a massive AI training set.72 Picking up on the link among operational images, data, and platforms, a key term (and featured in this book’s subtitle) is the invisual. Throughout the book, I refer to the pairing of the visual-­ invisual to draw this continuum as a territory of transformations that concerns images and their role as aesthetic–­epistemic agents. Focusing on how contemporary visual culture is operationalized in the automated procedures of digital platforms, Adrian Mackenzie and Anna Munster have proposed the concept of the invisual to help to shift the focus from the mere quantity of images to how they are being aggregated. This collective of images, their ensemble, is to be considered as material and generative forces, even if they take shape as image datasets. Mackenzie and Munster argue that they organize “all kinds of platform cultural actions and forms.”73 While this theme could be addressed as the gradual shift in complex systems on the coupling of perception and action—­and thus corresponds to a particular reading of cybernetics, image, and design74—­it is here also useful as a way to focus on how data and visuality are entangled in contemporary digital infrastructures of platforms. Invisual culture suggests that while the image persists in contemporary platform culture, it does not feature as a visual image in the optical or experiential sense. Mackenzie and Munster’s argument about invisual perception captures this transformation, which is furthermore the underpinning focus of this book too: Such a mode suggests that while visual techniques and practices continue to proliferate—­from data visualization through to LIDAR technologies for capturing nonoptical images—­the visual itself as a paradigm for how to see and observe is being evacuated, and that space is now occupied by a different kind of perception. This is not simply “machine vision,” we argue, but a making operative of the visual by platforms themselves.75

Introduction  23

What Mackenzie and Munster elaborate as the invisual—­that is, as “a new mode of nonrepresentational observation”76—­is in this book mapped through past and current modes of operationality. The tension between old and new is not returned to any ideas of natural vs. technical, analog vs. digital, but assumes perceptual techniques of sense and image guide and misguide, include and exclude, educate and miseducate. These images are organized in relation to a set of technological contexts or infrastructures, apparatuses, and instruments defined by operative ontologies of cultural techniques (see chapter 1). Thus the shift for the subject is not merely from “human to machine” (whether platform, AI, machine vision, or something else) but in relation to institutional sets of affordances that define what images as operative techniques are. The institutions can be military but can also be scientific organizations and discourses, colonial expeditions, corporate AI, massive scale national remote sensing and data platforms, and environmental sensing services, as well as different kinds of art and design practices. Contemporary digital platforms format and operationalize images as part of their invisual perception and aggregation. Mackenzie and Munster’s work will be one of the reference points for the development of the ins and outs of operationality, but I also want to emphasize that this book is not solely about (digital) data, AI, and platforms, even if it is my contention that writing any book on operational images after the emergence of “platform seeing” (Mackenzie and Munster’s term) cannot stay the same as it would have before such an ensemble of image techniques. This is why I stay focused not only on an exegesis of the concept of operational images as it has been specifically used by Farocki, Paglen, Hito Steyerl, and others, but how it becomes operational in a manner that is highly recursive:77 To investigate the turn to operational images, one reads contemporary forms of operationality through historical sources and cases that also inform the contemporary reading. The recursive operation of back and forth cross-­reading of different cases helps to detach from a linear assumption that would assign “the digital” a role (it often enjoys) as a totalizing historical shift or that would suppose “data” is a recent invention of big data infrastructures. Operational images, as a concept and an area of

24  Introduction

methodological questions regarding images, thus itself becomes operationalized: as per the above definition of one of the key aims of this book, it helps to understand how data and visuality, space (urban and nonurban) and image define each other, and how this image is temporally dynamic.78

A Manual I want to add a word on the methodological approach that tries to build some elements from the theme (of operations) into a guideline. The book is a sort of a manual that can take you to places but doesn’t give a full map of every detail. Instead of a historical analysis that would offer one holistic genealogy of the term “operations”— for example, how it emerges in Farocki’s work, or how it ties military imaging into a twentieth-­century story about machine vision, or how it fits with media archaeology of measurement and photogrammetry—­I work with the concept as a force of its own. What does that mean? First, the term “operation” is not merely employed to point to historical instances; it does not become a tool that, like a spotlight, moves across historical practices and technologies. Instead, the term itself is set to motion, too; the spotlight turns on itself. Second, this implies the already mentioned recursive task of how historical and contemporary examples and theoretical arguments are set to influence each other.79 “Operational images” is a heuristic term that helps to excavate media archaeological traits of past image practices in relation to contemporary data cultures, and in the process, it also changes pace and becomes respecified, retrained, reoperationalized. Third, picking up speed from Massumi again, we can refer to “conceptual persuasion,” which is not about a concept that is operationalized for particular tactical uses (such as to persuade) but rather carries its own force in the abstract so as to recognize the potential “strangely effective force immanently inflecting events, abstractly energizing their verging onto self-­completion.”80 Aptly, for Massumi, the note on persuasion speaks to his interest in “operative logics” such as preemption as a force of contemporary soft power of the military–­security complex as it spans across the field of social institutions. The idea that, for example, preemption is

Introduction  25

then a “force of attraction around which the field of power is bending” is intriguing.81 What would this mean for operational images? Not merely images placed in the social use of institutions, but as forces of attraction that bend fields of power, vision, sensing, subjectivity, information, data, and so on, differently? It is almost like the operational carves its own space as much as it acts upon someone else’s (institutional) orders: a vexing thought to follow throughout the book, as it introduces friction to the otherwise direct functional use of operation for specific institutionally defined purposes. Surrounded by such questions that set the mood of this book, the text is also a chance to explore how some of the theoretical propositions and arguments emerge from collaborative work and what I recognize as part of the methodological setup, as part of the “ways in which” and the “how” of the text’s procedure.82 Explication and speculation; historical examples and archival work; contemporary articulations such as those in the field of visual arts. This book sets in motion a “conceptual persuasion” that connects such instances into an argument that is added to, chapter by chapter. Many of my collaborations informing this book have been with artists and designers. To prioritize any of these collaborations is not to suggest that artistic knowledge is in some way superior over many other fields that work with images and define what an image can do but to recognize the role artist-­scholars have already played in defining this field and how it can be further articulated across historical and contemporary practices. Farocki himself is the key example, but we can name others, too, whose work discusses a similar territory of ideas in cinema, photography, and other practices of visual culture. Allan Sekula and the notion of instrumental image would be a good sparring partner in this discussion; it is another concept that has escaped beyond the field of photographic discourse and has been subsequently linked to investigations of contemporary logistics and critical forensic methods.83 Paglen and Steyerl have already made an appearance and will do so again. Both are examples of artist methods articulated in projects for galleries and texts that are not entirely academic but are influentially rearticulating the field of theoretical discourse of visual arts beyond the gallery and media art festivals or museums.

26  Introduction

In this book, besides referring to established work by such established artist-­w riters, I make space for contemporary practitioners to enter the debate as collaborators and co-­creators. In other words, some of the topics and projects described in the chapters, such as Geocinema’s Framing Territories, as well as my work with Abelardo Gil-­Fournier, are not merely presented as artworks for analysis. Instead, the theoretical concepts and ideas are informed by the dialogue, in various exchanges, and in different forms of objects (visual, data, concepts) that were part of those shared situations with the artist-­scholars. I had the opportunity to work with Geocinema, closely following their unfolding project, which also means acknowledging how they influenced my way of opening up operational images to concerns of contemporary practice (see chapter 4). Similarly, Gil-­Fournier and I have worked on art and other projects together, and our continuing dialogue has been part of writing this book too. The two of us are writing a joint book on environmental imaging, surfaces, and operational images, which also continues some themes appearing in this book. Many others could be mentioned: Rosa Menkman’s Impossible Images artistic research project has been running in parallel to my writing; Sasha Litvintseva and Beny Wagner’s audiovisual work on measurement is being finished as I complete the final edits of this text on questions of measurement; Aura Satz asked me the question, in passing, if there is a sonic equivalent to operational images while she is working on her current project on sirens. I never could answer that, but the question stuck in my mind as many other leads, thoughts, problems, and suggestions circulate in and through such interactions. I thus find it methodologically important to underline the broader network in which the conceptual work is situated. Such a network often becomes the invisible scaffolding for published work, but anyone engaged in institutional and extrainstitutional work knows about the force of such infrastructure. The concept of operational images becomes a discussion drawn from written and audiovisual archives of examples and encounters in contemporary situations of practice-­based making, including not just (fine) art but also design and architecture. For example, working as part of the Strelka Institute’s Terraforming program has become a part of composing this book through a specific shared space of concerns, concepts, and

Introduction  27

methods. Indeed, developing this concept not merely as a designation of a territory—­“this is an operational image, this is not”—­but as an articulation of relations of visuality and invisuality, agency, and labor becomes a way to reflect on the dynamics of concepts also in practice-­driven work. The operational image, as a concept, has a persuasive force that becomes involved in different studio, theoretical, field, and other practices across disciplines. Such a stance also expresses the idea that concepts are not merely stable positions that populate a generic map of “theory” but instead already traveling dynamic entities that transform material and semiotic situations—­while also being affected by such situations. Concepts are practiced in institutional situations (archival research or empirical studies, for example) and nonacademic artistic situations (experimental work with machine learning datasets, for example). Both examples already incorporate multiple kinds of competencies and techniques.84 Seeing concepts as a dynamic part of methodological work echoes some of the thoughts by Gilles Deleuze and Félix Guattari on the centrality of concepts for scholarly activity that shifts focus away from contemplating or reflecting. Against the shop window of ready-­ made and off-­the-­shelf concepts and methods, I am more invested in this constructive sense of concepts as operative in their own right as they mark a singularity: You will know nothing through concepts unless you have first created them—­that is, constructed them in an intuition specific to them: a field, a plane, and a ground that must not be confused with them but that shelters their seeds and the personae who cultivate them. Constructivism requires every creation to be a construction on a plane that gives it an autonomous existence. To create concepts is, at the very least, to make something.85

While Deleuze and Guattari offer a more detailed insight into how concepts are bordered by affects, functions, and percepts, we can follow this short lead to arrive at some methodologically useful map instruments (if a method is about a path and a road, we do not really want to know where we will end up, but rather what happens

28  Introduction

during those trips and travels). I do not claim the operational image is a concept for philosophy proper, but it is a situation of artistic practice and writing that produced a concept that shifts across disciplines; thus, the territory in question is, in this case, an aesthetic-­ epistemic-­conceptual one that then marks a specific entry into the trajectory of movement of the visual-­invisual. A similar trait is also clearly expressed by Mieke Bal’s development of conceptual work as its own methodologically significant form of inquiry.86 Bal’s language echoes the idea of boundary objects (as used in science studies and the history of science)87 but specifically in relation to concepts and their role in mediating between different disciplinary perspectives and creative practices. Indeed, the notion of a traveling concept is thus one that speaks to the dynamic multiplicity of the concept’s work—­never just an ordinary word but a delineation of a matter of forces and their territories, of situations and their expansive repercussions. Giuliana Bruno has built a whole methodological–­territorial complex around this idea of motion/ emotion in Atlas of Emotion: topographies of affect, concepts, film, and architecture are codetermining instances that are teased out by way of writing but that also emerge out of historically situated bodies.88 Incorporating the idea of traveling into a discussion of media and its materials, as well as of concepts such as the operational image as it moves across different disciplines, we can quote Bal: “But, after returning from your travels, the object constructed turns out to no longer be the ‘thing’ that so fascinated you when you chose it. It has become a living creature, embedded in all the questions and considerations that the mud of your travel spattered onto it, and that surround it like a ‘field.’”89 As noted earlier, also concepts change as they travel (not only between countries or disciplines but, for example, to and from archival sources); a concept can be a sticky enough surface that grabs hold of its encounters, leaving a smudge, a scar, a trace. Those encounters are part of its reformulation: it carries a history that impacts how it operates as an intervention into the contemporary field. In such a field of traveling concepts and dynamic, collaborative investigations, this book sets out to investigate the web of concepts and their living contexts of visual and invisual cultures. In this

Introduction  29

book, the operational image becomes a tool to investigate layers of practice; it establishes bridges across institutions and leaps into different discussions where this theme is present and where this concept is explicitly or implicitly being used: media studies, art and architecture, infrastructure and visual culture studies. While this book emerges from the Operational Images and Visual Culture project and is situated in the Department of Photography at FAMU at the Academy of Performing Arts in Prague, it has multiple institutional contexts. As a persuasive force, the term has a conceptual intensity that transforms what photography is or how it is being understood (and has a link to much of the discussions in post-­ photography and current networked and digital practices of photography),90 and this also leads to us asking questions about the “others” of photography in a manner familiar from media archaeological methodologies: Beyond visible spectra of light, what other practices of light are part of the histories of the present? Which practices and institutional epistemologies are implied in these situations? The chapters of this book respond to such questions. I have already noted some of the themes in this introduction, and a summary of what follows will suffice. After this introductory chapter, I turn to chapter 1, “Operations of Operations.” This chapter outlines some of the recent uses of the notion of “operation” in theory, with a special focus on operative ontologies. The chapter already starts to move toward some of the thematic directions of the book, including operations of capitalism and links between visuality and data (grids). But chapter 2, “What Is Not an Image? On AI, Data, and Invisuality,” especially dives into the topic. Here, the starting point is Trevor Paglen and others’ interest in the invisibility of technical images, especially in the context of artificial intelligence, which is laid out in my text through three different sites where image operations have been identified as a crucial part of politics of data: the platform, the dataset, and the model. The chapter also goes deeper into Mackenzie and Munster’s concept of invisuality, which becomes a useful reference point for those techniques where visibility, invisibility, and data also meet with questions of, for example, property and capitalism. If operational images are often considered as part of advanced military technologies, then how does the hypothesis that primitive accumulation and colonization across an

30  Introduction

intersectional sphere of force (land, class, race, sex) are part of what to consider as war add to our analysis and methods?91 Operational images as nonrepresentational forces of production are central to how the argument develops. Chapter 3, “The Measurement-­Image: From Photogrammetry to Planetary Surface,” investigates the so-­called measurement image from Farocki’s work to a short historical investigation of photogrammetry. Also, the planetary image as one of measurement is addressed by way of a quirky historical case of the two 1730s geodesic expeditions that measured the earth surface to draw maps to navigate. Questions of infrastructure and logistics feature across the chapters. Chapter 4, “Operational Aesthetic: Cinema for Territorial Management,” deals with some of these issues through operational aesthetics, not just as aesthetics in art but as preparation of capacities of sense and perception. As outlined in chapter 1, much of the recursive nature of operative ontologies and operational images is how they are images about images, and here artistic work, such as Geocinema’s, demonstrates the operations of large-­scale remote sensing and data as part of the operational images of geopolitics. Chapter 5, “The Post-­ lenticular City: Light into Data,” focuses on invisuality and examines transformations of urban sensing and images through a case study of lidar. Beyond being a single technology of laser scanning of the urban and nonurban environment—­used, for example, in autonomous vehicles—­it also links to an alternative genealogy of light: What forms of light off the visible spectrum but present as intensive pulses are nested in techniques of images and invisuality? How are the platforms of sensing and observation reformatting the city, such as those investigated through aesthetic and design methods such as Scanlab’s? The book ends with a conclusion, “A Soft Montage of Operations,” that returns to the night sky and beyond through Event Horizon, the black hole imaging project, while drawing together some final points about the centrality of operational images as a hinge that opens up cross-­disciplinary investigations. What the conclusion offers is also this, a line that I want to emphasize from the beginning: this book maps the histories and afterlives of the notion of operational images and, hopefully, offers insights on methods too.

ONE

Operations of Operations At first, operations seem similar to what we refer to as praxis or practice, but they emerge from a specific military–­technical legacy. Operations speak to questions of logistics in massive technical systems that work through the ability to abstract and optimize. Operations pertain to the algorithmic logic of software culture while also implying multiple scales of its applicability. And “operations” has become one key term for the tactical and strategic definition of territories and landscapes while remaining an abstract term with a history in scientific uses. Operationalism was, after all, defined in early twentieth-­century physics as an “ontological view about the basic constituents of empirical reality.”1 Later, in the 1920s, in the words of Percy Williams Bridgman, operationalism implied the importance of measurement operations, as “we mean by any concept nothing more than a set of operations; the concept is synonymous with the corresponding set of operations.”2 As such, the lesser-­known scientific use of the term refers closely to the centrality of instruments in the production of scientific knowledge and moves from experimental physics to social sciences. I am not seeking to apply the existing terms from the history of technology or science but acknowledging that they exist: any contemporary theoretical use of “operations” happens in the context of the term’s own cultural and technical history.

Operations, Operations, Operations A key proposal throughout this book is to follow the implications of the term “operation,” suggesting that it is, as a part of the term “operational image,” as central as the other part, “images.” It also pulls 31

32   Operations of Operations

operational images into resonating fields of concepts and themes in contemporary media theory and design and artistic practices. This chapter discusses how operational images can be read creatively across a theoretical landscape of concepts that have emerged from a suitably fertile soil of media and cultural theory scholarship. This mushrooming of references to operations can be read back to the historical role of the term as part of the cybernetic and even pre-­cybernetic realization of the centrality of information, feedback control, and data integrated into decision-­making. In addition to this historical track record, it has also served a role in different media theoretical discussions. While there is a backstory as to how “operational” and “operations research” gradually became a central part of technological rationality throughout the twentieth century from military to management, there is much more to these terms.3 Bringing operational or “operative” images into a conversation with operative ontologies as it emerges in the scholarship on cultural techniques (like Bernhard Siegert’s work), as well as with “politics of operations” (Sandro Mezzadra and Brett Neilson’s term) offers us an analysis of contemporary capitalism. A whole field of uses of “operative” and “operational” could be mentioned, though: Brian Massumi’s reference to the “operative logic” of power and Jimena Canales’s take on the (visual) operational art. We should not forget Wolfgang Ernst’s persistent focus on the operative either: according to Ernst, media are media only when operational, and thus “active”—­agential even.4 Recently, Kathrin Friedrich and Aurora Hoel proposed an “operational media analysis” that takes up the importance of operations. Focused on “image operations” within a broader ecology of “media operations,” they defined it as “a technologically mediated action or procedure where symbolic and virtual resources are gathered to effect changes in the physical environment.”5 Their example of image-­guided robotic radio­surgery is perhaps an unusual case for media studies. But, according to their argument, it is precisely for this reason that it is part of the necessary expansion of methodological possibilities in accounting for not only remote sensing but also operations that intervene in their surroundings. Such are the image and media operations we are interested in: “Technical objects enact a new resolution

Operations of Operations   33

of the human-­world system, a material reconfiguration of the world that releases new potentials for perception and action.”6 The “operational” sometimes appears as a qualifying word for “what kind of images,” and sometimes in the limelight for its own sake. However, we are after the operational definitions of material interactions, and in the spirit of the endless chain of turtles as the ontological premise of the world, this means defining the topic of visual culture and its transformation at the level of infrastructure. It takes operations to bring about operations. The cartography of this chapter helps us establish ways to read our main concept of concern through parallel tracks, turning the concept into a thematically interesting and methodologically useful one, and not merely a linguistic affiliation to the shared term of “operative/operational.” Therefore, we turn to a central part of this chapter and this book—­namely, a discussion of the operative ontologies that define the material chains of cultural techniques as they effectively frame the conditions of their own existence. Concepts circulate in theoretical discourse, but they also hit the ground in their situated uses, institutional affordances, and circulation across different disciplines. They also help make sense of what other things hit the ground, from military operations to management procedures, where capitalism specifies itself from a generic, hazy term to site-­specific violence and exploitation.7 This is entirely Farocki territory: operations of military targeting as part of the S/M (science and medicine, surveillance and military) histories of images8 and laboring, even enslaved bodies as conditions of existence of institutions of industrial images.9 This chapter develops an arc from a broader link between operative ontologies to the agential realism of Barad, then onto the specific questions where operational images are understood as instrumental grids that operate upon bodies, including a note on racial capitalism through the work of Simone Browne.

Operative Ontologies “Operative ontologies” is a central term in the field of cultural techniques that emerged from the Internationale Kolleg für Kultur­ technikforschung und Medienphilosophie (IKKM) research center

34   Operations of Operations

in Weimar, Germany, as well as from other scholars such as Thomas Macho, Cornelia Vismann, Markus Krajewski, and Sybille Krämer. In Berlin, the Hermann von Helmholtz-­Zentrum für Kulturtechnik10 has been active since the early 2000s, with research on “technical images” as one of its core areas, led by Horst Bredekamp,11 while IKKM has been developed under the guidance of and work by Bernhard Siegert (cultural techniques) and Lorenz Engell (ontography) with a multidisciplinary research program that featured film, media studies, art, (media) philosophy, architecture, and history of science, among the many disciplinary fields of the fellowship program and core staff. IKKM also included Farocki among its visiting fellows. The reason for me to focus on operative ontologies is not because of the coincidence between the terms, or even that Farocki was involved in the work of IKKM for a brief period (his death coincided with his residency period in 2014), but because I believe that operative ontology offers a necessary point of reflection for a rigorous description of operative chains that have a material force: a media theory of the operational. In Siegert’s definition, operative ontologies shift from metaphysical, ontological questions to more specific, even locatable, ontic operations.12 Methodologically, it does not start from universalizing idealizations, even if part of the work has been concerned with ontological distinctions often hardwired into the (Western) cultural mindset (inside/outside, masculine/feminine, sacred/profane, on/ off, signal/noise). Instead, work focuses on unfolding how those seemingly stable binaries have come about. How do they become hardwired into material reality in the first place, and how do further, complex layers of distinctions become operational in, for example, technical media culture? How do symbolic operations become effective as transformers of material reality? While the terminology includes a reference to the ontic/ontological pair that Martin Heidegger’s philosophy of Being was focused on, the emphasis and ensuing solution reflect a much more methodologically useful take for historical, situated, and material investigations. For Heidegger, the ontic/ontological was already an operationalization of the two terms in order to illuminate two aspects of the theory of Being; the ontic as the regime of the given (the spatiotemporally specific things in their materiality) and the ontological

Operations of Operations   35

as the (hermeneutic) horizon of understanding and orientation.13 The latter often takes precedence in the readings and applications of Heidegger. For media theoretical uses, the peculiar follow-­up term of operative ontologies—­like ontologies taking place, spatialized, temporalized, durational, and processual—­emerges as central. As noted by Siegert and philosophers such as Krämer, the hybrid term (oxymoronic to some, as it hints that ontology is in motion instead of a stable reference point) takes place as a seemingly self-­referential operation that bootstraps ontological investigations into concrete historical situations. We shift from lofty meditations that happen in the mind to an analysis of the ways in which (media) techniques are primary in the mobilization of objects and subjects in ways that become central to what only emerges secondarily as (metaphysical) concepts.14 The whole process is thus describable as a recursive chain of processes where the operations are serialized; they appear as a series, not as individual events. “So the operation of a door or a switch—­these are all techniques that are producing a difference and thereby creating what they differentiate. It’s not given before, but it’s created by these techniques.”15 As per the anthropological influence (not least of which includes Leroi-­Gourhan), bodies and technologies, habits, and their environments are interlinked and regulated and produced in specific, situated ways.16 They are recursively linked codetermining instances, where concrete things (organic and nonorganic) and abstractions switch places and regulate each other. In many ways, this outlines one premise of the method of cultural techniques and how it brings together anthropological concepts (operative chains) and contemporary cultural theory alongside how advanced technological systems of abstraction and complex feedback systems work beyond human bodily techniques. On the one hand, the term “cultural techniques” refers to the materiality of living practices that are irreducible to the concepts that refer to such practices; writing comes before conceptualizing writing, counting before the concept of a number, and so forth per Macho’s definition.17 This is the anthropological trait in cultural techniques, which, however simple it might seem, considers the complex technical abstractions that define our contemporary media culture, from planetary infrastructures of AI to large-­scale datasets and mass of

36   Operations of Operations

images. On the other hand, and in other contexts such as Vismann’s influential take on law and media, cultural techniques as operations, such as drawing a line or plowing agricultural land, are fundamental to producing subjects—­but also things like (property) law, as they mark territories, insides and outsides, and so on.18 So here we are, as per recent waves of critical legal theory, moving closer to a territory of law not as a text but as material dynamics existing in the world. And this can be further developed in relation to operative chains of data and abstract territories marked not only by tangible markers but also by immaterial regimes of control.19 Cultural techniques can thus, based on emphasis, relate to symbolic processes, including numbers or counting and then more complex mathematics and logical operations, as well as material transformations where the ontic becomes a literal placeholder for a dynamic set of operations that have the capacity to enact transformations in the real world.20 It becomes a hinge around which abstractions and concepts spin and amplify, and though ontological distinctions are being produced and maintained, they are also contested. In Geoffrey Winthrop-­Young’s words: Once again, cultural techniques refer to processing operations that frequently coalesce into entities which are subsequently viewed as the agents or sources running these operations. Procedural chains and connecting techniques give rise to notions and objects that are then endowed with essentialized identities. Underneath our ontological distinctions (if not even our own evolution) are constitutive, media-­dependent ontic operations that need to be teased out by means of techno-­ material deconstruction.21

All of this feeds into a methodology that addresses operations as ontogenetic forces that bring about realities: operative ontologies are understood as productive of material realities. They are thus, as observed later, instrumental to the dynamics of power too. Operative ontology seems to harbor a reference to the Heideggerian terms already mentioned, but this is not entirely necessary. I would go so far as to claim that despite the terms, the con-

Operations of Operations   37

ceptualization and ensuing methodology is not nesting (a secret) Heideggerianism, but speaks to other contemporary formations of material, posthuman, and politically interesting aspects of a situated form of emphasis that the operative/operational can help to frame, including mobilizing ontologies and specific instances such as “images” and, recursively, images about images, images about operational images, and operations that concern other operations. Indeed, other features in the existing literature on cultural techniques emerge from science studies and philosophy of science and technology, such as Bruno Latour and Michel Serres’s work and Karen Barad’s agential realism. For me, the latter is particularly interesting and will be referenced later in this chapter regarding the concept of onto-­epistemology. Such approving nods to Science and Technology Studies (STS) and new materialism are in place for various reasons, including reading complex human–­object relations, thus entering broader discussions of critical posthumanities.22 Like­ wise, terms such as “hybrid objects” start to populate the field in ways that seem familiar to decades of Actor-­Network Theory (ANT) and yet have particular methodological power in helping us think the pragmatics of operations. For Siegert, rehearsing some core theses of ANT, the term “hybrid object” helps move away from binaries of active subjects versus passive objects. It includes an appreciation of how quasi-­objects populate the ontic level of dynamic interactions of materials and signs, people and things. Furthermore, people and things can occasionally switch places based on situations and operations.23 Siegert refers to Serres’s arguments on the collective binding power of the quasi-­object: a parasitical object that stands as the third in-­between, but in so doing is also mobile and in circulation in dynamic ways, as Serres himself shows in The Parasite. Siegert brings these points into the discussion of media’s theoretically important underpinnings of the quasi-­objects’ powers. Hence cultural techniques and the operations that produce ontological distinctions are not merely referring to seemingly stable structures such as architectural elements (like doors opening and closing) or organizing principles (like the grid as a tool of design and projection, addressing and data); they are also mediating middles that circulate with a gravitational pull and an

38   Operations of Operations

antigravitational push. Operations are sustained by operations and their situated (but shifting) dynamics. Operations are meant to be persuasive. Siegert points out that, for Serres, this mechanism of circulation of the quasi-­object amounts to establishing a theoretical underpinning of the dynamics of collectives and sociability: no thing without a collective, no collective without quasi-­objects or things. To build on this point, the quasi-­object does the work that becomes central to the developing argument. In short, operative ontologies not only state and list the existence of ontological distinctions but are the engine that mobilizes them in the first place. In this mobilization, they thus become an approach that helps to map them as effective in institutional and other situations—­and the techniques themselves (as mentioned) as generative operations. Specific analysis of material practices as reality-­creating—­both in the material and symbolic sense—­has to (self-­reflectively) acknowledge that those practices themselves may be in movement. The symbolic and the real are in a tight, interconnected operational link, also reflected in specific case studies of things in literal movements, like the ships that are of constant interest to Siegert as sites of media and cultural techniques.24 From architecture to ships to infrastructure to visual design, such material objects are ontological agents that shift the world through their operations. Operative ontologies imply ontologically weighty operators, subsequently useful to realize both for methodological considerations and their material, ontological significance. Epistemology and ontology go hand in hand. The use of operative ontologies and the cultural techniques they refer to becomes ontogenetic in ways that are also intuitively present in the versions of STS and new materialism already mentioned. Of the various references mentioned in the development of what Siegert speaks of in terms of operative ontology, Barad’s agential realism has received less attention. I believe, however, it can be particularly useful as theoretical scaffolding for the topic at hand. Barad’s points are especially important as they feed into both a dynamic set of references of how “human and nonhuman, material and discursive, and natural and cultural factors in scientific and other social-­material practices”25 are entangled—­and how this, too, is defined by the onto-­ epistemological stance to cultural/media analysis. So, while the field

Operations of Operations   39

of new materialism is quite contentious—­often due to misreadings or lumping it together with all things “non-­human” or “vitalism,” or even Object-­Oriented Ontologies—­there is rigorous methodological advice to be found in how to treat matters of knowledge and knowledge of matter in operative ways.

The Operative Apparatus To briefly pick up the concept of agential realism, it is useful to note that it continues some of the same work as Deleuze and Guattari’s notion of agencement. That concept (translated as an assemblage that does not convey all aspects of the original) designates the material and mobile forces that construct and shape the world in concrete situations without being reducible to mere things here and now. Assemblages are not just tangible things but constellations of semiotic, social, and material flows26 that act, in some ways, similarly to what quasi-­objects do. Operative chains and agencement resonate in their ontological implications. As far as their pragmatics go, they enable a way to build up material ways of defining systems of knowledge that operationalize the material world (later examples in this book include geographical territories, colonial expeditions, smart urbanism of driverless cars, AI systems, and so on). Hence it is easy to see how agential realism links up with the onto-­epistemological stance, where knowing and being are inseparable. This is no longer the Heideggerian Being, but one that speaks to material practices in quantum experiments, the messy dynamic ontologies of apparatuses and the feminist materialisms of ethics and performative senses of the word. The scientifically and nonscientifically constitutive (cultural) techniques that have epistemological power are also materially present in this stance that helps to understand the stakes in mobilizing the terminology of operative ontologies: questions of how we know (including how we image, visualize, abstract, grid, map, and diagram) become part of the ontological register too: What objects and processes are involved in and produced by those operations of knowing? Barad’s pithy definition of onto-­epistemology is useful also in qualifying the “onto” in operative ontology: “Knowing is a material practice of engagement as part of the world in its differential

40   Operations of Operations

becoming.”27 Cultural techniques are defined by intertwining material and symbolic registers of knowing. While Barad’s take is most focused on the scientific instruments of quantum physics as onto-­ epistemological interventions, the notion of the apparatus that is at play becomes a concept that gains a life of its own. It does this to the extent, I claim, that it specifies the stakes of operative ontologies as far as they also concern operational images as apparatuses. To quote Barad at length: (1) Apparatuses are specific material-­d iscursive practices (they are not merely laboratory setups that embody human concepts and take measurements); (2) apparatuses produce differences that matter—­they are boundary-­m aking practices that are formative of matter and meaning, productive of, and part of, the phenomena produced; (3) apparatuses are material configurations/dynamic reconfigurings of the world; (4) apparatuses are themselves phenomena (constituted and dynamically reconstituted as part of the ongoing intra-­activity of the world); (5) apparatuses have no intrinsic boundaries but are open-­ended practices; and (6) apparatuses are not located in the world but are material configurations and reconfigurings of the world that re(con)figure spatiality and temporality as well as (the traditional notion of) dynamics (i.e., they do not exist as static structures, nor do they merely unfold or evolve in space and time).28

Naturally, this is geared more toward a science studies focus, but it implies a wider field of applicability of the terms as they travel from one discipline to another. This definition of apparatus that is rather different from the apparatus of classical film theory of the 1970s seems to have it all: apparatuses establish how the world is known but they also reconfigure the world; they produce differences while also being objects that can be known; they are dynamic and shape matter and meaning. This can work as a parallel definition of operative ontology as it starts to address instruments, images, diagrams, charts, tables, and more. In other words, we can start to refer to the operational image as one of onto-­epistemological operations as well as cultural techniques, but which have this particular quality,

Operations of Operations   41

specified in institutional operations when dealing with technical images. We will return to this topic in later chapters when focusing on AI and machine vision (in chapter 2), infrastructures of data and operational aesthetics (in chapter 4), and autonomous vehicles and operationalization of the city29 (in chapter 5) as one of the inter­ connected techniques of imaging and as skins, layers, and ecologies of traffic sensor data.30 Showing how these insights about the intertwined material and epistemic aspects are already nested inside cultural techniques literature can further emphasize some aspects that are useful to our question about the operational image. The particular pairing of aspects of real and symbolic, material and semiotic, is central to Siegert’s way of approaching the nature of techniques, which also includes discussions related to technical media, such as signal versus noise and how operations on the real/material turn into culturally relevant symbolic statements. While such themes are partly based on signal engineering, they are also questions that media theory has to deal with if we aim to understand quasi-­objects that populate the operational field and infrastructure of electronic media. Importantly, Siegert also addresses visual culture in various case studies of cultural techniques and their operative apparatuses through discussions of maps, diagrams, and grids (as well as paintings) that in many ways then facilitate the connection we must make between operative ontology and the operational image. It is true that one could sum this up with a reminder about the recursive nature of cultural techniques, that on a fundamental level, cultural techniques include writing about writing, singing about singing, but also images about images (referring to the operative chains that define the object/practice at hand).31 Operational images, as we encounter them, are images about images, a point that applies to many of Farocki’s works and Hito Steyerl’s videos too. This recursive pairing comes out also in this book: many of the examples (like the later discussion in chapter 4 concerning Geocinema’s Framing Territories) feature images about (operational) images about territories (that are made into images). This recursive chain skips between images and territorial interventions, where territory refers not only to land but to any planetary surface framed through Earth observation technologies, from satellites to sensors.32

42   Operations of Operations

Images can appear as concrete things to see and perceive in control rooms, in scientific papers, as screens, or as portable entities (as photographs have been on glass plates and other material surfaces). However, images also contain epistemic worlds beyond any representational function. They can mediate between concrete pictures and abstractions as diagrams and statistical visualizations do: an example could include visualizations of weather and climate data in analytical graphics since the nineteenth century.33 For Siegert, the abstractions enabled by particular kinds of imaging are where the operational nature of such media becomes most clear. This refers to images that are not necessarily looked at, like painting or art, but ones that are followed, offer guidance, and reveal multiple layers of scientifically relevant abstractions interpreted in pragmatic ways.34 For example, scientific, architectural, and even mathematical models could be seen as operations that are less about mimesis than they are about the pragmatics of following with a necessary “suspension of disbelief.”35 Another instance of following is maps, an example of the navigational image and a theme Farocki was interested in when engaging with computer graphics, games, and animation that facilitates a logistical way of dealing with bodies, media, and territories across virtual and actual space.36 As per Siegert, the map is the territory insofar as they are instruments that enforce epistemic order(s): “Maps are not just representations but also instruments. They are based on mathematical operations, and they constitute a substantial part of a cultural practice.”37 As per one of the core commitments of this book, the mathematics of the (operational) image starts much earlier than digital imaging. These commitments include a problematization of overly simple divisions between analog and digital practices of images. Looking at technical practices brings out this aspect well, articulating the interplay of continuities and discontinuities as the fundamental tension in our case.38 A contemporary architectural imaging environment that is able to integrate multiple scales of knowledge and interaction—­not just spatial scales and generative forms, but energy consumption models, land value projections, environmental variables, and so on—­is not of the same order as the projective calculation of a map or the photogrammetric measurement of an architectural façade circa 1900. But the longer-­

Operations of Operations   43

term history of the operational is of particular use when trying to understand its institutional role beyond technical capacities. In short, the operational image becomes understood through the frame of cultural techniques; it becomes understood as an operative ontology that establishes a pairing of epistemic and material, of symbolic operations that impact the world (or the “real”). Such a dynamic link between material and epistemic aspects plays in the discussion concerning the grid, a fitting example in the cultural techniques repertoire of operationalization of ontology as far as mathematics and materiality (visual image/representation) are concerned. The grid sits well as a technique of scale that can mediate scalar shifts of the very reference points it builds up as meaningful units of data while its own structure allows for this fluid, shifting role as an epistemic hinge. For Siegert, following Foucault and Deleuze, the grid exemplifies the key modern concept of order as a spatial arrangement in establishing space as “data and addresses.”39 As Siegert puts it, “In other words, it presupposes the ability to write absence, that is, to deal equally efficiently with both occupied and empty spaces.”40 This ability to deal with binary-­coded data or, for example, to encode space/territory into such terms, is an effective part of its pragmatic value. In media historical terms, the grid can be placed as part of the Renaissance-­focused history of zero, accountancy, and the perspective of the three-­fold of numbers, business, and visuality.41 In Siegert’s take, the focus is on later developments, after the Renaissance, ending up in a modern technological bureaucracy that rings familiar to anyone working with the operative grids of Excel, in those images not primarily meant for human eyes (and for many, for other reasons purely an inhumane practice of administrative media life). As is obvious from the previous note on scale, there are multiple kinds of grids and their differing kinds of epistemic orders, to echo Foucault in The Order of Things, mostly paying attention to the logic of tables from the Classical period of ordering visibilities to the Modern period of labor, life, and language.42 Siegert’s argument includes examples of different techniques where a quasi-­object, an operation that forms images, is employed: from the pictorial veil as construction of visual worlds (Alberti) to territorial control (colonial gridding in the Americas) to twentieth-­century administrative grids

44   Operations of Operations

(including spreadsheets). Indeed, Siegert acknowledges the underpinning question that I am keen to move forward to our discussion of operational ontologies and operational images too: “Can the expansion of Western culture from the sixteenth to the twentieth century be described in terms of a growing totalitarianism of the grid?”43 While this can be easily brushed aside as a generic rhetorical question concerning modern rationality (cue in “the age of the world picture”44), it implies a more specific follow-­up task: to investigate the specific techniques that sustain such forms of power, knowledge, exploitation—­a nd even extraction as capitalism and colonialism’s defining operations—­that hit the ground in concrete situations.45 Thus the notion of assemblage as part of the operative ontology comes to the fore. Indeed, to discuss grids as navigational tools (longitude and latitude) that hit the ground and reformulate colonial forms of marking absence (where there was none) already condenses a massive story of media, ships, violence, and subjugation into an operational figure where images as maps and diagrams play a central role. They occupy a place as concrete images and as marking effective follow-­up claims that are sometimes expressed as law and property, for example. Indeed, as Vismann points out, the Roman law and its variations of res nullius and terra nullius are also operative techniques of property as a special case of imaging—­as marking a territory, a scratch, a line, a cut. Grids mark a spot that can be marked as absent and legally to be owned, implying the physical data regimes of property (which returns in the immaterial property that pertains to contemporary digital images in other ways).46 Capital­ ism loves such legally binding (colonial) fantasies, especially those that can be projected into existing territories and bodies through onto-­epistemologically existing images. Contemporary legal theory on data and property would agree whether we deal with assetization or such productions of control regimes as NFTs as a special case of an operative ontology. These fantasies mobilize through symbolic actions, materially effective interventions, that, in the words of David Joselit, are the ready-­made reversed: “Duchamp used the category of art to liberate materiality from commodifiable form; the NFT deploys the category of art to extract private property from

Operations of Operations   45

freely available information.”47 Perhaps the same can be specified to “images” and not just “art.” Significantly, grids are produced while producing, combining the mathematical and the material elements significant for our case here concerning visual technologies and surfaces that work with or as data. They produce territories and places, as abstract and followed in the concrete—­accurate or not, misguiding or not. Grids are defined by their media material sites and institutional uses. In Alberti’s description, the veil is a hybrid at the crossroads of what unfolds as the media technologies of textiles and print.48 Similar cross­roads of vertical and horizontal grids define other hybrids that are important in the axis of operational ontologies-­cum-­operational images: the grid and topography, fantasy and territory, speculation and imaging, industrial production, and standardization. The and marks the spot of connection and combination, transformation and operation. Images work as more than content for eyes: they can have significant legal consequences (a special case of not just programmable images but images that program and format), and they have, as voiced throughout this book, a specific role in architecture and design. As Siegert points out, the significance of grids for modern architecture is demonstrated by such fitting crystallizations as Ernst Neufert’s Bauordnungslehre, “Architects’ Data” as the translation has it, published in 1943, in the midst of National Socialist Germany with Albert Speer’s endorsement. The book, a classic in its field, is shot through by the various operations of grids from drawing aids to space structures, perspective grids to planning grids.49 Standardization of units is a core feature in the history of rationalization of design, architecture, and construction. As Shannon Mattern points out, Neufert’s idea of standards might have emerged from his “experience as a bricklayer,”50 but it became applied to different materials, sizes, and scales. Standardization became a logistical view to space, meant to “facilitate the circulation of ideas, promote easy translation between disparate industries, save on storage space, and facilitate what he called ‘rapid design.’”51 Automation of architectural design becomes even further emphasized with platforms and software suites such as those offered by Autodesk, part and parcel of the contemporary versions of “rapid design.” The

46   Operations of Operations

ability to model and speculate with geometric forms becomes part of the ability to consider operational images among brick, paper, software-­based units, and broadly speaking, “data,” whether expressed as pixels or vectors, or underlying relational databases. Here, several types of lines and forms appear as potential commands, as (software) instruments for the composition of images: polylines, splines, rays, multilines, arcs, spheres, polygons, rectangles, points, helixes, donuts, and, for example, “modeling environments” such as in Blender, AutoCAD, or even SQL platforms for modeling data relationships. Operationality is thus not merely a quality of an image but of bespoke (design) environments too. As far as grids are concerned, their role as instruments of imaging and visuality feature in the long wake of aftereffects that can be written into a story of data-­driven operations: assigning a place for standardized things to be subjugated to, ordered to, removed from, guided toward, or potentially inhabited. Relational databases would be one interesting example in the history of grids, tables, and ordering that becomes integrated into contemporary data operations too. Furthermore, as a special case of what comes to haunt AI from chessboard grids to Go game grids, pixelated space becomes itself the hinge between different operational scales concerning space and temporality: The fusion of matrix grid and GPS has ensured the global presence of the operationalized deixis first conceived of in connection with the grid-­and register-­shaped settlements of South America. Indeed, what better way to describe some of the basic aspects of our media culture than to point to the mutual translatability of cartographic grid, topographic grid, planning grid, and imaging grid? Linked with the convertibility of these diverse grids and with corresponding scaling techniques, grids—­a formidable cultural technique—­have become the basis of a mediatization of space from which hardly anything can escape.52

Such are techniques of the visual, where things are not merely seen but placed and counted and resolved.53 It’s not even clear if we should call it visual anymore—­but let us leave that question for the

Operations of Operations   47

next chapter. Of course, much does escape from the gridded mediatization of space. But people are often convinced of the rational superiority of the grid in the manner of the earlier quip: as much as the map is the territory, so too is the grid the territory, demonstrating the material powers of epistemic operations, including that powerful technique of random-­access. Naturally, gaps between pixels, glitches in grids, and resolution hiccups reveal another layer in those visual operations. Maps and grids can be contested, debated, rejected even. Resolutions appear because of such instruments, but they are themselves produced as part of operative chains of resolutions that imply optical resolution and decisions made. In artist Rosa Menkman’s project “Resolution Studies,”54 a creative methodology of reverse-­engineering of the forms of standardized visuality leads into proposals for “impossible images”: images that work against the standardized units of reference and constraints of space, time, energy, and signal, or noise or cost. Some are speculative and hypothetical, and some are real in the scientific contexts of images, even if they are not resolvable to the human eye. So too has the notion of a table of calculation and knowledge undergone a shift when it comes to more recent dataset techniques. The grid of tabular data becomes instead a vector space mobilized in machine learning practices, for as Adrian Mackenzie puts it, “Vectorization remaps the grid of the table into the expanding dimensions of vector space,”55 which becomes a significant change in what data practices are effectively present in contemporary machines, and which ones deal with visualizable structures of data, or what is contained in particular representational surfaces: “In the vectorized and matrix-­form practices of vector space, machine learning produces for the first time a meta(s)table volume that cannot be surfaced on a page or a screen.”56 Operative ontologies are thus ways to understand these shifts of register between visuality and data—­different modes of addressability function in grid-­based and other visual data systems, forming one basis for computation.57 Instead of merely focusing on when operational images became digital, we are dealing with a legacy of techniques of power that operate on territories, (literally) hitting the ground. When they do, they operate on bodies. While chapter 3 deals with similar themes in understanding space through

48   Operations of Operations

established planes and lines, the final section of this chapter turns to the politics of operations, highlighting some existing work that has already tackled the political history of logistics of space.

Operations Hitting the Ground Much of the discussion on and around operational images reveals a distinct interest in topographic territories and living bodies as they are defined in images that are not considered mimetic or even representational. Instead, such instrumental images work according to the onto-­epistemologically effective apparatus of operations: guiding, putting in place, transforming into data, measuring, analyzing, diagramming, and so on.58 The history of photography has already mapped some of these operations in effective ways, from the Bertillon system of photographic identification for police to facial recognition urbanism, from military targeting systems to contemporary border surveillance drones, from hand-­drawn maps to the Global Positioning System, and from the nineteenth-­century invention of the statistical distribution of averages to the twenty-­ first-­century machine learning and training sets. The infrastructures of imaging become grounded in a history of media that is one of extended grids and addresses all the way to the bottom. Grids are, as Sean Cubitt suggests, “visualisations of proto-­d ata media.”59 Thus they are both concrete instruments as instrumental images—­“spreadsheets, databases and geographical information systems”60—­as well as infrastructures of knowledge. Furthermore, it is important to note the variety of other techniques of grids and resolution: as a special case of linking geography and climate data, climate model simulations, such as in the Intergovernmental Panel on Climate Change (IPCC) assessment reports, is a further example where data enables epistemic projections that do not lend themselves to be rendered on the axis of “good” and “evil” politics.61 A political history of operational images needs to address the specificity of the cases as per timelines and dates, uses, and institutions. Now images might be automatically time-­coded and stamped even if they circulate removed from a historical and located specificity, a theme noted by Farocki and even more so by Allan Sekula, who coined this as the fundamental characteristic of photographic

Operations of Operations   49

images: traffic in images that is also traffic in currency—­or as will be seen later in chapter 5 on autonomous vehicles, traffic as data and currency.62 As traffic concerns the interchangeability of goods and bodies, we see the potential to connect this body of research with analyses of racial capitalism and its longer histories, including the role of grids and diagrams as logistical media. This lineage of arguments can specify the (briefly mentioned) totalitarianism of the grid in ways that speak to a crucial area of interest to both operative ontologies and operational images. In the contemporary context, much of this work gravitates toward operations of surveillance. This is understandable considering the amount of attention facial recognition systems have received and how their design—­from training sets to institutional use—­has built upon existing racial discrimination and other forms of direct and indirect violence and “engineered inequity.”63 As is clearly articulated in Simone Browne’s Dark Matters, questions of diagrams and the broader visual culture of surveillance are to be read against the racialized backdrop that also anchors image practices. But these are also nonrepresentational techniques of operative ontologies and act, like in such infamous cases as the Brookes slave ship image of 1787, as diagrammatic optimization mechanisms, part of the logistical plan of the Middle Passage. As Browne underlines, this is where the “violence of slavery crudely reduced to geometric units”64 becomes clear through the organizing description of an image of a stowage plan. Enslaved black bodies, ships, and routes become easily abstracted into a logistical operation—­and a logistical image—­while at the same time, those logical operations (can) become historicized, locatable, and in this sense, embodied. The one does not have to dismiss the other; an analysis of images as they incorporate logistical operations can include recursive reflections on what sort of images about subjects are in play. To quote Browne: The Brookes diagram, “in serving the cause of the injured African,” offers an overview of the stowage plan of the slave merchant’s ship and forces me to reflect on my own surveillance practices in reading the archive of transatlantic slavery. The slave ship schematic is clinical in its architectural logic and provides an almost aerial viewpoint, overlooking the tiny

50   Operations of Operations

black figures set to represent the enslaved drawn “like so many cartoon figures,” as Spillers describes. What does it mean that I now look to this plan, but not from the elevated and seemingly detached manner as it was first intended to be looked upon?65

While operational images are not primarily representational, they can also turn to fundamental questions of subjectivity and operations of knowledge, as Browne shows. For example, she refers to Donna Haraway’s description of the universalizing God’s eye trick of the view from above that is often associated with aerial photography but is also part of other objectifying practices of taking distance, of remote sensing. There is, too, another layer that cuts through such diagrammatic images. Instead of optical surveillance, images such as one sees in various kinds of diagrams are part of the operative chain of logistics and already represent a shift from optics of surveillance to calculated measurement and optimization. This also implies that an important aspect is not only what is in the image (and its condensation of centuries of the violence of slavery) but also how it is institutionalized: images that define, for example, the role of merchant ships and routes across the Atlantic, the West and East India Companies, or then the subsequent national institutions or other arrangements part and parcel of the logistics of racial capitalism. Here, I am mostly narrating existing positions made clear by Browne, Ruha Benjamin, Stefano Harney, and Fred Moten. Harney and Moten have argued that the link between the (Atlantic) slave trade and the birth of modern logistics is a central part of how we should look at operational aspects of capitalism. Their argument refers to specific practices, but also to how this becomes the ontic operation that defines modernity at large: Modernity is sutured by this hold. This movement of things, unformed objects, deformed subjects, nothing yet and already. This movement of nothing is not just the origin of modern logistics, but the annunciation of modernity itself, and not just the annunciation of modernity itself but the insurgent prophesy that all of modernity will have at its heart, in its own hold,

Operations of Operations   51

this movement of things, this interdicted, outlawed social life of nothing.66

Without going into further detail, such arguments also resonate with others, including Anna Tsing’s writing about the plantation system as the production of colonial operations where enslaved people and plants were placed on grids of data, making them “self-­ contained and interchangeable units.”67

Grids, Grids, Grids Logistical media of capitalism takes place in different material formations. Some of them are ships as part of diagrammatics that relate people to space and vectors of transport. Some are surface territories, grids linking labor and agricultural production. Such operations of data are central elements in a political history of cultural techniques of operational images: the grid marks both a place and nothing. The operations of data define this particular logic of the image as it defines the movement of nothing where unformed objects and deformed subjects are held. The operative lies in that link, too—­the image as logistics, and the image as institutionally operationalized, as per one of the key arguments in this book. To be sure, this is not a claim that the logic of the diagram is determined and limited to racial capitalism or a generic issue with abstractions, but rather a statement that we can offer much more nuanced ways of understanding the shifting institutional uses for such concatenations that define how images and data are operationalized in a variety of historical practices. Aptly, Sandro Mezzadra and Brett Neilson coin their proposal to analyze contemporary capitalism as a politics of operations. And while their arguments emerge from a slightly different direction than this chapter, which has been working through the centrality of operative ontologies in media theory, I want to claim there is a resonance here that can be teased out to facilitate some moves across media theory and the cultural theory of racial capitalism and analysis of (images of) territorial formations. Mezzadra and Neilson argue for the usefulness of focusing on operations; it helps

52   Operations of Operations

to facilitate the back-­and-­forth between abstractions such as capitalism and their concrete events as they regulate, guide, and reconfigure bodies and territories. Their trope of the “image of capital ‘hitting the ground’”68 is central but in ways that imply that operations also format (reconfigure) the so-called ground instead of it being a metaphysical entity. The ground is ontic before it becomes ontological. Mezzadra and Neilson define the ground “in a sense that is at once material and prospectively constructed as an operative surface on which capital intervenes,” through different means that take into account “the specificity of spatial, social, legal, and political formations with which capital must grapple as it becomes enmeshed in dense constellations of flesh and earth.”69 The three main operations Mezzadra and Neilson focus on are extraction, logistics, and finance. None of these, as you can see, is particularly about images. However, Mezzadra and Neilson’s argument also becomes important when considering what the operative/operational is in the context of the broader landscape of contemporary theory. It maps onto the earlier discussions about operative ontologies and onto-­epistemologies and includes one of the available routes to expand on operations of racial capitalism. Certainly, some of this comes close to what has already been mentioned and what will be developed in terms of the logistical image70 while also laying out some other coordinates for us. Operations are complex. They do not imply a simple causal link or the application of generic principles to specific situations. On the contrary, operations incorporate feedback loops and complexity as they take place among a dynamic situation of living bodies and ground surfaces. Indeed, Mezzadra and Neilson argue that operations should not be “reduced to a device of ‘techno-­economical organization’ [they here refer to Jean-­Luc Nancy] as if it were a simple relation of cause and effect, or input and output.”71 While the conditional statement of “if/then” is easily seen as the backbone of the operational rationality present in the visual narrative offered by Farocki and cruise missiles, we cannot reduce the operational to being operationalized in such simple terms. Instead, as Mezzadra and Neilson continue, we need to see the operation as one that brings together activity and potential, facilitating a grounded view of the workings of capital.72 This also implies a certain notion of

Operations of Operations   53

agency—­an agential realism even perhaps—­which does not dispense with multiple levels of abstractions but looks for situated hinges where operations take place as agencement: “capital in action . . . between conceptual skies and empirical battlegrounds . . .” 73 For Mezzadra and Neilson, the argument about the politics of operations relates to discussions about capitalism both as an abstraction and a dynamic defined by local operations with planetary repercussions. This involves asking what the role of such abstractions is and how they relate to spatial and social practices and situations, arguments about geographical differences, and broader heterogeneity in ways that avoid universalizing notions—­even if operating across vast scales. For Mezzadra and Neilson, the focus on operations allows for recognizing capitalism as an engine of difference while acknowledging the usefulness of such general terms of reference. One can see there is a similar background drive, as per the idea of the ontic and the ontological, with the ontological being defined only through actual ontic operations. The operative powers of the ontic are not mere actualizations of the abstract ontological, but this allows us to appreciate the constitutive power of what an operation is: it has an existence of its own across spatial, social, and technical components. In other words, it also becomes historicized in ways that start to sound like an STS-­type of an analysis of the life cycle of a technical project or a dynamic of an institution (a stack of turtles that can be called infrastructure too): However automatic or given the results of an operation might appear to be, when we look at it from this perspective, there is always a back story, a drama of frictions and tensions in which the efficacy of the operation appears far more fragile and elusive than might otherwise be assumed. Approaching the operation from the heuristic point of view of this interval provides a kind of freeze-­frame that brings into relief the combination of social activities, technical codes, and devices that make an operation possible. Such an approach also allows us to look at the outcome of the operation without taking it for granted. For us, then, an operation is a process with a beginning and an end; a process that accomplishes something without necessarily yielding a material thing; and a process that impinges

54   Operations of Operations

on others, affecting possibilities and establishing multifarious and not necessarily predictable connections. This is why, in the analysis that follows, we are not merely interested in focusing on operations in the singular moments but also consider how they concatenate to link up and contribute to the fabrication of the world.74

The recursive links between extraction, logistics, and finance are discussed as histories of labor, territory, and capital. Could these histories be the ones that frame operational images too? Operational images of labor, territory, and capital? In any case, these are also the histories of conflict, underlining that frictions underpin the operational smoothening of logistics, optimization, grids, and scaling. One can also observe the politically useful emphasis on reading points of inoperation and breaking down as sites of struggle,75 and even the analytical angle to the concatenation is useful. The recursive links between these operative spheres can be seen as an operative ontology, one where the media theoretical notion becomes incorporated into an analysis of contemporary capitalism that then becomes itself one “ground” against which operational images are to be viewed and reviewed. In the broadest of terms, and what will come to define some arguments and angles later in this book, the trio of operations—­ extraction, logistics, and finance—­are defined by how geographical territories are reformatted in histories of extractivism that proceed both as material property and data-­driven operations, linking the two beyond metaphorical references: “New fields and quarries are opened in the landscapes and spreadsheets of contemporary capital.”76 Extractivism is linked to the production of material and legal territories in logistical terms giving rise to various spatial arrangements such as zones (e.g., Special Economic Zones) and (logistical) corridors that redraw borders and routes anew. The imprint of extractivism is doubled in the military and colonial histories of logistics that becomes a fundamental characteristic of the contemporary planetary condition, both in terms of material transport and the logistical media (data, software)77 underpinning this materiality. In Deborah Cowen’s words, “Logistics is not only about circulating stuff but sustaining life.” 78 Financial operations are involved not only in

Operations of Operations   55

the myriad ways discussed above but also with their own futurity as per particular trading and contract types like derivatives.79 Hence, it is fair to say that various spatial–­logistical formations are also conditioned by temporal futurity enacted in financial terms, continuing what Sekula referred to as the universalizing currency trope of “traffic in photographs” with the twenty-­first-­century logistical media and its operational images. We trade in operational images, which are increasingly an operational element in the financialization of earth grounds and undergrounds as commodities.80 Here, the point about operative ontologies becomes rerouted through images and finance as they hit the ground, where an ontology of “ground” is only established through operative techniques such as remote sensing coupled together with the scaffolding of finance and extraction industries tying a knot between cultural techniques research and the contemporary analysis of capitalism via operational images. Operations are where images of landscapes, property, and so on, are sustained and mobilized. Though this can be approached through the trio of extractivism, logistics, and finance, that does not need to be it: other operational spheres and scales of references can be included and addressed in the rest of this book. Forms of techniques (that are other than military targeting or the even broader aerial view and photography)81 are central in understanding the operative ontologies at play: images materialize at specific sites and spatial–­logistical formations, and they are operative in specific institutional practices (military, finance, extraction industries, scientific remote sensing, landscape surveys, architecture, and planning, to name some). In this sense, my argument concerns the mobilization of the terms “operation” and “operative” as a methodological hinge that facilitates a discussion of situated material practices with an epistemological force as per operative ontologies and the work of Karen Barad. In addition, it concerns how operational images are implied in a range of technological practices that define the contemporary moment. And while there is a need to look more carefully at the longer media archaeology of operational logic82 that emerges in the contexts of operations research and scientific “operationalism,” we already face ongoing and intensive debates about the specific uses and range of the operational: to measure, to guide, to analyze, to

56   Operations of Operations

mention a few of the characterizing features where the operational image functions as a logistical apparatus, one that reformats territories and bodies to its own onto-­epistemological needs. Dealing with operative ontologies as a frame for operational images helps us to build a stronger theoretical case for further discussion: the notion of operation has become a central term in the long twentieth century of military tactics and scientific management, and it has featured in several cultural and media theory contexts over recent years. This chapter proposes that operational images can be understood as onto-­epistemological knowledge instruments, able to trigger repercussions across a multitude of scales by combining epistemic abstractions with site-­specific impacts, such as those involved in the remote sensing of extraction sites.83 Moreover, ontic operations are part of all sorts of contemporary mechanisms and institutions such as trade and transactions, finance and property (contract operations), thereby linked to those genealogies of (racial) capitalism already mentioned. These are also the institutional contexts where one can start to draw out some of the contemporary versions of cultural techniques: if anthropotechnics is fundamental to establishing any reference point of the ontological “human”; temporal techniques of calendars, measurement, synchronization for establishing “time”; and grids, ruling, scales, and measurement (again) for establishing “space,”84 we can add complex recursive layers—­and we will continue to formulate this in the next chapter—­that also pay attention to contemporary forms of posthuman agency, data regimes, complex computational infrastructures, and territorial management that have both material and informational dimensions.

TWO

What Is Not an Image? On AI, Data, and Invisuality Not seeing anything intelligible is the new normal. —­Hito Steyerl, Duty Free Art: Art in the Age of Planetary Civil War

What Is Not an Image? This chapter is on the operations of invisibility and invisuality. Of those two terms, the first seems to prescribe an optical situation of (not) seeing, while the second pertains to mathematical functions, data, and platforms. Considering how often seeing, vision, and other metaphors are used when discussing observations relating to and based on data, this division is not entirely clear-­cut. What is clear, though, is that speaking of images in such terms helps to avoid a binary division between analog and digital images. Some of the terminology discussed in this chapter is also meant to open up to questions of data. There is a wonderfully useful phrase from Georges Didi-Huberman: “The entire history of images can thus be told as an effort to visually transcend the trivial contrasts between the visible and the invisible.”1 We can add that this history also needs to transcend trivial contrasts between data and the visible too. Of course, knowing what is trivial and what is not demands a significant ability to evaluate and describe changing material, semiotic, and social contexts of images. As far as photographic theory is concerned, this question was central to how the material existence of an image becomes invisible when the image as representation emerges: to have an image is to forget there is a technical 57

58   What Is Not an Image?

mediation that is the apparatus of capture and visualization. In Roland Barthes’s words, it is the pesky referent that gets in the way of the photograph itself: “Whatever it grants to vision and whatever its manner, a photograph is always invisible: it is not it that we see.”2 The same idea was voiced by Erwin Panofsky, too, in relation to painting since the Renaissance: “The material surface upon which the individual figures or objects are drawn or painted or carved is thus negated, and instead reinterpreted as a mere ‘picture plane.’”3 The material mediation, then, is what enables seeing and becomes unseen. Furthermore, the history of photography—­a nd the history of theories of photography—­has been, for the most part, about what sort of visible or invisible rays, what wavelengths of radiation, are perceived as relevant and valid. What sort of light counts and literally also counts, as calculational images, as programmed images? Tell me what spectrum of electromagnetic radiation you are concerned with, and I will tell you what your discipline is. Is it the visible spectrum that registers people, objects, and landscapes? Or is it the invisible spectra observed by other remote sensing systems on satellites and the ground that register people, objects, landscapes—­ but also a multitude of other layers of reality from temperature to humidity, and from mineral resources to the chemical composition of vegetal surfaces (see Figure 3)? These themes condense many of the approaches in visual theory as it is reflected through measurement instruments that sometimes are cameras, sometimes hyperspectral remote sensing devices, sometimes telescopes capturing the night sky as photographs of luminosity filtered through the spectroscopic lens. Some of the questions of the history of technical sensing and imaging resurface in contexts of data. In other words, if photographic theory has often been premised on the idea that it is a system of signs (and thus information), we can propose that images are systems of signals and noise and thus susceptible to interpretation and pattern recognition. A statistical distribution of characteristics becomes a base unit of an “aesthetic,” a recognizable trait for machine/computer vision and other systems that mobilize such an aesthetic as epistemically relevant. This also applies to remote sensing: classification and perception of different material characteristics from vegetation to soil

FIGURE 3. Operational images emerge and work along the pipeline from data acquisition to data analysis while occupying a special way of tapping into wavelengths of (in)visibility. Here visualised are “spectral characteristics of energy sources, atmospheric transmittance, and common remote sensing systems.” Diagrams redrawn based on the source, Lillesand, Kiefer, and Chipman, Remote Sensing and Image Interpretation, 2 and 11.

60   What Is Not an Image?

and from sand to asphalt, snow, clouds, and water is often perceived as a spectral signature that actually is a spectral response pattern; not necessarily so much an “absolute and unique”4 representational entity as much a varying characteristic. These might indeed be quite technical dilemmas, but they relate to the broader sense in which questions of sensing and images are integrated into a different epistemology than that of unique objects in representation. This starts to trouble what is visible and invisible as it becomes a question of masking/unmasking reality. It also includes how to reframe these questions in the age of cloud analytics, a term used by Louise Amoore to refer to the “capacity to extract patterns and features from data to open onto targets of opportunity, commercial and governmental.”5 In such descriptions that shift the frame of knowledge from optics to software and data, from pictures to models and patterns, we are not so far away from what Wendy Hui Kyong Chun argues is one essential feature of visibility/invisibility in the computational culture of software and interface design. Complexifying notions of transparency, visibility, and opacity, Chun dissects the relation of seeing as knowing, while arguing that we need to understand the (computational) system as “a conduit that also amplifies and selects what is at once real and unreal, true and untrue, visible and invisible.”6 This, one can argue, is also where data systems—­ operationalization of images as data in platforms—­involve a production of what is counted as real and unreal, true and untrue, visible and invisible. Of course, many would be quick to add that the power of withdrawing from view and becoming invisible—­or worse, natural—­is the work of ideology.7 But it works the other way round, too, as per Foucault’s quip that “visibility is a trap” that is materialized as architectures; this trap is designed as observational events bootstrapped in existing racialized and gendered systems of power.8 This chapter outlines how the concept of operational images helps to map different regimes of invisibility and invisuality of images by providing three partly overlapping angles to the issue: a platform, a dataset, and a model. These offer insights into the current discussions about the transformation of visual culture (and, for example, issues related to labor of AI images) and the aesthetic and epistemic context in which images become something other

What Is Not an Image?   61

than pictures. In simplest terms, the focus of the chapter is how visual culture studies (broadly understood) deals with invisuality as it is articulated through questions of data and machine learning techniques. Trevor Paglen on “invisible images”9 and Hito Steyerl’s “How Not to Be Seen” video essay (2013) are exemplary of the practices through which invisibility is not just unseen, but an entity produced, framed, and regulated as a cultural technique. These approaches also articulate the transformation of optics of vision and images in relation to a digital culture defined by machine vision. The previous chapter’s points about operative ontologies also help us understand how invisibility and invisuality are produced in specific media techniques. These terms, these three sections, are also ways to engage with the material epistemology of operational images in contemporary digital culture. Transparency and opacity, intelligibility and noise entangle on surfaces of images. Some can also become an affordance, switching the focus from the value of visibility to the primacy of opacity. Informatic opacity is coined as a technique of resistance of biometric regimes of imaging the (vulnerable) body in Zach Blas’s practice-­ led work.10 In Stephanie Syjuco’s 2019 installation Dodge and Burn (Visible Storage), digital photographic techniques of a chroma-­key green and Photoshop transparency checkerboard pattern entangle with imperial techniques of making invisible traditional patterns and garments, histories, and bodies. Grids appear at multiple scales of space, image, and bodies. Many other examples could be quoted, especially the question of camouflage, invisibility, and opacity that feature heavily in surveillance-­driven discourse that also makes use of the terminology of the operational image.11 Much of the use of invisibility and opacity as terms are part of the methods of figuring out what forms of perception relate to operations of power in digital culture and how transparency and opacity are also designed (in the sense Chun underlines as the function of a medium that is the conduit of the visible and the invisible). None of this was invented as late as digital culture, even if digital images are very specific kinds of nonrepresentational, rendered objects—­even models instead of representations.12 The longer photographic history of the invisible focuses on scientific imaging in physics, biology, and astronomy: of objects that are too fast to be

62   What Is Not an Image?

FIGURE 4. Atmospheric dust made visible. Microphotographies du laboratoire de Louis Pasteur, 1870–­1890. Bibliothèque nationale de France. ark:/12148/ btv1b3200016d.

sensed (separating sensing and imaging) or too far (such as the astronomic photograph, like the opening examples—­the Frauenhofer line mentioned in the preface and the astronomic analysis of luminosity at the Harvard observatory—­of this book). Microphotography at Louis Pasteur’s laboratory picturing minuscule atmospheric dust (see Figure 4), Ernst Mach’s photograph of a bullet in motion (1884), August Strindberg’s celestographs (1893), and Ottomar Volkmer’s statement (1894) that “the progress of the natural sciences, and more specifically, the progress that has been made in the fields of photographic technology and electricity in the past decade, is so marvelous that even today it allows us to fix a number of invisible facts in

What Is Not an Image?   63

nature in photographic images”13 testifies to this theme that runs through several decades of epistemic and aesthetic experimentation. Peter Geimer offers a detailed glimpse into this history, from such examples of scientific use to the emerging artistic practices in the first half of the twentieth century. Furthermore, in another context, he proposes a closely related question. Geimer’s suggestion to switch the question from what is an image to what is not an image is particularly useful for a book focused on describing and conceiving such images that are primarily not seen (perhaps not even meant to be seen, but watched?), but which take an important instrumental part in an operation. His term for the nonimage or even the un­ image narrates entities that cannot be seen as having been destroyed through archival deterioration or some other related process affecting old daguerreotype plates and other photochemical images.14 Such images at the threshold of their disappearance become part of the opening note on the materiality of the medium as its invisible support, although in this case, the ontological point becomes part of a temporal process. For Geimer, the genealogy of unimages includes photographs that have vanished or eroded, that are defective and glitched. Besides errors that expose the photograph itself, this genealogy also includes operations where the photographic plate intersects with other practices of light (visible or not): chemical effects on surfaces and skin, X-­rays, radioactivity, electricity, infrared, and so forth. Such an example is René Colson’s late-­nineteenth-­century physical experiments, related to the broader question of sensibility with the photographic as both an instrument and object of knowledge that prescribes realities that do not have an image of what they should look like but are diagrammatically traced into existence.15 Such operational ontologies of invisibility launch experimental probes into what is not visible, but that becomes a model for what can be known through the operations involved. They bring this invisible, even sometimes seemingly impossible view back as an epistemic and aesthetic object of manipulation into an apparatus of images.16 The function of images and instruments as experimental systems and models will be featured later in this chapter. While the question of what is not an image, in Geimer’s take, refers to archival and historical situations, photochemical affordances,

64   What Is Not an Image?

or liminal cases of photographic images, it is through this diversion that a broader question also emerges: it also implies all the cases of other (than) pictorial images. Unimages come to the fore, and unseeing becomes practiced in a version of Paul Virilio’s “sightless vision.”17 The operations of photographic chemicals and signs implies thus a broad field that is also the concern of this book: worlds that are not merely seen, as Paglen notes, but not necessarily only invisible either. The one that becomes a diagram, even if not exactly an image; the one that measures the world and produces a table or some other form of data media. Invisibility is thus not merely a sphere of existence that remains unseen; it concerns a praxis. Before it becomes ontological, it is operative. Here, Geimer’s points are useful to quote in full as they concern the institutional and technical operations of knowledge in what becomes a discussion close to the operative ontology that underpins aesthetics of knowledge. First, photographic visualization was not the transfer into the image of a phenomenon that had always already existed, though it had been latent; it was, first and foremost, the fabrication of an image. Second, as a general rule, it was not evident what this image showed; it needed to be decoded, stabilized, and isolated from an amalgam of facts and artifacts. And third, the “invisibility” that preceded it was not nothing but a sort of structured darkness that, despite its persistent obscurity, was interspersed with experiences, expectations, or imaginations.18

In something that resembles a response to Didi-­Huberman’s note about transcending triviality, Geimer proposes to see invisibility as an interplay of material operations and cultural imaginaries instead of an ontological reality outside the human senses. Furthermore, as a discussion of images as entry points to knowledge objects for manipulation (an operationalization of invisibility), it brings to light the sort of extended angle to images that this book is engaged in through the notion of operational image. Visible or not, it is operations—­platforms, datasets, and models—­a ll the way to the bottom.

What Is Not an Image?   65

Platform Operations First, dear reader, a thank you for your patience: you have made it this far, and I am moving deeper into the concept of the invisual. After all, the term was promised to be central for the developing argument, and so far, I have given only brief glimpses into the term’s use. Consider the postponement of this discussion to here, the second chapter of this book, as a kind of a watershed moment, for the argument of the book oscillates back and forth between visual and invisual culture. There is additional potential in the term “invisual,” which might be more useful in the long run than keeping with the invisible. It is a peculiar neologism, as prepositional twisting often can be: the shift from the visual to its negation also includes the metaphorization of space that becomes helpful, too.19 It is both a negation (prefix) and a spatialization (preposition) of the visual, and it includes an expansion from optics of vision to logistics of data, but entirely “in” space—­literally. Not that this shift from invisible to invisual is one of linear historicization: “the invisual” becomes a heuristic term, and like “operational image,” helps to reevaluate historical sources in a different light. This resonates with some media archaeological methods that read recursively in and across historical periods. The term “invisual” developed and used by Adrian Mackenzie and Anna Munster describes “platform seeing,” which offers a useful, interesting, and smart way of dealing with the dilemma of images in computational platforms. At what level can we claim that we are addressing images even when they do not cater to traditional forms of human seeing or observing? This is a question that is implied in the legacy of operational images as well: the computational is one threshold for what defines visuality, but even more specifically, digital platforms as environments of aggregation and operationalization of images become a more specific place where this focus condenses in relation to the economy and ecology of power. It is also on platforms where images turn into diagrams that can be operationalized in the technical sense and legal and other ways. If operational images always already were images defined by operations of measure, scaling, quantification, comparison, and analysis,

66   What Is Not an Image?

in which ways does the platform become one key site that continues, as well as transforms, that legacy? Mackenzie and Munster use the term “invisual” in their discussion of assembly and operations of images in corporate platforms from Facebook to Nvidia. This view to platforms includes multiple components from software and services to hardware and image processing. It is not restricted to platforms in the business or legal sense (although there are clear implications at stake). Their approach underlines an important lesson about images in technical and logistical terms. Images always come as series, that in Sekula’s words, constitute a traffic in images since the emergence of photography: quantity and value in circulation.20 Mackenzie and Munster are focused on the formatting and preparation of images as part of the operational status of the platforms, not on what happens or does not happen to the photographic image. This pragmatic, even methodological stance asks: What is being done to the image, and how does this logistical operation define its usefulness and function? The image as an aggregated series is a central starting point to what the image is and how it is being captured and observed: “Their operativity cannot be seen by an observing ‘subject’ but rather is enacted via observation events distributed throughout and across devices, hardware, human agents and artificial networked architectures such as deep learning networks.”21 Observation and sensing become the central tropes for this operation that shifts the stakes further from visual culture while also addressing the mass-­image as defined by Cubitt. Observation is sensing in the broadest sense (a smart city of sensors and other data capture points, for example) and includes mathematical functions as (partial) observations of data.22 Besides the accumulating quantity of images captured and mobilized in platforms, the invisual mass-­image is an example of administrative data media that is found in relational databases with operations of locating, identifying, recognizing, and measuring among the attributes of interest: Relational databases like these are more interested in metadata—­ geolocation, personal identifiers, device identifiers, date stamps, facial recognition, distances and dimensions recorded by auto-

What Is Not an Image?   67

focus functions, upload addresses and the kinds of measurement prized by astronomical and meteorological databases. Commercially operated databases combine such data with data from searches and likes, swipes, shares, tags and other interactions, to construe from the relations between images a mass image of the world in constantly evolving organisational diagrams that only machines can read.23

When anything can be made into an image—­a series of patterns for analysis or an organized set of pixel relations presented on a digital screen—­the constitution of the image as “soft” has been argued to be a central aspect of this transformation that also includes operational images. The claim that we are dealing with algorithmic processing (to paraphrase Hoelzl and Marie)24—­and not geometric projection—­as the image’s primary quality is radically developed further to include the preparation of material that can be processed as if an image but for various ends other than human perception. Images can have other autonomous functions regulating, ordering, controlling, and relaying. Instances of data visualization since the eighteenth century could also be considered as part of this lineage of “proto data media”:25 the emergence of charts, graphs, tables, and diagrams as central features of modern statistics (consider William Playfair’s Commercial and Political Atlas and Statistical Breviary of 1786). Mackenzie and Munster write that invisual perception goes further than recapping machine vision. What is meant by observation and sensing is more than a substitution of human agency with machines or data: Such a mode suggests that while visual techniques and practices continue to proliferate—­from data visualization through to LIDAR technologies for capturing nonoptical images—­the visual itself as a paradigm for how to see and observe is being evacuated, and that space is now occupied by a different kind of perception. This is not simply “machine vision,” we argue, but a making operative of the visual by platforms themselves.26

This making operative of the visual is where the relation to the main theme of this book emerges: it is not the image but the platform

68   What Is Not an Image?

where (distributed) agency lies in this shifting regime of (in)visuality. What is recognized as an image is also an operation of data, commerce, law, and so on, defined by the complex ecology of platforms. There’s no individual observer or one individual image as the process of pattern recognition is automated through different techniques such as convolutional neural networks that are central for image analysis en ­m asse.27 The fantasy space of pattern recognition resides in a particular assumption about the epistemic world of data: there’s always more; to work beyond the intelligible is the multi­scalar promise that, instead of vast, invisible worlds of a micro-­or macroscopic optical regime, is about the endless permutations inherent in data, training datasets for machine learning, relational databases, and their operations across platforms that reformat their surroundings.28 Image ensembles feed forward to the creation of (statistical, mathematical) models for various bespoke tasks. The example used by Mackenzie and Munster of the DeepMind company’s AI model for the game of Go shows the operationalization of and through the invisual domain: a world that is approached as patterns, diagrams, series, and data. The AlphaGo program was widely discussed in popular discourse between 2015 and 2017 as it beat several leading Go players, demonstrating how effective the neural network had become in playing the infamously complex board game. Using the Monte Carlo tree search in mapping and identifying the effective decision patterns functioning at the back of the neural network also implies interesting points about the centrality of cognition as perception, as observation. In provocative terms, Vladan Joler and Matteo Pasquinelli describe “the history of AI as the automation of perception,” where the automation of perception is understood as “a visual montage of pixels along a computational assembly line.”29 Here too, the (so-­called) machine intelligence that goes into playing Go has a special version of pattern recognition as a new kind of cultural technique30 that sorts invisual observations (from images to remote sensing data) into operational actions. Go game squares are also a visual pixel grid. Such grids can be operationalized in a logistics of images that can be adjusted into multilayered neural networks and thus inserted into training sets: game boards like Go can thus become a 19 × 19 observational ma-

What Is Not an Image?   69

trix. Here, cognition is observation and is further processed as images integrated as heterogeneous forms of data into platform operations. This does not imply an image in the traditional sense but the logistics of platforms that create, organize, and mobilize data and datasets. Furthermore, from a 19 × 19 pixel space, the scaling up to millions of pixels across millions of images becomes the space where these problems are computationally challenging and wonderfully emblematic of an operative ontology of multi­scalar data. As Mackenzie and Munster write: We draw attention here especially to the primacy of image ensembles; the model trained via the DeepMind platform “learns” by observing many images. AlphaGo acts in the world to the extent that local spatial correlations can be associated with actions and rewards for those actions. The development of these systems centres on many cycles of observation followed by action. This cycling through observation and action constitutes the “training” of the model; a training that seemingly requires very little “prior knowledge” on the part of the model since it only receives pixels and game scores as input.31

Pixels in, models out. So-­called images in, so-­called statistical models out. The transformation enabled by such invisual platforms seems to produce one form of operationalization that describes not only a technical but also a political reformatting of the world that is one version of what Benjamin Bratton coins as platform sovereignty operating “within territories that are composed of intersecting lines, some physical and some virtual.”32 This new political geography of data rearranges planetary visibility according to a different logic than traditional cartographic projection. Go could be claimed to be just another abstract and formal example that the history of (symbolic) AI has mobilized so many times, but in fact, platform invisuality adjusts to multiple kinds of live situations too. Many examples of research projects with cloud robotics, machine learning, and other experiments with making the real world into an observed “image” testify to this: consider, for example, the work on Dex-­Net cloud robotics and point cloud data of 3­D object models. Moreover, the perceptual capacity of machine vision,

70   What Is Not an Image?

learning, and platform invisuality is not restricted to “ready-­made” pixel arrangements: platforms are involved in different dynamics, from urban life to environmental earth observation to agricultural innovation—­even “platform ruralism.”33 Such powers of operationalization define its particular grip on different material sites of reality: the smartphone, the image sensor inside the smartphone, urban and nonurban sensors from access cards to facial recognition systems to thermostats to driverless cars (see chapter 5), a long list of systems that are often covered under “surveillance” but include much more specific cultural techniques that are nested with other cultural techniques (such as pattern recognition). Platforms thus both distribute (image-­events of observation and capture across various dynamic situations of people and things) and integrate (synchronizing, synthesizing these events into series, into mass-­images in datasets). It’s a push and pull of images and data, of sensor capture and projective modeling that drives these operative ontologies of the visual/invisual.34 In other words, invisual images demonstrate one updated version of the pairing of the visible/invisible as a shift to the observation of data relations and patterns. The operational image is integrated into platforms as well as machine learning procedures that take images as training sets and models. The refashioning of surfaces that can be read as images can be seen to work as one form of abstraction, but it is in all forms and ways real. These abstractions and invisual events effectively establish interfaces to those material worlds. This platform-­based way of observing and modeling is also about integrating material worlds into data operations and producing handles that can change those worlds: “Unlike other geographic projections, the interface is not only a visual representation of an aspirational totality; it is an image of a totality that when acted on also instrumentally affects the world.”35 In addition, operations of statistical modeling are actively involved in real-­world situations, thus becoming one form of a technique of intervention. Abstractions are not unreal, and they are not separated from material surfaces. As you can see, this meshwork of terms maps how a transformed notion of the image relates to questions of abstraction regarding platform operations. But in our case, the main focus is still on the coupling of operations and images; what sort of an understanding of

What Is Not an Image?   71

images and operations is adequate in discussions of data and digital culture. As acknowledged throughout my discussion, the questions are not entirely new, and there are many answers already that relate to software, interfaces, and interactive screens. The argument about images as part of the broader programmable, executable culture was recognized earlier in software studies with different emphases by Alex Galloway, Wendy Hui Kyong Chun, and others. For Galloway, like in games, computational images are primarily geared toward execution and action.36 Chun opens up the question of software as execution to the genealogy of what is narrated as visible and invisible in computer history from aspects of gender and programming. Furthermore, following on from discourses of computers as information machines, this relates to how transparency is being sustained as the ideological backdrop for a machine that is primarily meant to make visible.37 This theme features in contemporary big data analytics and data visualization, too; how to draw patterns and make visible the otherwise unseen from datasets substantially bigger than the usual corpus of, for instance, humanities research.38 So while we could continue describing the various technical formulations of platforms and images, it is clear that the stakes of those techniques and humanities-­focused discussions about data and (operational) images are not the only context. Some of it concerns methodology and how we address images in contemporary technical humanities; some concern “images in the wild,” which implies their political economy as much as their political ecology. Here, the platform as a central feature of capitalism, political geography, and digital culture39 is not a mere extension of the operational image but a site where it is executed with additional force, adding to the early 2000s context of discussions about software and visibility. While we have inherited a rich set of ideas about technical images as mobilizing an ontology of the invisible and bringing it to play a key part in various institutional forms of knowledge (and coercion) over the past 150 years, the platform describes a more recent way images are formatted into platforms and how they format the world in patterns of value, knowledge, control, and more.40 This can be discussed in terms of a “political economy of digital data.” As defined by cinema scholars Ruggero Eugeni and Patricia Pisters, it pushes us to analyze “the logics of production, circulation,

72   What Is Not an Image?

and transformation of light (both from an optical-­geometric and a physiological perspective); also, it drives us to evaluate the intertwining of these logics with that of the production, circulation, and algorithmic transformation of digital data—­without forgetting the principles of sovereignty that govern these dynamics and determine their trends.”41 In short, and of particular interest for visual and invisual studies, the question of the image is what mediates this shift between the political economy of light and data. As a more specific unit for a “political economy of digital data,” platforms are a condensation of economic transactions and, as such, set the stage for what Sekula articulated as the link between photographs and money. Now, though, it’s images beyond photography, and data, that are linked in this way. Here, the platform is not merely a replacement of the market as a meeting place of providers and customers topped up with various access-­for-­data arrangements but an intensive apparatus for extending logistics of data to a variety of urban and nonurban situations. In short, it is not only facial recognition as surveillance, but agricultural solutions, landscape surveys, spectral signatures, and many other things that are aggregated and, in that process, turned into a particular kind of a mass-­image. Furthermore, the platform turns the world into its own image. Like a recent collection of speculative texts probing the transformation of cities into platforms asked: What is the city as Uber, the city as Instagram, the city as Palantir, as GroundTruth, as Amazon? 42 Here, the question of images as money is not merely about the circulation of photographs but the platforms that enable property and labor regimes that can be characterized by their operational invisuality. In other words, in the bundle of images, things, and people, platforms are also legal arrangements that assign positions of visibility while being operationally invisual: sellers, buyers, (gig) work, images, copyright, and other regimes of intellectual property through which the world is shaped in a particular anamorphic fashion, to follow Matteo Pasquinelli’s use of terms.43 But whatever we might refer to as (anamorphic) distortion, we can also refer to in terms of an expanded vocabulary concerning design in platform capitalism. Engaging platform invisuality as posthuman property is an insightful way to focus on the materiality of data as an operative force

What Is Not an Image?   73

in concrete spatial contexts.44 This term, mobilized by Jannice Käll, helps us to understand operational platforms in relation to broader spatial contexts of digital technologies. Autonomous vehicles and their way of (in)visualizing urban space is one example that I will follow up in chapter 5. But already, I should add that such an approach to spatial justice refers to how space is formatted by platforms, continuing the earlier points by, for example, Sarah Keenan on the law as production of space and place “from nation-­states to public parks to eruvs.”45 Such a merger of law and invisuality is useful in helping us understand how platforms reorganize the world. The operationalization of images is effective in contemporary contexts of data and where visual studies are not the only route to such an operational reality of observations.46 The operational image can be seen working in various contexts outside military targeting and vision; it is integrated into various other sites of execution where images do not anymore look like images. Or, even if they do, their primarily operational value might be as property, for example, one form of the continuation of war (a more frequently recurring reference point for operational images).47

Data Shots Invisual culture is introduced as a core thematic and conceptual mediator that grounds much of the discussion in this book. The invisual is not to be mistaken as synonymous with the invisible, but they do stem from a shared concern about the transformation of images in different knowledge and aesthetic practices. Mackenzie and Munster’s analysis of the platform becomes a helpful scaffolding for many of the other ways I want to engage with operational images beyond the earlier input of Farocki and others. Investigations of invisuality—­such a peculiar term that holds on to the legacy of visuality while denouncing it—­are helpful as insights into the continued question about images and operations. But the answers as to where to address this question are somewhat different: some concern platforms (and thus legal and economic forms of the operationalization of the image), and some concern diagrams (thus inserting themselves into a long history of visual practices that have

74   What Is Not an Image?

a particular role to play in mathematical, statistical, and relational forms of knowledge). Thus the concept of the invisual becomes a central way of understanding what operational images are in the contemporary context. In this way, they add an important qualification to the book’s argument: a shift from visual to invisual practices where images are not merely bundled up as datasets but operationalized across different institutional uses through platforms. Two implications come to the fore: First, as per Hito Steyerl’s already quoted words: “Not seeing anything intelligible is the new normal.”48 Put differently: images don’t look like images at all, not that they are (anymore) invisible either. It is not even seeing, perhaps. Images are, rather, invisual. And second, images in and as data (sets) focus on patterns and models operationalized across platforms of aggregation, ordering, and training. It’s a complex set of nuanced transformations where “images” are sometimes anachronistic terms used for data but are still, in some cases, also a process of operationalization of the history and archives of existing photographs and other images. Images are treated not as individual pictures but as entities on a different scale that does not fit in easily with a traditional vocabulary that assumes a human seeing subject who recognizes an image as an image.49 Images come in masses, as a mass-image that is increasingly the site of verification and validation of the world as it concerns various practices of power and knowledge.50 The image might be part of a logic of a target (war and administration, as in addressing), but it is also a statistical distribution governed by one (as is the case in digital photography and computation of images already in-­camera). Abstraction and photography have a long history already. The data-­intensive operations upon images continue some of that legacy,51 which could be said to include not only various practices lumped under surveillance, extraction of data, and commodification of that data, but also other techniques that take their aim at the face, the expressions, gestures, and actions: datasets such as Moments (MIT-­IBM Watson AI Lab) of three-­second videos of people, animals, but also objects and natural movement, Kinetics (consisting of YouTube clips), and AVA for “spatio-­temporal localization of atomic visual actions.”52 Many of these then fulfill several earlier dreams of archives

What Is Not an Image?   75

of cinematic gestures and bodies that were of interest to many twentieth-­century thinkers and cinematographers and also act as media datasets of techniques of the body. Picking up the thread from the previous chapter on cultural techniques, here is a good moment to remember the line from Marcel Mauss describing how bodies walk, climb, swim, wash, soap, eat, and more. Eschewing the binary of nature and culture, Mauss also observed what cinema had done to bodies (with a particular gendered bias in his observation): A kind of revelation came to me in hospital. I was ill in New York. I wondered where previously I had seen girls walking as my nurses walked. I had the time to think about it. At last I realised that it was at the cinema. Returning to France, I noticed how common this gait was, especially in Paris; the girls were French and they too were walking in this way. In fact, American walking fashions had begun to arrive over here, thanks to the cinema.53

While cinema, according to Mauss, was one form of archiving and then educating (forming and regulating) the body in a transnational fashion—­enacting a particular form of a cybernetic feedback loop before cybernetics54—­the contemporary datasets as categorized and categorizing snippets of bodies in movement constitute a second-­ order continuation, often extracted from existing media archives, whether from Hollywood or YouTube’s eclectic collections. Many archives-­cum-­datasets thus continue a particular form of institutionalization of the body that recircuits forms of micropolitics of the body. We can also call this the operationalization of the body registered and labeled in particular typologies, statistically reproduced in such models that return to format how the world of bodies is perceived.55 The operation is situated at one instance of perception-­action and across a chain of operational links in an infrastructural system. Any singular body, image, action, and reaction seems to give way to the aggregated mass image that fuels platform invisuality. This pertains to both the operations of datasets and the ensuing ways those are mobilized. As in, for example, the broad field of computer vision where “seeing is superseded by calculating probabilities” and

76   What Is Not an Image?

where “vision loses importance and is replaced by filtering, decrypting, and pattern recognition,”56 as Steyerl argues. Once again, the image is decisively seen as something beyond represented or pictorial while it is tightly tied together with a statistical distribution of patterns regarding objects, people, and events. What a paradox this seems at first. We see things that refer not to things but statistical distribution. We observe people, but we do so based on statistically conditioned fine-­tuning on the level of digital images, or in the case of AI, statistical models. We might still watch pictures, even moving images that feed back to a gestural repertoire available, but our gestures are already integrated into datasets that feed into training sets and model what gestures become meaningful. Hence the shift from a discussion of digital or networked images57 to datasets of images is relevant in terms of our topic and the site where operational images function. The operational image is located not necessarily in a particular shift of ontology of image that is often specified as “digital” but in how images are operated upon and become operationalized through aggregation, algorithmic analy­sis, and the ensuing questions of data-­driven mobilization of the mass-­image. This chapter revolves around platforms, datasets (such as ImageNet), neural nets (and other machine learning techniques), and models in order to understand operational infrastructure.58 And although they are not really images as such, they are at the center of questions of operational images. To continue focusing on datasets and the mass image, the ImageNet, also famous for the ImageNet Large Scale Visual Recognition Challenge competition, has become a go-­to place for discussing photography and visual culture changes. In addition, it has become a recurring example in discussions on the role of images in the pipeline of machine learning techniques and computer vision. But many other image datasets could be mentioned too, corporate and noncorporate, urban and rural, human and nonhuman. Such examples range from Microsoft Common Objects in Context (or COCO) to Open Images (CC BY 2.0 license), from Fashion-­MNIST (of Zalando online shop items) to FieldSAFE (for agricultural object detection), from LILA BC (dedicated to biology and conservation, such as datasets about glaciers and other land cover) to National Agriculture Imagery Program (NAIP) datasets. These examples boot-

What Is Not an Image?   77

strap different approaches to the use of images—­and what constitutes an image in the first place. Of course, ImageNet is, for many good reasons, a central dataset practice of the past ten to fifteen years where changes in (in)visual culture become visible. Its extensive work in the compiling and naming of images consists of a vast logistical operation where the image—­and the model processed from images—­is created. The tens of thousands of Amazon Mechanical Turkers doing this work are but one component in this logistical operation, alongside other elements that speak to the centrality of taxonomic categories as taking place where images used to be. At the back of a longer history of principles of machine learning and computer vision, which is outside the scope of this little book and my argument,59 it is fair to say that ImageNet’s status in the 2010s helped to understand some aspects of the changes in photography through computer vision.60 This especially concerns the evolution of datasets, as Fei-­Fei Li, the leading figure behind ImageNet’s work at Stanford University, argues. Furthermore, as Nicolas Malevé points out, this shift points to the move from the algorithm-­centric world of image processing and computer vision to the way “modeling functions in contemporary machine learning systems”61 through large datasets compiled from a variety of sources—­like in the case of ImageNet from the internet (Flickr). Access to useful and usable large datasets and the availability of efficient machine learning techniques such as convolutional neural networks become the technical operation that starts to define a particular invisuality of images. Also, the ways of looking at them no longer allow us in any meaningful way to refer to them as “visible” or “visual” only.62 The automation of seeing, vision, and invisual operations incorporates a different sense of a subject and how knowledge about images is handled. Analyzing what an image is and what is in an image is not simply about the decomposition of the constitutive elements, even if this is often how feature extraction is presented. To quote Malevé for a useful explanation about machine-­learned images and models: The developer assembles a data set reflecting the variations of the domain under study and utilises automated means to

78   What Is Not an Image?

calculate an optimal function that treats the features of the data as parameters. In computer vision, this technique, at its most simple level, uses large visual databases in which discrete units such as pixels can be considered as data points. Common techniques of machine learning in computer vision are said to be “supervised,” which means that the data is curated . . . to provide examples from which the machine learning algorithm extracts regularities: the software “learns by example.” To come back to the case of the cat, in the data-­oriented paradigm, the developer does not try to decompose the animal in distinct shapes and explicitly summarise their relations. Instead, she curates a large series of photographs, where the cat is displayed in various positions, and lets the algorithm detect the regularities traversing the various samples. Through this phase of “learning,” the algorithm produces the model of the cat.63

This is both an analytical and, significantly, a synthetic operation as it concerns the automation of modeling as an assembly of large (training) sets. Through these series of aggregated datasets for training, models are produced for use and reuse in different applications. This process is specific to how machine learning works but think of it in relation to Mauss’s techniques of the body and cinema as a circuit of capturing, registering, educating, and reinforcing. Moreover, not only a generic category is observed: a cat, a body, even an identifiable movement. The aim is to see the mass-­image as not about a taxonomic thing but as consisting of experimental processes of weighting perceptual parameters for particular features. This amounts to finding the right operational balance between any cat and a group of cats struggling in the wind at five in the evening.64 The latter is not so much a quantified example of a general group but a singularity in its own right. Regularities are extracted, which is very much the stuff of the promise of AI, but the complex meshwork of this perceptual apparatus is not evacuated due to this promise of standardization.65 Any mass-­image in a dataset or the ensuing trained model is always situated in a world of singularities, and observation goes beyond recognition of a thing in the world. Even the statistical model that reproduces and reformats the world is a

What Is Not an Image?   79

singularity in its onto-­epistemological being: one technique with the operative ontological force that shapes the world. Photographs—­a lready transformed from a picture to a compression format (such as JPEG) optimized for digital circulation66—­a re thus part of the infrastructure of datasets that become the more effective and operational end of images in this regime of (in)visuality. Yet, as has been argued by Malevé, Kate Crawford, and Paglen, but also by Daniel Chávez Heras and Tobias Blanke, the photographic persists as remediated as well as a black-­boxed element in machine learning datasets. The photograph is structured as a historical reference point of (computer) vision while the complexity of the photographic beyond a representational collection of things, people, and situations is easily ignored. Indeed, this mediation—­often voiced as naive realism of photographs as if they were just pictures copying reality—­is one place where the rich historical reality of diverse practices and materials becomes invisible and is seen primarily through the mechanisms through which an image is made relevant. (Cue Barthes and Panofsky from the beginning of the chapter, but updated into the question of the medium of machine learning and datasets.) The photograph is thus constantly redefined in relation to contemporary practices, which more or less bundled it and nested it inside their own mechanisms of analytic definitions, thus engaging in a technical and discursive redefinition of what an image is.67 Chávez Heras and Blanke argue that a lot of this work black boxes the photograph itself as a historical array of materials, lenses, and visual practices of taking pictures. In addition to research that argues that the labeling of training data images is a central site of politics (including politics of distributed global labor of Amazon Turkers), Chávez Heras and Blanke point out how we need further “archaeologies” of images too: to excavate what forms of mediations are already bundled up in (photographic) images. Unpacking machine learning procedures must also involve unpacking the assumptions about images. Hence, computer vision techniques as packaging particular kinds of image-­t heories—­a s Leonardo Impett has demonstrated68—­can be extended to alternative, even conflicting, genealogies of photography where debates about representation, nonrepresentation, measurement, calculation, and materiality are

80   What Is Not an Image?

implicitly part of the landscape of images as data. Finally, Chávez Heras and Blanke argue that forms of optical computation—­“glass” computers as they coin it—­are embedded in the camera lens, which is already a form of technical mediation and automated synthesis.69 I am not after accounts that explain machine learning through optical processes, but it helps to understand that different forms of computation are already bundled up in the operational scaffolding of visuality. Thus, it is useful to include such details if one starts to examine the assumptions about images in machine learning. Furthermore, Chávez Heras and Blanke’s methodology for doing this is immanent to the techniques concerned: their alternative engagement thus consists of assembling “a dataset with which to analyse photographic practice, and then use it to train a bespoke focal length classifier as a proof-­of-­concept for a system designed to investigate the optical perspectives implied in MbM [“Made by Machine: When AI met the Archive” dataset].” 70 Thus to rephrase the earlier points: machine learning is a technical purification and standardization of what an image is that also models the image in ways that exclude much of what it has been as a material practice. As a process of layered selections, decisions, and training, this is instrumental to how the pipeline of AI takes place. This concerns datasets specifically constituted about images, such as photographs, and/or employs what is assumed as a so-­called photographic image. The operationalization of photographic images includes both making invisible their rich nuances and historical development (let alone the variety of experimentation of what constituted the photographic in the first place) and the praxis of working with images without looking at images (or hiring cheap labor to do the looking instead). This is not to promote a kind of nostalgic take that we should merely critically look at images as individual solitary pairs of eyes, like a particular stance historically assumed, as museum objects, or as other forms of (an)aesthetics. Instead, I propose that this approach runs through a longer genealogy of images as administrative media where images are instruments involved in bureaucratic ordering procedures. Sekula’s work is one early example of recognizing this aspect in the “instrumental potential in photography,”71 where even without a full-­fledged review, one can pick up the cues

What Is Not an Image?   81

in his discursive takes through the focus on photography as it shifts parameters of social ordering, and presents a “new juridical photographic realism.” 72 This is where the linking of photography and “archival rationalization” was “designed to contribute directly or indirectly to the practical transformation or manipulation of their referent.” 73 This manipulative, formatting power of operations persists as our reference point throughout the attention paid to distinct practices of ordering that present visual reality but not necessarily for the aesthetic gaze. Paglen’s famous coinage that an “overwhelming majority of images are now made by machines for other machines, with humans rarely in the loop,” 74 is likely to be correct, but to be fair, most images were never meant to be looked at in the sense of aesthetic consumption. Instead, they were part of administering data: about people, nature, territories (like geological, geographical surveys) as well as in laboratories of chemistry and physics involving images with a very different eye to aesthetics.75 The same applies to contemporary practices of manipulating worlds through data and computational operations. Images take place, but they also disappear. As Malevé points out, from Flickr to a dataset such as ImageNet, a lot happens between, with all sorts of exclusions already taking place at that stage. Again, this could be construed as a version of Barthes’s argument about the disappearance of images in the process of fabrication of datasets. For Malevé, though, this is the paradox resulting from the photograph’s dual existence and disappearance in relation to operations of data and machine learning. “The paradox is that to resolve a photograph as data, all the heterogeneity that pertains to the medium, its apparatus, and its circulation needs to be repressed. The photograph needs to be made a transparent vehicle. And to make it transparent, computer scientists need to engage in the production of their own apparatuses and produce their own alignments.”76 But only one kind of image disappears, while another image surfaces. What was called “photography” might disappear, all sorts of data relations and scales might appear. This disappearance is not a romanticization of historical modes of hermeneutic understanding. Instead, it acts to point to the centrality of the image as it becomes a measured surface flatness, a function that was already present in much earlier uses of photogrammetry (see chapter 3). This is, in and of itself, incredibly

82   What Is Not an Image?

impactful when it comes to images taking a role in intervening in the world, which is to say that it is also part of a long nonlinear history of images taking action. Obviously, there are many direct political stakes at play. Many of them also include images as their central hinge of operations that hit the ground, hit the dataset, hit the bodies involved. Multiple forms of bias have become a central reference point for evaluating the social impact of AI in this sense of the mass-­image. Representational forms of analysis of visual culture are clearly insufficient in dealing with the pattern recognition and data-­operations at play that still affect how and what kind of bodies are at play, visible in which ways and at what scales, as a target or as labor, as a collective pattern of statistical data.77 Algorithmic techniques take a central role in those operations. Different infrastructures (classification mechanisms, datasets) become a site of activism. Joy Buolamwini shows us so in her work Gender Shades, which maps existing skin-­t ype distribution in datasets used for facial recognition training in relation to the Fitzpatrick Skin Phototype Classification (FSPC). But besides highlighting the underrepresentation of black individuals, she also created a new bespoke dataset, “the Pilot Parliaments Benchmark (PPB), to achieve better intersectional representation on the basis of gender and skin type.”78 Cross-­reading the Buolamwini project in relation to work on machine learning and politics of statistical distribution of patterns, we arrive at points articulated by Ruha Benjamin, Matteo Pasquinelli, and others: building on existing “world bias” where “datasets [might] reinforce race, gender and class inequalities” 79 normative power can further continue the institutional bias in statistical and algorithmic forms that can significantly amplify such distortions, to return to Pasquinelli’s point mentioned earlier.80 The question at hand is not so much a perspectival distortion observed by a gazing subject but a distortion at the level of invisual statistical distribution: it concerns collective distributions, data, and forms of operationalization in and out of images. Furthermore, for a consistent politics of statistical distribution of patterns, we should acknowledge that “art and media are fundamentally about the deception of sensory organs,”81 which would just put data technologies in that infamous lineage of evil media:82 persuasion instead of negotiation, trickery instead of representation, fooling instead of

What Is Not an Image?   83

communicating. Or, to put it differently: instead of saddling ourselves with “bias or no bias” as the two alternatives to choose from, the stakes are more clearly formulated in the question of what forms of justice are built into the various computational, institutional, and statistical techniques.83 Laboring people, too, disappear or are (re)morphed in operations of datasets under capitalism.84 The (racialized) arrangement of bodies and operations of extraction, logistics, and finance are central engines in the circulation of what this book discusses in and through and as operational images. Read thus, the dissection by Malevé and others regarding the Amazon Mechanical Turk work in those terms: what was once in the early nineteenth-­century phases of photography described as the new medium’s defining characteristic—­ objective accuracy—­is being transferred onto a laboring body. It is the body that needs to be accurate and coordinated and to see according to preset labels and taxonomic categories.85 To look at masses of images at high speed in order to annotate them, the mechanisms for reaching consensus among the laboring Turkers about annotations is a peculiar assemblage of eye, hand, and cognitive coordination in relation to the platform environment. The mass-­image is constructed in and through platforms that anonymize labor as a version of the micropayment gig economy. In Malevé’s words, which describe in fascinating ways the amount of cognition—­embodied and platformed—­going into the image process: To annotate at speed does not consist of a mechanical response issued from a passive subject. To understand the annotator’s contribution, it is fundamental to understand the process as one of elaboration which goes beyond rational choice and explicit judgement. The epistemic contribution consists in embodying a scale, figuring out rhythms and levels, understanding and refraining involvement. To attend to the process of elaboration means to avoid concentrating exclusively on the semantic decision. The elaboration is not limited to a pivotal moment, where the annotators assert the meaning of a photograph. It includes the complex methods through which they synchronise within an alignment and embody a scale.86

84   What Is Not an Image?

An active image. An activated body. An image that might be programmable. But so are bodies in their rhythms and scales. The earlier assemblage of chronophotographic data visualizations of bodies, work, and movement defined the connections from pioneers of images as recording instruments such as Marey to pioneers of management as intervention such as Taylor. This pair of recording-­ intervention and scientific measurement-­management/administration recurs in many instances ever since, like a media archaeological topos of sorts.87 Many of the themes in those feedback loops of labor and operative images return in the distributed set of microlabor. The labor of annotation exemplifies thus also the algorithmicizing of the body or, in other words: it is not only the images that might or might not be operational, but the bodies that are instructed and operated upon accordingly to clean, organize, and produce the (mass) image. This point relates closely to Joanna Zylinska’s term “undigital photography” that aims to probe the long tail of practices, including human labor, that defines technical images in the age of AI,88 but one can also point to various avant-­garde art historical precedents of programming the performer, and instructing the operator.89 Beyond avant-­garde arts, the work of technicians of the image and laborers paid to look at images is at least as important as the usual histories of innovators and inventors. The history of operational images is, in this sense, the history of anonymous images and anonymous labor, too, as they are essential in the preparation of what becomes visible. If one wants a historical counterpoint to contemporary preparation of images into datasets into models, one example that can be quoted is the meticulous work that went into the preparation of photographic astronomy images from glass plates into reproducible copies in the late nineteenth century: a technique of mechanical accuracy was underpinned by a long infrastructure of trained hands making halftone prints to ensure the invisible became visible. The little story about operations that format the astronomic observation of celestial objects of light into an image is one where the political economy of light and labor, of scientific work and objects of knowledge, intermesh. While it shows that the technical, even scientific, images had relied for a long time on various kinds of hand-­eye coordination in preparation of such images—­in the case of scientific accuracy in depicting nebulae and stars, this meant pay-

What Is Not an Image?   85

ing attention to brightness, contrast, and other values as they would stand out against the pitch blackness of empty space background—­ there are of course differences that furthermore emphasize the point. The relation to and respect of such manual labor was different, a point that is somewhat summarized in this anecdotal quote from 1911 by astronomer Charles Perrine, who “reminded his colleagues dealing with slow engravers of ‘the advisability of not forcing them to rush work on fine subjects.’”90 Similarly, the work of female computers at Harvard Observatory is one strand in the gendered history of operational images, as noted earlier.91

Intermezzo: Labor on a Single Platform The (gendered) labor that prepares and calculates images is often invisible and outside of the image itself. What would these scenes, described above, look like if they were a Farocki film? Consider describing the image being labored into a dataset as a version of Farocki and Antje Ehmann’s project Labour in a Single Shot. Started in 2011 and featured in workshops and as exhibitions, the project is organized as what sounds like a simple instruction of filming a work sequence in a one- to two-­minute shot in ways that capture aspects of the sequence and the gestures of the body. A cinematic reference point is Workers Leaving the Lumière Factory (Louis Lumière, 1895), but now these short scenes feature contemporary work: cobblers, cooks, waiters, window cleaners, nurses, tattoo-­artists, or garbage workers. But most of the work activities happen behind closed doors. Often labour is not only invisible but also unimaginable. Therefore it is vital to undertake research, to open one’s eyes and to set oneself into motion. Where can we see which kinds of labour? What is hidden? What happens in the centre of a city, what occurs at the periphery? What is characteristic and what is unusual with regard to each city? What kinds of labour processes set interesting cinematographic challenges?92

The above is a workshop description of the brief for Labor in a Single Shot. We can follow up with what might sound a peculiar question:

86   What Is Not an Image?

What would be the single shot of a process that is distributed through the Amazon platform, coordinated across multiple workers doing the same task, and resulting in images being interpreted so that they disappear in the process? 93 Can this shot be taken? Does it have a specific relation to cities or some other geography, territory, or architecture? Perhaps it is a collection of single shots, of distributed works, coordinated into a collection of multiple images about multiple images being worked on and annotated. The image being operated upon—­through labor and as a technical process—­ disappears twice, but the shot itself would frame this disappearance through proxies that inscribe and trace the reframing of the images (mass-­image) as they are readied to become models.94 This could be called a model image or an image-­model.95 In either case, the stakes are clear: the images are prepared to feed into a model that feeds back to what is captured in perception. Such models are not necessarily visual nor tangible in the architectural sense of “modeling” but are statistical and mathematical functions in machine learning. This also means that the image becomes a statistical distribution while being delivered and maintained by labor and the platforms that organize that labor. While platforms synchronize logistics as bodies, software, routines, data, and production of images (such as in the example of the ImageNet and other uses of the Amazon Mechanical Turk), this “labor on a single platform” (to misquote Farocki and Ehmann’s project) is one aspect of what is being referred to as the invisual.

Model Images Images are compiled from datasets to platforms in various institutional, para-­institutional, and cross-­institutional uses. As Estelle Blaschke has shown, photographs are carriers of images. As images, they are also carriers of data (which admittedly can also be metadata such as image texts, annotations, etc.): standardized, scalable, and “driven by the desire to make images ‘more’ exploitable for a variety of uses—­from business and industry to scientific research and knowledge transmission.”96 Images include various levels of taxonomic and other data that help to format their usability for a multitude of purposes, and where the image itself becomes an entry point

What Is Not an Image?   87

to data aggregation that leads into the synthetic modeling of the world. The work put into perceptual AI and robotics indicates the challenges where sensor-­based realities of action and movement are entangled with questions of operational images: the gap between sensing and images becomes minimal. In addressing the administration of operational images, I want to underline that significant techniques, operative ontologies, and practices of invisuality help us understand some of the stakes in data platforms and computational contexts such as machine learning. “The model,” “the platform,” and “the dataset” are some terms that speak to the operational image without a primary reference point in optics or the pictorial. As you might have seen by now, this is not to ignore historical discussions but to argue that those are also to be re­evaluated in light of data and instrumental imaging questions that stretch the notion of “image” toward some nonimages like charts, tables, and diagrams.97 To address, even briefly, the “model” in this context brings together many aspects at play, from images as instruments of intervention to the mobilization of data in and beyond the sensorial. Images have acted as instruments in the history of various laboratories in experimental psychology, plant physiology, physics, and biology, as well as in the pipeline of production of models in contemporary machine learning. In scientific practices, photography of the invisible shifted the ground of verification of what is (and is in) the image.98 The status of evidence could be questioned when there was no way of referring to a “ground truth” outside the apparatus. Synthetic ground truths emerge much earlier than in recent AI-­training data practices, and reality had to be approached through models of reality. While photography of nonphenomenological physical or astronomical worlds might have triggered this decentering of ground (truth), it became more widespread in the context of twentieth-­century knowledge practices: the quantum measurement apparatus as elucidated by Barad was already quoted (in the previous chapter) as an example of this. Arguably, one can say the same of earlier, much simpler optical visual technologies such as Galileo’s telescope in the early seventeenth century. In a way, that apparatus concerned an early example of a calculated data practice of visuality: “The telescope creates the

88   What Is Not an Image?

senses anew: it defines the meaning of vision and sensory perception, turning any and all visible facts into constructed and calculated data.”99 According to Joseph Vogl, the telescope established its aesthetic–­epistemic argument not only by empirically seeing more but in how this apparatus operated upon its own conditions of what counts as seeing and knowing. The assemblage that brings together sensing, images, heavens, skies, planets, observation, and data is also a creation of a particular world of mediations and, in Vogl’s words, also of limitations. It produces a bespoke model of what perception is, its own model image of what counts and can be coined as anesthetics: The critical point of the historical analysis of media is not to be found in what a medium makes visible, tangible, audible, readable or perceptible; it is not so much located in the aesthetic of the data and in-­formation provided by a medium but rather in the anesthetic side of a media process. Again, what does Galileo Galilei see when he turns his telescope toward the sky? What exactly are the visibilities that Galileo observes, then captures in his texts and drawings—­t he lunar surface, unknown fixed stars, the Milky Way, the moons of Jupiter? Sidereus nuncius leaves no doubt: Galileo sees, newly perceptible in his telescope, not just sun, moon, and stars but the difference between the visible and the invisible.

In other words, this operation upon the invisible also produces the main limitation of the epistemic sweep brought about by the telescope: to realize that because of this assemblage, “every visibility now bears a stigma of provisionality; every visibility is surrounded by an ocean of invisibility. Everything visible remains contingent, forever encompassed by the imperceptible and the unknown.”100 Later, (photographic) images take a place as models of invisibility. As Geimer points out, late nineteenth-­century and early twentieth-­century scientific photographs function as models in the fundamental sense that gathers more epistemic force gradually over the twentieth century in relation to physical, mathematical, architectural, and computational models.101 Here again, the question of

What Is Not an Image?   89

what counts as proof and evidence—­and what this implies about the invisible, inconceivable, even invisual—­stands out: The pictures may figure as evidentiary material or indispensable illustration in a demonstration, but their existence as such is not already proof “that something invisible may really exist.” Their function more closely resembles that of a model. “An experimental model system has always something of the character of a supplement in the sense Derrida confers on the notion. It stands for something only the absence of which allows it to become effective.” . . . The model never fully coincides with what it represents, and this systematic misprision is the source of its productivity. Applied to the case of photography, this means that the visualization of an “invisible fact” did not straightforwardly undo its “invisibility.” What the photographic image presented to the eye was not the spoils of a foray into the darkness of the invisible world, hauled back unscathed into the world of visibility.102

Instead of merely capturing the visible and the invisible, the focus on models helps to consider images as experimental systems. This concerns photographs and other kinds of images closer to data visualization too: “Camera obscura tracings, photographs, and the inscriptions of self-­registering instruments”103 can be named as part of what Daston and Galison coined as the regime of mechanical objectivity of (roughly) the nineteenth century. More recent experimental systems are likely to include machine learning and cloud analytic practices.104 But to address telescopes and photographic (and related) techniques also includes the awareness of them as instruments that produce epistemic reality, which is why they became integrated into the scientific practices and institutions that have a vested interest in images as invisual operators. Invisuality starts early. Think of Marey’s chronophotographs as one instance where such expanded visual instruments are essential to capturing data105 (see chapter 3), but consider also the above-­voiced suggestion to see photographs as models in their own right. Will this pairing up of experimental and instrumental uses of images also help to figure out the scope of

90   What Is Not an Image?

the operational image in contemporary data culture as a particular model of what an image is? A jump from telescopes to images and contemporary datasets, and machine learning training and other techniques, is not a linear one though. But the point about the provisional epistemic conditions that work through seeing/images is relevant. To follow this sketch about models also implies that every dataset is provisional, every pattern recognition implies a different possible pattern, every training set implies an alternative statistical model that could be created. In this operative ontology of the image, the roles are reversed. If anything that exhibits a pattern to be read and reconstructed is an image, it is the technique that produces the world as an image. It is not that images are being analyzed, but that the analysis produces potentially anything as an image—­a never-­ending cascade of patterns, images, perceptions. In this sense, part of the book’s argument is that operational images are scattered across models, diagrams, and graphs functioning in various assemblages of knowledge. Indeed, in the contemporary context, we are likely to find such image operations outside the laboratory and the military in all sorts of assemblages of machine learning and image (training) datasets. They are also found in the multitude of sensors that capture faces, body postures, gestures, and more: environmental sensing and subsequent imaging, corporate practices for urban and nonurban uses from policing to agriculture, gaming and other forms of motion capture and interaction design, and a set of other things that propose a radical detachment from what images in the photographic context were. The aim of producing images as standardized, scalable, and ready-­made for operational (administrative) uses underpins images as models also beyond scientific institutions. Images as models (and data that help build models) emerge gradually in the history of technical media, even if this aspect becomes much more obvious later in the computational and digital knowledge practices. In other words, a model in the historical sense outlined by Geimer or Vogl does not suggest the same thing as a model that emerges as a core feature of deep learning in contemporary computational practices where “the model is the statistical representation of a large and diverse training dataset into one file.”106 Nor is it that what’s a model in the

What Is Not an Image?   91

photographic sense is the model that emerges in the quest to map the hitherto invisible scales of climate change as a mass aggregation of data.107 But the question of the model becomes a site of investigation of the change of images, as it becomes a site of investigation of data and knowledge, which is why it warrants placement alongside the discussions of platforms and datasets. Model, as such, also becomes an epistemic device to investigate the operations of images and operations of formatting the world in their own invisual image. Such models are not visualizations of the world but operational interfaces through which changes are enacted, whether in the contexts of urban planning, military operations, or earth observation of land use and land cover, vegetation, agricultural crops, erosion, and so on. In many ways, this focus on the model turns the tables: the contemporary practices help to see images as models, diagrams, platforms, machine vision. Such experimental systems have more than a rhetorical relation to invisible worlds. In other words, the invisible is not merely what is not seen, but it is also what is produced as part of the practices of establishing what is knowable. This also features in contemporary practices of data and machine learning from cultural techniques of pattern recognition108 to techniques of probabilistic modeling and the centrality of datasets as they are mobilized in the operationalization of the image. Knowing and seeing become probabilistic, and the models that are produced are rehearsed as rhetorical “seeing” from the identification or misidentification of terrorist subjects to border control to other situations with specific geographical and spatial characteristics. This kind of operational image is not merely an image that targets; it is a statistical model that builds an invisual environment as a model of the world through which actions can take place.109 The world is formatted for such epistemic conditions of actions; platforms are one example of this capacity to format. Gradually, our understanding of image, visuality, and perception changes as those terms become part of machine learning experimental systems. For example, statistical models can become a multidimensional vector space110 representing a very different link to “visual” than one of “images.” An investigation of graphical notation systems in mathematics and visual operations of diagrams as part of

92   What Is Not an Image?

a media archaeology of technical images would be one way to build a longer historical argument about depictions of contemporary machine learning and data culture as an interface of statistics, images, and material territories.111 We are far from Galileo’s telescope in historical time and technological logic. Still, one point of similarity remains: a world is created anew as it is being observed, and not one of heavens and planets only, but across a different kind of a vast field of data operations. Here, models have a mediating function while becoming, recursively, one example of operational images as something else than “just” images.

Still Not Just Images This chapter addresses different variations on the themes of invisible and invisual images and, in the process, ends up in machine learning and datasets. Part of this has meant mapping the discussions on operational images in relation to contemporary AI and data culture work. In so doing, the chapter has focused on images as data operations in a world that is urban, dynamic, modeled, simulated, and layered in metaphorical and technical ways (such as neural nets producing models based on training data). Images do not feature as images but as platforms, datasets, and models. Building on the earlier chapter, I proposed that we look at the productive processes and valorizations, the operative ontologies, of what is at stake in this invisual visual culture.112 Some of the stakes of those operations concern value creation. Some are in existence in the institutional arrangement of knowledge creation. Some are emblematic of the forms of power (including law) that distribute, as Foucault would have it, what is visible and sayable in the first place.113 Much more could be said—­a nd has been said by other scholars—­on all of these issues. For example, in chapter 4, I will continue these arguments in relation to work on the scientific data platforms of the Chinese Digital Belt and Road initiative and how the artist duo Geocinema has engaged with that aspect of remote sensing and imaging. In chapter 5 the points are continued in relation to the urban traffic of data as one example of corporate data practices of operational images of mobility and lidar scanning.

What Is Not an Image?   93

As this chapter comes to a close, it is tempting to ask: So what is an image if it is lost somewhere inside the machinations of photography and datasets and machine learning? Some decades ago, scholars could still respond by saying that images are like language. For example, W. J. T. Mitchell suggested this in the mid-­1980s: “Instead of providing a transparent window on the world, images are now regarded as the sort of sign that presents a deceptive appearance of naturalness and transparency concealing an opaque, distorting, arbitrary mechanism of representation, a process of ideological mystification.”114 This move was meant to help discuss the constructed and even political nature of images, moving away from the naive realism that images depict things. In some ways, critical insights to software and execution articulated an updated route: programming languages are a special case of visual knowledge, as Chun pointed out, where politics act not only through natural languages. What would be an even more updated version? What is an image if it is not just an image but also data? What is an image if it is not just an image but a model? What is an image if it is not just an image but also a platform operation of distributed machine vision across multiple devices and observations that can aggregate those observations under a platform logic (and business model) and execute those as real changes in the world (e.g., in legal terms such as property and control)? Such contemporary answers are no less political than Mitchell’s suggestion about the language of images.115 But, likely, many of the answers are now implied in the work of scientists and engineers dealing with images. “Images began to function at least as much as a tweezer, hammer, or anvil of nature: a tool to make and change things.”116 This point by Daston and Galison about the history of scientific practice is apt in the broader sense, too: much before and after poststructuralism, images were something else than just images or language. Many of these responses, though, are at first sight pragmatic: Which datasets of images can be mobilized in particular convolutional neural network environments, and how to find computational power for dealing with training sets of high-­resolution images? For example, the deep learning neural network from machine perception to image (dataset) to statistical model-­pipeline is practically

94   What Is Not an Image?

processed as GPU processing power that incorporates invisual operations inside the computational architecture.117 Ranjodh Singh Dhaliwal refers to this moment of (in)visuality in terms of rendering as computation, underlying the centrality of parallel processing (GPUs) and platforms such as NVIDIA as part of a refashioning of contemporary digital capitalism. In short: it is not only that the invisible worlds are being rendered visible for any perceiving subject but that rendering is considered a central part of the computational apparatus—­a political ecology as much as a political economy—­in which operational images also take a central role. Jacob Gaboury’s work on the archaeology of computer graphics makes this point in detail: such digital images are far from images in the representational sense and are instead complex computational events of rendering that draw upon invisual data to produce visible surfaces.118 Observation and seeing are peppered across a heterogeneous set of events in the world which are integrated onto a platform, which feeds on the anamorphic perspective to images themselves: twisted and transformed, designed and designing, they model the world according to their own distortion (or programmed rendering).

THREE

The Measurement-­Image From Photogrammetry to Planetary Surface 1736, 1858, 1989 The previous chapter dealt with the fundamental question of what is visible against the backdrop of the invisible and the invisual. The brief genealogy moved from scientific photography to questions of data and AI. Along the way, addressing data became more central than the image in the representational sense, although it is not always clear that the two are separate. Data can be extracted from the depictive and the picturesque, which is historically part of the colonial legacy of the modes of representation as they proliferated across vast territories from the Caribbean to India and Southeast Asia, for example, alongside other data operations such as trigonometric surveys.1 In many instances and across gradual development, computation and calculation become operations of the digital image that not only were rendered but also were rendering the rest of the world as data and images.2 As usual, much more could be written and said about that particular topic, reflecting the centrality of AI for operationalizing both images—­as datasets, for a range of applications—­a nd different surfaces, planes, plans, territories, and volumes of the world as images. In the previous chapter, this was discussed as the question of formatting but resonates with the account about technical vision articulated by Donna Haraway already in the 1990s: both organic eyes and technological sensors build in “translations and specific ways of seeing, that is, ways of life,” that are developed into an account about vision as organization: “There is no unmediated photograph or passive camera obscura in scientific 95

96  The Measurement-­ Image

accounts of bodies and machines; there are only highly specific visual possibilities, each with a wonderfully detailed, active, partial way of organising the world.”3 The point should be read literally: vision systems program possibilities of action. Technical images work as instrumental interfaces to that which cannot be grasped by fingers, hands, unmediated human perception, and as Vilém Flusser argued, focus on how “elements such as photons or electrons, on one hand, and bits of information, on the other hand, [are made] into images.”4 Images as ways of partitioning the world are based on apparatuses that “grasp the ungraspable, visualise the invisible, and conceptualise the inconceivable.”5 In this fundamental sense, images are interfaces where the epistemic and the operational coalesce. In other words, it is not only the images that we may manipulate but the data and/or territories which those images stem from and return to. Hence, operational (see Figure 5). They are formatted and, in the process, become instruments of reformatting. Image instruments operate beyond anthropocentric life. They can also be about the amplification of, for example, processes of growth in a precision farming system or the integration of blockchain into agricultural practices.6 While the topic concerning agriculture, farming, and growth is picked up in a book that I am currently writing with Abelardo Gil-­Fournier—­on operational images, vegetal surfaces, and territorial management and data—­the tie-­in between images that are programmable and executable becomes a core concern already in the present volume. Images are part of the operational infrastructures of executing and programming different material surfaces and events, alongside the list of potential actions that the “operational” in the operational image can incorporate: targeting, analyzing, comparing, tracking, navigating, extracting, trapping, projecting, forecasting, measuring, and quantifying. I am sure others could be added to this list that features slightly different variations in different chapters. So when Pantenburg writes about Farocki’s operational images that “in this capacity to ‘act,’ . . . could be called the cynical inheritors of the aims and hopes that political or educational filmmaking has always fostered,” 7 we can continue his line of thought. Technical and automated images are politics by other means: in large-­scale technical systems, in institutional work

FIGURE 5. The Mars Perseverance robot imaging is one example of remote sensing and operational territories even beyond the earth. The Mastcam-­Z being installed is “a multispectral, stereoscopic imaging instrument that will enhance the Mars 2020 rover’s driving and core-­sampling capabilities,” which also takes 3D images at very high resolution. Images and quotation from NASA/JPL-­Caltech. https://mars.nasa.gov/.

98  The Measurement-­ Image

of execution of commands, and in ways that are often read as part of the logistical mapping of the earth but one determined by the partial view that Haraway mentions. Thus, as per Farocki’s interest and ours, we ask: What is the political and educational image-­ making format that pertains to this ecology of automation? Could operational images in this key also educate the political potential of such systems, including measure and quantity? This chapter focuses on measuring by starting with some of Farocki’s work, which channels genealogies of measurement shifting between the photographic and the cartographic, between measuring earth surfaces and measuring that is established by images. Farocki’s assemblage of moving images, the soft montage,8 becomes a starting point for a media archaeological argument of sorts. An early warning for those desiring a straightlaced story of technical images: while we begin from the story offered by Farocki regarding vision and photogrammetry (in photographic terms), this chapter will venture in a slightly aberrant direction, including geodesic measurement of the planet, one that I argue is a constitutive part in the longer history of planetary imaging and design. Instead of offering comprehensive historical narratives, the following sections are meant to link to contemporary work on measure, data, and quantity in technical images and photography. The question of measurement will be discussed in terms of the multi­ scalar planes (image planes or planes made into images) that are at play in our historical examples that I hope will also frame the question of the multiple historical layers of the operational image circa 1736, 1858, and 1989. Those dates might seem random, but they have a structuring heuristic function: Farocki (1989), photogrammetry (1858), and geodesic measurement (1736). At the back of the cultural technique of measurement is the paradoxical push and pull between the seemingly straightforward quantification brought about by the precision of measurement and the de facto conflict that measurement involved, such as in the early case of the measurement of the earth: “Precision of measurements was a matter to be interpreted, attacked, defended, and represented.”9 Measurement implies standards, scales, instruments, and methods; it implies precision and mediation of that precision. This is also where images come into play: to mediate precision through images (such as in photogrammetry) and make the earth itself into an

The Measurement-­Image  

99

image by establishing it as measured lines (trigonometry in geodesy and later more advanced cartographic techniques). To claim maps as “platforms of calculation interface”10 becomes the link to the operational aspect of predigital instrumental images too.11 However, I want to start with Farocki himself and the rather famous scene in his film Images of the World and the Inscription of War (1989), which introduces central ideas of operational images.

The Measurement-­Image Sandwiched between shots of scientific wave modeling and historical photographs of unveiled Algerian women in 1960, Farocki inserted the story of Albrecht Meydenbauer, often cited as a pioneering figure of (architectural) photogrammetry (see Figure 6). Farocki narrates the near-­fatal accident Meydenbauer had in 1858 when measuring the facade of Wetzlar Cathedral in Germany. Even Meydenbauer himself annotated the picture of the cathedral with the spot he almost fell from. This story is often mentioned as a historical anecdote and in commentaries related to Farocki’s operational images, but it is worthwhile to voice it again. According to the legend, and what Farocki tells us, this episode led to the desire to formalize a way of reading measurements from photographic documents of buildings in order to reshuffle questions of proximity and distance, objectivity, and personal danger—­and to avoid the necessity of climbing up dangerous facades or cathedral rooftops when you can just take pictures. But for Farocki, it was also the broader context of the double aspect of such a peculiar image that was of interest: such measuring images can record what is destroyed, while images that measure are also used for military targeting. In Farocki’s voice-­over: The first major scale measurement based on photography was achieved in 1868 at the fortress of Saarlouis. Later, Meydenbauer initiated the establishment of memorial archives, which creates a correlation, in the sense that the military destroy and the curators of monuments act to preserve. He wrote: “Perhaps some would find it incredible, but it is a fact proved by experience: in a scale picture one does not see everything, but one sees many things better than on the spot.” This capacity to see better is the reverse side of mortal danger.12

FIGURE 6. Stills from Farocki’s film Images of the World and the Inscription of War (1989). Reprinted with permission from Antje Ehmann. Copyright Harun Farocki GbR.

102  The Measurement-­ Image

Meydenbauer’s subsequent work on the use of photogrammetry consisted of a technical element—­a specialist camera for the purpose of making such measurement-­images (Meßbild)—­and an institutional basis in cultural heritage. The latter concerned a Cultural Heritage Archive (Denkmälerarchiv) that was to be designed to keep data of buildings destroyed or otherwise lost to time. This archive plan was steeped in a nationalist celebration of German culture (and its classicist European narrative) that was first recorded and saved as buildings and then recorded and saved as data about the design. Hence this was meant as a collection of “reliable images” and “geometrical drawings”13 where photographic technology proved a reliable instrument. As Meydenbauer narrates, what was imagined in perspectival terms as a technique could be reverse-­engineered with the help of the measurement-­image that photogrammetry provided: technical images could provide a more accurate infrastructure of data records than the history of art historical drawings with their possible mistakes in the representation of measures and ratios. However, as a technical image, this was no ordinary photograph for that very same reason: it was built to measure and reproduce measures accurately. To be sure, the attention to detail of the architectural plan and the facade had been a central part of the photographic discourse at least since Arago, emphasizing indeed, precision: “All these pictures [of Parisian monuments] could be examined with a magnifying glass, without losing any precision.”14 The portability of precision and detail15 became part of the features of this data practice of storage and analysis that also had colonial uses (as I will argue later), putting a twist to the idea of the cultural heritage archives underpinned by colonial expeditions, such as the French in Egypt. As such, the photogrammetric images are best seen in the lineage of scientific instruments, but they incorporate many other layers too that tie them into a discussion of operational images before and beyond digital computation and automation.16 Strictly speaking, automation does feature in Farocki’s frame for Meydenbauer too: the opening minutes of Images of the World also include the plotter printer that draws an architectural facade, establishing a link between the 130 years of historical distance from Meydenbauer

The Measurement-­Image  

103

to contemporary image practices: facades not only measured from an image but also reproduced in automated procedures of drawing a (vector) line, which now, in 2022, seems low-­tech compared to current rendering practices used in architecture and planning (see Figure 7). Such include a broader infrastructure of computational power and the ability to mobilize a much wider digital array of data, “image libraries of objects, scenery and materials,”17 facilitating this virtual life of architectural operational images. But as far as our focus on measurement, lines, and plot(s) in images and territories goes, the link is apt, and it becomes a channel that reroutes us in the right direction. Before Farocki starts to use the term “operational image” in the military context of targeting, pathfinding, and navigation, this reverse side of measurement and reconstruction features as part of the cinematic narrative. It builds a suggestive historical case closely related to then-­circulating discourses about the Enlightenment, rationalization and visual culture, simulation, and modeling. This sequence of images and narrative includes thus the cultural heritage data collection by Meydenbauer as it does scenes of the unveiling of Algerian women for identity card pictures, as well as references to the Enlightenment—­in German as Aufklärung, which includes the then meaning of the term as a period of rationalization as well as a military term referring to (aerial) reconnaissance (Luftbildauf­ klärung).18 What’s implied is that the Enlightenment starts from techniques of vision in the spirit proposed by operational ontologies: abstractions are bootstrapped into the world by specific operators at play that establish key material and epistemic divisions; the aerial image, the survey, the measurement-­image could be quoted as such operative procedures. Farocki’s work condenses multiple cases of archives, data, photography, and measurement in ways that become instructive of a whole genealogy of operational images that operate by way of manufacturing planes or, in Sybille Krämer’s words, cultural techniques of flattening: the ability to create such artificial objects as flat surface planes of different materials that enable diagrams, maps, measures, and other examples of operational graphics. Such planes bring together images and numbers under the auspices of measurement.19

FIGURE 7. Stills from Farocki’s film Images of the World and the Inscription of War (1989). Reprinted with permission from Antje Ehmann. Copyright Harun Farocki GbR.

106  The Measurement-­ Image

Meydenbauer’s late nineteenth-­century architectural and topographical measuring surveys are one part of the genealogy of remote sensing as they establish ways of remote capturing data that can be processed and synthesized later. The image is also a data-­ transporting device. The measuring-­image works with the ability to capture scales, which guides the development of the bespoke instrument built in 1867 by Emil Busch in Rathenow.20 In Joerg Albertz’s summary, the photogrammetry apparatus included: - the definition of the image plane by means of a mechanical frame, against which the photographic plate is pressed before exposure; - the integration of an image coordinate system realized as crosshairs that are imaged on the photo plate during exposure; - a compact camera design with a fixed focus to define the principal distance (or calibrated focal length); - mounting on a tripod with the possibility to adjust the camera axis horizontally, the image plane vertically, and one of the image coordinate [axes] again horizontally.21

Analyzing this apparatus in the context of the operational image demonstrates a shift from focusing on the image to the operations that produce this particular kind of measurement-­image. As Meydenbauer emphasizes: these are not just any photographs, but this plane is integrated with a bespoke measure of its own that can be read through an analyst’s eyes. While the technique soon developed into its own service—­the Königlich Preussische Massbild-­ Anstalt at Schinkel-­Platz 6 in Berlin—­for production (40 x 40 cm), copying, and enlargement, the implications of the technique went beyond this particular price catalog.22 Measurement, as a core cultural technique at play that also includes auxiliary techniques like flattening, quantification, and standardization, is executed both through such techniques that integrate image planes with mechanical frames and, in doing so, establish a bespoke coordinate system for “navigation” of this plane. This way, the facade, the building in its right proportions, or the territory/ landscape becomes integrated into an image, as an image. The pic-

The Measurement-­Image  

107

torial qualities become secondary to the mathematical proportions, the descriptive geometric dimensions. In Lacan’s words: “What is an issue in geometric perspective is simply the mapping of space, not sight.”23 Not that the two are always opposite: the pictorial can be measured and be operative, as many of the techniques featured in this chapter, and this book generally, show. Data and the pictorial can similarly meet in the very same images in question, but the epistemic, aesthetic questions and the operations executed become the separating thread. My interest is centered on this: What ways of life can the image establish? What possibilities does it prescribe, enable, and disable? Which routes are implemented into an image that becomes a special case of cartography—­a navigational image indeed.24 Also, the other aspect of cultural techniques is here visible too: What operations does the subsequent image make possible, which institutional context is it executed in, and does “automation” mean digital automation or the sort of algorithmic power that institutions hold through their tasks, priorities, orders, and goals. This would then be the other way of reading through Farocki’s sequences as well: scientific facilities, painting, and architectural studios, but also, the Auschwitz concentration camp and the later (re)analyzed aerial reconnaissance pictures revealing the details of the chimneys of a crematorium. The images operate in a multitude of different ways: as technical images or measuring images but also later reread as different forms of evidence that Eyal Weizman has included in his work developing a methodology of counterforensics.25 What Weizman brings out in forensic architecture’s work is the double aspect of a photographic image: “At the threshold of detectability, both the surface of the negative and that of the thing it represents must be studied as both material objects and as media representations.”26 This kind of methodology could be seen as a special case of architectural imaging employing operational images and a nod to the question above about the slightly less cynical political and educational uses of operational images. As material objects that become integral to cultural techniques of measurement and information, photographs are an example of what in this book counts as technical images.27 In other words, I am keen to backtrack from the usual preference of focusing on digital

108  The Measurement-­ Image

images (in discussions of operational images) to some of the constitutive ways geometry, data, measurement, and navigation are implemented onto picture planes in so-­called analog images too, including in drawing but also in practices such as topographic surveys.28 Photographs and drawings are not the same—­far from it—­but there are some constitutive commonalities in relation to questions of measurement that will be brought to bear on the question of the operational image. Although I am not going to rehearse in detail the well-­known history of linear perspectival painting as perhaps the best-­known example of mathematics of Renaissance images that still operates at the back of many of our assumptions of the construction of images that construct patterns of vision. But in reference to this story from Brunelleschi to Alberti to Galileo’s “perspective tube” and its later variations in astronomic imaging, Samuel Edgerton summarizes the stakes well as far as this genealogy is one of invisuality as much as visuality and as far as it concerns technology as much as “art”: “Indeed, linear perspective may have been more important to the history of modern technology than to art and science. With this remarkable tool [of the grid] it was now possible to plan complex machines by means of exacting pictures to scale, thus allowing each connecting part to be previewed and tested from every aspect on two-­dimensional paper before being manufactured into irrevocable three-­d imensional fabric.”29 Although Edgerton adds that “even modern computer modeling is based on the rules of linear perspective,” I instead emphasize it the other way round: such mathematics of the image constitutes the earlier media archaeology of modeling, simulation, testing, and measurement that find their multiple instances in some of the examples in this chapter too.30 Mario Carpo also makes a similar point in his archaeology of architectural practices: from Leonardo and Alberti onwards, qualification of precision through lines (and color for Leonardo) was also a way of carrying “measure” through painting techniques and material existence. In such ways, the invention of perspectival techniques was also a notation system and an effective data compression system that operated upon the flat surface.31 Hence, significantly for our purposes, as image surfaces, photographs contain geometric ratios that can be reverse-­engineered not only for such purposes of data collection and analysis that

The Measurement-­Image  

109

Meydenbauer mentioned but also for more advanced pattern recognition. You could call it a version of an encoding–­decoding scheme,32 but this one deals less with meanings than it does with quantifiable information. As pattern recognition, it also becomes a later iteration of what can be read from the informational surface that we call “an image” (see chapter 2), whether it is seen as an information-­rich entity in its singular existence or as it is serialized into a dataset. More on that soon below. Paradoxically, in many ways, the measurement image is an information filter that reduces some information while bringing out particular geometric details. In Mary Ann Doane’s words: “There was an overcrowding of detail in the photographic method,”33 which becomes both an issue (how to see anything meaningful) and a promise (multiple hidden and invisual layers to be discovered even with the risk of data apophenia: seeing patterns where there are none.)34 This resonates with the points raised in the previous chapter about invisibility, disappearance, opacity, and data. True, Doane was talking of another kind of data visualization than what we were discussing: Etienne Jules Marey’s chronophotography as scientific imaging. But the points apply here, too, and constitute a central issue in different administrative and scientific practices of the late nineteenth century (and onward): the realization of an information-­rich world captured in information-­rich photographic plates demands a way to mediate between image and classification, identity and variation (or even falsification). The ones in charge of different metric practices—­such as a combination of photography and anthropometrical techniques—­k new this very well, such as Alphonse Bertillon being occupied with judicial matters and mass photography of his period in the 1890s: “Thus, the solution of the problem of judicial identification consists less in the search for new characteristic methods of individuality than in the discovery of a method of classification.”35 The image became a way to capture measures and proportions (but also, in Marey’s case, temporal patterns and movement) and links the practices to a particular operationalization of technical images. The photogrammetric measure of Meydenbauer and the measurement-­lines (Hilfslinien) included in Marey’s chronophotographic studies represent a layer of production of data teased out

110  The Measurement-­ Image

with the help of the image.36 Or, more accurately, such image practices in the late nineteenth century become helpful when trying to capture what later practices of quantification and analysis can do with the measured-­image. In this manner, photogrammetric images and other measurement-­images are part of the techniques of machine-­readability and “insights not available to the eye alone.”37 Instead of eyes, however, there were images.

Measurement Lines Marey’s chronophotography is for a good reason a go-­to example for a lot of photography and cinema scholars. It stands out both in media archaeological terms when considering the early cinematic experiments in the light of scientific imaging and how movement becomes captured as aesthetics and as data.38 While the understanding of sensation and observation went through an enormous transformation in the latter half of the nineteenth century in experimental psychology and related laboratory sciences, the role of images was not restricted to being mere auxiliary instruments. Instead, the image and what it framed had built up its own reality, stretched between two poles: the linear perspective and the subsequent capacities of technical images by the late nineteenth century. The frame of an image had become a central unit of reference that was meant, in Marta Braun’s words, “to enclose a temporal and spatial unity.”39 The organizational capacity of the frame was also to synthesize so that it, as a cultural technique, defined “a single instant in time and in a single space”—­a lthough this is precisely what became problematized. The frame included multiple layers of movement, a “series of positions in space,” but even more so, it included a multitude of data points on the information-­rich surface. This becomes obvious when we look at the more abstract images Marey produced: graphs, diagrams, and data visualization and notation systems (see figures 8, 9, 10) that not only were produced from captured human and other animal movement but also presented another answer to the question, What can the image frame (or in the context of photography, what can the plate contain) as it shifts away from representational content to an administrative entity of

The Measurement-­Image  

111

FIGURE 8. Marey’s practices of imaging produced different forms of data inscriptions of movement. The lines have become an iconic part of the media archaeology of cinema, but as data visualisation they offer an alternative view to already mentioned cultural techniques of bodies in movement. Bibliothèque nationale de France, EO-­91-­BOITE FOL B.

the quantitative kind? This is also part of the genealogy of the logistical image (see chapter 4) as far as it concerns enumeration. Here, frames, plates, and lines are ways of showing more than the image contains at first sight. They measure and thus produce a second-­order data reality, both productive and limiting, both filtering and amplifying simultaneously. As Erika Balsom puts it: “The integrality of the body vanishes into an array of lines. Paradoxically, to analyse the motion of a body, the body itself must disappear, as it is replaced by symbolic representation.”40 We can refer to this also as a production of measurement and, based on measurement, a production of data. Lines crisscross bodies; What could be a more modern scene of an image even if a fully-­fleshed out story would be likely to start in painting and the Renaissance techniques of measurement and data compression? 41 Lines help decipher the metric specification of the disposition of a body in movement while the image itself is less and less understandable as a visual product: it features more naturally in the realm of the invisual as the previous chapter outlined. Lines organize territories into areas, even grids (see chapter 1) and thus are also part of the history of data organization from paper maps to GPS.42 Usually, we think of territories produced in such a way—­as

FIGURE 9. An abstraction of animal movement into a notation system. A data visualization from Marey’s La machine animale, locomotion terrestre et aérienne (Paris: G. Baillière, 1873), 152.

The Measurement-­Image  

113

FIGURE 10. A picture of a horse and its data trail. An image from Marey’s La machine animale, locomotion terrestre et aérienne (Paris: G. Baillière, 1873), 170.

“plots,” a term central for questions of property—­but the plotting appears on and in images too. The production of bespoke measuring images from photogrammetry to Marey and much in between also implied the reverse to be true. In fact, much of the discourse around photogrammetry, metrophotography, and measurement assumed as much about the readability of lines, angles, and measures (see Figure 15). In other words, measurement images picked up on a range of existing techniques that had dealt with ratios in painting and (land) surveys and thus condensed a lot of that knowledge gradually since the mid-­ nineteenth century. The point was not only to start producing a particular kind of an operational image but also to start understanding existing capacities of deciphering the world in those terms. Artists might be a special case of trained ability to draw lines according

114  The Measurement-­ Image

to meticulous mathematical rules that guide the hand, but mathematics could also provide tricks such as trigonometric functions and technical aids. For example, Kittler mentions Lambert’s late eighteenth-­century contribution to the early history of photogrammetry in this light, for as someone who turned to Leonhard Euler’s trigonometrical functions: “He no longer determined perspectival geometry as relationships or proportions between lines, like Euclid or Pythagoras, but rather as transcendental functions of an angle of vision.”43 Geometry became one stop on the route that led to a particular kind of imaging practice. As a branch of descriptive geometry, historical sources about photogrammetry repeatedly feature mathematician Gaspard Monge’s work, which can be contextualized as part of the history of technical imaging,44 as Carpo writes: Technical drawings evidently require an easier and more direct way to notate spatial measurements. In a famous passage of his treatise On Building, Alberti recommends that architects should avoid perspective and use instead other kinds of nonforeshortened, scaled drawings, similar to what modern designers would call parallel projections in plans, elevations, and side views. Mathematically formalised by Gaspard Monge only at the end of the eighteenth century, parallel projections remained the primary notational tool of all design professions almost to this day. Monge’s method used two sets of parallel projections to univocally notate the position of any point in space onto two planes that, if needed, can be drawn on the same sheet of paper: descriptive geometry is a brilliant mathematical invention.45

Monge’s work promised to do exactly this: to offer an infrastructure of notation as a guideline. It was meant as a helpful way to understand plots of land (as surveyed) in relation to plots on paper (as images and diagrams). The opening page of his Géométrie descriptive gives the whole game away in a matter of paragraphs, explaining the aim to be what we would now refer to as the operational image that deals with data: to be able to store on two-­dimensional surface representations that information about a multidimensional world

The Measurement-­Image  

115

and its qualities, and then to be able to read all kinds of operational truths from those image surfaces.46 In Monge’s universe, any natural surface can be “considered as composed of points,”47 which are then plotted into both the abstraction of those points and their relations (lines). What we now call data points are, of course, a variation of this thesis that still in Monge’s time was tied to a particular transfer operation between the world and its representation on the surface of an image (a drawing, a map, and later, a photograph). In such technical images, one reads points and their positions like one would read a map of locations, which is exactly the point when one realizes this as one point in the archaeology of invisuality, or how the shift from the primacy of figurative, pictorial visuality to operations of data is executed in relation to such image practices. This focus in the late eighteenth century and early nineteenth century is phrased as the “new question of how the optical actuality of the world can be reconstructed from the data available to the eye as sensations,”48 but we can be even more precise when it comes to some of these discourses and their institutional uses: measurement is put to particular use in describing anything as data. In short, none of the measurements was unmotivated but part of a lineage of military and colonial uses of operational images as they plot territories. Consider the lineage of such practices of precision in the French context. Monge’s contribution to art and science was tuned to the national interests from the military design of fortifications to colonial expeditions such as Egypt in the late 1790s, where his expertise in surveys and measurement came in handy.49 Descriptive geometry was for a long time nothing short of a military secret.50 While the scientific aspect of measuring and imaging was already a central part of the military–­colonial complex in France, it was later reinscribed into an imaginary past of photography, thanks to Arago. As Ariella Azoulay argues, the imperial premise of photography related to this expedition and its later reflection in photographic (data capture) in Arago’s often quoted words from 1839: While these pictures are exhibited to you, everyone will imagine the extraordinary advantages which could have been derived from so exact and rapid means of reproduction during

116  The Measurement-­ Image

the expedition to Egypt; everybody will realize that had we had photography in 1798 we would possess today faithful pictorial records of that which the learned world is forever deprived of by the greed of the Arabs and the vandalism of certain travelers. To copy the millions of hieroglyphics which cover even the exterior of the great monuments of Thebes, Memphis, Karnak, and others would require decades of time and legions of draughtsmen. By daguerreotype one person would suffice to accomplish this immense work successfully.51

Arago imagined what Egypt would have looked like through photographic images. Even if photography was a later technical invention, the expedition included draughtsmen with camera obscura devices, natural scientists with other instruments,52 and Monge armed with his descriptive geometry. All were on board the trip already; these represent the inklings of an operational image of capture and violence that often surrounds actual military operations in this extended sphere of Operations Other Than War: an operational aesthetic. During the expedition, a lot was measured, and a lot was drawn. The camera obscura was used as a technical aid for measurements to “obtain exact perspectives of monuments.”53 The camera lucida was integrated into survey techniques and apparatuses such as the telescope and included, for example, patents such as the graphical telescope in 1811 by Cornelius Varley. Such techniques were employed while photography was starting to gain ground toward the middle of the century, but as Xinyi Wen argues, the point was not a straightforward imitation of the qualities of assumed objectivity of mechanical imaging.54 In the ecology of operational practices in relation to institutions of survey, military, and expeditions, the instruments of precision were anyway prescribed in relation to what sort of measurements could be produced, including for very pragmatic ends. Instead of one apparatus of technology that would have enabled stored images “to be transmitted across space and time and then sent again to another point in space and time,”55 scientific institutions trained people to see, measure, and draw in ways that could be brought home with relative confidence in the accuracy—­ and usefulness—­of the drawings. From colonial institutions like the

The Measurement-­Image  

117

East India Company to the scientific context of the Royal Society of London, training into epistemic, even kinds of “forensic,” images was a central concern for operational needs.56 Thus the route of reading the history of technical drawing and imaging takes us to interesting institutional insights. The French Egypt expedition—­as well as many other French, English, and other scientific expeditions—­is an example of the operational images in colonial use, as a particular colonial visual epistemology even. Also, the charts of the British and the French and the East India Company provide another context where the exact measurement of location had been for a long while of crucial importance, as the “overall investment in science would also bring substantial returns in colonial trade and international geopolitics.”57 Nautical charts, after all, were a crucial intersection of data that had become operationalized in the form of a navigational image that was about not merely territorial contours but also invisible forces such as magnetism. Edmund Halley’s 1701 Sea Chart would be an early example.58 Hence it is clear that some of the “investment in science” was for very practical reasons. In histories of photogrammetry, the occasional mentions of hydrographic engineer Charles-­François Beautemps-­Beaupré are for reasons less to do with individual images than to do with techniques for formalizing instructions for their reproduction. This institutional operational knowledge was handed down to other naval officers and hydrographic engineers as “the principles of the picture-­measuring method.”59 The importance here, besides a particular chart of an archipelago in the South Pacific, or the coastline of Australia, is found in the method. Naturally, what had to be brought back were accurate topographic maps of colonial coastlines of interest. Such methods were also brought back that could be codified for further use. One can describe it as the operationalization of geometric descriptions, measurements, drawing techniques, photography, and other apparatuses (graphic telescope, camera lucida) for specified institutional purposes. The description, reproduction, and employment of such methods thus become part of the institutional pragmatics of the image that is not purely scientific but relates to questions of measure and precision, navigation, and surveying. For example, in Beautemps-­Beaupré’s

118  The Measurement-­ Image

case, seemingly far removed from photography but essential to the measurement-­images and apparatuses, he “produced many sea charts by making not only angular observations from both boats and the shore (essentially using the intersection method to plot prominent points) but also, importantly, sketches and views of the primary topographic features to avoid errors” (see Figure 11).60 You would not want to send erroneous data back as it would just escalate later errors, whether in navigation or other procedures. You had to trust your instruments and your images (maps); the navigational calculation of paths was, at its core, a need for such drawings and the distances they helped to cover. Empires and nations run on the accuracy of maps, and they run on the ability to manage distances.61 Management of distance was one form of remote sensing of sorts. The preface to the English translation of Beautemps-­Beaupré’s An Introduction to the Practice of Nautical Surveying and the Construction of Sea-­charts emphasized the importance of seeing and measuring accurately at a distance as concerning “The mode of conducting a survey on a coast without landing, the most difficult and delicate of all hydrographic operations.”62 In short, this also concerns the implications of taking distance that also featured Meydenbauer’s anecdote about photogrammetry in Farocki’s account: a theme of significant importance across an ecology of practices of measuring images. To observe features like coastal lines at a distance, to mark them down accurately, and thus to bring them back across a geographically vast distance was a key feature that has persisted in how we describe these types of images, from aerial observation to remote sensing. Managing distances was already a key feature of such pre-­photographic ways of surveying that also included techniques of managing the relation between the pictorial view and geometrically exact maps. The same theme repeats and builds up in accounts across the nineteenth century. Colonel Laussedat’s influential work picked up on Beautemps-­Beaupré’s but brought it from the sea to the land and moved it to the air. A combination of descriptive geometry and technically enabled ways of seeing from a distance, such as with aerial balloons and photography (or drawing), started to mark a specific task where images were seen as one element in the operative chain of measuring (see figures 12, 13, and 14). Laussedat and others’ work

FIGURE 11. An image from Beautemps-­ Beaupré’s An Introduction to the Practice of Nautical Surveying and the Construction of Sea-­charts, reproduced in Recherches sur les instruments, les méthodes et le dessin topographiques, par le colonel A. Laussedat: Tome 2 (Paris, 1898–­1903), 16. Bibliothèque nationale de France, département Sciences et techniques, 8-­V-­27824 (2).

120  The Measurement-­ Image

FIGURE 12. A stereoscopic view and the apparatus for construction of topographic data images as drawn in Recherches sur les instruments, les méthodes et le dessin topographiques, par le colonel A. Laussedat: Tome 2 (Paris, 1898–­1903), xv and xvi. Bibliothèque nationale de France, département Sciences et techniques, 8-­V-­27824 (2).

can be described as a description of methods, as training manuals of how to produce and read operational images, and as such part of the broader history of “the diagram as a form of visual instruction.”63 Furthermore, their existence was specific to what now would be called (in almost mundane terms of academic jargon) interdisciplinary image practices. The metrophotographic and photogrammetric practices were indeed recognized no longer only for topographic and physical geographic measurement but also for their importance in other arts and sciences, such as meteorology, oceanography, and

The Measurement-­Image  

121

FIGURE 13. A “test table” for preparation of stereoscopic images for operational uses in Recherches sur les instruments, les méthodes et le dessin topographiques, par le colonel A. Laussedat: Tome 2 (Paris, 1898–­1903), xvi. Bibliothèque nationale de France, département Sciences et techniques, 8-­V-­27824 (2).

astronomy, as explained by Laussedat, in hindsight, at the end of the nineteenth century.64 Later, it became part of aerial urbanism executed in photographic images in the early twentieth century and onward.65 The focus on images about imaging becomes a central feature of operational images as they are understood in this book: both the ways images participate in cultural techniques such as measuring and how to institutionalize certain practices are understood as part of the story of operational images. From Renaissance and Early Modern diagrams describing how the mathematics of linear perspective functions to diagrams in contemporary science and engineering

122  The Measurement-­ Image

FIGURE 14. Construction of an aerial view. In Recherches sur les instruments, les méthodes et le dessin topographiques, par le colonel A. Laussedat: Tome 2 (Paris, 1898–­1903), x. Bibliothèque nationale de France, département Sciences et techniques, 8-­V-­27824 (2).

papers that describe the computational procedure of AI techniques from training sets to models, there’s at least one stop between the images where abstraction takes place through measurement-­lines and shapes. Like in manuals demonstrating how to use photogrammetry, the image content aptly fades into the background, and abstract lines take precedence. The production of photographic images follows as part of a production of aiding lines that helped decipher the abstract shapes in and of an image as particularly useful. This genealogy of reading the image as and through grids, lines, curves, angles, and proportions persisted in different ways across any dividing line of drawing versus photography. Laussedat, but also later Meydenbauer and his architectural and topographic interests, was part of the more institutional aim to turn photographs into a set of lines; to turn the photographic image into a drawing of recognizable, calculable relations; to filter it into a particular logic of lines and shapes of geometric value. The abstraction of the moving body in cases much more often

The Measurement-­Image  

123

FIGURE 15. The image plate as a site of abstraction that connects the history of photography, measurement, and engineering drawing such as descriptive geometry. From Jacques-­T héodore Saconney’s Métrophotographie (Paris: O. Doin et fils, 1913), 202. Bibliothèque nationale de France, ark:/12148/ bpt6k98188646.

discussed, such as Marey’s work on chronophotography, was historically parallel to techniques that managed to look at concrete images of facades and landscapes in abstract ways. This example in Figure 15 from an early twentieth-­century photogrammetry (or “metrophotographie”) manual would be a case in point in readjusting what we consider the frame and the photographic plate in relation to its potential abstract layers with the help of guidelines. Lines, forms, and geometric data overflow the frames of photographic plates. Saconney’s list of examples in the 1913 manual

124  The Measurement-­ Image

Métrophotographie is a good technical condensation of those issues that characterize a particular stance to the image. Images are one thing whereas it is the institutional context where the technics become truly operational: cultural heritage data of architectural facades might be one, but as is clear by now, the emerging possibilities of aerial observation (for example, for the French military) offered by photography elevated by balloons, kites, and airplanes were one context of development. The history of photogrammetry as part of technical drawing and other images relates, in many ways, to the demonstration of those principles as a midway point between what linear perspective was in terms of calculating the image and ratios of surface planes and what is forwarded from the nineteenth century to current architectural imaging, for example. Geometric lines and shapes, triangles of trigonometric calculation, lines of all sorts across drawing, maps, and photography—­the point and the line persist as part of the construction of the image as data. There is an entire media history of the line from the diagrams of Renaissance painting and rays of light “being subordinated to a Euclidean geometry of the ruled line,” as Sean Cubitt argues. From Albrecht Dürer’s studies of images of measure (such as “Underweysung der Messung” of 1525) to the industrial lines of twentieth-­century animation (from horses to a famous mouse) to vector graphics, Cubitt demonstrates the fundamental productive logic of the line: it is fundamental to cultural techniques of plotting and grids. Cubitt argues that “as the instrument of composition and the medium of abstraction, line’s central job is to abstract form from the manifold of perception, to select and translate reality and the experience of it into signs, and in composing the schematic of the image, to extract edges from what can actually be seen: colour and texture.”66 We can also add to the mix the lines that emerge outside drawing and other arts: the multitude of scientific, mathematical lines. One special case is graph theory, as it emerges in Leonhard Euler’s work and as a solution to the Seven Bridges of Königsberg problem in 1736. Here the abstraction of urban topography of the city into nodes and edges also serves as the creation of graphs across different flat planes—­that is, planar graphs and subsequently all the multidimensional ways nodes and edges become practiced through more

The Measurement-­Image  

125

recent data science. A complex living city is calculated as optimized routes. Lines and points turn into data (points and graphs). In some cases, it also happens the other way: tabulated data turns into instrumental maps—­such as the data visualization in the nineteenth-­ century case of isothermal maps that Birgit Schneider connects to the emerging and influential turn of planetary climate cartography that makes invisible forces—­like statistically ordered temperature averages—­visible across vast geographical terrains. The drawn and continuously fluctuating horizontal lines mark a different planetary space than visible terrain.67 Lines are vanishing mediators that make areas, forms, and objects stand out as measured entities. What is tracked as thresholds of visibility and invisibility are also, in another register, thresholds of visuality and invisuality: things seen contain operative potentials as data. The line as material and textured entity might disappear in this linear operationalization—­a point that Tim Ingold makes68—­ but when disappearing, it also brings into existence other things through this artificial geometric procedure. For Ingold, the line becomes a key instance of knowledge practices. Straight lines have a particular role in the institutions we are dealing with in this chapter (not that topographic lines are necessarily straight, even if relations could be calculated as geometric proportions). Both as plotlines and guidelines, they organize space and surfaces in ways that relate to Krämer’s argument about cultural techniques of flattening, including producing surfaces with the help of lines and then lining up those surfaces in ways that become constitutive of different (graphical) practices of knowledge such as diagrams. To quote Ingold: The straight line is implicated in this vision in two quite distinct ways: first, in the constitution of the surface itself; secondly, in the construction of the assembly to be laid upon it. For the first, imagine a rigid line that is progressively displaced along its entire length, in a direction orthogonal to it. As it moves, it sweeps or rolls out the surface of a plane.  .  .  . For the second, imagine that the plane is marked with points, and that these points are joined up to form a diagram. This, in a nutshell, is the relation between our two manifestations of the

126  The Measurement-­ Image

straight line. One is intrinsic to the plane, as its constitutive element; the other is extrinsic, in that its erasure would still leave the plane intact. In what follows, and for reasons that will become evident as we proceed, I shall call lines of the first kind guidelines, and those of the second plotlines.69

Diagrams, graphs, tables, charts, maps, and other administrative and calculating media that constitute the “other” of the photographic image are described as cultural techniques of the line that also operationalize the surfaces they fabricate. Here the recursive operations described in chapter 1 are again in full view. As guidelines, they produce surfaces; as plotlines, they manufacture mathematical, geometrical, even statistical data. This data then defines those surfaces produced, and in a chain of operations, the material, aesthetic, and epistemic elements define each other. And consequently, such operational chains become a central trait for technical images from Meydenbauer to Marey and a shared part of the history of topographical and geographic surveys and measuring images that is the final section of this chapter.

The Geodesic Lines of Earth: A Planetary Measure What if we invert the problem of lines that leads into Farocki’s cinematic condensation of images, data, operations, and measurement? What if we invert the order of priorities in our discussion that has so far been about images, but perhaps should be about operations, and in this case, about lines that plot and guide?70 In other words, besides the genealogy of images equipped with lines that help us to see what the composition of images and decomposition of measures from images is about, we look at lines as the primary instance of constituting what becomes an image. Lines across photographic or other images are an instance of the measurement-­image as it emerges in photogrammetry in France and Germany across the nineteenth century, leading up to figures such as the clumsy Meydenbauer risking his life. But I want to propose that we look at lines that constitute the earth as an image already before instances of photographic lines and much earlier than digital images: thus, lines that constitute plots, work in territorial surveys, and become the link between to-

The Measurement-­Image  

127

pography and the (visual and invisual) image are here instances in the long media archaeology of measure and quantity. For the discussion of the operational image, I dived straight into the world of Meydenbauer’s photogrammetry through Farocki’s film. What followed was the broader question of the photogrammetric view to areas and images. This entry through Farocki helped to identify the thematic strands that also feature in this chapter: distance, measure, institutional techniques, visual production, and the military with its logistical pragmatics. Thus, Images of the World and the Inscription of War is an operating manual in its own right. At the same time, it articulates the centrality of measuring lines that relate not only to dominant state practices but also to such methodological discourses as forensic architecture.71 However, the constitution of lines is also present in several historical instances, if one were to write a proper historical outline of photogrammetry from Johann Heinrich Lambert’s 1759 book Perspectiva liber (The free perspective) to Laussedat’s work on constructing a world of measuring lines that he called métrophotographie and that included both drawing and photographic techniques. As mentioned, the inversion of the line is an interesting way to start looking at planetary techniques such as the plotting (of territories) as operational ontologies at the base of architectural and property practices. In addition, looking to geodesic arc measurement as an element in establishing grids is a central part of the story in mediating planetary surfaces, landscapes, architectural facades and what we see as (photographic and other) images. Hence, I am interested in how the question of lines defining measurement-­images becomes replaced onto different scales and institutions of imaging—­for example, concerning geometric, planetary data. In Peter Sloterdijk’s worlds, maps and other depictive, planimetric media present “globalisation as an image,” which should be taken in the most literal terms of cultural techniques of making flat planes in the service of navigation and other operations. “Imperialism is applied planimetry,” he continues, “the art of reproducing orbs as surfaces and worlds as charts,”72 which here applies to naval charts as much as topographic surveys, as well as the transformation of the image into a picture plane that helps to measure territories and other dimensions too. This resonates with the point mentioned

128  The Measurement-­ Image

earlier: maps are interfaces for calculation, and the maritime map is one example in history where “The map is indeed like a 2D slide-­ rule that incorporates in a precalculated format huge masses of information.” 73 In other words, the planet is operationalized as a map-­image with plotlines and guidelines; navigation is not about a mimetic relationship between the world and representation in the image (map) but a coordinating movement on surfaces of different materials and variables of terrain.74 While navigation is a cultural technique related to such questions of the porous border of outside and inside, world and image, it is underpinned by the existing precalculated and codified measurements. In other words, the question of who decides on which measures are to be used resonates closely with Azoulay’s point about “the right to dissect.” This refers to the right to plot and extract that becomes integrated into a history of photography as imperial techniques: “The right to dissect and study people’s worlds—­the Napoleonic expeditions are a paradigmatic example—­ and render their fragments into pieces to be meticulously copied is taken for granted.” 75 This should serve as a particularly apt example that will offer ground for the argument is the French geodesic expedition in the 1730s, as it transports the book’s themes to the question of the large-­ scale image (as large as the planet, in fact). Similarly, many of the later expeditions and projects would be of interest even if outside the scope of this book. Geodesic practices link to different threads of relevance: the establishment of the standard of the French meter as one ten-­millionth of the terrestrial meridian bears the traces of the geodesic measurement that feeds back into the standard through which territorial measurements become comparable.76 The link among the practice of measurement, establishing standards, mediating images, and sensing or calculating practices is why geodesy is interesting for the discussion of operational images. Some of that becomes even clearer in the context of the twentieth century as the shift from (paper) maps to radio navigation to GPS offers a narrative of technical media and nonrepresentational images. For example, the ability of GPS to deal with the issues of earlier triangulation and trilateration by way of its satellite-­based signal-operated measurement was to provide not only accuracy but a new kind of “intan-

The Measurement-­Image  

129

gible knowledge space of electronic points that shares space with the physical world but does not refer to it.” 77 The interconnection of different institutional needs becomes characteristic of the GPS and global coordination as an infrastructure for operational uses “integrating day-­to-­day surveying of property boundaries or engineering work with intercontinental war, without any special effort.” 78 At the back of that point about our current satellite-­based targeting and measurement, the 1730s expedition might seem rudimentary. It was, after all, still tied to the surface of the earth instead of being able to calculate the basis of the navigational image from the orbit. But, in much more painstaking conditions, the expedition was launched to measure the exact shape of the earth—­a question of not merely metaphysical but practical interest for the purposes of navigation and those themes of long-­distance colonial control79 that were already mentioned in our discussion of naval hydrography, for example. What Mary Louise Pratt has called “Europe’s first major international scientific expedition”80 and the launch of a new kind of “planetary consciousness”—­a longside the classificatory system published by Carl Linnaeus (or later von Linné) in 1735—­the expedition consisted of two teams whose comparative perspectives were to resolve the debate that had lingered on for years across both sides of the Channel: whether Newton’s account regarding the shape of the earth (as an oblated spheroid) or the competing (erroneous) version promoted by the astronomer Jean-­Dominique Cassini (a prolated spheroid) was correct. The French–­Anglo debate was about methods, calculations, and observations, and it resulted in the two parties being sent to measure how the earth’s surface had laid its lands. While De Maupertuis’s trip via Sweden to Lapland81 did their trigonometric measurements for the Arctic Circle view, the comparative results were gained from the rather infamous, slightly earlier launched trip by La Condamine, Godin, and Bouguer to Peru.82 In this manner, by way of the first instance of planetary comparative measurement, the latitude lines could be measured closer to the North Pole and the equator so that both measurements would reveal that clinching detail about the shape. Or, as de Maupertuis puts it in 1738 in La figure de la terre, after having returned from Lapland in August the year before: to be able to see and position oneself on the globe and maps made of

130  The Measurement-­ Image

grids of latitude and longitude is useless unless one knows the accurate length of their degrees. And those vary based on where you are.83 This expedition found its way there and back in a substantially shorter time and with fewer problems than the other party, which had years of issues, not least because of the unruly behavior of some of its leading scientists. The goal in the back of the head of Maupertuis, prepping for the wintery conditions and to record interesting anthropological details of Sami indigenous peoples’ culture, was to prove that “the length of the degree of latitude actually increased as one moved northward.”84 This amounted to saying Newton was right by empirically measuring what he had proposed through calculations. The whole debate between Cartesians and Newtonians demonstrates multiple layers in the narrativization of scientific arguments around the early eighteenth century—­w ith such figures as Voltaire with his rhetorical might heavily involved—­and are a highly fascinating case study for historians of science. Indeed, as Mary Terrall puts it, it involves how participant writers and scientists “turned observations, calculations, and theories into . . . representations, and then tried to mobilise consensus among their colleagues and among a broader public”85 in order to gain consistency for those representations. Here, the theme of “representation” is also taken literally as about the operational image of the earth and its measure—­both as a scientific and geopolitical issue. To record and compare the exact measure of the length of a degree of the arc of the meridian is one version of the problem of lines that cuts the globe into grids and plots. The image/frame of the grid becomes essential for all sorts of purposes obvious in contexts of data and cartography, but to establish the grid in the first place was an operation not any less significant for the history of imaging. The technical operations that mapped the earth into a particular shape are thus here a significant precursor to the late 1960s and early 1970s revelations of the earth as seen from space—­and one milestone in the project of planetary design scaffolded on a photographic image and the imaginaries it triggered—­but the surface-level images are at least as significant in our story. While the techniques of such a combination of manual measurement by chains and trigonometric calculations have been replaced by the GPS that circulates the

The Measurement-­Image  

131

planet and senses it into a dynamic composite image, we are still articulating things across the three-hundred-­year period of a media archaeology of operational images. Remote sensing takes two kinds of forms out of which images are produced. You can build instruments that observe signals from afar, or you can travel great distances in order to bring back the necessary data for an image. Sometimes images are the things that do the carrying. This, too, is “remote sensing,” even if by other means and perhaps with scare quotes needed to emphasize the extended meaning of the term.86 While this geodesic work of measuring the true shape of the earth is an apt example of the work of “immutable mobiles” that Latour refers to as the function of data capture in maps, diagrams, and other portable objects of knowledge, it is also one example in the logistics at the back of establishing “centres of calculation.”87 The production of images, maps, and (geometric) lines played a central role in the double work of operational guidelines and plotlines (see Figure 16). In addition to producing measurement data and images (mostly of methodological value), the trips presented an interesting mix of instruments, observations, and calculations. To quote Terrall detailing the 1730s expeditions: All parties agreed on the general design of the technical operations, which included triangulation for determining terrestrial distances and stellar observations for determining latitude. Meridian observations of the angular distance of designated stars from the zenith, taken at two different observing stations, defined the difference in latitude between the two locations. Their angular separation could then be compared to the linear distance between them, computed by triangulation. The end result is a comparison of latitude, expressed in degrees of arc, with length on the ground. On a sphere, the length of a degree would be equal all along the meridian. On an elongated sphere, the length would decrease from equator to pole, and conversely for an oblate sphere.88

The instruments for the Northern expedition were designed and made in England and France: a transit telescope, a pendulum, a

132  The Measurement-­ Image

FIGURE 16. Two illustrations from de Maupertuis’s La figure de la terre (Paris, 1738). Bibliothèque nationale de France, ark:/12148/bpt6k625769.

clock, quadrants, and a zenith sector. The use of the telescope, itself part of the story of visibility and invisibility as briefly mentioned in the previous chapter, was in this case used in combination with the zenith sector to provide exactitude in relation to the position of stars. This enabled the observer to be positioned on the earth’s surface and in relation to stars.89 In subsequent scientific quarrels (and there were several), the validity and proper use of the instruments were also raised.90 But besides a story about scientific communities, it stands as one of locations: Is the world seen as an image constructed by local cartography (in this case, the French observatory and trigonometric measurements restricted to the French territory in Europe) or by comparative expeditions that, by traveling to a remote place (from the perspective of the imperial powers of Europe), also established the possibilities of future travel: “improved methods of navigation”91 and “colonial expansion.”92 Hence, the word “utility” used for the geodesic measurement is repeated in their own

The Measurement-­Image  

133

words, such as Maupertuis’s report from 1738, as it is observed in later writing analyzing “the promise of extending the Crown’s reach to distant corners of the globe.”93 Meanwhile, in Peru: When in 1738 the news about Maupertuis’s success reached the scientists in the other part of the world, it did not entirely mean their trip became superfluous, despite the disappointment about how delayed they were. A spoiler: the project took several more years to finish. The Northern expedition confirmed the key result already: a degree of latitude at the Arctic Circle was 57,437 toises compared to the measure in Paris of 57,060 toises.94 Game over for the race to measurement. Luckily for Bouguer, La Condamine and others, besides the shape of the earth, the comparative results from the Arctic Circle and the equator would be useful for navigational purposes, which still made the other trip worth the while. While I could continue narrating the various peculiar and fascinating details, I want to focus on the scalar image that was produced. The multi­scalar vision coded into measured areas as cartographic images, as planimetric media, consisted of the comparative arctic–­ equator bifocal vision. Also, another kind of comparative, bifocal view was part of the measurement: a local site could be matched to a planetary scale. In short, to map and measure a region (length of a degree of latitude) as an image was to map and measure the planet in the same go. This local image (see Figure 17) was mathematically the baseline for what the planetary “looked” like, not as a representational image so much as a navigational i­mage of size, form, and possible directions. The trigonometric measurement on the surface in relation to the astronomical comparison of location helped establish a surface-­measured view yet abstracted across vast distances predating the later composite images of Earth from space.95 Later versions of images of Earth, from the Earthrise to the Blue Marble, are heavily circulated and quoted as iconic planetary images, but in many ways, the earlier ones, such as the technical data produced through the 1730s expeditions, should not be forgotten either. The precise area south of the city of Quito (nowadays Ecuador’s capital) is both the concrete location of the measurement lines (or chains) as well as the site of abstraction of the mathematical measurements that coordinate a simulation of the earth.96 Thus, below

FIGURE 17. Triangulation of Quito as a local base line area, both an abstract and concrete ground truth in cartographic image. From Bouguer and La Condamine, La figure de la terre (Paris, 1749). Bibliothèque nationale de France, département Réserve des livres rares, V-­7546. ark:/12148/bpt6k1051288w.

The Measurement-­Image  

135

is the description of the geodesic procedure while also containing the principles of this simulated, operational image where the lines measured on the earth’s surface enable this abstraction: After the expedition confirmed the length of the chain of triangles, routine astronomical observations and mathematical calculations were all it should take to determine the length of a degree of latitude. Once they had the overall length of the chain, the scientists would take simple star sightings to establish the latitude at each end. Dividing the length of the chain by the difference in latitudes would produce a single number, the length of a single degree of latitude at the equator. When this was compared to the length of a degree back in France, they would know the true figure of the Earth for the first time.97

This describes something that is not merely a drawing and predates photographic and contemporary satellite techniques of earth observation. Yet the mix of observations and notation functions as an element in this very short sketch of the media archaeology of the operational measuring image. The episode is but one in the history of many influential expeditions that followed—­such as Humboldt and Darwin’s voyages—­but it serves as an apt example in linking geodesy and topographical surveys as the backdrop to what comes out as photogrammetry. The Peruvian case is interesting and contrasts to the much smoother-­ running Lapland expedition. Instead of the Swedish military lending a helping hand, Bouguer and La Condamine and others relied on forced labor and slaves, and the whole trip could also be characterized as mathematical precision and measurement under conditions of altitude sickness (in the heights of the mountainous area) that the European scientists suffered from. This would then present a further context for the earth-­departing abstractions and the labor of operational images, which also depended on colonial slavery, among many other aspects. The planetary images are also labored images, an observation that returns us to the core elements in Farocki’s work too. But instead of ending with Farocki, I want to recall a point made by Ariella Aïsha Azoulay on photographic imperialism that I hope helps us connect to themes that will emerge in the next chapter:

136  The Measurement-­ Image

In a split second, the camera’s shutter draws three dividing lines: in time (between a before and an after), in space (between who/what is in front of the camera and who/what is behind it), and in the body politic (between those who possess and operate such devices and appropriate and accumulate their product and those whose countenance, resources, or labor are extracted). The work of the shutter is not an isolated operation, nor is it restricted only to photography. If shutters in the service of petty sovereigns were limited only to cameras and were not operative in other domains—­wherein the violence perpetrated by the sharp movement of their blades hits bodies at a greater proximity—­the departure of the camera and the photographer from the scene would not necessarily be part of a devastating regime.98

The description is a powerful way of accounting for the practices of drawing and photography that this chapter has discussed. But the shutter was not the only means of division, and the photographic is only one part of a lineage of operations. Indeed, the shutter is premised on dividing lines: temporal, spatial, and embodied. On the one hand, those lines are part of various institutional practices interested in topography, coastal lines, navigation, and the militarization of descriptive geometry (as per Gaspard Monge); on the other (even if very much related), they are lines of operative ontologies through which abstractions and simulations start to take place. The lines that return in more recent imaging, from cartographic techniques to GPS and the composite world views such as Google Earth,99 are of a different technical order but relevant when outlining the question of the operational image. While these sorts of practices were often backed up with the imperial prestige of scientific or military institutions, we should not forget how flawed they were: the idea of a hegemonic practice of measurement as the practice of power must be complemented with the persistence of errors, stupidities, drunken mishaps, and sheer incompetence too. Consult the Bouguer and La Condamine expedition as a case in point. Imperial operations—­and operational images—­are full of errors and are far from the disembodied rational gaze they are (sometimes) projected

The Measurement-­Image  

137

to be. A different book should be written on this topic alone: the inoperative image. Human errors and technical glitches included, I am tempted to bring into play some further themes that relate to contemporary digital imaging practices that work by way of measurement, triangulation, and laser scanning.100 But I will return to those kinds of data worlds in chapter 5. What is at stake here are the operational possibilities of measurement as a way to move from three-­dimensional worlds onto two-­dimensional planes, and in some cases, back into three-­dimensional models and simulations (for example, in architectural and other spatial, urban practices). Computer graphics is one later variation—­and quite a significant one—­in such a media archaeology where images and data are in productive tension.101 Hence the argument about the measurement line returns both to the filtering and productive force of the line. Those forces are integrated into several capture mechanisms, including surface territories and image surfaces across the eighteenth and nineteenth centuries. However, even if this chapter has more or less focused on such earlier cases, it is tightly connected to both the broader argument about operational images and, as the next chapter will show, contemporary practices of Earth observation (and Earth rewriting). Next, earlier points about invisual images and data and the planetary composite images are discussed in relation to the operational aesthetic.

This page intentionally left blank

FOUR

Operational Aesthetic Cinema for Territorial Management This is a book about visual and invisual cultures where images are operationally part of institutions, platforms, and thus also automated procedures. The chapters, s­ o far, h ­ ave outlined the case for operational images primarily as operations and only secondarily as images. Images turn into data and data into (statistical) models, which then play a role as mechanisms that can format environments in their own image. The mobilization of the idea of invisuality is not only to conclude that contemporary images are invisible (which they are, in a certain register, if one tries to see them through the eyes of the embodied perceptual system of humans) but also to say that this situation concerns the circulation and operationalization of images in the logistics of platforms, not replicating what is being seen. Images have a central role in the epistemological sense of having become part of capturing data, and in how they participate in designing environments: this can be seen to apply in the restricted (but influential) sense of, for example, architectural imaging and its software platforms that circuit data (e.g., terrain data) into the redesigning of urban and nonurban landscapes. The media archaeology of planetary images and measurement is one example of this: a history of calculation and rendering that defines the predigital production of the earth just as it does in excessively more complicated ways in the contemporary data platforms of invisual Earth observation.1 In Leon Gurevitch’s words, this constitutes a historical lineage from Dürer and Stabius’s world map of 1515 up to the Google 139

140  Operational Aesthetic

FIGURE 18. Dürer and Stabius’s map of the world from 1515: “Terrestrial map of the eastern hemisphere.” Woodcut, British Museum (“Presented by the Imperial Library, Vienna, November 1848.”), CC BY-­NC-­SA 4.0. Courtesy of the Trustees of the British Museum.

Earth platform as data visualization “of the earth’s systems under the auspices of digital media’s simulative effects.” A history of globes and images of the Earth is not the primary concern of this book. Still, it does relate to the practices of the symbolic and operational rendering of Earth surfaces for a variety of purposes, from targeting to surveying to management to other forms of control that are usually investigated as part of the satellite age while they emerge in earlier periods, such as in Operations Research. Such images concern what is in them, what they depict, and where they are mobilized—­and in which place they redefine the place and enter a reconfiguration of that space. Images, thus, are also cultural techniques (see chapter 1) in ways that become clear in a variety of cases presented throughout this book. Formatting is one of those operations. It is also a term familiar to media operations beyond digital platforms: format, reformat, prepare, and standardize. While formats have specific meaning in

Operational Aesthetic   141

practices and studies of media (surrounding questions of content), they also relate to the specification of standards. They can be said to apply to the preparation of channels of interoperability that pertain to hardware, protocols, and interfacing with users. They can also apply to multiple aspects of media archaeologies of images, including in photography. As Roland Meyer puts it, writing about the operationalization of portraiture: Rather than simply having a format, pictures are subjected to repeatable and standardizable processes of formatting: of establishing, specifying, and standardizing material and symbolic frames, arrangements, divisions, and dispositions. Such processes of formatting organize and structure pictures as products of media technologies and embed them within larger technical ensembles such as albums, archives, and databases.2

But what if formats don’t apply only to images or media products? What if images are mobilized as instruments of formatting environments, territories, landscapes, subjectivities, and social relations? This is the implication raised by considering platforms and their invisual mobilization of images as argued in the discussion of (digital) platforms and invisuality in chapter 2. Formatting thus starts to apply as one specific operational technique that characterizes a relation to human and nonhuman environments. Here, operations of images are understood in terms of a cascading set of related cultural techniques that are not restricted to visual culture: measuring, targeting, addressing, formatting, analyzing, synthesizing, calculating, standardizing, projecting, forecasting, and so on. Such would be among some of the terms in any glossary of operational images. And, anyway, these are among the terms that structure this book across its chapters. Why then make images in the first place? Is there any use for pictorial or representational practices if all that matters is the operational part of the equation? Is there a cinematic or visual practice of invisual culture? Or is that oxymoronic? What is the relation of artistic methods that deal with these cases of invisual operations? Farocki’s work, as Thomas Elsaesser has argued, can be cited as one guideline that makes sense of the new invisibilities produced

142  Operational Aesthetic

as operative parts of contemporary power.3 Here, camouflage features as a central part of visual production in ways that exceed the usual military contexts of such techniques of misperception and making invisible: processes of power in their routines—­and Farocki was a master of cinematically framing routines from prisons to factories to shopping—­a re often traced only through the proxy of bodies in training. Operations come to define subjectivity—­or perhaps the operational is how subjectivity is formed (cue in a big part of Foucault’s oeuvre). Thus images are featured not merely in fully automated worlds of technical operations but in the broader assemblage that makes humans work—­sometimes even work for operational images and technical images (to create them). Consider, for example, the Parallel I–­IV series about animation images in this sense, or the Creators of Shopping Worlds (2001) that itself concerns formatting of space (malls). Hito Steyerl’s How Not to Be Seen (2013) is a further classic video piece that by now is a regular reference point for understanding the operational production of invisibility and misperception. As a didactic (or a “fucking didactic,” as the subtitle promises), the educational aim is to demonstrate what image surfaces function as sites of intensive production of worlds while implying centrality of techniques such as calibration for production of resolutions. Rosa Menkman’s recent White Out video installation (2020) and her artistic research project Resolution Studies pick up similar themes: narratives of seeing and not seeing, and of resolution approached not only as a technical feature but as an investigation of the standard image and how fabrication of standards is also a practice of exclusions. From the core courses of cultural theory: what is left out matters as much as what is framed. “A resolution also always involves the inherent compromise as to what other ways of rendering are not rendered actual. And it is through these other ways—­not (yet) implemented or supported resolutions—­that we need to train to see, run and formulate our alternatives.”4 As far as artistic and aesthetic practices are concerned—­and what I have offered here is not a full review of all works of significance—­ operational images are often images about images. They frame change of imaging practices with images that are not themselves operational but depict and narrativize the institutional, technical, geopolitical, and material situations. This is not to suggest an un-

Operational Aesthetic   143

necessary disconnect between the artistic framing of such scientific, military, and technical uses, but to employ the realization of recursive chains of imaging that I discussed in chapter 1. As cultural techniques, these are images about images as second-­order semiotic and material techniques. This chapter taps into this process through contemporary art and moving image practices. Many contemporary artists, like Armin Linke’s work on technologies of ocean sensing and Geocinema’s work on the Chinese Digital Belt and Road Initiative (DBAR), are good examples of visual practice that deal with invisual cultures: how signals and data are processed to form images and models. Both show that invisual cultures take place on different institutional levels, including as administrative events: policy decisions are taken, technical resolutions feature side by side discussions of questions of infrastructure; scientific projects from remote sensing to autonomous systems are part of both symbolic nation-­building (case in point, the DBAR) and geopolitical strategy. Much is at stake in terms of financial value and scientific results, soft power, and its material aftereffects in terraforming the planetary surface. Such technologies are often best understood as large-­scale systems or infrastructures that show multiple levels of impact on the territories and regions where they operate.5 I have worked with artist-­scholar Abelardo Gil-­Fournier on geographical datasets in ways that also engage with the notion of an operational aesthetic of territory, where machine learning techniques such as next-­frame prediction become a way to reformat territorial futures, incorporating a sense of cartographic knowledge that concerns different temporal futurities: next-­territory prediction. After this framing of the problem, I continue with the notion of “operational aesthetic,” followed by the practices of Geocinema and other artistic works that help us understand both the geopolitical implications of the employment of operational images and the link to infrastructural and logistical images. These practices speak broadly to operational images and Earth observation, and related data practices involving environmental and political sustainability, thus moving beyond the military sphere of interest and operations. Furthermore, they are part of the operative sphere of the imaging of climate and weather, environment, and ecology that Birgit

144  Operational Aesthetic

Schneider also outlines in related terms: an aesthetic organization of data that becomes part of not just “seeing” but establishing the base for operations (or framing process and operations).6 We can thus reroute some of the concerns in Farocki’s artistic method and moving image installations with contemporary works that deal with other aspects of what the operational image implies as aesthetic organization and administration. This helps to articulate the idea of operational aesthetic not as one specific style but as a method of training: to train to understand the various scales of the infrastructure of images, their epistemic value, institutional role, and the historical shifts. This pairs up with the point made in the previous chapter about the methodological function of discourse about measurement-­images. Here, to train to see how operational images feature an idea about multi­scalarity and the institutional context of artistic and design practices is very different. But images about operational images drill us into a technical mindset of an action at the base of the image. The base of the image as infrastructure, and the infrastructure in territorial situations, are here the sites of operational images. This field of the operational aesthetic thus includes multiple layers from operators to screens, a site of operation to the distribution of operations across large infrastructures and vast territories.7

Operational Aesthetic: Take 1 It might sound too simple to suggest a specific aesthetic of the operational image. It is not, after all, a specific genre or style, despite the connotations to the aesthetics of surveillance, administrative screen space, situations like the control room,8 or “think tank aesthetics” as a fusion of operations research and twentieth-­century modernism, as Pamela Lee shows in her work.9 Instead, while aware of these historical, aesthetic, and architectural contexts, I see “operational aesthetic” as a heuristic term that helps to connect specialist image practices and technical imaging into a broader discussion of the notion of operationality that emerges from the earlier history of photography and calculative images to the digital images and automation that is at the back of military applications, AI and platform

Operational Aesthetic   145

culture, and the gradual shift to the centrality of invisuality as per the argument in chapter 2. Operational images can then be seen as a particular conceptual switch in what we consider an image and what we consider as users and operators in such aesthetic practices. Action/acting takes precedence, which is evident when, for example, Elsaesser argues that a redefinition of the aesthetic takes place in such combinations of looking, operating, and clicking (or interface actions more generally): One can go even further and claim that operational images function as instructions for action—­[and] are the new default value of all image making, against which more traditional images, i.e. images meant merely to be contemplated, to be watched disinterestedly or for their aesthetic qualities are being redefined as specialized instances of operational images—­and I’m not even primarily talking about advertising, propaganda or pornography. After all, there has grown up a generation of users who, instead of wanting to look at images, are more and more expecting to be able to click on images on their Facebook feed or on Instagram.10

Aesthetics of operations become about processing the bodies involved and formatting their actions, as well as the various other proxies related to the spectrum of the what and where of operations. While much of the above seems to relate to a discussion concerning digital culture and the image as interface, Tom Gunning’s, Ilka Brasch’s, and others’ take on the operational aesthetic, shifts the focus to an earlier context of cinematic culture as well as popular depictions of technology in print media.11 Brasch focuses on late nineteenth-­century fiction and early twentieth-­century cultural expressions in outlining a particular narrative and visual style. Drawing on existing historical research on a variety of media formats, from P. T. Barnum’s “public hoaxes” to Edgar Allan Poe’s literary writing to early cinema, Brasch develops an argument that concerns the relation between technological systems and the public display of technology. The domestication of new technologies is mediated through an aesthetics of how things work, where the observation of

146  Operational Aesthetic

detail led to an appreciation of industrial society in action. Brasch quotes Neil Harris on the operational aesthetic in U.S. culture of the nineteenth century: Machinery was beginning to accustom the public not merely to a belief in the continual appearance of new marvels but to a jargon that concentrated on methods of operation, on aspects of mechanical organization and construction, on horsepower, gears, pulleys, and safety valves. The language of technical explanation and scientific description itself had become a form of recreational literature by the 1840s and 1850s. Newspapers, magazines, even novels and short stories catered to this passion for detail.12

Such displays can be seen as a version of the cabinet of curiosity in the emerging age of technical systems but with much room reserved for deceit and trickery.13 A passion for detail went together with a variety of affects concerning new technologies, from anxiety to enthusiasm. Brasch expands on Harris’s work to show the various ways that the operational aesthetic shifts across genres and institutional contexts. It becomes effectively seen as both a narrative fascination with the workings of systems—­questions of observation, detail, and process—­and training how to read and appreciate systems, a recursive operation upon the reader/viewer as such. Reading Poe is thus contextualized in this broader technological culture. It becomes itself indexical of what the genre of print serials teased out: detective fiction unfolding through the newspaper and magazine serials, thus becoming a processual genre. Operational aesthetic homes in on the processual side of technological systems. For example, one could cite philosophical toys and visual technologies that demanded a relation between touch and process,14 such as the phenakistoscope and zoetropes. Still, there is also a broader continuity across different technologies and what they imply by way of operational process and the position of the user. The operational aesthetic becomes understood in Brasch’s analysis as the narrative format of “this is how that works”: while at first seeming distant to our topic at hand—­video installations by Farocki, contemporary media arts, technological images, and the

Operational Aesthetic   147

operations after operations research and cybernetics—­it helps to focus on the narratives and images that make sense of technological systems. In an early cinematic context, as Gunning points out, the operational aesthetic comes to concern a “fascination with the way things come together, visualizing cause and effect through the image of the machine, bridges the end of the nineteenth century and the beginning of the twentieth, shaping many aspects of popular culture”15 while reserving a specific spot for the inoperative too: the failing of a systemic, even programmed procedure exemplified by various hiccups in Chaplin’s and Keaton’s films. A hint of the inoperative image, again. Two sides to the operational aesthetic, then: “This is how” as showing, picturing, and demonstrating that make sense of operational procedures; “how that works” as referring to the advanced systems where images are featured in operation, the military and scientific systems that exhibit nonhuman automation (as much as the training of operators for particular tasks). This show and tell is even more interesting for complex bundles and layers of interface systems. Marianne van den Boomen has referred to this as the built­in, designed depresentation that comes as a feature of hiding “procedural complexity” that features in computational environments. As a term, it is useful to read against the backdrop of aesthetics of demonstrating a process: “The icons on our desktops do their work by representing an ontologized entity, while depresenting the processual and material complexity involved. This is the way icons manage computer complexity, this is the task we as users (in tacit conjunction with designers) have delegated to them.”16 The managerial task of interface design is somewhat related to the above-­mentioned design of looking, operating, and clicking. Thus this also concerns the design of complex systems that can be packed into simpler shells for user interaction—­or operator interaction—­and then feature as the other pole (of what scholars of operational aesthetics in cinema and literature outline). These are not contradictory but, in peculiar ways, complementary traits. To show and tell in order not to see all that there is.17 Such detailed historical reading of late nineteenth-­century and early twentieth-­century public cultures of operations represented in a variety of formats comes to imply that the representations are

148  Operational Aesthetic

a form of training. In ways that echo Walter Benjamin and other media scholars’ arguments about the drilling and training of modern subjects to understand technology, the idea of the operational aesthetic as both domestication and as training the operator becomes clear.18 While the operator is a central figure of modern large ­technological systems, it also takes place on other scales of the interface with technology (and not merely traditional media technologies). The operator is created both for the home and the workplace. As Brasch notes, mid-­nineteenth-­century advertisements by I. M. Singer & Company “addressed them not as purchasers of a commodity but as operators of a sewing machine.”19 The shift in gender focus of “operating” becomes an important part in appreciating the different histories of the operational aesthetic that includes questions of public and private labor too: The reassurance of sewing as a woman’s task then impacts the notions of the machine and of the operator, both of which move from the public and work spheres into the private sphere of the Victorian home. Whereas technological mechanisms were established in farm work or in industrial facilities, both times performing functions that in one form or another translate into monetary value for its operator, sewing machines provide their home-­based operators with surplus time instead of money.20

While the sewing machine had an important impact on various forms of the display of technology and the empowerment of female work with machines, at another scale, the telegraph, and then subsequently the telephone, presented an interesting dilemma for the operational aesthetic of machinery. As Brasch points out, such systems were not restricted to any one site where they reside as machines: they relied on a distributed network to function. As such, the operator—­with the fine-­tuned specialist skill, haptic touch, and sensibility—­became understood through her/his/their work that interfaced with large-­scale systems and distributed infrastructures of communication. As Brasch sums it up, “Human agents are trained to become operatives, just like they were when using a sewing ma-

Operational Aesthetic   149

chine.”21 As operatives, they are both at the site of the machine and somewhere else, as the system spans vast geographical territories and becomes a structuring of a different sense of space. This, indeed, was a key realization of early media and technological theory too: space and time, not residing only at one place at a time, distributing a set of communicative affects, capacities, and sense of subjectivity.22 Is the operational aesthetic, in this sense, an aesthetic of training? Not an aesthetic only in the sense of what training looks like but aesthetic as guidance of sensorimotor activities toward particular forms of action and operation. Images that function as manuals that show labor and knowledge, whether packed into a human (habits, gestures, memory) or a machine (system). Elsaesser seems to imply this in his discussion of Farocki’s work on labor and the new invisibilities of control systems, or labor in “control societies,” in the manner Gilles Deleuze outlined. Training thus is both cognitive and affective, goal-­oriented even if potentially unconscious. In many ways, this then applies to an endless chain of recursive operations of images that train: images that show images that train humans to operate images and images that increasingly operate without human intervention or even beyond human perception. The operational aesthetic though, when dealing with large-­scale systems that emerged in the late nineteenth century and populated the twentieth-­century technological landscape and problems of operational research, are then faced with the same task: how to understand, cognitively, the out of sight and out of place that is integrated into a formatting of the world in its own image, an image that programs potentials of action. How should we understand the nonlinear yet connected operations of telecommunications, remote sensing, and remote targeting, or the work of infrastructures that enable displaced communication to take place and reformat physical territories?

Geocinema: Digital Earth Images One section of Geocinema’s Making of Earths narrates the collection of minerals and other ground samples in Tengchong, Yunnan province, China, in the 1970s. The airborne experiment was central to the emergence of new remote sensing applications in China, and it

150  Operational Aesthetic

started from the ground up, mapping and organizing ground truths into a taxonomy: 15,000 rock, soil, vegetation and water samples had been taken from its mountainous landscapes as part of the inaugural 1970s Landsat remote-­sensing experiment. Their spectral signatures folded into databases which were crucial in training satellites to see not only pollution from urban and oceanic topographies, but also the mineral, petroleum and coal deposits underneath the nation-­state’s territories.23

From airborne photogrammetry to spectral imaging and many other techniques, remote sensing operates as productive cross-­ reading of ground truths and invisual technologies that prepare images in order to prepare data in order to prepare policy from extraction to sustainability. Grounds are formatted (organized, divided, classified) for epistemic purposes to be used as training data and establish an operational chain: ground to satellite to ground again to data for further imaging. This operational chain is extremely useful for depicting large-­scale areas, whether those that are hard to reach (such as remote sensing and AI systems used to identify ocean-­floor ecologies from cable infrastructure to biota) or just widely distributed where the clustering of particular objects or resource types facilitates management of those areas.24 Such operational images are thus not located only on the screen but in the circulation of information and data from sensors to machine learning processing. The opening note about turtles all the way down also applies to this operational infrastructure of remote sensing and its role in contemporary geopolitics that formats surfaces and reinscribes earths (and oceans).25 Geocinema’s cinematic installation and the broader research project Framing Territories look at what they call planetary cinema and the “making of earths” through their research on the Digital Belt and Road initiative (DBAR). Asia Bazdyrieva and Solveig Suess focus on the Chinese data infrastructure for sustainable development through artistic methods. These methods become a way to tie in contemporary audiovisual work with insights into large-­scale media infrastructures that address the technoscientific basis of framing

Operational Aesthetic   151

FIGURE 19. The Making of Earths, 2020, Geocinema. Still from a video. Courtesy of the authors.

climate and bring another layer of imaging into view through their experimental take on remote sensing (that itself produces another layer of visibility of planetary-­scale changes). From “Planetary Cinema” to “Calibration” to “Future Forecast,” Geocinema’s themes highlight both spaces where spatial data is constructed for use and spaces where infrastructure operations enable circulation of all forms of entities (roads, railroads, data, and people, including data about ethnic minorities). The themes also highlight epistemic processes that emerge from their work with DBAR as well as with historical archives such as the Jesuit histories at the Zikawei observatory in Shanghai that formed a part in a network of observatories. Other historical reference points include the early nineteenth-­century case study of the weather and astronomical observations of Mongkut the Great, monarch of Siam (currently Thailand) read side by side with questions of calibration that refers to remote sensing techniques of ground-­truthing but also the other levels of work of standardization as it features in DBAR and other policy work as a transnational glue. This creative recursive structure of themes in many ways reproduces the recursive logics of power at play in DBAR and its role in broader Chinese politics of sensing.26 As such, their work presents us with a useful case for a contemporary version of operational aesthetics: it deals with the multi­scalar

152  Operational Aesthetic

reality of scientific imaging and large-­scale infrastructures while moving across different sites and contexts of production of weather and ecological data: from satellite ground station and ground station security guard to interviews with policymakers and scientists to operator control rooms and screens of graphs and diagrams. In this sense, the work is set in a similar context as, for example, Armin Linke’s installation, moving image, and photography works that focus on seabed mining and the planetary management of oceans, such as the Prospecting Ocean project.27 Likewise, Susan Schuppli’s Ice Cores (2019) video features similar themes of the transformation of material samples into data and climate science relevant knowledge, even if the institutional spaces in question are an archive (Canadian Ice Core Archive) and a lab (OSU Ice Core and Quaternary Geochemistry Lab in the United States). Another contemporary example of a mix of artistic and architectural aesthetics is the Anthropocene Observatory.28 Here the notion of an observatory as “a composition of documentary practices and discourses”29 allowed for a way to understand the links among knowledge institutions, standardization of information, and engagement with planetary-­scale issues in ways that tapped into infrastructures of information about environmental and territorial changes: from bureaucratic administration to measuring instruments to the management of ecological data. To underline the centrality of curatorial practices as sites of crystallization of different disciplinary practices, ZKM’s Critical Zones (curated in 2020 by Bruno Latour and Peter Weibel) can also be quoted in this context.30 It is also relevant both for its metaphoric use of the notion of “observatory” and for the specific topic at hand, as Geocinema’s The Making of Earths was part of that exhibition alongside many other contemporary practitioners. In what follows, I primarily refer to the version of the Framing Territories project installed as The Making of Earths for ZKM’s Critical Zones exhibition; this version consists of sections that underline the narrative choices of aesthetics of sites, logistics, labor, and historical layers in particular ways. The different themes for The Making of Earths (“Footnotes to the Making of Earths”) speak of remote sensing and remote actions that correspond to technological protocols, policy, and bureaucracy. In addition, they produce a se-

Operational Aesthetic   153

ries of images of operations that tie together images, data, satellites, people, ground stations, and scientific conference dinners. The specific sections of The Making of Earths produce a series of multi­scalar surfaces of images with nonlinear structure. This cinema is composite on many levels, including the exhibition outputs. While their work can be contextualized in a variety of contemporary contexts, from experimental (moving image) media arts to speculative design, I want to bring this into a discussion of infrastructural and logistical images. As such, their work (indirectly) elaborates Brasch’s idea about operational aesthetics in the context of large-­ scale data infrastructures that consist of physical, informational, and human elements, namely working bodies on-­and off-­screen. In Geocinema’s case, two elements stand out, and they are intertwined in ways that elaborate a particular planetary angle to our topic at hand. First, the DBAR plans are a diagram of a planetary image, the Digital Earth, as it has been symbolically coined since the 1990s. Second, the cinematic method elaborates systems that are too large to be grasped at one point of reference, image, or frame; in one situation; or by way of interface-­operator experience. Geocinema’s Framing Territories was underpinned by the Digital Earth program 31 fellowship that allowed them to conduct research, interviews, shooting, and data collection “on the ground” in both Beijing and other locations where the DBAR program hits the ground. Their focus on the earth observation network and its remote sensing and environmental data operations was entangled with their interest in cinematic expression and planetary infrastructures. Their work as part of the Strelka Institute’s program had already triggered the methodological line of considering “planetary-­ scale sensory networks—­cell phones, surveillance cameras, satellites, geosensors—­as distributed cinematic apparatus,”32 which resulted in an extended and speculative sense of what a camera is. As a postlenticular device (see chapter 5), the camera here is perhaps more of a metaphorical residue when it comes to the actual employment of scientific sensors that operate in various locations on the ground and in orbit, but their interest in moving images persisted as central to the methodology. Indeed, this large-­scale sensory operation is described as nonlinear and fragmented with multiple elements of data, paper, and people: “Sensing fragments of the earth

154  Operational Aesthetic

FIGURE 20. The Making of Earths, 2020, Geocinema. Still from a video. Courtesy of the authors.

their operations generate terabytes of raw data, infrastructural architectures, obscured labour, dissonant weather, governmental policies, scientific management, environments and situations.”33 The DBAR program—­as an infrastructure and policy project and even a platform—­is a fitting focus for the research into planetary infrastructure. It could easily be seen as an interesting and expanded case study for an STS project on science infrastructures such as geopolitics. However, in this case, it is approached through cinematic practice and methodological considerations as to what role images play in this constellation of scientific concerns about remote sensing, ground infrastructure, and projects that span across national boundaries. As a science program, DBAR has an interesting history related to the birth of remote sensing projects in China since the 1970s and systematic investment in realizing the symbolically celebrated Digital Earth projects discussed by Al Gore in 1992 as part of a run-­up to the vice ­presidency in the Clinton government, where the broader agenda of “information superhighway” became a popular trope of policy discourse. Interestingly, the Digital Earth was picked up in the Chinese context, and it started to resonate with a variety of foreign policy (and other) goals over the past twenty years. By the end of the 1990s, the Chinese Academy of Sciences had already hosted “the first Inter-

Operational Aesthetic   155

national Symposium on Digital Earth” (1999) with the first Beijing Declaration on Digital Earth followed up in ten years with the next Beijing Declaration on Digital Earth in 2009.34 There, Digital Earth becomes manifested as the term that includes work toward infrastructures and information resources of sustainability, questions of “global commons,” and a general range of security Operations Other Than War: “Digital Earth should play a strategic and sustainable role in addressing such challenges to human society as natural resource depletion, food and water insecurity, energy shortages, environmental degradation, natural disasters response, population explosion, and, in particular, global climate change.”35 Digital Earth can be seen as a conflation of the lineages of cartography and meteorology.36 It picks up on the “digital” as a period of big data from 1972 Landsat Earth Observation data to the twenty-­ first-­century information infrastructures. It establishes the ability to integrate and visualize data as a central engine of influence and impact. The integration of Gore’s vision of a “three-­dimensional visual representation of Earth that would help humankind take advantage of geo-­referenced information on physical and social environments, linked to an interconnected web of digital libraries”37 structures Digital Earth as an international project from satellites to ground surfaces with an air of (political and scientific) neutrality. Yet, at the back of the several institutional connections and methods, it connects to geopolitical planning and presents us a case of techniques of images and imaging as they work on (transnational) territorial management.38 Even if outside the scope of this chapter, it is useful to read the Digital Earth and the subsequent DBAR in relation to the Belt and Road initiative focused on physical infrastructure projects and legal frameworks that reformat territories across vast regions in Asia and beyond. Architectural, infrastructural, legal, and economic tools—­ such as debt regimes—­are part of Belt and Road’s different investments from ports for maritime routes to railroads to dozens of new Special Economic Zones “modeled after the Shenzhen Special Economic Zone.”39 A good example of such projects is the “estimated $60 billion China-­Pakistan Economic Corridor, a collection of projects connecting China to Pakistan’s Gwadar Port on the Arabian Sea.”40 Belt and Road, and also Digital Belt and Road, are thus apt

156  Operational Aesthetic

examples of politics of infrastructure understood in the broad sense while also supported by other measures of geoengineering in and out of Chinese territory: these include hydrogeographic maneuvers that are not restricted to terra firma, but include atmospheric factors too, such as weather modification.41 Bazdyrieva and Suess’s summary helps to understand the scale of operations across the world: Similar to the Belt and Road Initiative, the geographies of its digital counterpart remain not clearly defined yet ever expanding. Having twenty-­seven satellite missions, super computers in Jinan and Wuxi, and four satellite ground stations within China’s borders, the initiative had already launched eight representative centers outside and across four continents before the end of 2018; first in Bangkok, then Helsinki, Potenza, El Jadida, Peshawar, Moscow, New York, and Lusaka, with many more to come. Made of an ensemble of coordinates, the project relies on creating networks between nation-­states with more than sixty regional partnerships and transnational agencies such as the United Nations. Less explicit is how these locations coincide with particular infrastructural partnerships, where along with “capacity building” for scientific exchange, heavy projects laying down fiber-­optic cables have been enabled across regions in Eurasia, Africa, and the Arctic. This includes the Lusaka economic zone, a China Smart Tech City in Morocco, with similar contracts in Thailand, Pakistan, and a 10,500 km fiber-­optic link across the polar seabed through Russia into Finland.42

While the Belt and Road initiative has become a key feature of China’s foreign policy branded by infrastructure development and financing, the Digital Belt and Road is defined as a science program that is closely aligned with the sustainable development goals of the UN—­with an effective link to territorial plans across vast areas of the earth. The sustainability plans (including achieving carbon neutrality by 2060), material epistemologies, and infrastructures have significant geopolitical stakes built into them: climate modeling and their infrastructures are forms of political closures and transformation through which contemporary planetary governance

Operational Aesthetic   157

is taking place. Infrastructure for data and data-­based services can intensify geopolitically significant asymmetries while still working within generally agreed upon global sustainability goals. As Bazdyrieva and Suess narrate from one of their field investigations into DBAR networks: “Many of the projects proposed over the duration of the conference included ways to accrue sets of data from ‘developing countries’—­nation-­states which did not have the capacity for their own satellite ground stations or supercomputers. For example, Bangladesh and Myanmar were promised on-­demand satellite images of their own territories through an online gateway. Funding provided by the Chinese state was to allow the UN Environment Program to implement air and water monitoring systems in Nairobi, Cape Town, in Addis Ababa.”43 DBAR interest in ecological environments, natural disasters, land resources, and other geoscientific data and modeling44 is part of its grand science infrastructural plan: global, quasi–­real time, based on mass-­scale data acquisition, integrating space–­a ir–­g round observation, with “high spatial, temporal, and spectral resolutions.” Data about the earth is constantly constructed anew according to different plans, some with good intentions and some more cynical, some a mix of both. The territorial exchange becomes transposed onto DBAR services and other relevant, structured dependencies. Computers, satellites, and data become part of implicit geopolitical soft power. To note, not all of this power is particularly soft either as it works as an example of internal colonisation: it is important to consider how many of the Chinese “data-­intensive and predictive technologies . . . were first tested and developed in Xinjiang,”45 as Shaoling Ma notes, drawing connections between persecution of minorities such as Uyghurs and the broader state-­led programs for sensing and sustainability. In the context of this book, the question of images comes to the fore. While the idea of a Digital Earth conjures an image (or imaginary) of a virtual globe—­a theme sometimes used in the rhetoric too—­it consists of a variety of sources of Earth observation and other data that is Ready to Use (RTU), managed through various software platforms. While the visual media history of Digital Earth would be itself interesting—­from early modern woodcuts (see Fig­ ure 18) to techniques in cartography to Landsat satellites to Chinese

158  Operational Aesthetic

satellite remote sensing and big data to early 2000s Microsoft 3D geographic teaching software46 and onto Google Earth—­the focus here is on the operationalization of particular images (see Plate 4). This means also focusing on what role images play in the context of territorial transformation and management as enfolded in data. Indeed, one could track the trajectory from the “computer and 3D graphic technologies supported by the film and video game industries”47 to the ways virtual globes are an element in how platforms of scientific data produce operational images—­namely, as models and simulations. Multiple genres of images are at play: photographic, cartographic, and data-­driven visualizations are integrated on the platform. Satellite data to various forms of RTU products such as “digital orthophoto maps, regional image maps, top of atmosphere reflectance, land surface reflectance, top of atmosphere brightness temperature, land surface temperature, normalized difference vegetation index, ratio vegetation index, global environment monitoring index, normalized burnt ratio, normalized difference water index, and pixel quality attribute”48 function alongside “other big Earth data” that provide a wealth of standardized and thus combinable and comparable material. Measurement stands at the beginning of production of environment as data. The Digital Earth at the back of DBAR is thus driven by a question of “massive, multiresolution, multitemporal, multitype Earth observation data”49 that is then mobilized along with the platform for various physical models (climate change, Earth system) and AI models (related to chapter 2 but here on a scale of multi­scalar Earth models as images). Geocinema’s The Making of Earths visually dissects the theme of the globe as it features as an architectural ornament in DBAR plans and science buildings. The globes featured in posters and a very large 3D model at the Remote Sensing and Digital Earth Institute entrance in Beijing (see Figure 21) are reminiscent of revolving globes found in school classrooms, except these are expanded to loom above the height of an average human. The model pairs up with the globe-­ shaped cinema theater inside the building as a walk-­in world in its own right. One can discuss these as cultural techniques of inside and outside, space and projection, and cartography and cinema. Other

Operational Aesthetic   159

FIGURE 21. A globe in the entrance foyer of the Remote Sensing and Digital Earth Institute in Beijing, outside a globe-­shaped cinema theater. Making of Earths, 2020, Geocinema. Still from a video. Courtesy of the authors.

subtle shifts across the binary divisions of inside/outside are shown: a passage to inside the fabrication of cinematic continuities across a heterogeneous set of input data; a physical model, transformed into a digital screen; a paper-­based cartographic representation of territory juxtaposed with a data map. As a cinematic version of engaging with operative ontologies, it shows various elements that make up the scaffolding of the DBAR: the stitches that hold the symbolic and the material infrastructures together, something that could also be seen as a “cosmographic project” in the sense that historian of science John Tresch has described the term. In Tresch’s use, it refers to the concrete spaces, images, and practices where cosmologies are materialized, “ideas of the order of nature that are enacted, embodied, elaborated, and contested in concrete settings, institutions, representations, instruments, and practices.”50 Inconspicuous details from the series of globes on posters, cinema theater shape, and other recurring symbolic objects are thus also part of how the cosmology is being inscribed beyond written manifestos and other grey literature51 of remote sensing and Digital Earth. Or, to phrase it even more tightly: DBAR is a project that feeds off grey literature and its material repetitions across architectural space that resonates with infrastructural space.

160  Operational Aesthetic

Cosmographical insight to abstract sets of ideas (planetary technologies, infrastructures—­cosmologies even) is thus on par with operative ontologies, and here they are placed in relation to material scientific systems. The planetary as it is being built as architecture, as symbolic system, as infrastructure—­even as “sculpture”52—­is opened up by Geocinema’s camera movement across the space as it links up with a pan across the image surface of a composed, meteorological earth-­image. In other words, the camera views the cosmographic details of architectures of institutions, images, instruments, and practices, and the Making of Earths is framed through these scenes. The project frames the production of invisual images through its own cinematic images that show the process unfolding. No tricks, no underlying secrets are assumed or revealed, just the surface process that emerges at sites, symbols, and technical explanations. As a “making of” film, it fits into the description of operational, procedural aesthetics as “how is it done”—­but one that does not give out a simple explanation of a deus ex machina. Instead, a multitude of processes unfolds where the operational is not merely one type of an image—­not necessarily even one concrete picture—­ but rather about the institutional links and articulations and sometimes casual observations, as in many cases of the interviews peppered across the episodic film. The operational refers thus to a vast, distributed capacity to sense, register, and digest the world as images and spit it back as models. The shift between data and image, so central to much discussion about operational images, comes through in an interview Geocinema filmed with Professor Li of the Academy of Sciences. A didactic explanation of the pipeline of data follows, underlining a cosmographic world-­building exercise with knowledge: from signals to pictures to data to information to knowledge, the material flow of remote sensing input becomes one form of “immutable mobiles” that travel from one infrastructural site to another thanks to the work of standards. One could with justification refer to the various sites filmed by Geocinema as “centers of calculation,” to echo the term proposed by Bruno Latour: those nodes of scientific power that can accumulate data, processing it by working through “maps, account books, drawings, legal texts, and files, that cartographers, merchants, engineers, jurists and civil servants get the edge on all

Operational Aesthetic   161

FIGURE 22. A diagram of the DBAR platform, Geocinema. Courtesy of Asia Bazdyrieva and Solveig Suess.

the others.”53 The ability to combine, superimpose, reshuffle, and summarize are essential operations that feed forward back into the world. Signals transform into pictures, then into data, then into information, then into knowledge, sent out to the world. Now, though, such centers of calculation might be located somewhere between spatial and scientific institutions and their platform effects; or, that the platform itself becomes one form of an operative center of calculation that accumulates data and processes it and reformats worlds in and through logistical operations. Professor Li’s explanation in the section on “Algorithm” in Making of Earths is one specific point where images and pictures are specifically talked about: “The satellite is just a signal for us, it is not a picture,” he explains and continues, “it’s a signal, and we process it to be a picture.” A technical detail turns more than technical. It turns into particular moments of transformations in the logistical operations of DBAR. As such, the tricky relation between representation and the invisuality of data is somewhat implied in how material is gathered in Geocinema’s work; the signal is the base reference point for much of DBAR’s scientific work, but images that it mobilizes are not visible in the traditional sense. Hence, the operational aesthetic of large-­scale signal and data systems is focused on

162  Operational Aesthetic

proxies that are cosmographic details and operative ontologies; they are inscriptions in and around centers of calculation that DBAR establishes that are the geographic locations for the audiovisual piece and the larger Framing Territories project: the Chinese Academy of Sciences, the Sri Racha ground stations, the climate research center of Bangkok, conference halls in the Tengchong volcanic district, and the risk-­management rooms of the oil conglomerate Gazprom in Omsk, Russia.54 In each case, the operational aesthetic at play is about choosing, selecting, and framing which image is adequate for the job and which scene and architecture at play speak to how territories are being formed in this mass-­scale circulation of signals and materials. Both the installation Making of Earths and the Framing Territories project venture into different sites and situations to map the geocinematic apparitions of contemporary entanglement of data and territory. One could justifiably argue that it stretches the term “cinema” way beyond many scholars’ comfort zone but also returns to what Thomas Elsaesser called the s/m version of a media archaeology of cinema: science and medicine, surveillance and military, and so on.55 And it also speaks to what Ute Holl demonstrates as the laboratory origins of cinema: scientific sites of research of living beings such as in experimental psychology, measurement and information management, feedback and reaction tests, and other places where moving images played a role.56 Now, though, the operational aesthetic is focused on different sorts of scientific sites such as climate and environmental sciences and Earth observation but that same lineage is still articulated in relation to its own conditions of existence. In this case, it is about data and logistics.

Logistical Images Geocinema’s depiction of indoor places like the Sri Racha ground station is full of stillness. Monitors next to monitors occupied by no one in particular, computer fans humming, lulling the already passive room to sleep; “Idle staff keep watch over labouring machines as they download petabytes of raw data from satellites and atmospheric sensors.”57 The scene is preceded by a more official note on the interoperability of data, then cut into scenes populated

Operational Aesthetic   163

FIGURE 23. The Making of Earths, 2020, Geocinema. Still from a video. Courtesy of the authors.

primarily by screens and monitors; in an alternative take that was not part of this installation version for the ZKM exhibition, the camera lingers even longer in the sleepy room of monitors humming, emphasizing a disconnect between labor that is visible, and labor that is distributed across automated systems and their maintenance. Operational aesthetics that engages with distributed systems of such invisual scales becomes a different thing when faced with proxy-­hopping that tells a story of infrastructure. Here, proxy-­ hopping is meant to indicate this set of sites, infrastructures, interviews, people, papers, screens, images, and others that define the logistical operation that is DBAR—­not that the film is meant to show what DBAR is. In my view, it is meant to show what operations it consists of, fragments of moments where calibration and synchronization, technicalities, and assumptions occur. The mundane, even boring scenes of control rooms feature military operations rooms with experts plotting logistical movements of troops (enemy and friendly) or remote control interfaces from space flight to drone strikes, smart cities to vehicles on Mars. Some of this coalesces with the “ecology of operations”58 of radar as coined by Bernard Geoghegan, but it also stretches further than the specific screen and logistical technology of radar. This is one aspect of the twentieth century lingering on to twenty-­first-­century digital

164  Operational Aesthetic

smart cities and their architectures of governance, even if the implication of one command center observing across a space—­like panopticon—­is as such quite mistaken.59 But the spatial scenes of operators and operating interfaces take us to Farockiesque themes of labor, knowledge, and control. Observation is tightly regulated and full of standardized ways of behaving to avoid missing a thing and knowing what to report.60 Not much that is visible necessarily emerges at all. In a project conversation I had with Geocinema, they pointed to a historical document, written in the late nineteenth century and archived in the Zikawei Observatory in Shanghai, on instructions given to a weather observer: how the body must comport in order to observe and annotate the weather reliably. The historical document comes out as an anecdotal archival find, but it started to gain gravity in its own right as it was set against the other themes of observation in the advanced networks of data and labor. It formed part of a discussion about the operative chains of standardization and regulations and of calibration and interoperability as part of the logistical nature of observation, which is never solitary even when done alone—­the paradox of operating as part of large-­scale systems, centralized or not. Sometimes operators of operational images turn out to be monitors themselves, on standby mode, performing labor that itself is invisible only due to its seemingly passive nature. Action-­packed images—­documentary photographs or fictional descriptions—­a re often different from the nature of labor in the operational aesthetic (think of the earlier-­mentioned Ehmann and Farocki project “Labour in a Single Shot” too). While the images might have a similar quality as fiction—­to synchronize multiple data points to feed forward such synthesis toward potential actions—­the action itself is out of sight and distributed across a multitude of geographical regions. Earlier I mentioned in passing the notion of the logistical image, but I want to return to it briefly as it reveals in more detail these points outlined above. Toscano and Kinkle’s use of the term that stems from work by Allan Sekula, Fredric Jameson, and others connects questions of political cartography to visual artistic practices that are less about contemplation and more about understanding

Operational Aesthetic   165

the relations and abstractions of class, social reality, and technology: “The logistical image—­whether the particular domain of logistics is military or commercial is of minor importance here—­is to be considered primarily in its informational functionality, as an element in a concatenation of actions, or in a flow.”61 The notion hovers between such images, still or moving, that help to understand the vast sphere of logistical operations and are themselves conditioned, dependent on logistics. The notion can be best described in relation to cognitive mapping as per Fredric Jameson: What are the possibilities of mapping the scales and abstractions of capital in relation to lived worlds of experience? The two are of fundamentally different scales of reference and somehow implied in each other. What forms of aesthetics—­a nd cartography—­a llow us to visualize this in ways that stay true to the reality of abstractions while understanding the political necessity of maneuvering in such a peculiar space, in such a peculiar definition and control and framing of bodies and territories? As Toscano and Kinkle point out, cognitive mapping emerges from urbanist Kevin Lynch’s The Image of the City (from 1960) that addressed questions of urban “legibility” to its inhabitants. These points then develop into Jameson’s political question of what navigation means, and further to a discussion where the logistical image is both defined by logistics and tries to make sense of its own conditions of existence. To quote Toscano and Kinkle: A consideration of the logistical image can thus open onto an aesthetic and political inquiry into the conjunctions between circulation and abstraction, the traffic in photographs and their abstraction from use, the role of images in logistical flows (military, productive, financial) and their modes of exchange and commensurability. Attention to the traffic in and of photographs, to their integration into logistical apparatuses of production and destruction provides a critical counterpoint to the lures and impasses of images and representations of “traffic,” of the circulation of goods and people.62

This perspective, resonating with the point about invisual operations as a particular kind of logistics (in and through platforms)

166  Operational Aesthetic

also underlines the need for nonrepresentational accounts of images. Instead of merely reading the applicability of concepts to audiovisual practices, the inverse relation can be considered: How does Geocinema’s Framing Territories project and its installed versions—­as one example among many that could be cited—­help to understand the contemporary operational logistical image processes? It becomes useful to consider it as one take on the logistical image where the primacy of operations becomes a way to frame the image as nonrepresentational. Their images of invisual processes help to understand the platformatting of territories across a vast distributed space and a longer historical period of the emergence of remote sensing; especially when their moving images of DBAR infrastructure of tech and people, process and administration, are read in relation to their research narrative, the point about logistics and geopolitics becomes clear. The earlier note about remote sensing and inventory of mineral signatures follows through into contemporary projects where the “visualizing of land and air” becomes a formative element in DBAR’s Digital Earth as a cascading set of logistical operations based on past data and future predictions. As Geocinema’s Asia Bazdyrieva and Solveig Suess elegantly outline, the commercial underpinnings of DBAR’s sensors and future-­modeling data are a central part of the project: Clients here range from insurance companies to real estate developers, and those constructing projects such as ports, railways, or economic corridors. With the underlying claim that further ecological certainty will be possible from the harvest and interpretation of datasets from atmospheric particles, “big data” becomes a mode of sociopolitical and economic ordering. These migrating modes of governance are trained along “mining of implicit and non-­obvious patterns, rules, and knowledge behind big data” made possible with financial promises and more practically through to standardization of scientific protocols. . . . Translating these datasets into possibility curves, statistical future projections, and graduated maps, environmental risks are spoken into the vocabulary of GDP, political negotiations, and urban management.63

Operational Aesthetic   167

Such a logistical image—­one where nothing visible necessarily happens but a massive concatenation of actions is implied—­is at the back of the conjunction of governance, value, and invisuality. Hence the operational aesthetic—­the logistical image and the aesthetics of cognitive mapping, the processual and its interplay of visibility, invisibility, invisuality—­at play here refers not merely to a geographically distributed “this is how it works” sort of show and reveal but to the difficulty of revealing anything. Indeed, the notion of invisual is again central as such infrastructures are also platforms that can integrate multiple logistical operations and scales of activity and spew them forward. Furthermore, in this case, it is glued in a particular style of Digital Earth’s holistic rhetoric and relation to environmentalism such as natural hazard prediction and management alongside agricultural management, observation of environmental conditions, and so on. The discussion could also be expanded from DBAR to other platforms of environmental data management. Geocinema responds to the assemblage of signal, data, and their architecture in two ways: both in relation to the space where those operations take place (indoors, outdoors, construction sites to board rooms) and in relation to which spaces are being framed and managed, both now and in and through predictions of a multitude of possible futures.64 Operational images are bureaucratic procedures, but they cascade on multiple material levels. Indeed, as operational aesthetics of invisual platforms and data, this form of a logistical image of infrastructures responds to Sean Cubitt’s neat summary: “The purpose of data is neither to know nor to picture but to manage and predict,”65 which itself leads to the temporal underpinning of the framing of territories as data. They are contemporary in the fundamental sense of the term,66 occupying multiple different temporalities where past patterns are aggregated into datasets for modeling futures and for prescribing what futures are perceived, captured, and reinforced. This is one way to understand contemporaneity (as per AI and machine learning); another one is the accompanying aesthetic sense of diagrams that visualize, and sometimes propose or even impose, futures. It can be a nightmare to be caught up in someone else’s future, even in their image of one.

168  Operational Aesthetic

Territorial Prediction Forecasting plays a key role in the operational aesthetic presented in moving images by Geocinema. It occupies one part in outlining the multiple temporal levels of the aesthetic of images—­both operational and framing them in Geocinema’s work. In the lineage of the visual and cinematic arts of logistics such as Allan Sekula and Noël Burch’s film The Forgotten Space (2010) about planetary ocean transport logistics and capitalism,67 Geocinema’s work attempts to synchronize its own images to deal with infrastructure, logistics, and data in ways that are beyond representational. The sequences feature the process of images being born, resonating as cosmographical inscriptions like the multiple earths and globes, maintained by labor, actuated in a variety of paperwork and administration. As explained above, forecasting features both as a conceptual figure in their research narrative pertaining to DBAR and as one platform of management and prediction. In Bazdyrieva and Suess’s words: “With Earth Observation data annexed to capital, techniques of measurement are inseparable from infrastructures of governance, where interpretations of how the earth is changing are feeding back into political modeling across the Belt and Road.”68 This feedback loop, a version of what I have in other contexts coined as “medianatures” to imply the recursive, cascading, complex, logistical continuum of media and nature,69 characterizes how particular material epistemologies of knowledge and management pick up on a prescriptive material force when they hit the ground. In parallel to Geocinema’s project, artist-­researcher Abelardo Gil-­ Fournier’s work on territorial prediction as image prediction speaks to a similar set of concerns with other means and artistic methods. In his approach, framing territories is approached through geographical datasets and a next-­frame prediction technique that become both a tool of creating imaginary landscapes based on historical data and working with time-­lapse imagery to investigate the invisual process of territorial forecasting and prediction. Here, the work is methodologically interesting in how it pairs up with Geocinema’s project and other recent works that engage with AI imaginaries of “fake” geographies.70 Even if coined “fake,” these fabricated image worlds are central to understanding real and specula-

Operational Aesthetic   169

tive questions of territorial management: a particularly apt example of a recent art project in this vein is Asunder (2019/2020) by Tega Brain, Julian Oliver, and Bengt Sjölén. Asunder pulls satellite images and runs them through the Community Earth System Model (CESM)—­a climate model—­to simulate possible futures for selected regions from Silicon Valley to Dubai to arctic regions. Without being able to go into further detail about Asunder, it connects to our discussion here by focusing on creating images that have a particular relation to material transformation—­even terraforming, of sorts. Responding to the discussion on platforms and invisuality in chapter 2, it also creates simulated scenarios through images that reformat territories. “The result is a fictional ‘environmental manager’ that proposes and simulates future alterations to the planet to keep it safely within planetary boundaries, with what are often completely unacceptable or absurd results.” 71 Gil-­Fournier’s “Surface Value” workshop72 at the transmediale festival in 2018 engaged with a closely resonating register of “absurd results” of automated systems. Surface Value engaged with the geographical, the geological, and the planetary surface from the perspective of aerial vision since the early twentieth century.73 He applied these approaches to contemporary art and design practice contexts with geographical datasets and technological images. Such images that extend from a representational status to operational functions and invisual applications as platforms become essential in the mediation of territories. Gil-­Fournier linked some theoretical discussions about operational images and surface forensics with a crash course on machine learning. He offered a tutorial on the basics of machine learning, from installing the necessary software and plugins to a discussion of the Google Earth Engine platform. While the workshop was focused on data and geographical formations, the method was to think through images: to investigate the time-­lapse as one technique of creating sequences out of historical changes in a particular selected region and then feeding it forward to possible sequences that might follow. The task was to perceive how images can take up a life of their own in an AI version of Paul Klee’s note: a line (is a dot) that went for a walk. In practice, it meant treating large-­scale

170  Operational Aesthetic

territories and regions as sequences of images and image-­like features where “change” is prescribed (in a rather blunt way) as a description of aggregated patterns across historical datasets, projecting future changes accordingly. Some examples of the large-­scale use of datasets mostly deal with different monitoring and research uses, focusing on forest-­ and water-­coverage changes, agriculture and urbanization, and animal habitats. With available satellite data and relevant Application Program Interfaces (APIs), projects such as Global Forest Cover Change and Disease Surveillance and Risk Mapping are good examples of how data is used for prediction and where satellite imagery participates in the operational definition of territories.74 Gil-­ Fournier’s focus was to investigate these techniques as test cases concerning the media operations that can be mobilized to speculate with imaginary landscapes and reflect on those (image/data) tools and techniques of territorial prediction. In many ways, it was also an experiment with invisual image practices. They look like geographic territories, and yet they act like digital data operations. In Gil-­Fournier’s brief, the participants were tasked to address the planetary images as video images, as moving sequences of description and tracking of change across a geographical surface. As the workshop text explains: Next Frame Prediction names the experimental set of Machine Learning techniques aimed to algorithmically produce the frame that follows a video stream of images. This approach to video-­prediction makes use of convolutional neural networks that, relying on large databases of videos, are able to identify temporal patterns and behaviours within a sequence of frames. This analysis provides them with the capacity of generating future movement and, in consequence, of extending videos with plausible futures. This project is conceived as a platform to research on the application of techniques of video prediction to sequences of historical satellite data. Starting from the available timelapses of archived aerial images, this platform aims to generate future visions of the surfaces of the Earth where, flattened as video sequences, terraforming activities such as deforestation or ur-

Operational Aesthetic   171

banisation might be scrutinised as visual feeders for predictive algorithms.75

Gil-­Fournier uses next-frame prediction as an example technique for the production of ensuing fabricated images. This technique mobilizes video ­prediction and convolutional neural networks as forms of temporal pattern creation that starts to resemble simple abstract forms on-­screen, which can (still) be connected to the operational sense of an image: they are meant to speak of plausible futures (urban growth, deforestation, or any identifiable surface-­level change). As such, the link to Geocinema’s investigation of DBAR and Asunder’s imagined landscapes of an automated “environmental manager” are two points of resonance for such a practice, where next-frame prediction becomes de facto, a fabulated next territory prediction. There is already a range of operations explicitly mentioned in the workshop description: aerial images and flattening (as one part of imperial power), terraforming (as an aspect of contemporary geoengineering but also including speculative proposals beyond geoengineering),76 predictive algorithms (as mediated governance), and then, of course, video too: a technique of temporal modulation but now also data visualization and simulation. Video becomes part of the discussion of artificial intelligence and data; terraforming opens up as visual activity. The assemblage concerns the operational loop between visuality, geological operations, urban developments, landscape changes, and what can be called the planetary-­scale image factory. Gil-­Fournier also asked what sort of mechanisms create plausible futures that emerge from the historical set of data about geographical and geological patterns, urban and nonurban development, and the surfaces through which these are analyzed. However, before one rushes to answer, it is the question or definition of plausibility that is at stake. What form of criteria, institutional decision, or rationality decides on plausibility concerning something that by definition should be contingent? In this case, plausibility—­itself by design a rather vague notion unless specified into an institutional work of standardization of expectations and parameters—­can be seen, for example, as part of such futures that are useful for the

172  Operational Aesthetic

futures market and various other value (capture) mechanisms that work with immaterial/speculative assets based on access to data and how to execute the processing of data in relevant ways. Besides its use for environmental monitoring, prediction stands at the center of various techniques of speculation that tie together contemporary institutional operations and modeling of futures. There is a long history of how speculation becomes a technique to pin down futures—­firmative speculation, as coined by the “Uncertain Commons” collective.77 Speculation grows into a multifaceted discourse of potentials and imaginaries and their capacities to be enacted. Furthermore, the predictions based on machine learning readings have also become a service (even if this aspect works at a different scale than that of Gil-­Fournier’s experimental small-­scale workshop). In other words, the mechanism for this data-­landscape ­prediction is already becoming fine-­tuned as one aspect of input into the turbulent financial machine, the data-­service industry, and their links to different operations of planning, construction, geopolitics, and data platforms. Melinda Cooper argues that the sort of futures methodologies that we recognize from climate change and other modeling and simulation services share a temporal emphasis with finance: “the multiple future worlds attendant on alternative actions in the present.”78 Actions feed forward to a future, or a few futures, that feed(s) back. Such techniques map and measure Earth’s surface changes while participating in transforming those surfaces: the images are potential avenues to intervention—­whether environmentally focused policy changes or capital investment that terraforms regions in its own image. Multiple economic, technological, and data-­ intensive feedback loops complexify this picture. Indeed, as Sean Cubitt noted in a different context, relevant here too: “The concept of environment is then an abstraction that nonetheless acts as if it were a reality.” 79 In some form of an indirect response, such projects as Asunder and Gil-­Fournier’s work(shop) show that environments are calculated as abstractions that then take place as reality in the operational chains of territorial planning, decision-­making, and reformatting. As art and speculative design projects, they also investigate the invisual side of logistical images as one form of the operationalization of futures.

Operational Aesthetic   173

Operational Aesthetic: Take 2 If our first take on the operational aesthetic stemmed from a historical focus on questions of procedural framing circa mid- to latenineteenth century (whether real or faked, a trick or actual workings of a mechanical process), what does the above discussion of contemporary media artistic methods imply? The operational aesthetic that spread across vast geographical distances in the wake of such remote communication technologies as the telegraph already implied a peculiar relation of contemporaneity as synchronization of bodies in relation to absent bodies: we are in touch across distances, and this implies a particular aesthetic of operations across distances too. But as Christina Vagt notes, “Computer simulations belong to a long history of action at a distance through models but also through concepts,” which opens up a slightly different lineage to consider “aesthetic procedures that create their own specific objects of study.”80 The vast planetary-­scale infrastructures such as DBAR present even a more complex set of demands for an operational aesthetic that needs to engage with the logistical image in ways that map spatial and temporal distances while engaging with the data w ­ orlds of simulation and modeling. The reaching out to futures—­the forecasting—­is one case of techniques of speculation that is fully integrated into financial, environmental, and geopolitical mechanisms of working with (remote sensing) data. This operational aesthetic has also moved from the fairground (P. T. Barnum) to the battleground (Control and Command), if one follows a cue from Jimena Canales’s take on operational art, not only as it is used in military parlance but also with implications for contemporary art and visual culture, including Farocki and Paglen.81 Canales narrates the “not what but how” of the operational aesthetic through a scene from Dumas’s The Count of Monte Cristo (1844) that recounts the pros and cons of tele-­technologies—­namely the optical telegraph. Quoting Canales: The Count of Monte Cristo, in one of his elaborate plots to accumulate wealth and sink his enemies, went to visit a telegraph—­a technology that during the first half of the 19th century worked by transmitting light signals across long distances. “I have been

174  Operational Aesthetic

told,” Monte Cristo told the humble operator, “that you do not always understand the signals you repeat.” “Certainly, sir; and that is what I like the best,” the operator responds. “Why do you like that best?” “Because then I have no responsibility; I am a machine and nothing else.” Upon hearing those words, Monte Cristo recognized a moment of opportunity and seized it, bribing the operator into repeating signals that he did not understand. The signals, announcing that King Don Carlos, who had been imprisoned in France, escaped from Bourges and returned to Spain, caused Monte Cristo’s enemies, holders of Spanish bonds, to lose a million francs. Reflecting on his success, he exclaimed: “I have just made a discovery for 25,000 francs, for which I would have paid a 100,000.”82

The operational aesthetic turns to cunning ends as a manual of manipulation of knowledge by means of plots and tricks. Money can be gained, but also other things. The plot defines one form of a trick that features as a central design element in unfolding the logic—­even logistics—­of technological operations.83 Canales’s discussion focuses on the links between Trevor Paglen’s photographic art as a version of the logistical image—­for example, the work on CIA rendering flights as images show logistics in operation, organizing space, knowledge, secrets through an infrastructure of people and planes and schedules. She also links this work with a broader discussion of operational art as it relates to a military discourse of Operations Other Than War, somewhere between tactics and strategy. In contemporary technical terms, that sphere is one of the technical operations taking the place of subjective agency, evaluation, and decision-­making, bringing operational art into an aesthetics of images that have the capacity to concatenate actions. It could also be seen as part of the already mentioned history of Operations Research as it defined the mobilization of specialist scientific analysis and teamwork of experts from military operations to society at large: problem-­focused observation and analysis that was to aid executive decision-­making in firms and organizations. This becomes integrated into different forms of scientific management during the Cold War period, and it becomes one aspect of “Think Tank Aesthetics” from the Ops Room to the modernist art of “pat-

Operational Aesthetic   175

tern recognition,”84 radically expanding the regime of the image too (as argued earlier in this book). Beyond modern art and architecture but not disconnected from it, Operations Other Than War is the sphere where many operational images are said to take place and where art has a double meaning, both of military craft and artistic practice. Pattern recognition persists as an art of detection, observation, and image ­analysis beyond the gallery or the studio. As a version of the how, not what that relates to the operational sphere framing of territories as aesthetics/art, it is visible on two interconnected layers: operations that imply the centrality of images—­of technical and even automated images—­as one center of agency; and then, the contemporary media art methods that bring together cinema, video, photography, and digital techniques that make sense of such logistical procedures and their mass impact on territories. In outlining this argument, I want to add that this chapter’s discussion is not a full-­fledged definition of art or the aesthetics of operational images. The survey of artworks and practices is not meant to be complete. But the chapter has shown the various strands of concern that relate to questions of process and image, spatial framing and infrastructures of data, labor with data, and how this labor engages in the logistics of images. Next, I continue this discussion through urban traffic, where operational images continue their navigation through the city, mapping the urban realm (and beyond) as a dynamic sphere of light—­a different site of invisual culture than what is implied in traditional photographic terms. Some of the themes in this chapter carry on with a discussion of technologies of lidar and other practices of light that feature another take on the nonrepresentational operations of images. The city takes center stage as it is driven through by autonomous vehicles and is defined by invisual images: sometimes seen through interfaces and sometimes automating the traffic of cars, data, and people.

This page intentionally left blank

FIVE

The Post-­lenticular City Light into Data If one takes the images seriously the way Farocki does, the distinction between aesthetic and information value recedes. The question rather becomes what these images do and what can be done with them. If there is a metaphor that anchors these images, it is that of circulation and transport itself; a concept that acts like a gravitational centre and that applies to the city’s infrastructure as well as to the traffic of images circulating. —­Volker Pantenburg, Manual: Harun Farocki’s Instructional Work

Urban Light A fitting opening scene for this chapter: The sky over Shanghai was illuminated by a coordinated swarm of drones, drawing an image. Such marketing stunts had been used in the past, several times already, but this image was a QR code, an executable “image” hovering over the urban landscape. Part of a marketing campaign by the Chinese media and video-­streaming company Bilibili, the QR code was linked to a new role-­playing game Princess Connect! Re:Dive.1 As advertisements, the rewritten sky is one version of a much more low-­tech driven point that Lawrence Grossberg made about interstate highway billboards in the 1980s: billboard ads relate specifically to the highway logistical system of traffic and, as such, to their primary task of advertising, which is why a billboard is “not a sign to be interpreted, but rather, a piece of a puzzle to be assembled.”2 It 177

178   The Post-­lenticular City

is also why to look at the sky as it transformed into a puzzle and assembled into patterns and the interpretation of “what is that” turns quickly into “what does it do” as you point the mobile phone to read the screen of a sky. For Grossberg, the point relates to how one reads popular culture without prioritizing what it “means” in the traditional sense: “I want to suggest that interpreting the effects of popular culture, and its politics, is less like reading a book than like driving by the billboards that mark the system of interstate highways, county roads and city streets that is the United States.”3 The primacy of circulation, navigation, orientation, and clicking through a variety of interfaces—­ readable or executable—­persists even if in a different register. The executable marketing image assembled by a swarm of drones is not like reading a book. It is one form of an operational image that takes on different functions as per the main threads across the book: not just next to a highway road, but in other logistical, infrastructural, and mediated situations, it fills the sky as QR codes now fill up urban facades and other places, marking them as potential spots of executable space that disguises itself as an image. QR codes are a prime example where the operations of the city are amplified and channeled through machine-­readable surfaces. Hence the Shanghai stunt is perhaps best read as a way to realize the radically increased read/write surfaces that define a city. As N. Katherine Hayles has argued, QR codes are part of the broader augmented reality of the city as such image markers that “unlock gates to physical locales.”4 Christian Ulrik Andersen and Søren Bro Pold coin a similar pattern as the urban metainterface that “semiotizes” space anew through, for example, app-­enabled property regimes such as Airbnb, Uber, and TripAdvisor.5 However, whereas they speak of the city as text, I refer to an operationalization of the city as a material process of imaging that triggers multiple levels of non-­textual processes (often referred to in generic terms as “automation”). Either way, such arguments about reformatting the city underline many of the themes in this book, both about action and execution (as per the briefly earlier mentioned themes in software studies) and about digital forms of control, including property control, where images are central structuring elements in the sphere of environmental computation in and out of the smart city.6

The Post-­lenticular City   179

Images are temporary roadside stops for attention in a circulation of information gathered as data for further modeling and execution. They are also sites of data aggregation (detection and capture) or channeling traffic (opening gates, executing remote orders, etc.). Contemporary scholarship also has to observe how those networks are embedded in forms of architecture at the back of the work on the networked image, where the snapshot is in circulation, and the image appears on digital screens and, even more so, as invisible (or invisual) data on platforms.7 Sometimes referred to as augmented reality8 and strongly resonating with N. Katherine Hayles’s analysis of the nonconscious cognition of automated technical systems,9 it relates to the multi­scalar processes where images play a part in control and mobility, actions, and guidance on and off actual image devices. Here, the (photographic or other) image is distributed, and the whole operational apparatus is, in fact, one interlocking recursive system. The two previous chapters have paved the way for this one. Chapter 3 focused on the measurement-­image in practices of reading territorial planes, and image planes provided a way to move on to the question of the operational aesthetics of large-­scale surfaces: operational images featured in photographic terms, technical drawings, and more recent remote sensing and machine learning techniques that both analyze and synthesize geographical areas as territorial planes and data. The planimetric management of the planet as an image ran through the discussion built on the argument of operational images as navigation, as processual aesthetic, as a technique of infrastructural management. The operational aesthetic proposed in the previous chapter addressed a method of training related to scales and infrastructures of images and what their institutional role tells us about their territorial situations or the sites of operational images. In this chapter, we build on other creative examples that can be read in relation to what the operational aesthetic implied: a process-­ focused, dynamic insight into such images that have a fundamental impact on modeling the world and that help us to understand the links between image and infrastructure, spatial data modeling and its different links to devices. In other words, we move from the rural worlds and earlier colonial expeditions to operational images

180   The Post-­lenticular City

to contemporary networked images and urban events coined as “operational city” recently by Mark Andrejevic (a term he uses to discuss “smart cities”).10 We will build on earlier notions of platformed invisuality, navigation, and traffic, as well as images as data points, while using selected design and artistic practices to define contemporary images’ broad range of operationality. Light is a crucial reference point insofar as it relates to the techniques of switching from visible light to invisual data, which is the operational element distributed across surfaces, vehicles, and other relays of action and executability. Cities are and have been full of light and images, and the urban sphere of light and technology has been a vital part of the emergence of a contemporary link between media and subjectivity. Photography brings things and events to light, and the city is itself full of light and movement, something captured in writing, theory, images, and films from Siegfried Kracauer, Dziga Vertov, and Walter Ruttman, among many others. One such take is Farocki’s on the operational, logistical city of transport in Counter Music (2004) as well as the interior space design depicted in Creators of Shopping Worlds (2001)—­in essence, a film about data, urban space, design, and affective capitalism. Early twentieth-­century writing on the nexus of media, space, and the city focused on the rescaling and shifting perspectives, a spatiotemporal reorientation in views deemed phenomenologically impossible at an earlier point. The city was one site for dynamic transformations of speed, perspective, scale, and experience. Technical media thus feature not only in the scalar operations that brought the invisible to visible manipulation but also across the city as an audiovisual (design) lab of its own. As per Kracauer: “For the first time, the inert world presents itself in its independence from human beings. Photography shows cities in aerial shots and brings crockets and figures down from the Gothic cathedrals. All spatial configurations are incorporated into the central archive in unusual combinations which distance them from human proximity.”11 It was cathedral facades, photographs, and photogrammetry that featured in the chapter 3 discussion about Meydenbauer, who was interested in a “central archive” of the cultural heritage of architecture—­of things measured and recorded as data. Here, in the technological city, though, data and signals take precedence. From them,

The Post-­lenticular City   181

one also reconstructs the city as it has happened, as it is happening, as it will (perhaps) happen from capture to prediction. Similarly, as the whole book is concerned with the various sites and situations where images and data switch into each other, here too the city is a place where light becomes an image, an image becomes data(set), a signal becomes a visual model, and a model becomes a prescription of the operational elements of the city from the control rooms to the distributed computational events. There would be two kinds of ways—­or questions—­that define what I am after in this chapter. The first one would be to ask a question that sounds peculiar: What kind of light defines the city? Scholars of urban vernacular photography might have their response already in mind; theorists and architects investigating urban lighting since (at least) electric light and modern media technological spectacles of the neon city might have theirs; artists defining facades through urban projection practices might have many ideas to contribute. But here, the focus is on a particular light-­pulse and laser-­scanning technology of lidar that triggers computation of the city as a massive distributed light event: it sends either near-­infrared laser light pulses or “water-­penetrating green light,” and the returning signals are captured with a GPS receiver.12 The second ensuing question would be about the computational images that follow from the realization of this particular kind of light that becomes scannable, data-­prepped, and formatted, which relates to the broader machine-­readability of the city. How is light operationalized in laser scanning, as data and modeling, and thus also then the platforms that replace visible light? Here, the question of the visible or invisible is again retuned to the invisuality of those platforms on which seeing is distributed even beyond singular machine vision technologies. Designer-­artist groups such as ScanLAB also feature in this chapter, presenting their own terminology through visual imaging practices, coining the “post-­lenticular” landscape. Visual theory and photographic studies have only begun to deal with the discoveries of the early nineteenth century, such as the infrared (by Frederick William Herschel) and the ultraviolet (by Johann Wilhelm Ritter) as visual realities that form an alternative genealogy to the technical media of the past two hundred years.

182   The Post-­lenticular City

This points to an alternative genealogy of machine vision and remote sensing, where different light spectra can be registered and used in different institutional contexts (see Figure 3 in chapter 2). The implication is this: tell me which spectra of light you are interested in, and I will tell you what your discipline is. What are the inroads for critical humanities and digital culture studies to these other realms of light, these other realms of seeing and detecting that have become far from marginal? Hence, mapping such a variantology of media is still an interesting task to understand what else, other than just pictorial images, has been at play in scientific practices of images.13 In addition to computer-­generated images, data, and measurement, they constitute forms of nonrepresentational imaging.14 But, of course, the two lineages are not separated, even if photographic theory and history have tended to focus more on capturing light through things we call “cameras” instead of, for example, sensors.15 In more specific terms, following Jennifer Gabrys, we can continue this line of thought about invisual culture premised on sensor data: These shifts in the practices of processing image data as sensor data are productive of sensor environments that create distinctly different engagements with imaging, not necessarily as an a priori fixation of visuality, whether as an epistemological or disembodied register, but instead as a processual data stream that irrupts in moments of eventfulness and relevance across data sets comprised of multiple sensor inputs.16

The interplay of images and data is thus a problem specific to the history of science and new forms of imaging. It becomes relevant to how images are institutionalized in data culture across different realms of property and governance: to own and guide, navigate and control. Sensor instruments might not always create images, but even data can be quickly converted into images—­and vice versa.17 Furthermore, this is not merely a question of traditional images that depict a thing or two but the platforms that constantly read and write space and produce data clouds and visualizations, models and mediations as operational “guidance”: this is how you should see

The Post-­lenticular City   183

and perceive, this is where you go, this is how you go there.18 Commercial augmented reality of indoor spaces is one example of the operationalization of space as an invisual model that takes the platform model as its modus operandi. From actively taking pictures to constantly sensing the environment, for example, smartphones are thus an active part of modulating the space in ways that incorporate those key elements of platforms tackled earlier in this book. In this vein, Google’s earlier Project Tango would be a property operation of modeling that works as a Geographical Information System (GIS), expanding what we think of as images.19 Following Tango, the ARCore platform (Google) builds a possibility to read/write the world through motion tracking, environmental sensing (detecting “the size and location of all type of surfaces: horizontal, vertical, and angled surfaces like the ground, a coffee table or walls”), and light estimation.20

The Pulse of the City This chapter also continues earlier themes of the book: images are also understood as part of the larger subset of “measurement,” whether this concerns photogrammetry (see chapter 3) or pattern recognition (see chapter 2), which form a fundamental background for the contemporary context of platforms and AI. From the singularity of an image, we shift to the mass image as it concerns the database and the dataset as infrastructures that rewrite the urban and nonurban space.21 A persistent and yet peculiar dilemma that haunts so much of Farocki’s interest in the “standard image”22 of contemporary culture: Why are there still images in computational culture? Why do we see anything? Hence the two hundred years of machine vision 23 is an assemblage that has prepared some responses for this question that was asked so many times in the various operations and instruments of making images: What kinds of surfaces record visible light, what surfaces record invisible light, what forms of rays travel as straight lines and which ones are bent, and what is the form and format of aesthetic that fits this expanded sense of the sensing?24 What are shadows in images, and how do interplays of shadows and light turn into data that can be transported across distances as photometric tables or other forms of operational images?

184   The Post-­lenticular City

Measurement of the intensity of light—­photometry—­forms a less-­discussed part of the visual–­invisual spectrum, but the point is clear: light is not merely there for us to see but to make things seen and measured. From photographic photometry as a technique central to astronomical and other scientific measuring to the use of experimental techniques for measuring the higher atmosphere, detection is a central part of this sensor-­oriented way of understanding practices of light.25 For example, already in the 1920s and 1930s, certain kinds of “scanning” used searchlight beams and photoelectric measurement of returning radiation to establish scientific models of the sky.26 While searchlights have a central part in the military history of sensing,27 this work demonstrated the awareness of creating such artificial “scenes” of light that were not merely a Nazi spectacle of the 1930s but illuminating the sky so as to measure otherwise imperceptible wavelengths of light. Edward Hutchinson Synge’s pioneering work is often mentioned as an early form of lidar imaging and thus “laser” scanning; the link is mostly due to work developing theoretical ideas and design plans for telescopic devices and tapping into wavelengths below visible light.28 This capacity had already been part of some practices in astronomical imaging, too, with chemical photographic images being able to capture more than human visible light in some cases. Hence the task was not only making visible that which was at stake but also detecting data from the wavelengths and other realms through extended practices of light. Insights into making visible are part of how we relate light to spatial contexts of illumination—­real and metaphoric—­but some of the operations of light are very much bound to their temporal underpinnings evident in techniques such as the flash and the pulse. It is in temporal conditions that things are perceived. Both distance and scale became crucial in techniques of working with light in many kinds of ways that extended the geographic measurement and developed new forms of photogrammetry. We can track a particular period of invisual culture reliant on, first, visible light and its dynamic signal-­based capacities (and detection capacities) that lead to a genealogy of signals and pulses of light and sound; and second, contemporary sensor and data technologies that map the city as one intensive, complex landscape of dynamics and navi-

The Post-­lenticular City   185

gation. This same genealogy links to research into light waves and far-­away planetary objects by Christian Doppler in the 1840s, with the later (1920s, 1930s) realization that the physical world is a (often light-­emitting) broadcasting station of sorts.29 Later in the 1930s, the invention of radar as the technique for synesthetic transformation of sounds to images foreshadowed the recent discussions about lidar imaging and autonomous cars as a light echo the pulse of the city. Alternative contexts of sensing or seeing, with a specific focus on technologies of pulse and light, have emerged as one key type of time-­critical and posthuman visual measurement and observation.30 Of the multiple technologies of pulse and light, lidar stands out as one of the most discussed examples in architecture and urbanism. It has become a widely used technique across a range of fields in scientific measuring, including architectural modeling. Since both professional-­level lidar and the scanner have been incorporated into iPad versions since 2020, the technique allows geospatial data production for building information modeling (BIM), surveys, and contexts where accurately measured 3­D images are needed. Thus from 1960s aerial surveys to more recent specialized uses like “land management and planning efforts, including hazard assessment . . . , forestry, agriculture, geologic mapping, and watershed and river surveys,”31 it has become a technique that is a part of the shared lineage discussed in chapter 3. As measurement structure and shape, it is used for surface and atmospheric particles and the built environment. Lidar might be of a different scale than techniques of cartography such as GPS, but it is part of the shared, broader aim of making “space legible and governable”32 through accurate measurement. Here, knowledge of location and features is related to potentials of action—­thus such operations of “actionable geographic knowledge” became detached from mere national projects of mapping and integrated a new transnational, multilayered sense of space as an affordance. In short, and as William Rankin puts it, the territory is “defined by practices of knowledge,”33 a theme not unrelated to the earlier discussion on geodesy and photogrammetry too. However, if in that discussion (chapter 3) the issue was how to model and understand the path of light rays that can then help to measure space in

186   The Post-­lenticular City

and from (also pictorial) images, here the procedure is automated: incoming signals beyond visible light are used to model space in algorithmic software environments. Furthermore, the case of lidar presents an excellent way to look at operational images through techniques of scanning. Beyond semantics, this refers to techniques capable of capturing data and forming quantified, statistical, and reproducible images of territory (a page of a book, a 3D human body, or topographic features of a landscape).34 Lidar imagery is mobilized in technical uses from modeling to scanning, design, and surveys but also in several contemporary art projects investigating the actionable urban space and its matters of sensing. This is machine vision, but not only: it also concerns the visual processing of the world as data and as such concerns the broader computational ecology—­a nd platforms—­of those operations. Furthermore, 3D scanning is incorporated into many contemporary data operations already in place. It becomes defined and redefined in alternating institutional operations. As Mario Carpo narrates, the Microsoft Kinect device for gaming included “an ingenious depth sensor that worked by triangulations, using a laser projector and a camera.”35 The Kinect is, thus, a de facto scanner that then became repurposed from gaming to architecture, too, where the sensor data could be used in CAD software platforms. Here, scanning becomes a way to sense space as time, signals, and calculation with different technical solutions to the issue of measurement. Depth-­sensing technologies are evolving quickly: some use traditional laser or infrared beams, and calculate distances based on the time of rebound (also known as “time of flight”), or, increasingly, by reading the difference in phase between outgoing and returning beams; some use triangulations, like the earlier Kinect machines did, with a laser beam sending a marker to the target and a camera, in another vertex of the triangle, to read it (variants of this method are known as structured-­light depth sensors); and so on.36

In many ways, it is also clear that this form of imaging is not about visuality per se, but about a particular temporalized relation to space, even navigation: these are not images to be seen, but ter-

The Post-­lenticular City   187

rains to orient oneself in whether the navigation concerns vehicles, bodies, or other objects in dynamic interrelationships.37 In this context of images that echo across the city, to speak of the pulse of the city is not a metaphor but a technical description: lidar as light radar is a technology of millions of directed (ultraviolet or near-­infrared) light pulses per second, where the returning signal is then recorded and modeled accordingly. Synge’s scientific work on capturing wavelengths smaller and smaller becomes the backbone of this invisual city. An image of the city, an image of clouds, an image of complex formations; each is measured in front of our eyes and fed into a different atmosphere than the one high up above our heads. Instead, the cloud-­based, platformed computational capacity to capture light becomes a central element of the operational landscape. Consider the city as an already complex formation; it is a pattern of dynamics upon which media technologies attempt to build their sensing, modeling, and imaging networks. Cities are full of signals and light; they are full of cameras and techniques of observation; they are full of sense and sensing, of surfaces that reflect and refract light. The city is continuously being seen, registered, and measured from shop windows to closed-­circuit televisions to motion tracking and remote sensing. However, an increasing amount of the technological seeing and observation that takes place in the large-­scale visual landscapes of the city works to question traditional photographic modes of understanding visual power, introducing different genealogies of what the city is, as an assemblage of materials and seeing, as movement and large-­scale dynamics. The city is thus a perfect test case—­a laboratory even—­for technologies and images that both relate to a genealogy of machine vision and reveal new aspects of it. Ecologies of sensing exhibit a vast multi­scalar complexity that speaks to many changes in human perception; however, it is more likely to be “discorrelated”38—­in Shane Denson’s words—­as multiple levels of events that do not add up into a single take on what is being sensed but instead are an ecology of networked circulation of images and data that prescribe different affordances of action. Even action changes in the sense of what was considered central to photography: instead of an image formed of a singular shot, a constant background sensing models space and action in real time and in predictive time. The pulse operating in lidar

188   The Post-­lenticular City

images is also a good conceptual route to understand the diversity of temporal operations that take the place of the earlier conceptualizations (or build on them), such as networked images. The city becomes a site of posthuman forms of pulsated sensing. This refers to the various autonomous systems that now process things we still call “images” for the sake of convenience a­ nd familiarity. For example, suppose the theoretically tuned media archaeology of the camera has been, until now, focused on the detachment of the seeing eye from both the human body and from the act of seeing.39 In that case, we can articulate how this camera-­eye got fixed on the (autonomous) car, where it sees in ways that are not just seeing but modeling, mapping, measuring, predicting, and a range of other cultural techniques that pave the way for a wider set of infrastructural implications. In other words, the forms of observation now introduced—­ whether through WiFi signals that can model space into an image or then lidar, radar, or camera as measurement instruments—­are part of the operative ontologies. These do not focus on the act of seeing at the site of the device or the body of the perceiver, but in the connected networks where multiple feeds are part of the dynamic formation of an image in real time. Again, the double aspect of operations is at stake: images that operationalize their environment and images that primarily operate and only secondarily represent something, if anything.40 Here is precisely where we shift our question from What does the computer see? to What does the sensor and the scan do in the context of sensing and movement? Any discussion of the digital that obsesses with the isolated ontology of representation based on technical qualities of the image is somewhat misguided—­at least when it comes to the attempt to understand the transformation of images.

Post-­lenticular Imaging “The modern lens is no longer tied to the narrow limits of our eye,”41 László Moholy-­Nagy voiced in an earlier period of experimental practices in the 1920s and 1930s. But now, it might not be even a lens where the capacities of the eye are narrowed or expanded. Even more so, it might not be an organic eye that is at stake. While the

The Post-­lenticular City   189

term “post-­lenticular” could be seen as describing a wider shift in visual technologies and photography, it is also embedded in the title of a work, Post-­lenticular Landscapes (2016–­17), by ScanLAB Projects. In 2016, they traveled to Yosemite National Park to produce a 3D hologram model of Yosemite Valley using laser-­scanning technology. From lenticular photography to laser-­based pulses, from scanning to modeling, the work was contextualized as part of the history of landscape photography, continuing Eadweard Muybridge’s famous Yosemite series of the 1870s and the series of various professional and amateur images produced since. Besides the history of the site itself, the work is situated in relation to the legacy of photogrammetric modeling of geological landscapes where Edouard Deville’s large-­scale experiments “with dry plate cameras within the Rocky Mountains beginning in 1887”42 is one of many significant reference points for the surveys and models. Such references help contextualize where post-­lenticular landscapes fit in the lineage of operations of measurement and surveying. The format of the abstract landscape produced from the trigonometric measurement of the earth’s surface and shape from the eighteenth century to later abstractions of ratios from the photographic image are part of the said operationalization of landscapes as territories (see again chapter 3). While measuring produces possibilities of standardization and comparison—­image relations as data and currency—­ they encapsulate different histories too. These histories are about not only technology but the codetermining ecologies where operations become meaningful as they hit the ground. It is, however, important to note that the location of Yosemite and the history of “wilderness” (produced through the photographic) is part of the troubling legacy that can be considered as part of colonial “extractivism” and also the consumerism of landscapes that erases native inhabitants from the picture.43 And while I remain uncertain whether the notion of “extractivism” can be extended to apply smoothly to operations of data (and images), it is very clear that such practices were also part of a production of the hallucination of a terra nullius of sorts, a violent erasure of indigenous lives and traditions in the midst of particular settler colonial ideals of conservation.44 For ScanLAB, the cameraless 3D scanning of the national park becomes a test case for the technology that emerges from aerial

190   The Post-­lenticular City

FIGURE 24. Equirectangular Landscapes 05: Nevada Falls (After Muybridge), based on 3D scan data captured in Yosemite National Park in 2016. ScanLAB Projects; reprinted with permission.

imaging and large-­scale surveys. The abstract measurement of what is contained in an image frame (or photographic plate) is transposed onto the more advanced techniques that model space to register the returning (invisible) electromagnetic signals. This is specific to lidar and scanning, and it also links to transformations of imaging at large. As John May put it, “All imaging today is a process of detecting energy emitted by an environment and chopping it into discrete, measurable electrical charges called signals, which are stored, calculated, managed, and manipulated through various statistical methods (Bayesian, Gaussian, Poissonian).”45 It seems fair to put laser scanning in that same repertoire of techniques. Laser scanning moves the focus from the apparatus of seeing to the sensors, processing, and infrastructure in which imaging is produced. While a longer discussion would be able to relate it to the legacy of cameraless photography, here we focus on the contemporary contexts of making images that hover between aesthetic sights and epistemic operations beyond lenticular apparatuses. In Peter Ainsworth’s technical summary: Unlike traditional optics-­based camera recording, there is no lens involved in the apparatus. What is recorded at the moment of capture has no fixed application beyond the retrieval of data from a given directional sensor, and the apparatus is not

The Post-­lenticular City   191

solely focused towards image-­making—­it measures distance between the sensor and the terrain in the creation of a cloud of multiple convergent points. These point clouds, however, are often also combined with a traditional photo-­based process, whereby the lens-­based camera is positioned in the same space as the laser scanner, with the images subsequently mapped onto the surface of the polygon mesh—­t he computationally joined-­up point cloud—­in post-­production. However, in the Post-­lenticular Landscapes, we are solely presented with the experience of the cloud—­a hyper-­detailed pointillist rendering of landscape in stark black and white.46

Large-­scale landscapes and urban dynamics become sites and time-­critical events of measuring and modeling integrated into contemporary contexts of the multi­scalar ecology of sense, data, and light. As ScanLAB articulates the use of laser scanning, this shifts the focus from the camera to scanning, but it also returns to “Muybridge’s original endeavour to capture the scenes in three dimensions as stereograms.”47 It also relates to the operational images of photogrammetry articulated in relation to photography across the nineteenth century and early twentieth century. It harkens back to before technical images (primarily drawing but also camera obscura and lucida) and after photography: scanning. Over two weeks, ScanLAB traversed the valley to scan the terrain with the 4x4 vehicle they had turned into a digital base camp to process images. The stereogrammatic view was technically updated, including a particular awareness of the logistics and infrastructure of imaging. In their words, even the transport of “such high tech equipment into a comparatively inaccessible environment formed a major part of the re-­enactment, mirroring the epic nature of the early pioneer photographers.”48 This links to the recurring theme in this book about technologies of imaging as fundamentally about logistics and infrastructure. The focus on the image gives way to the primacy of operations that establish the possibility of any image. This point becomes even more clear in the context of lidar applications of contemporary urbanism and autonomous cars, which, as we know, are techniques of movement and navigation. The traffic of images becomes a literal part of the infrastructural arrangement.49

192   The Post-­lenticular City

FIGURE 25. Post-­lenticular Landscapes. Installation view at Hyundai ARTLAB, Seoul 2018, originally commissioned by LACMA. On view are Urban Diorama (holographic vehicle), 2016, Post-­lenticular Landscapes (4k Animation), 2017, Equirectangular Landscapes 01–­06 (Prints on aluminum), 2018. ScanLAB Projects; reprinted with permission.

However, it also points out that traditional genres of imaging and photography, aesthetics, and cultural history are entangled through the many and speculative uses of new technologies. In such experiments by ScanLAB Projects, photographic and rhetorical tropes such as landscapes of Romanticism shift gear and site. As Geoff Manaugh writes: from “extreme landscapes—­as an art of remote mountain peaks, abyssal river valleys and vast tracts of uninhabited land,” the new Romanticism is one not seen but sensed through “autonomous machines.”50 A romanticism of wilderness gives way to a new romanticism of the platform city. A romanticism of nature gives way to a romanticism of the automated artificial city. ScanLAB’s Post-­Lenticular Landscapes is in direct relation to their project the Dream Life of Driverless Cars, where lidar takes a central role as the standard technology of imaging, processing, and transmitting movement in movement. Embedded vision systems are an integral part of different autonomous vehicles (from cars to drones) and describe the link between the reformatting

The Post-­lenticular City   193

of space through their operational images and the broader data analytical infrastructure. Here, the invisual event of sensing and navigation becomes distributed across different components that make up these computational systems. Several technical solutions such as the “Myriad 2 Visual Processing Unit” have already been mentioned in earlier scholarship as examples of embedded vision systems. As McCosker and Wilken write, “The chip is designed specifically for processing machine vision and integrating data from multiple sensors to allow the device to make sense of surroundings, avoid obstacles or track and follow objects.”51 Myriad 2, Intel Movidius VPU, and the other Graphics Processing Unit (GPU) cases are where much of the transformation of capacities of imaging is taking place. This argument is outlined in detail by Ranjodh Singh Dhaliwal in his take on computational infrastructures reliant on the parallel processing power of contemporary GPUs.52 Furthermore, invisual practices and mobility are integrated on platforms such as those developed by NVIDIA Drive, which consists of multiple hardware and software elements where operational images are not only technical aids but part of the reformatting of urban sensing. NVIDIA Drive Hyperion is a hardware testing platform that consists of “a complete sensor suite—­including 12 cameras, nine radars, and two lidar sensors—­and the Orin-­based AI computing platform, along with the full software stack for autonomous driving, driver monitoring, and visualization.”53 At the software end, the multiple kinds of sensing, perception, and prediction processes are highlighted well: to be able to integrate map and perception data, to be able to predict movement paths and their relation to the vehicle, to monitor the surroundings and the in-­cabin situation of the operator or passenger who also becomes a data point concerning their levels of “attention, activity, emotion, behavior, posture, speech, gesture, and mood.”54 The multi­scalar operations of past, present, and near-­time future are integrated into the different input sequences of which lidar is merely one in modeling the dynamic environment in relation to the networked AI system of the car is where the “operational” stands out, not only as an image but as an environment of synthesis that produces not one accurate representation but the real of variations.55

194   The Post-­lenticular City

The intense demands on the sensor and neural network infrastructure related to autonomous cars are due to the required capacity to process, in situ, a massive amount of incoming data. The environmental dynamics is one aspect of this multi­scalar operational image that is not found on the screen (only) as it is distributed and embedded in world objects. At the same time, the vehicle simultaneously transmits data in real time with the wider platform of which it is part. In this sense, autonomous cars are both an issue of data transmission while reserving a place for images and movement navigation that is not merely the case of here-­now, but the capacity to construct navigable futures via the PredictionNet. What’s more, this concerns a reading capacity and an agent in traffic, a (re)writing capacity as per the feedback loops of prediction systems with the vehicle’s decisions. Back in ScanLab’s city, the world becomes detected in different light pulses; as a scan in movement, it also becomes a recursive image for navigation: As the scanner moves through the city, slowing for speed bumps and stopping in traffic, the city map created warps and extends depending on the speed at which we move. Stuck in traffic a Routemaster bus becomes an elongated, narrow corridor, broken only by the shadow of a passing cyclist. Turning the corner into Parliament Square duplicates Big Ben as we observe the tower for a second time.56

The Dream Life of Driverless Cars offers a ghostly apparition of the scanned city, shifting through street scenes without humans or other organic life, passing by semitransparent buildings with their collective architectural layers. The scanner moves through the city as it observes, tracks, and facilitates the vehicle among the city’s multiple other mobile and immobile parts. As a second-­order record of movement moving, it senses movement within an ecology in movement.57 Movement, sensing, and calculation intersect, pointing to the invisual operational image ecology that reformats the city. City surfaces and neural networks are in computational dialogue. Light exists as traveling energy that can also be data; light is produced so that it can be registered as mass image datasets. Geoff

The Post-­lenticular City   195

Manaugh proposes that ScanLAB’s Dream Life of Driverless Cars is a form of the transformation of cityscapes and visuality, images and data, movement and seeing movement, echoing some of the concerns that Paul Virilio had tracked as a central part of the transformation of visual culture since the early days of modern technical images. Manaugh also outlines the case of autonomous cars as a way to understand the visual change of environments: the use of 3D scanning lidar sensors in autonomous cars relates to a particular navigational way of mapping the city not merely as one image but as an ecology of machine-­flickering signals that captures “extremely detailed, millimeter-­scale measurements of the surrounding environment, far more accurate than anything achievable by the human eye.”58 Such multi­scalar light emissions are premised on a particular ecology of time when millions of light bursts per second echo back their pulsations at great speeds. In many ways, this is the contemporary version of the Doppler effect, named after Christian Doppler’s mid-­nineteenth-­century investigations, insofar as it includes some early ideas about the echo-­pulse principle of measuring objects and their movement through light signals. Doppler’s influential research, which can be quoted in relation to multiple engineering and research innovations that characterize later technical media and remote sensing, was focused on planetary movements and measuring light. According to Doppler, the color, being dependent on the frequency of light, led to the observation that objects in movement emit a different frequency in relation to each other. The remote sensing analysis of spectra of light quoted in the preface already through the Fraunhofer apparatus and then in the Harvard astronomy photograph analysis is also continued in this case. Doppler’s point implied that a transmitted pulse signal returning from a measured source could communicate the location of that object based on its frequency, leading to all sorts of implications for navigation and observation. So, while the 1842 paper “Über das farbige Licht der Doppelsterne und einiger anderer Gestirne des Himmels”59 was about the extraplanetary scale of the measurement of light and movement, it also began to speak to measurements at other scales too: the world is measurable as a function of signals and their echoes, of pulses and their reverberations. As you have by now

196   The Post-­lenticular City

noticed, throughout this book the astronomical is interfaced with the surface level through measurement, calculation, and comparison. Analysis of light is an analysis of remote objects and movement, which comes to specify ways of interacting with movement, too; an operational invisuality. Twentieth-­century variations of the Doppler-­t ype modeling of space (outer space and down on the Earth’s surface) include the military technologies that observe and see without eyes, such as synesthesia, where sound turns into image (radar) and echoes pulse signals into visual perception. As Ryan Bishop and John Phillips write, “Radar technology allows soundwaves to see,”60 which crystallizes a larger trait that runs through modernist aesthetics and the military–­ technological infrastructure that reorganized forms of perception, visibility, and invisibility across the twentieth century. While the ping-­pulse enabled the reading of large-­scale environments, the screen was only a part of the wider network of sensing, with control and warning stations peppered across strategically important locations. As one gridded and gridding response (see chapter 1) to the problem of spatial control, the radar screen and its subsequent development into interactive computer screens was a way of turning incoming signal pulses into possibilities of intervention, a “repertoire of latent actions.”61 These actions, while receiving pulses on the screen are, however, related to an understanding of the broader territory as one of screened space, as Bernard Geoghegan argues: Radar operates according to a nested series of screening operations. The first screening typically happens at the level of the environment itself: radar transformed hundreds of miles of open space into a surface for the reflection of radio waves traveling at a rate of 186,000 miles per second before reflecting back to a precisely tuned receiver. Airplanes and other objects in open space became screens that, like an inverted X-­ray photograph, return waves to their point of emission.62

From screens as technological entities to a screenable environment, from the interpretation of incoming signals to the networks of actionable intervening, the territorial scale of operative images becomes clear when reading it through a genealogy of the radar.

The Post-­lenticular City   197

Furthermore, it relates to how any signal can be modeled into an image or a model and how pulsating worlds form the backbone of their aesthetic–­epistemic modeling. Beyond a “natural” pulse that can be modeled, devices can send out pulses that shed light on where one is, where they are going, how they should respond. Lidar, beaming its millions of pulsing light bursts per second across urban surfaces, is one form of an environmental screening. It also acts as the upgraded laser version of flash photography and the probe light, each of which had a revelatory impact on the history of photography: artificial light that enables seeing by a pulse quicker than the human eye. The invention of flash introduced such light that escapes even the blink of an eye, producing an extension of photography and what would now be called an active sensor: the production of light to record and measure light.63 While it allows seeing, the flash itself remains beyond the register of the eye’s reaction.64 However, the autonomous car’s lidar system is not the flash of photojournalism but one of laser-­based scanning operating at a different intensity. These laser images are at the center of ScanLAB’s work and include both the modern version of Muybridge’s Yosemite images and new photoscapes of cities as large-­ scale systems, where seeing is not limited to the humanly visible and images capture much more than that limited spectrum.65 However, the scanned lidar visualization is not of the usual scale of an image made of light; instead, the millions of tiny light bursts form the technical (and operative) ontology of this sort of imaging practice. As already mentioned, it also relates to the complexity of the technological laser scan that records both as a movement across the city and as a sensitive way of dealing with light that itself is a proxy for the complexity of the multiple surfaces of the city. The city is an image; the light is a proxy; the capture of light is a capture of a city as it is alive on different levels of the electromagnetic spectra. This is also where part of the architectural appeal of this sort of imaging through energy comes from: it helps to map and model the complexity of dynamic surfaces of built environments (as well as so-­called natural environments). Besides the apparent technological accuracy, these are also systems of imaging that are particularly vulnerable to over-­seeing,

198   The Post-­lenticular City

where all sorts of things like “complex architectural forms, reflective surfaces, unpredictable weather and temporary construction sites”66 can confuse sensors, misperceiving the cityscape in surprising, accidental ways. Sensors need to be constantly adjusted, calibrated, in a self-­reflexive, slightly paranoid loop of am I seeing this correctly? It’s a technical feature necessary for the various vision/invisual/sensor mechanisms of such vehicle systems as well as a fundamental feature of images as models (see chapter 2): the intervention into the vehicle’s movement and thus the environment is based not on a representational image but on a model that is not necessarily a future that will happen but according to which the system operates. They are a codification of possibilities that are the predicted future upon which action occurs.67 But this reflective reality, the over-­seeing of light, the modeling of futures, also hints at what Manaugh aptly describes as “a parallel landscape seen only by machine-­sensing technology in which objects and signs invisible to human beings nevertheless have real effects in the operation of the city.”68 Invisibility is not merely about the unseen signs and signals but also statistical predictions: the invisual that lurks not in the shadows of the optical spectrum but the neural network calculated models of the thousands of tiny futures just a fraction ahead of direct perception. “Now light, where it exists, can exert an action, and, in certain circumstances, does exert one sufficient to cause changes in material bodies,”69 wrote William Henry Fox Talbot in the Pencil of Nature in 1844. The changes referred to the registering surfaces of light—­such as emulsion-­covered glass or paper—­but imagine this as a narrative that prioritizes exactly that: light exerting an action. Bodies react and are transformed, even guided accordingly. Sensors pick up light and respond. Light can also exert unusual responses. Accidents of over-­seeing can cascade and be multiplied, and they can be taken as guidelines for investigations into contemporary visual, photographic, scanning technologies. In other words, as glitches, they become ways to understand the functions of this form of imaging turned imagining. Dream lives can be imaginaries, but they can also be hallucinations with effective epistemic uses, such as in the machine learning calculation of what happens.70 Lidar here is only one part of the ecology of screening in actual territorial situ-

The Post-­lenticular City   199

ations. Indeed, ScanLAB’s Dream Life of Driverless Cars was meant not as a technical demonstration of the accuracy of lidar as a stand-­ alone technical feature but rather as an experimental framework for a scanning device that also records its conditions of existence: Their goal, Shaw said, is to explore “the peripheral vision of driverless vehicles,” or what he calls “the sideline stuff,” the overlooked edges of the city that autonomous cars and their unblinking scanners will “perpetually, accidentally see.” By deliberately disabling certain aspects of their scanner’s sensors, ScanLAB discovered that they could tweak the equipment into revealing its overlooked artistic potential. While a self-­driving car would normally use corrective algorithms to account for things like long periods stuck in traffic, Trossell and Shaw instead let those flaws accumulate. Moments of inadvertent information density become part of the resulting aesthetic.71

A range of methodological implications come to the fore. On the one hand, contemporary uses of lidar rearticulate a relation even to lenticular histories of photography, thus investigating the relations and frictions with post-­lenticular approaches. On the other hand, the alternative genealogies of imaging are one way to explore lidar as the echo and the pulse of the city, all the way from Doppler’s research into spectra of light to the forms of flash and photography that define the technological configurations of images in the age of photography. The pulsating screening of environments is also a way to understand that the projected plan of a smart, intelligent city—­whatever you want to call it—­starts already on the level of the existing city: its sentience, its materialities, its flickering lights.72 Or, as Bratton puts it, “The sensing and thinking systems are located not just in the valuable subjects and objects rolling around, they are built into the fabric of the city in various mosaics.” 73 Thus what is at stake is not a machine vision of space as a representational entity of what has been seen. Instead, the ecology of models and operations starts in the fabric of the city, which itself is sensorial in the multiple meanings of the term: a multitude of computational sensing is matched up with materials and bodies of sense.

200   The Post-­lenticular City

Sample Cities and Speculative Futures Machine vision, technologies of scanning, and the broader non-­ human spectrum of agency of images have already entered the focus in the scholarship of photography and visual culture, questioning divisions between media-­specific analyses such as cinema versus photography.74 This applies to the book you are reading as well. Operational images are not specific to only cinema or photography or, for instance, digital images. The term cuts across such distinctions in the attempt to understand the role of technical images in reformatting space as well as the role in transformations between images and data. That, already, is quite the world-­embracing broad topic. Still, the anchor here is the operational image, as both a situated concept (of the late twentieth and early twenty-­first century) and a potential field of forces that also facilitates media archaeological excursions, as we briefly saw in earlier chapters. As such, it has methodological force. In intersecting discussions, theoretical and historically sourced positions cut across individual technologies and genres of representation, which have started to broaden the scope of what is understood by machine vision.75 A similar move is presented here through the case study of lidar that ushers us to the broader sensorial regimes of the city and what kind of images are being formed in that ecology of sensing. Machine vision is expanded from a set of algorithmic technologies of vision to material events of the city that are not necessarily computational in the traditional sense and the platformed invisuality that works with visual material (optical and nonoptical, lenticular and post-­lenticular) but is “not of the order as visual,” 76 as Mackenzie and Munster have demonstrated. Instead, traditional reference points of visuality and ordering such as “eye, lens, sensor, file, screen or database” are remixed in the “invisual image ensembles”77 and combined anew in the diagrammatic functioning of the platforms. In this book, the notion of operation refers to this dynamic in new transversal connections and functions of images. In more specific terms, the pulsating light problematizes the idea of perception or visuality as a stable registering of forms situated in space as if it were a container instead of a field of dynamic signals. The signal-­

The Post-­lenticular City   201

FIGURE 26. “WiFi antennas as sensors for person perception. Receiver antennas record WiFi signals as inputs to Person-­in-­WiFi. The rest [of the] rows are . . . images used to annotate WiFi signals, and two outputs: person segmentation masks and body poses.” Image text from original source, Person-­ in-­WiFi: Fine-­Grained Person Perception Using WiFi. Courtesy of Dr. Dong Huang, head of the DeLight lab at Carnegie Mellon University.

based registering of the world means a move from representations of things to modeling events. Practices of modeling have a slightly different relation to the epistemic power of what is considered real; this is done less for representations than for prediction, monitoring, and intervening purposes. Besides lidar, other examples could be used. In passing, I already referred to “WiFi” as such a site of visuality/invisuality, which illuminates the argument I am making. In a recent paper, “Person-­ in-­WiFi: Fine-­Grained Person Perception Using WiFi,” a team of researchers suggested that WiFi antennas and signals are, in principle, sensors capable of detecting body postures and movement. This form of seeing beyond the visual spectrum is introduced as an alternative to camera-­based, radar, and lidar (light detection and ranging) technologies that have already been used in the context of “people perception.”78 In a city—­nowadays, almost any city—­that is cut through by an extensive range of wireless signals between signal stations, this means that the city is continuously forming images. These images, however, are based on the sensing, registering, and

202   The Post-­lenticular City

algorithmic modeling of WiFi signals where what is usually considered transmission or communication turns into images, and where signal processing turns into spatial modeling in a reminder that everything that exists as a signal can also exist as an image. Theirs is not the first paper to claim the usability of WiFi as a technique of sensing. The authors list earlier work in which 1D signal space has been reconstructed into “2D fine-­grained spatial information of human bodies,” 79 questioning the theoretical focus on the visual spectrum through the mapping of opening doors, keystrokes, dancing, and even static objects. Out of this broader ontology of urban “wirelessness”80 emerges the possibility of mapping, visualizing, and modeling space, humans, animals, objects—­anything that reflects the signals from WiFi traffic and antennae. This quirky detail also reveals how vision, or even a metaphorical extension of “seeing” in this technological context, is not only a distinct sensorial capacity to be understood in relation to other human-­based sensoria (hearing, touching) but also a technique of modeling according to sampling rates: “Lidars have sampling rate[s] in the range of 5–­20 Hz, which is much lower than other sensors such as cameras (20–­60 Hz) or WiFi adapters (100 Hz).”81 The image, then, is a sampling of a spatio­temporal situation: a constantly produced entity that cuts across the dynamics of a city as part of an operational processing of what is being seen, at what time, in which relations, and to what ends. The image is the platformed compilation of such signals into useful units of reference, whether diagrams, models, or something more pictorial for an operational purpose. In short, an image might not resemble much (or at all) what we expect an image to be like, which is why so many of the questions of art theory and aesthetics revolve around the question: What are our objects of investigation when they have switched from the visible to the invisible, and from visual to the invisual? If maps, QR codes, or for that matter, diagrams, are treated like images, the term both broadens its scope and demands a specific refocusing—­ that is, I claim, about the operational and infrastructural apparatus. Images are gateway drugs to data. Or, if you prefer, interfaces.82 The echo light pulse systems of laser scanning become one aesthetic channel to the complex systems of contemporary mapping of the urban world. While ScanLAB Project’s work turns toward tech-

The Post-­lenticular City   203

nological and aesthetic questions, a similar short film operates in the imaginary near future of laser scanning and corporate surveillance. Liam Young’s audiovisual work Where the City Can’t See (2016), written by Tim Maughan, relates to the same bundle of issues that unfold when considering these images as interfaces. Young’s speculative design version of these visual/invisual interfaces is introduced as “the first fiction film shot entirely through laser scanning technology” and is “set in the Chinese owned and controlled Detroit Economic Zone (DEZ), in a not-­too-­d istant future where Google maps, urban management systems and CCTV surveillance are not only mapping our cities, but ruling them.”83 Speculative fiction about automated governance of cityscapes continues the discourse about scanning, but with a particular focus on urban politics. Moving beyond QR code cityscapes, scanning is understood as a form of control, emphasizing how this survey of a landscape and this scan of a city are part of the reordering and rethinking of what images do, positioned somewhere between the traditional visual image and its role as digital measures of relations, movement, events, and management of dynamic entities across large-­scale territories. Young’s film also engages with what could be coined as post-­ lenticular subcultures: .

Exploring the subcultures that could emerge from these new technologies, the film follows a collection of young factory workers across a single night, as they drift through the smart city in a driverless taxi, searching for a place they know exists, but that the map doesn’t show. They are part of an underground community that work on the production lines by day, by night adorn themselves in machine vision camouflage and the tribal masks of anti-­facial recognition, enacting their escapist fantasies in the hidden spaces of the city. They hack the city and journey through a network of stealth buildings, ruinous landscapes, ghost architectures, anomalies, glitches and sprites, searching for the wilds beyond the machines.84

Similar to ScanLAB’s investigation of laser scanning as a technology that records its own glitches in the perceptual field/model, this speculative fiction unfolds an unstable city of moving bodies and

204   The Post-­lenticular City

FIGURE 27. Still from Where the City Can’t See. Directed by Liam Young and written by Tim Maughan. Reprinted with permission.

rhythms, of perception that hovers on the thin line of continuity and discontinuity. The city is pulsing. In the glitching of a cityscape, the WiFi transmits and registers; signals and sampling rates reveal the shifting shapes of a city. Specific techniques are used for specific operational ends. Lidar is effective in catching the multi­scalar movements and events from the size of an insect to the large-­scale buildings. The city is itself a connected entity of sensing and sensors, but what stands out is the transversal interconnection of mobile and immobile bodies: the question of visible and invisible is not, as argued already in chapter 3, articulated on the axis of visible light but in the infrared spectrum and in the invisuality of data that relates to the multiple platforms that form the augmented reality of the city as rewriting (as property, as experience, as value). The images that are being formed are a modeling of the incoming pulses. They are entities that enable and hinder perception as they hide and seek in the midst of ecologies of machine sensing. Paradoxically, to understand the shift of visuality and the photographic into the register of the invisual, one needs to look away from images and toward their infrastructural coupling with large-­scale systems of sensing and computation. Where the city can’t see is also the operational sphere that supports sensing, observation, and the production of images.

The Post-­lenticular City   205

Conclusions The operational is not merely of the scale of an image.85 One of the points I am making as I build on and with the work of others is that machine vision should be understood not merely as a special case of vision or seeing, but (at least as much) as an operation of navigation and movement that is infrastructured across different layers of urban and nonurban environments. For example, in terms of autonomous vehicles, this relates to data pooling from the camera, light detection, lidar, radar, ultrasonic sensor, and vehicle motion data feeds. Furthermore, this data pool of combined sources—­that may or may not be images in the pictorial sense—­is not only for on-­site processing but also for various other scales of uses, such as simulation and modeling.86 As a network platform, this system that gathers sensor data into meaningful models of the world-­in-­action is also part of the design and engineering problems of storage and transmission and includes rather mundane-­seeming engineering dilemmas such as how to transmit such a massive amount of data in real ­time that consists of several moving entities in an environment that itself is dynamic on multiple scales, including the reflective surfaces of the city registered by laser scanning. This sparks an argument about images as well: to discuss such operational images as part of a bundle of other forms of measure and infrastructure where imaging is actioned. The image becomes entangled, even conflated, with real-­time processes of dynamic sensing, which in turn act as the necessary prism through which even histories, media archaeologies of the image, and the photographic too can be seen anew. In other words, I argue that the photographic discourse about the operational and the instrumental image must be updated to include the infrastructural image as was already discussed in the previous chapter in the context of operational aesthetics. Although these are not necessarily images that represent or depict infrastructure, their mode of existence as environmental media87 is premised on how they act and trigger actions in particular situations: this might be an event of execution at the moment, in the microseconds of decision time in traffic, or it might be a slowly unfolding intervention through a forecasting model. In this vein, the aesthetic of the infrastructural depicts, represents, and operates in that broad circulation of image-­data-­environment.

206   The Post-­lenticular City

This chapter has offered some potential ways to respond to this question, especially in the context of lidar, but this can and should be extended beyond this specific technology of scanning. From the artistic works of ScanLAB Projects to those of Liam Young and a host of others, a common line runs through these practices that track images that track, observe images that observe, and try to gather a sense of how images make sense outside the scope of standardized human vision. This lineage speaks to posthuman photography and its multiple variations across the non­visible spectrum of pulse and light.88 Hence, in addition to Young’s film, the discussion of Geocinema and the Digital Belt and Road system in the previous chapter is but one example of what I mean. Here, I have continued with different aspects of operational aesthetics. Such a form of operational aesthetic is also the educational arm of methods for investigating images of this sort: screening, measuring, intervening, operating. We do not read them like books, nor do we look at them like usual images, but we relate to them like assembled parts of a logistical system of traffic. And if you want one more film reference for this mix of epistemic and aesthetic images, turn to Farocki’s Counter-­Music again. Its multiple juxtapositions and transitions of bodies, cities, and operational images speak to the centrality of circulation and transport, as Volker Pantenburg’s epigraph of this chapter already told us: urban infrastructure and traffic of images conflate.89 Following from the above discussion, we can claim the following: while these contemporary images are part of a media archaeology of scientific photography and different multi­scalar technical measures, they also take a particular role in contemporary computational cityscapes as scans, as infrastructure, and as interfacial skins90 that are not merely singular instances of images, even such peculiar ones as QR codes, but entire environments of sensing and action.91 Consequently, they are folded into a network of operations in large-­scale systems. In this sense, these images continue the legacy of not only photogrammetry but also plotting as computational solutions to problems of measurement, analytics, and rewriting of surfaces at different scales.92 In the contemporary situation, the extent and scale of computation are of a different order, and the

FIGURE 28. Two stills from Counter-­Music (2004). Courtesy of Harun Farocki and Antje Ehmann. Copyright Harun Farocki GbR.

208   The Post-­lenticular City

processing of images becomes automated as part of these operative chains. This is amplified with the automation of events from image sensors as part of the distributed network of computing (of which the mobile vehicle is only one part) that tries to keep up with the already existing complexity of the moving landscape. The city itself is a large-­scale synthetic, even synthesizing, unit that presents a particular case for ecologies of images embedded in multiple sites, functions, vehicles, and passages for a multitude of programmable sequences that take the form of intervention. And it goes beyond the city, too, including the range of automated image and machine vision systems that function in large-­scale, embedded, and dynamic systems, from cityscapes to meteorological systems, seafloors to ground terrain, military scanning of landscapes to development of experimental ways of environmental visualization. As Florian Sprenger has aptly shown, lidar and related sensorial systems sit as part of the history of environmentally sensitive robotics, which is now also part of large-­scale systems.93 Adaptive systems that are continually producing information about their environment have provided one solution to problems of maneuvering and movement in autonomous systems. From the experimental robotic systems of the 1950s to the new robotics turn in the 1980s, these forms of autonomous systems became conceptually and technologically dependent on sensing: instead of trying to upload a map of the environment into the machine, the goal is now to create multiple layers of perception and sensing that situate the system in its location.94 As Sprenger argues, autonomous systems and their forms of scan-­based sensing, including lidar, can be seen as special cases of robotics designed to be context-­aware and reliant on adaptation to environmental conditions. This also helps to address the argument about contemporary forms of sensing and sensors as mobile systems. Indeed, as mentioned earlier, the case for understanding these imaging systems as inherently about navigation becomes clear.95 Even the event of sensing is premised on the latent possibility of a movement, of coordination, orientation, and mobility. In our case, the laser-­scanned images and landscape are directly related to the instrumental96 role these systems play in coordinating (in) a city, its multiple levels of agents and events, and in other contexts, other forms of navigation in complex ecologies. Hence we

The Post-­lenticular City   209

have to look beyond the image, the QR code, the screen, the vehicle as a stand-­alone unit, and instead understand that the image is, at best, an interface97 that allows a kind of access to other scales of infrastructural action that mobilize multiple kinds of knowledge of large-­scale, dynamic systems including maps, information systems, AI, sensors, data transfer, and so on. Environmental perception and localization in relation to external data and maps become a form of synchronization that adds to the work of actual sensing and imaging. All of this can be seen as one crystallization of what images have become in the twenty-­first century’s complex and distributed large-­scale autonomous systems. This infrastructure can be understood as an operational bundle of technologies that is also part of a political economy of innovation that aims to reformat the city according to its ideals of the “city as computer.” As Shannon Mattern argues, this trope assumes—­and rhetorically (re)produces—­a frictionless programmability of the city as its modus operandi.98 One can also track this infrastructure through its corporate financial attachments that mobilize their own views of operative ontology through the employment of data feeds, maps, location-­systems, and other variations on the theme of “seeing” and “imaging” in the smart city: digital data platforms connected to technologies of self-­driving cars, to multiple scales of navigational infrastructure, to urban technologies (for example, Sidewalk Labs), and to various forms of maps, robotics, engineering, and expertise that form the links between technological discourse and corporate valorization mechanisms. Spatial data rendered as maps or other sorts of images become a component in platform invisuality: this is where questions of data, operational images, and property regimes meet up and hit the smart city ground. While it concerns the production of value in for-­profit systems and often under corporate control, this is not necessarily “extraction” as has become customary to discuss (in contexts of data colonialism) so much as the production of different worlds of sensing that are prone to capture, control, and commodification.99 Data is not a finite resource, but it is produced in conditions of finite resources. To be clear about extraction helps us to seek ways out of the current logic of platform capitalism toward alternative forms of the capture and distribution of value.

This page intentionally left blank

CONCLUSION

A Soft Montage of Operations The Introduction to this book started with astronomy and the capture of the light of the celestial night skies that became an object of calculation and comparison. Photographic glass plates detecting light turned into data about celestial objects; images triggered calculations that produced further (nonphotographic) images such as diagrams and measures from images (such as light values in comparative data tables). Thus it feels only natural to round out the book by returning to the sky, beyond human vision but constantly calculated, some of which even returns as an image. This is also a parallel story of a more recent date that provides a diffraction pattern of these two perspectives: different sites of practice, both about astronomy and images, more than one hundred years apart, yet both dealing with technical images and photography in ways that have a specific relation to data and institutions, measurement and invisual visibility. So, fast-­forward 120 years to offer a fitting parallel development that concerns photography, data, and the planetary dimension of observation. In 2019, a peculiar image was announced. Something made available for the wider public as a spectacle of a different kind and not merely for astronomers to analyze. Something that was even said, in some commentaries, to be a photograph, even if it could not possibly work like terrestrial photographs usually do, conditioned by sunlight. If photographs entered the world as a technology for capturing the solar spectrum of light that became quantified, compared, and managed, this one seemed to do the opposite: to capture the absence of light as it pictured a massive black hole at the center of the Messier 87 galaxy. The rim of energy around this entity 211

212  Conclusion

FIGURE 29. “First Image of a Black Hole.” EHT Collaboration.

that is 6.5 billion times bigger than our own Sun functioned as the proxy signal that black holes exist. As such, not exactly visible in the mundane sense of the term, but something that can be seen through measurement, parsed together as a composite image, fabricated based on a signal frequency captured by multiple radio dishes. Such an image lends itself to all sorts of philosophical theories concerning perception, absence, and other fantasies about the real and the imaginary, possible and impossible. But in this case, even the (seemingly) simple question of how this became possible is fascinating as an entry point to a different way of understanding not only the astronomical perception and its scientific instruments but also the function of photography. A more media-­theoretical and technical account might simply refer to it as a large-­scale Fast Fourier Transformation operation, where an image is determined in relation to the frequency response of the black hole. As Robert Pietrusko explains, an analogy for this procedure would be a sonogram of a sound wave, but one wherein parts of the signal spectra are missing.1 Operational images are not an aberration, but the standard case of images—­and operations that produce images, alongside other entities such as datasets, models, and more. The Event Horizon Telescope emerges not only as an astronomical science project but as a way to rethink images. This ex-

Conclusion  213

periment with imaging is premised on a distributed high-­tech scientific apparatus, training, and lessons from the history of science where aesthetic techniques have been mobilized for various ends. Harvard-­based historian of science Peter Galison, involved in the Event Horizon and Black Hole Imaging project, explains that we are dealing with a peculiar perception, measurement, and registering of light that bends around the black hole: “Black holes are surrounded by orbiting gases that glow with heat. Still, the challenge of imaging is immense: not a pixel would exist without the combined efforts of telescope operators, instrument designers, theorists, data experts, and engineers, from early-­career to senior scientists.”2 Therefore, part of the team’s work was to engage with the history of scientific imaging and notions of objectivity from drawing exercised in scientific atlases to photography and to the “trained judgment” that emerges as a particular skill of interpretation over the twentieth century.3 Such questions helped to guide the methods of how to form an image of something in traditional terms invisible but still potentially representable through a synchronization, analysis, and computational interpretation of massive digital datasets of signals (each telescope site of the project produced about 350 terabytes of data daily, stored on “high-­performance helium-­filled hard drives”4). As an exercise in massive-­sized interferometry, coordinating and synchronizing with the help of atomic clocks, the received radio wavelength signals formed one operational backdrop to the complex process of constructing the relevant spectral signature of the black hole. At the same time, part of the problem persisted in how to compose, from those signals, an image of a thing that has never been verified in visual terms. Earlier theoretical models in physics exist, and now a measured spectral manifold too. But how might we build a plausible image that conforms to this space of expectations? In other words, there is no “ground truth” against which to compare the accuracy of the reconstruction. Moreover, as synthetic datasets are becoming more and more central to machine learning practices, there was an additional dilemma in relation to this comparative process of what signals indicate: What might they picture? No wonder, then, that artist Rosa Menkman includes the black hole image in her catalog of “impossible images”: these are images

214  Conclusion

that have pushed the boundaries of signal ­processing, and they linger on the horizon of visible/invisible, visual/invisual. The Event Horizon Project visualization sits in the category of “images of objects that can be only inferred, transcoded, or otherwise ‘perceived’ involving a significant signal/noise bounty” alongside other categories and examples of images that were once impossible (X-­rays and others), that will become possible (Medipix particle imaging), or that remain fully speculative (an image of dark matter). As phantom shots, impossible for living humans, they occupy a central role in technical ways of seeing and knowing.5 A key technique examined in the Event Horizon Project, the CHIRP algorithm (Continuous High-­resolution Image Reconstruction using Patch priors) engineered by Katie Bouman at MIT (again, we are back in Cambridge, Massachusetts), was presented as one solution to molding, interpreting, and preparing data into an image. It includes coordinating the various received signals and cleaning them from atmospheric data, as well as modeling the massive amount of input so that it helps the later process of building plausible models of what kind of an image could be composed in that image search space. Here, the image is algorithmically composed and premised on training sets in machine learning techniques. Bouman used a machine-­learning algorithm to identify visual patterns that tend to recur in 64-­pixel patches of real-­world images, and she used those features to further refine her algorithm’s image reconstructions. In separate experiments, she extracted patches from astronomical images and from snapshots of terrestrial scenes, but the choice of training data had little effect on the final reconstructions.6

The six observatories around the globe (in Chile, Hawaii, Mexico, Arizona, Spain, and the south pole) that were part of the registering of signals are a significant upgrade from the glass plates sent from Peru to Harvard. Furthermore, instead of only (human) computer analysis of annotation and comparison, computational methods helped to synchronize the perspectives from the various rotating data-­recording places of observatories in order to get a rich set of signal-detecting angles, so to speak. Of course, it was not merely

Conclusion  215

studying automated algorithms that enabled us to see phantom views. As Galison outlines, the project’s focus on what kind of a data process leads to a reliable image was based on training four teams with different methods to work with a sort of reverse engineering of visual datasets back to the test image: First, all four teams received test data sets created from unknown images. Some were astronomical objects, others were simulations, one was even an image of Frosty the Snowman. Running blind, each team had to reconstruct the original test image. When all teams could successfully pass the tests, then—­ and only then—­did we tackle real data. Still in isolation, the teams began peering, in awe, at the first real black hole images. A month later (July 2018), the four teams gathered to compare their catch, pixel by pixel.7

From early expert reading of photographic images to such later teamwork that consisted of developing algorithms, methods, and more, at least one thing remained similar: the ability to stitch together input from around the planet in order to enhance the accuracy of the analysis. Pictures, to say the least, are processed and fabricated in complex operations on the ground of the Earth, in institutions, in and through such operational apparatuses that are more than just one machine. Distributed operational aesthetics indeed. In the case of Event Horizon, argues Benjamin Bratton, the whole planet is turned into a massive camera, the planet becoming an integrated sensing surface that enables data transmission and synthesis.8 However, to be more accurate, the actual operations consisted of much more hard work in the synchronization, fine-­tuning, shipping of massive datasets on physical hard drives (delivered by human couriers on ships and planes), with the center of operations (still) in Cambridge, where data was parsed together in the interferometric operation. But as an operation of integration of scales, we can still refer to “new economies and forms of knowledge,” as Orit Halpern puts it, that are at play. In these operations of scale, surface, and imaging, she continues, particularly acute questions of “how we currently govern and manage life on Earth through computation”9 emerge.

216  Conclusion

These high-­tech images are made in teams, they are constructed out of signals and other data, and they are modeled (with machine learning techniques of optimization and search) to fit preexisting ideas of what an image might be like. Such images are born thanks to Bayesian heuristics (among many other layers already mentioned). Even if operational, they are thus often pictorial, even in contexts of advanced big science (including for reasons having to do with gaining public support and attracting funding, etc.).10 Moreover, images—­and this applies to not only astronomical imaging but also contemporary data culture—­are constantly formatted and calibrated in relation to their technical platforms of use as well as what is expected of a picture; all of this shifts the primary focus from what is being represented to their operational preparation.11 And there is an interesting feedback loop between specialist practices and our contemporary theoretical work that asks: What even is an image, especially in the light of things seen and unseen, calculated and synthesized, measured and rendered.12 Remember: tell me what spectrum of electromagnetic radiation you are focusing on, and I will tell what your discipline is. I have tried to avoid such juxtapositions that early photographic images are more naive ways of picturing and that now, with digital (big) data platforms, we are forced to shift from the pictorial to the nonrepresentational. Historians of photography would be quick to point out how inadequate this sort of a linear narrative is and how early on particular analytical qualities of imaging were at least as, or even more, central than just depicting 3D things on 2D flat surfaces. Instead, seeing, representing, and sensing has taken many twists and turns in ways that build up to the complex layers that make up the media archaeology of operational images.13 We could, thus, take Elizabeth Eastlake’s mid-­nineteenth-­century words as a good guideline for such investigations: photography is “neither letter, message, nor picture.”14 I have a personal fondness for this short quip. I would go as far as to claim that it is more interesting than many things said about images and photographs within the (commercial) industry discourse of AI and image datasets. Technical images are not just pictures, messages, or communication. They demand a revised vocabulary to address them. Hence: operations. The concepts, approaches, and ideas that have been discussed in

Conclusion  217

this book, from photographic glass plates circa the 1890s and fin de siècle and then to the computational images around the 2020s, inform the end result of not only what but also where operational images come to mean: grids, platforms, (training) datasets, AI models, urban remote sensing, histories of photogrammetry, and measurement alongside contemporary examples of art and design that practice invisual culture. Operational images consist of layers upon layers, recursive chains of techniques, turtles upon turtles of technologies, practices, people, training, legal and financial instruments, and many other aspects—­operations that hit the ground. This approach has meant building on Farocki’s original work and diverting from it to investigate the concept as a force of its own; we can refer to this as conceptual persuasion, to echo Brian Massumi’s terminology mentioned in the introduction. At the beginning of this book, I hinted that I am more focused on operations than images. And yet, these two cannot be positioned as opposites either. Instead, they are codetermining instances: operations that build up possibilities of an image; images that have operational capacity and meaning in different institutional situations. I have, though, wanted to map an operational approach to images and investigate what the term “operational images,” after Farocki, has come to mean. This has been somewhat less an exegesis of its multiple uses (even if I have tried to include many references to earlier work and applications, including also the past ten to fifteen years of artistic practices) than a scoping for potential; What can it do? How can it function? How can the concept itself operate? In an interview with cinema theorist Thomas Elsaesser, Farocki had confirmed that considering cinema as part of “the history of other techniques and technologies of surveillance, measuring, calculating, automation”15 is an apt characterization. Elsaesser followed up on this line of thought in his theoretical work, not least in the already mentioned notions of S/M operations of cinema, including science and medicine, surveillance and military (and one could add: security and money, statistics and machine learning, stonks and metaverse, and others).16 Beyond cinema, operational images have been discussed in this book as those instances that are about images and “other techniques” which in some cases also problematize what is visible in the first place. An ontological example such

218  Conclusion

as black holes is in this sense intriguing and influential, but many sites of social visibility and power (of capture, targeting, control) on the earth’s surface would be too. Farocki, writing about the U.S. carceral system, echoes Deleuze’s point about societies of control: “With the increase in electronic control structures, everyday life will become just as hard to portray and to dramatise as everyday work already is.”17 Another point about operational aesthetics raised throughout this book: that which pushes much of visual culture to the sphere of invisual operations that mobilize different techniques of observation does not necessarily look like anything special. It is in the intermedial relations, the techniques of imaging and observation, and their operative stance to contemporary formations of territory, extraction, finance, and the production of data (and more), where some of the structured dynamics of the terms “operational aesthetics” and “operational images” emerge. In many ways, I have tried to keep in mind Farocki’s audiovisual method of soft montage when writing the various entry points to elaborate the potential scope of operational images. This has not meant that Farocki’s words or works are to be taken as some magical key that unravels the core truth about the issue—­such a logic of the secret and solution is far from what conceptual work should do in this instance. But as a method, it facilitates how material can be structured in nonlinear yet informatively articulated ways. It demonstrates one mode of thought that maps concepts, historical sources, and contemporary discussions into an operational loop. As instances of “multi-­dimensional thinking”18 as well as “loop-­like series,”19 the soft montage of images forms a series where relations of combination stand out, but so do the gaps in between: Farocki himself discusses his poetics under a different heading, not metaphor but montage. His montage takes two forms. One is as a sort of meta-­commentary, traversing especially the early films like a steady murmur, repeating the need to “separate and join.” The other type of montage is embedded in the movement of the thought, as its structuring dynamic, but verbalised, if at all, only as the cut, the gap and what becomes visible “in-­ between.” Where film theorists speak of segmentation, Farocki

Conclusion  219

(or his characters) discuss the difficulty of thinking things together at one level, while at another, making distinctions and keeping things apart. Only when the two levels are aligned, are the preconditions of new knowledge present: making connections on the basis of having taken something apart is thus where the rhetoric of metaphor meets the technique of filmic montage.20

I am not making a film, though; this is, after all, a book. But the method is still useful in answering the question: What forces compose the concept of the operational image as a particularly apt investigatory tool of contemporary culture, and how are operational images part of a broader landscape of epistemic, military, aesthetic, and data-­driven institutions? The answer varies based on context, the connection between contexts, and sometimes what falls in-­ between. The loop and the recursive serialization that this book presented suggests that operational images can be mapped into a specific historical period of media theory and technological culture (not least since the 1980s to early 2000s as the period when a lot of influential theory from Flusser, Virilio, Kittler, and Baudrillard took shape), but it also resonates with the broader context of “operations” as they are featured in contemporary theory, such as “operative ontologies” (chapter 1). In addition, questions of data and invisuality (chapter 2), measurement and territory (chapter 3), operational aesthetics across contemporary institutions of science and data (chapter 4), and in urban and nonurban contexts of lidar and machine sensing (chapter 5) form the essential answers to the arguments and hypotheses posed at the beginning of the book: that the “operational image” is a term that helps to understand the hinge between data and (in)visuality, space and image, and temporality and image. In doing so, it becomes a methodological tool for investigations that proceed via a series of cross-­disciplinary leaps: media theory in combination with art studies, in conjunction with architecture and critical data studies. And perhaps this is needless to state, but I will do so anyway: the particular loops and series in this book are not meant as the final word on the topic, but as (hopefully) helpful guidelines that deepen the often-­mentioned point: we live, work,

220  Conclusion

function, and imagine in the midst of such images that do not primarily represent but operate. We are, persistently, mapping techniques of reaching the phantom­like invisible that increasingly transforms into the platform-­like invisual. Nothing less is the task of analyzing mediations in and beyond media studies proper, from visual to invisual cultures.

ACKNOWLEDGMENTS

Books are not only about the author whose name is on the cover, but also the people who help out in affective, intellectual, friendly, and loving ways. Here’s a list of you who have supported this book in one way or another. Thank you to Antje Ehmann, Tomáš Dvořák, Michal Šimůnek, Tereza Stejskalová, Josef Ledvina, Martin Charvat, Trevor Paglen, Ryan Bishop, Shannon Mattern, Sunil Manghani, Giuliana Bruno, Francisco Alarcon, Bernhard Siegert, Doreen Mende, Jane Birkin, Athina Stamatopoulou, Abelardo Gil-­Fournier, Stephen Cornford, Samir Bhowmik, Benjamin Bratton, Sean Cubitt, Larry Lynch, Joasia Krysa, Geoff Cox, Sasha Anikina, Aura Satz, Beny Wagner, Daphne Dragona, Robert Pietrusko, Rosemary Lee, Miha Brebenel, Chris Speed, Eda Sancakdar, Xinyi Wen, Xuefei Cao, Sheung Yiu, Mika Elo, Tuomo Rainio, Jesper Olsson, Jannice Käll, Andrew Fisher, Alejandro Limpo, David Benqué, Ned Rossiter, Asia Bazdyrieva, Solveig Suess, Ranjodh Singh Dhaliwal, Sebastian Scholz, Pasi Väliaho, Rebecca Birch, Lindsay Smith Zrull, and Sarah Lavellee. And thank you to Elise Misao Hunchuck for her expertise and help in editing the book’s language and expressions. I also want to collectively thank the institutions that supported this book along the way: the department of photography at FAMU at the Academy of Performing Arts, Prague; the department of Digital Design and Information Studies and the Digital Aesthetics Research Centre at Aarhus University; and the Winchester School of Art (at the University of Southampton). I want to thank the organizers and audience members of talks at the University of Edinburgh (Design Informatics), Bath Spa University, the University of Arts in Helsinki, the Royal College of Art, the London South Bank University’s Centre for the Study of the Networked Image (CSNI), the Internationale Kolleg für Kulturtechnik­forschung und Medien­philosophie (IKKM) research center at Bauhaus University in Weimar, and Aalto Uni­ versity in Helsinki. Thanks to conference organizers at the Media 221

222  Acknowledgments

and Cultural Studies Association (UK) annual conference in 2020 in Brighton; Narrative 2020 (The Inter­national Society for the Study of Narrative) in New Orleans; the Digital Arts: Multiple Realities, Ludic Realities; International Congress of Art, Science, Technology (2021) at UEMG (Belo Horizonte in Brazil); the 2021 Estonian Graduate Winter School of Culture Studies and the Arts (Uni­ver­ sity of Tallinn); and Zone 1: Re-­T hinking Methodologies at Inter­ sections of the Arts and the Humanities (Linköping University). The staff at the University of Minnesota Press were amazing all the way through. Thank you to Doug Armato, Zenyse Miller, and everyone else involved throughout this process. The anonymous reviewers offered several good suggestions. Thank you to Fred Kameny for the index. This book’s research and writing were supported by the Operational Images and Visual Culture project, funded by the Czech Science Foundation’s EXPRO grant 19-­26865X.

NOTES

Preface 1. This refers to Brian Massumi’s interest in how the contemporary form of ontopower operationalizes “the nature of time, perception, action, and decision.” See Brian Massumi, Ontopower: War, Powers, and the State of Perception (Durham, N.C.: Duke University Press, 2015). See also Pasi Väliaho, Biopolitical Screens: Image, Power, and the Neoliberal Brain (Cambridge, Mass.: MIT Press, 2014). 2. A much more detailed discussion of the aims of this book comes in the introduction that follows. 3. See Frederick Suppe, “Operationalism,” in Routledge Encyclopedia of Philosophy, n.d., https://doi.org/10.4324/9780415249126-Q077-1. 4. Carl Sagan, Broca’s Brain (New York: Ballantine Books, 1980), 343. 5. Geoffrey Winthrop-­Young, “‘More Things in Theory than in Heaven and Earth’—­A Conversation with Geoffrey Winthrop-­Young by Melle Kromhout and Peter McMurray,” text available at https://digitalpassage.wordpress.com /2014/12/22/winthrop-young-more-things-in-theory-than-heaven-and -earth-are-dreaming-of-interview/. 6. Michelle Henning, Photography: The Unfettered Image (London: Routledge 2018), 18. 7. Myles W. Jackson, “Fraunhofer and his Spectral Lines,” Ann. Phys. (Berlin) 526, nos. 7–­8 (2014): A65, https://doi.org/10.1002/andp.201400807. 8. Susan Schuppli, Material Witness: Media, Forensics, Evidence (Cambridge, Mass.: MIT Press, 2020), 278. 9. Alongside Fraunhofer were other nineteenth-­century scientists, such as Robert Bunsen and Gustav Kirchhoff, working on and with the spectroscope in the 1860s. 10. “Fraunhofer placed the six lamps behind a shutter, 1.5 Bavarian inches (36.0 mm) high and 0.07 of a Bavarian inch (1.68 mm) thick, which was pierced by six narrow slits less than 1.5 inches high and 0.05 of a Bavarian inch (1.20 mm) wide. Each lamp was placed .58 of a Bavarian inch (13.92 mm) apart and directly behind a slit. The light from the lamps would travel 13 Bavarian feet (3.7 m) to a prism, which was made of flint glass with an angle of approximately 40°, where it would be refracted and decomposed into colors. The dispersed light then traveled through a second slit placed directly behind the prism, which accordingly blocked a portion of the emergent beam. Some of the rays were channeled to the site of a theodolite located 223

224   Notes to Preface

in Fraunhofer’s laboratory at the very great distance of approximately 692 Bavarian feet (199 m) from the six lamps. The six-­shutter mechanism controlled the angles at which light from each lamp struck the surface of prism A, thereby determining the locus of the corresponding spectrum.” Jackson, “Fraunhofer and His Spectral Lines,” A66. 11. Thomas Elsaesser, “Simulation and the Labour of Invisibility: Harun Farocki’s Life Manuals,” animation: an interdisciplinary journal 12, no. 3 (2017): 226.

Introduction 1. Annie J. Cannon, “Biographical Memoir of Solon Irving Bailey,” National Academy of Sciences 15 (1932): 193. 2. J. Donald Fernie, “In Search of Better Skies: Harvard in Peru I,” American Scientist, September–­October 2000, 398. 3. Fernie, 396. 4. Dr Raimondi, quoted in Solon I. Bailey and Edward C. Pickering, “A Catalogue of 7222 Southern Stars Observed with the Meridian Photometer during the Years 1889–­91,” Annals of the Astronomical Observatory of Harvard College 34 (1895): 28 5. Michelle Henning, Photography: The Unfettered Image (London: Routledge, 2017), 7. 6. Edward Pickering, quoted in Dava Sobel, The Glass Universe: How the Ladies of the Harvard Observatory Took the Measure of the Stars (New York: Penguin Books, 2017), 38. 7. Annie J. Cannon and Edward C. Pickering, “Spectra of Bright Southern Stars Photographed with the 13-­inch Boyden Telescope as Part of the Henry Draper Memorial,” Annals of Harvard College Observatory 28 (1901): 129-­P.6. 8. Kelley Wilder, Photography and Science (London: Reaktion Books, 2009), 34. 9. Wilder, 31. 10. Wilder, 34. 11. Wilder, 34. Furthermore, it would be possible to expand to the longer history of engineering drawing where questions of measurement, techniques of descriptive geometry, and photogrammetry play a central role before such technical images as photography. In this sense, a media archaeology of operational images would become more about Gaspard Monge than François Arago or other pioneers of scientific photography. See Peter J. Booker, A History of Engineering Drawing (London: Northgate, 1979). 12. Harriett S. Leavitt, “1777 Variables in the Magellanic Clouds,” Annals of the Harvard College Observatory 60 (1908): 87. 13. Leavitt, 107. 14. John Durham Peters, Marvelous Clouds: Toward a Philosophy of Elemental Media (Chicago: University of Chicago Press, 2015). 15. “Les lames de plaqué préparées par M. Daguerre blanchissent au contraire à tel point, sous Faction de ces mêmes rayons et des opérations

Notes to Introduction   225 qui lui succèdent, qu’il est permis d’espérer qu’on pourra faire des cartes photographiques de noire satellite. C’est dire qu’en quelques minutes on exécutera un des travaux les plus longs, les plus minutieux, les plus délicats de l’astronomie.” Arago, “Rapport de M. Arago. (Séance du 5 juillet 1839),” in Historique et description des procédés du daguerréotype et du diorama, ed. Louis-­ Jacques-­Mandé Daguerre (Paris: Susse, 1839), 23–­24. John Tresch, “The Daguerreotype’s First Frame: François Arago’s Moral Economy of Instruments,” Studies in History and Philosophy of Science 38 (2007): 445–­76. 16. Wilder, Photography and Science. 17. Jimena Canales, A Tenth of a Second (Chicago: University of Chicago Press, 2009). 18. Lindsay Smith Zrull, “Women in Glass: Women at the Harvard Observatory during the Era of Astronomical Glass Plate Photography, 1875–­1975,” Journal for the History of Astronomy 52, no. 2 (2021): 115–­46. 19. Zrull, 117. 20. Zrull, 117. 21. There are echoes in some aspects of John Tagg’s argument: “Photography as such has no identity. Its status as technology varies the power relations which invest it. . . . It is a flickering across a field of institutional spaces.” Tagg, The Burden of Representation (London: Macmillan, 1988), 63. 22. On the technical image and style, see also Horst Bredekamp, Vera Dünkel, and Birgit Schneider, eds., The Technical Image: A History of Styles in Scientific Imagery (Chicago: University of Chicago Press, 2015). 23. Patrick Maynard, The Engine of Visualization: Thinking through Photography (Ithaca: Cornell University Press, 1997), 124–­29. To be fair, this ability to build a whole infrastructure of images around light/absence in gradients was at the back of what photography was meant to be in Daguerre’s (marketing) parlance: “The plate is exposed to light and at once, whatever the shadow projects on this plate, earth, or sky, running water, the cathedral lost in the clouds . . . all things big and little engrave themselves instantly.” Quoted in Maynard, 155–­56. 24. John Bender and Michael Marrinan, The Culture of the Diagram (Stanford, Calif.: Stanford University Press, 2010), 7. 25. Bruno Latour, Pandora’s Hope: Essays on the Reality of Science Studies (Cambridge, Mass.: Harvard University Press, 1999), 53 26. Latour, 53. 27. Latour, 55. 28. H. S. Leavitt and E. C. Pickering, “Periods of 25 Variable Stars in the Small Magellanic Cloud,” Harvard College Observatory Circular 173 (1912): 2. 29. Both “operative images” and “operational images” appear in English literature on the topic. The German term is operative Bilder. 30. For work on Farocki, there are many excellent places to start. See, for example, Volker Pantenburg, Farocki/Godard: Film as Theory (Amsterdam: Amsterdam University Press, 2015); Antje Ehmann and Kodwo Eshun, eds., Harun Farocki: Against What? Against Whom? (London: Koenig Books, 2009);

226   Notes to Introduction

and Thomas Elsaesser, ed., Harun Farocki: Working on the Sight-­Lines (Amsterdam: Amsterdam University Press, 2004). 31. See Florian Sprenger, Epistemologien des Umgebens: Zur Geschichte, Ökologie und Biopolitik künstlicher Environments (Bielefeld: Transcript, 2019). On the somewhat connected notion of the “navigational image” that Doreen Mende has developed based on Farocki’s work, see Mende, “The Navigation Principle: Slow Image,” E-­Flux lectures, November 29, 2017, https://www .e-f lux.com/video/176025/e-f lux-lectures-doreen-mende-the-navigation -principle-slow-image/. 32. Harun Farocki, “Phantom Images,” Public 29 (2004), 18. 33. Farocki, 18. 34. Farocki, 18. 35. There is a wealth of work on military technologies of perception that, of course, includes work by Virilio, but also studies such as Ryan Bishop and John Phillips, Modernist Avant-­Garde Aesthetics and Contemporary Military Technology: Technicities of Perception (Edinburgh: Edinburgh University Press, 2010); Antoine Bousquet, The Eye of War: Military Perception from the Telescope to the Drone (Minneapolis: University of Minnesota Press, 2018); Jeremy Packer and Joshua Reeves, Killer Apps: War, Media, Machine (Durham: Duke University Press, 2020); and Anders Engberg-­Pedersen’s forthcoming book, Martial Aesthetics: How War Became an Art Form (Stanford, Calif.: Stanford University Press, 2023). On aerial imaging, see Caren Kaplan, Aerial Aftermaths: Wartime from Above (Durham: Duke University Press, 2017); Lisa Parks and Caren Kaplan, eds., Life in the Age of Drone Warfare (Durham: Duke University Press, 2017); Emily Doucet, “‘The Idea Was in the Air’: Nadar’s Aerial Media,” Grey Room 83 (Spring 2021): 112–­37; Michael Richardson, “Drone Cultures: Encounters with Everyday Militarisms,” Continuum 34 (2020): 858–­69; and Hendrik Bender and Max Kanderske, “Co-­operative Aerial Images: A Geomedia History of the View from Above,” New Media & Society 24, no. 11 (2022): 2468–­92. On Farocki in art theory, see also Hal Foster, What Comes after Farce: Art and Criticism at a Time of Debacle (London: Verso, 2020). 36. Sharon Ghamari-­Tabrizi, The Worlds of Herman Kahn: The Intuitive Science of Thermonuclear War (Cambridge, Mass.: Harvard University Press, 2005), 47. 37. Ghamari-­Tabrizi, 48. 38. Patrick M. S. Blackett, “Operational Research,” Operational Research Quarterly 1, no. 1 (March 1950): 3. 39. Blackett, 3. 40. William Thomas, Rational Action: The Sciences of Policy in Britain and America, 1940-­1960 (Cambridge, Mass: MIT Press, 2015), 216. Thomas’s book is in general a recommended entry point to the history of operations research. 41. A much longer bibliography could be listed, but examples of feminist data studies and related fields include Catherine D’Ignazio and Lauren F. Klein, Data Feminism (Cambridge, Mass.: MIT Press, 2020); Jacqueline Wernimont,

Notes to Introduction   227 Numbered Lives: Life and Death in Quantum Media (Cambridge, Mass.: MIT Press, 2019); and Nanna Bonde Thylstrup, Daniela Agostinho, Annie Ring, et al., eds, Uncertain Archives: Critical Keywords for Big Data (Cambridge, Mass: MIT Press, 2021). 42. Harun Farocki, “Cross Influence / Soft Montage,” in Ehmann and Eshun, Harun Farocki, 64–­79. See also Jussi Parikka and Abelardo Gil-­ Fournier, “An Ecoaesthetic of Vegetal Surfaces: On Seed, Image, Ground as Soft Montage” Journal of Visual Art Practice 20, nos. 1–­2 (2021): 16–­30. 43. Paul Virilio, The Vision Machine, trans. Julie Rose (Bloomington: Indiana University Press, 1994), 59. 44. Harun Farocki, “Industrialization of Thought,” Discourse: Journal for Theoretical Studies in Media and Culture 15 (1993): 76–­7 7. 45. Virilio, The Vision Machine, 59. 46. Aud Sissel Hoel, “Operative Images: Inroads to a New Paradigm of Media Theory,” in Image—­Action—­Space: Situating the Screen in Visual Practice, ed. Luisa Feiersinger, Kathrin Friedrich, and Moritz Queisner (Berlin: De Gruyter, 2018), 11–­27; and Paul Virilio, War and Cinema: Logistics of Perception, trans. Patrick Camiller (London: Verso, 1989). See also Bousquet, Eye of War; Packer and Reeves, Killer Apps. 47. See Georges Didi-­Huberman, “How to Open Your Eyes,” in Ehmann and Eshun, Harun Farocki, 38–­50. 48. Rosi Braidotti and Matthew Fuller, “The Posthumanities in an Era of Unexpected Consequences,” Theory, Culture & Society 36, no. 6 (2019): 7 (emphasis in the original). 49. Braidotti and Fuller, 8. 50. Brian Massumi, Ontopower: War, Powers, and the State of Perception (Durham, N.C.: Duke University Press, 2015), 69. 51. Massumi, 69. 52. Tom Holert, “Tabular Images. On the Division of All Days (1970) and Something Self Explanatory (15 x) (1971),” in Ehmann and Eshun, Harun Farocki, 92. 53. Raymond Bellour, “The Photo-­Diagram” in Ehmann and Eshun, Harun Farocki, 146. On Foucault and invisibilities, see also Gilles Deleuze, Foucault, trans. Seán Hand (Minneapolis: University of Minnesota Press, 1988). I return to the visible/invisible relation and diagrams in chapter 2. 54. The Harun Farocki Institute website is at https://www.harun-farocki -institut.org/en/. 55. Thomas Elsaesser, “Simulation and the Labour of Invisibility: Harun Farocki’s Life Manuals,” animation: an interdisciplinary journal 12, no. 3 (2017), 219. For a more comprehensive take on computer graphics and images, see also Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021). 56. Elsaesser, 219. To someone like me, trained as a cultural historian, the word “always” usually sets off alarm bells, but Elsaesser’s underlying argument is valid. It detaches operational from merely the technically determined

228   Notes to Introduction

algorithmic form of automation that emerges in smart systems, robotics, and artificial intelligence since, roughly, the 1990s, and restores a particularly interesting trait to the term. 57. A perfect example of a detailed analysis of the operationality of remote sensing as part of scientific teams is found in Janet Vertesi’s Seeing Like a Rover: How Robots, Teams, and Images Craft Knowledge of Mars (Chicago: University of Chicago Press, 2015). As she writes based on her fieldwork: “The scientific image itself does not so much document the object out there as document the work of different communities of knowing subjects that enable, produce, and constrain knowledge of the world” (103). 58. Elsaesser, “Simulation,” 223. 59. This point can be also very detailed, as Janet Vertesi demonstrates in her fieldwork focusing on the Mars Rover imaging teams: such scientific images produced by the robotic imaging units transform a “natural object as an analytical object, inscribing a value into the very composition of what that object is and what makes it interesting, so that subsequent viewers and image makers will see, draw, and interact with that same object in the same way.” Seeing Like a Rover, 103. Here, operationality is not merely a characteristic of the technological machine but becomes one form of guidance between the community of analysts, operators, and the like. Furthermore, on the broader implications of biopolitics of visual culture, see Pasi Väliaho, Biopolitical Screens: Image, Power, and the Neoliberal Brain (Cambridge, Mass.: MIT Press, 2014). 60. Antje Ehmann and Kodwo Eshun, “A to Z of HF, or 26 introductions to HF,” in Ehmann and Eshun, Harun Farocki, 206 61. On machine learning and its learners, see Adrian Mackenzie, Machine Learners: Archaeology of a Data Practice (Cambridge, Mass.: MIT Press, 2017). 62. See Joanna Zylinska, “Undigital Photography: Image-­Making beyond Computation and AI,” in Photography Off the Scale, ed. Tomáš Dvořák and Jussi Parikka (Edinburgh: Edinburgh University Press, 2021), 231–­52. 63. In reference to the IKKM Weimar conference. 64. Farocki in Eye/Machine I video (2001), quoted in Volker Pantenburg, “Working Images: Harun Farocki and the Operational Image,” in Image Operations: Visual Media and Political Conflict, ed. Jens Eder and Charlotte Klonk (Manchester: Manchester University Press, 2017), 49. 65. Joanna Zylinska, Nonhuman Photography (Cambridge, Mass.: MIT Press, 2017). 66. Cf. W. J. T. Mitchell, What Do Pictures Want? The Lives and Loves of Images (Chicago: University of Chicago Press, 2005). 67. Aud Sissel Hoel and Frank Lindseth, “Differential Interventions: Images as Operative Tools,” in Photomediations: A Reader, ed. Kamila Kuc and Joanna Zylinska (London: Open Humanities Press, 2016), 178. 68. Trevor Paglen, “Invisible Images (Your Pictures Are Looking at You),” New Inquiry, December 8, 2016, https://thenewinquiry.com/invisible-images -your-pictures-are-looking-at-you/.

Notes to Introduction   229 69. Nathaniel Tkacz, Being with Data: The Dashboarding of Everyday Life (Cambridge: Polity, 2022), 173–­76. Tkacz is here drawing on Karin Knorr Cetina’s work. 70. Paglen, “Invisible Images.” On the quantity and scale of images, see Dvořák and Parikka, Photography Off the Scale. 71. This resonates with past years of design theory, especially Benjamin Bratton’s work on the vertical diagram of planetary infrastructures and political formations. For more, see Benjamin Bratton’s The Stack: Software and Sovereignty (Cambridge, Mass.: MIT Press, 2015). 72. Alberto Toscano and Jeff Kinkle, Cartographies of the Absolute (Winchester: Zero Books, 2015). 73. Adrian Mackenzie and Anna Munster, “Platform Seeing: Image Ensembles and Their Invisualities,” Theory, Culture & Society 36, no. 5 (2019): 4–­5. 74. See Orit Halpern, Beautiful Data: A History of Vision and Reason Since 1945 (Durham, N.C.: Duke University Press, 2015). 75. Mackenzie and Munster, “Platform Seeing,” 6. 76. Mackenzie and Munster, 7. 77. On recursive methods and media theory, see Geoffrey Winthrop-­ Young, “Siren Recursions,” in Kittler Now: Current Perspectives in Kittler Studies, ed. Stephen Sale and Laura Salisbury (Cambridge: Polity, 2015), 71–­94. 78. I also want to note other recent proposals use the notion of “operation,” or “operationality,” as a methodological concept in media analysis. For a take on “operations as units of analysis,” see Kathrin Friedrich and A. S. Aurora Hoel, “Operational Analysis: A Method for Observing and Analyzing Digital Media Operations,” New Media & Society (March 2021), https://doi .org/10.1177/1461444821998645. 79. This recursive methodology is outlined in more detail in Geoffrey Winthrop-­Young, “Siren Recursions,” 71–­94 80. Massumi, Ontopower, 209. 81. Massumi, 209. 82. Here, the work by Celia Lury et al. is relevant in how they outline the question of interdisciplinary methods through concepts that speak to activities of methods of composition of problems and maneuvering in that space of problems. In Lury’s words, outlining a vocabulary of methods as grammatically gerunds, they want to “identify the potential of interdisciplinary methods to compose problems as interruptions of the (historical) present. . . . The aim is to consider how interdisciplinary methods might constitute some aspect of what is given, the present—­in all its geo-­political complexity—­a s a problem, which is to say, as a situation that may be methodologically activated in specific, precise ways.” Celia Lury, “Introduction: Activating the Present of Interdisciplinary Methods,” in Routledge Handbook of Interdisciplinary Research Methods, ed. Celia Lury et al. (London: Routledge, 2018), 3. 83. Thomas Keenan, “Counter-­forensics and Photography,” Grey Room 55 (Spring 2014), 58–­7 7.

230   Notes to Introduction

84. Mackenzie, Machine Learners. 85. Gilles Deleuze and Félix Guattari, What Is Philosophy? trans. Hugh Tomlinson and Graham Burchill (London: Verso, 1994), 7. 86. Mieke Bal, Travelling Concepts in the Humanities (Toronto: University of Toronto Press), 2002. 87. Susan Leigh Star, “This Is Not a Boundary Object: Reflections on the Origin of a Concept,” Science, Technology, & Human Values 35, no. 5 (2010): 601–­17. 88. Giuliana Bruno, Atlas of Emotion: Journeys in Art, Architecture, and Film (New York: Verso, 2002). 89. Bal, Travelling Concepts in the Humanities, 4. 90. See some of the work emerging from the project in Dvořák and Parikka, Photography Off the Scale. See also Andrew Dewdney, Forget Photography? (London: Goldsmiths Press, 2021), which offers many arguments that resonate closely with this book in terms of mapping the historical legacy of photography vis-­à-­vis its radical transformations beyond photography as an epistemic figure. In addition, we should pay attention to the curatorial work that has been significant in reframing questions of photography. Here I am thinking of Katrina Sluis’s important work—­but many others could also be mentioned. 91. Éric Alliez and Maurizio Lazzarato, Wars and Capital, trans. Ames Hodges (South Pasadena: Semiotext(e), 2016).

1. Operations of Operations 1. Frederick Suppe, “Operationalism,” Routledge Encyclopedia of Philosophy, https://www.rep.routledge.com/articles/thematic/operationalism/v-1 /sections/behaviourisms. Quoting Suppe: “Examples of an operation would be the procedure of laying a standard yardstick along the edge of a surface to measure length or using psychometric tests to measure sexual orientation.” 2. Percy Williams Bridgman, quoted in Suppe. See also Jimena Canales, Bedeviled: A Shadow History of Demons in Science (Princeton, N.J.: Princeton University Press, 2020), 152. 3. A more detailed account of a “media archaeology of operations” would pay specific attention to the 1920s and 1930s as the period when the scientific use of the term, the military field of “Operations Research” (OR) and Martin Heidegger’s pair of ontological/ontic operations (discussed later in this chapter) appear in those separate fields. However, that systematic mapping has to appear somewhere other than in this book. For an insight into OR, see William Thomas, Rational Action: The Sciences of Policy in Britain and America, 1940-­1960 (Cambridge, Mass: MIT Press, 2015). 4. Wolfgang Ernst, Digital Memory and the Archive, ed. with an introduction by Jussi Parikka (Minneapolis: University of Minnesota Press, 2013). 5. Kathrin Friedrich and A. S. Aurora Hoel, “Operational Analysis: A Method for Observing and Analyzing Digital Media Operations,” New Media & Society, March 29, 2021, 4, https://doi.org/10.1177/1461444821998645.

Notes to Chapter 1   231 6. Friedrich and Hoel, 7. 7. The phrase “hit the ground” is here used in a similar manner as in Sandro Mezzadra and Brett Neilson, The Politics of Operations: Excavating Contemporary Capitalism (Durham, N.C.: Duke University Press, 2019). 8. The notion of S/M borrows from Thomas Elsaesser’s use: see Elsaesser, “Early Film History and Multi-­Media: An Archaeology of Possible Futures?” In New Media, Old Media: A History and Theory Reader, ed. Wendy Hui Kong Chun and Thomas Keenan (New York: Routledge, 2005). 9. Drilling, training, and educating bodies is one special case of institutionalization of algorithmic procedures executed by humans. See Geoffrey Winthrop-­Young, “Drill and Distraction in the Yellow Submarine: On the Dominance of War in Friedrich Kittler’s Media Theory,” Critical Inquiry 28, no. 4 (2002): 825–­5 4. See also Allan Sekula, “An Eternal Esthetics of Laborious Gestures,” Grey Room 55 (Spring 2014): 16–­27. 10. Website at https://www.kulturtechnik.hu-berlin.de/en/. 11. Horst Bredekamp, Image Acts: A Systematic Approach to Visual Agency, trans. Elizabeth Clegg (Berlin: De Gruyter, 2018); Horst Bredekamp, Vera Dünkel, and Birgit Schneider, eds., The Technical Image: A History of Styles in Scientific Imagery (Chicago: University of Chicago Press, 2015). 12. Bernhard Siegert, “Öffnen, Schließen, Zerstreuen, Verdichten: Die operativen Ontologien der Kulturtechnik,” Zeitschrift für Medien-­und Kulturforschung 8, no. 2 (2017): 95–­113. 13. Sybille Krämer, “Die Rettung des Ontologischen durch das Ontische? Ein Kommentar zu ‘operativen Ontologien,’” Zeitschrift für Medien-­und Kultur­forschung 8, no. 2 (2017): 125–­41. 14. As Geoffrey Winthrop-­Young puts it, “Indeed, one pithy way to describe the rise of Kulturtechniken in German cultural theory is to label it part of a largescale, albeit largely uncoordinated, Heidegger update. . . . The older Heidegger came to oppose philosophy to Denken (thinking); the study of cultural techniques provides a kind of flanking manoeuvre by relating the thinking of Sein (Being) to the processing and operating of bits and pieces of Seiendes (beings).” Geoffrey Winthrop-­Young, “Cultural Techniques: Preliminary Remarks,” Theory, Culture & Society 30, no. 6 (2013): 10. 15. Siegert, quoted in Geoffrey Winthrop-­Young, “Material World: An Interview with Bernhard Siegert,” Artforum, Summer 2015, https://www .artforum.com/print/201506/material-world-an-interview-with-bernhard -siegert-52281. For a resonating account drawing on Simondo, see Friedrich and Hoel, “Operational Analysis,” 9. 16. The idea is echoing André Leroi-­Gourhan and “operative chains” of techniques: “Techniques involve both gestures and tools, sequentially organized by means of a ‘syntax’ that imparts both fixity and flexibility to the series of operations involved.” Leroi-­Gourhan, Gesture and Speech, trans. Anna Bostock Berger (Cambridge, Mass.: MIT Press, 1993), 114. Cf. Marcel Mauss, “Techniques of the Body,” Economy and Society 2, no. 1 (1973): 70–­88. 17. Macho, quoted in Winthrop-­Young, “Cultural Techniques,” 8.

232   Notes to Chapter 1

18. Cornelia Vismann, “Cultural Techniques and Sovereignty,” trans. Ilinca Iurascu, Theory, Culture & Society 30, no. 6 (2013): 83–­93. 19. See, for example, Andreas Philippopoulos-­Mihalopoulos, Spatial Justice: Body, Lawscape, Atmosphere (London: Routledge, 2015); and Jannice Käll, Posthuman Property and Law (London: Routledge), 2022. 20. In Heidegger’s prose: “Mathematics also is not a reckoning in the sense of performing operations with numbers for the purpose of establishing quantitative results; but, on the contrary, mathematics is the reckoning that, everywhere by means of equations, has set up as the goal of its expectation the harmonizing of all relations of order, and that therefore ‘reckons’ in advance with one fundamental equation for all merely possible ordering.” Martin Heidegger, “Science and Reflection,” in The Question concerning Technology and Other Essays, trans. W. Lovitt (New York: Harper & Row, 1977), 170. 21. Winthrop-­Young, “Cultural Techniques,” 11–­1 2. 22. See Rosi Braidotti, Posthuman Knowledge (Cambridge: Polity, 2019). 23. Siegert, “Öffnen, Schließen, Zerstreuen, Verdichten,” 101–­3. 24. See, for example, Bernhard Siegert, “The Chorein of the Pirate: On the Origin of the Dutch Seascape,” Grey Room 57 (Fall 2007): 26–­47. 25. Karen Barad, Meeting the Universe Halfway: Quantum Physics and the Entanglement of Matter and Meaning (Durham: Duke University Press, 2007), 89, 26. 26. Gilles Deleuze and Félix Guattari, A Thousand Plateaus, trans. Brian Massumi (Minneapolis: University of Minnesota Press, 1987), 22–­23. 27. Barad, Meeting the Universe Halfway, 89. It is important to note that Barad’s take ends up using the even more specific term “ethico-­onto-­ epistemology” to underline the ethics of knowing, which, while not being emphasized in these passages I am writing, is not to be dismissed either. 28. Barad, 146. 29. See Mark Andrejevic, Automated Media, (New York: Routledge, 2020), 94–­112. 30. See Benjamin Bratton, “The City Wears Us. Notes on the Scope of Distributed Sensing and Sensation,” Glass Bead, 2017, https://www.glass-bead .org/article/city-wears-us-notes-scope-distributed-sensing-sensation/. 31. Here an interesting relation to the question of Mitchell’s metapictures is to be noted as they appear as images that “might be capable of reflection on themselves, capable of providing a second-­order discourse that tells us—­or at least shows us—­something about pictures.” W. J. T. Mitchell, Picture Theory: Essays on Verbal and Visual Representation (Chicago: University of Chicago Press, 1994), 38. 32. Jennifer Gabrys discusses in more detail techniques of observation and the ideals of invisibility and nonintervention vis-­à-­v is the broader field of “becoming environmental of computation,” where “earths and environments are programmed.” Gabrys, Program Earth: Environmental Sensing Technology and the Making of a Computational Planet (Minneapolis: University of Minnesota Press, 2016), 267.

Notes to Chapter 1   233 33. As Birgit Schneider writes: “Climatology was developed in conjunction with the methods of analytical graphics, there is a broad history of diagrammatic methods of visualization behind the history of climatology.” Schneider, “Climate Model Simulation Visualization from a Visual Studies Perspective,” WIREs Clim Change 3 (2012): 188 34. The two are not at all disconnected in the history of cartography as demonstrated by Christian Jacob’s significant research. See Jacob, The Sovereign Map: Theoretical Approaches in Cartography Throughout History, trans. Tom Conley (Chicago: University of Chicago Press, 2006). See also William Rankin, After the Map: Cartography, Navigation, and the Transformation of Territory in the Twentieth Century (Chicago: University of Chicago Press, 2016). 35. Matthew C. Hunter, “Modeling: A Secret History of Following,” in Design Technics: Archaeologies of Architectural Practice, ed. Zeynep Çelik Alexander and John May (Minneapolis: University of Minnesota Press, 2020), 59. To quote Hunter in full: “Unlike older views that posited modelling as fundamentally mimetic or governed by a criteria of resemblance, this recent literature stresses the cognitive utility and practical necessity of models that represent indirectly, through highly conventionalized techniques, and with substantial suspension of disbelief.” 36. On the navigational image, see also Doreen Mende, “The Navigation Principle: Slow Image,” e-­flux lecture, November 29, 2017, https://www.e-­flux .com/live/165096/e-­flux-­lectures-­doreen-­mende-­t he-­n avigation-­principle -­slow-­image/. 37. Bernhard Siegert, “The Map Is the Territory,” Radical Philosophy 169 (September/October 20): 13–­16. See also Roger Paez, Operative Mapping: Maps as Design Tools (Barcelona: Elisava & Actar, 2019). 38. On the pairing of continuity and rupture as the fundamental productive tension of media archaeological method in (new) film history and cinema studies, see Thomas Elsaesser, Film History as Media Archaeology: Tracking Digital Cinema. Amsterdam: Amsterdam University Press, 2016. 39. Bernhard Siegert, Cultural Techniques: Grids, Filters, Doors, and Other Articulations of the Real, trans. Geoffrey Winthrop-­Young (New York: Fordham University Press, 2015), 97. On addressing, see Ranjodh Singh Dhaliwal, “On Addressability, or What Even Is Computation?” Critical Inquiry 49, no. 1 (Autumn 2022), 1–­27. 40. Siegert, 97. 41. This follows the argument focusing on zero as the placeholder, the imaginary spot, as articulated by Brian Rotman: “The assertion that zero and zero-­like signs permeate several, very different, signifying codes and artefacts of the Renaissance is not unexpected if one thinks of changes in these codes in historically materialist terms: the historical emergence of mercantile capitalism rode on the vector of trade, business, commerce, finance, money. And money required a system of writing, which included book-­keeping and calculation, to enable it to function as an international medium of exchange. It was precisely to meet this need that double-­entry

234   Notes to Chapter 1

book-­keeping and Hindu numerals, both written in terms of zero, were introduced in Italy at the beginning of the Renaissance in the thirteenth century. Zero then was a principal element of Renaissance, that is to say mercantile capitalism’s, systems of writing from the beginning.” Rotman, Signifying Nothing: The Semiotics of Zero (New York: St Martin’s Press, 1987), 5. 42. Michel Foucault, The Order of Things: An Archaeology of the Human Sciences (New York: Vintage Books, 1994). 43. Foucault, 98. 44. Martin Heidegger, “The Age of the World Picture,” in Heidegger, Question Concerning Technology, 115–­36. 45. Mezzadra and Neilson, Politics of Operations. 46. See Margaret Davies, Property: Meanings, Histories, Theories (Abingdon: Routledge-­Cavendish, 2007); and Käll, Posthuman Property and Law. 47. David Joselit, “NFTs, or The Readymade Reversed,” October 175 (Winter 2021): 4. 48. Alberti quoted in Siegert, Cultural Techniques, 99: “[The veil is] woven of very thin threads and loosely intertwined, dyed with any color, subdivided with thicker threads according to parallel partitions, in as many squares as you like, and held stretched by a frame, which [veil] I place, indeed, between the object to be represented and the eye, so that the visual pyramid penetrates through the thinness of the veil.” 49. Ernst and Peter Neufert, Architects’ Data, 3rd ed., ed. Bousmaha Baiche and Nicholas Walliman (Hoboken: Wiley Blackwell, 1987). 50. Shannon Mattern, Code + Clay, Data + Dirt: Five Thousand Years of Urban Media (Minneapolis: University of Minnesota Press, 2017), 106. 51. Mattern, 106. 52. Siegert, Cultural Techniques, 120. Jeffrey Moro develops important insights into the media techniques of grids in his take on weather prediction: “Grid techniques enable weather prediction to make equivalent real atmospheres and virtual models, further enabling computation’s conceptual and material control over the planet itself.” Moro, “Grid Techniques for a Planet in Crisis: The Infrastructures of Weather Prediction,” Amodern 9 (April 2020), https://amodern.net/article/grid-techniques/. 53. These techniques of grids become a central element in computer graphics, including their role in enabling random-­access, which features as a central part in mediating between the difference of material screens and (digital) images. Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021), 55–­86. 54. Rosa Menkman, “Beyond Resolution” website https://beyondresolution .info/ABOUT. 55. Adrian Mackenzie, Machine Learners: Archaeology of a Data Practice (Cambridge, Mass.: MIT Press, 2017), 73. Mackenzie defines vector space as “a hyperspace of indefinite dimensions generated by the project mapping of data variables or features into distinct coordinate dimensions” (221).

Notes to Chapter 1   235 56. Mackenzie, 221. 57. On addressability and computation, see Dhaliwal, “On Addressability.” 58. Allan Sekula’s argument about the instrumental images of aerial surveillance is also apt in the broader sense of institutions of rationalized analysis from science to military: “The meaning of a photograph consisted of whatever it yielded to a rationalized act of ‘interpretation.’ As sources of military intelligence, these pictures carried an almost wholly denotative significance. Few photographs, except perhaps medical ones, were as apparently free from ‘higher’ meaning in their common usage. They seem to have been devoid of any rhetorical structure. But this poverty of meaning was conditional rather than immanent. Within the context of intelligence operations the only ‘rational’ questions were those that addressed the photograph at the indexical level, such as: ‘Is that a machine gun or a stump?’ In other words, the act of interpretation demanded that the photograph be treated as an ensemble of ‘univalent,’ or indexical, signs, signs that could only carry one meaning, that could point to only one object. Efficiency demanded this illusory certainty. Thus codes were developed for translating an ambiguous two-­dimensional image into an unequivocal knowledge of its three-­dimensional referent and back again into conventionalized and unambiguous two-­dimensional signs on a drawn map. A triangle stood for a dump; a circle with a central dot stood for a trench mortar. A terrain was reduced to a set of coded topographic features, ‘grounded’ by the digital logic of the grid. With the development of camouflage, a low-­level language game evolved in which the indexical status of the sign was thrown into question, thereby inflating the suspicions of the photo-­interpreter.” Sekula, “The Instrumental Image: Steichen at War,” in Photography against the Grain: Essays and Photo Works, 1973–­1983 (Halifax: The Press of the Nova Scotia College of Art and Design, 1984), 35 59. Sean Cubitt, “Image + After I: Photography as Print and as Scientific Instrument,” Fotomuseum Winterthur, September 1, 2017, https://www .fotomuseum.ch/en/series/processing/. 60. Cubitt, n.p. 61. Schneider, “Climate Model Simulation.” 62. Allan Sekula, “Traffic in Photographs,” in Photography against the Grain, 77–­101. 63. Ruha Benjamin, Race after Technology: Abolitionist Tools for the New Jim Code (Cambridge: Polity, 2019). 64. Simone Browne, Dark Matters: On the Surveillance of Blackness (Durham, N.C.: Duke University Press, 2015), 47. 65. Browne, 49. 66. Stefano Harney and Fred Moten, Undercommons: Fugitive Planning and Black Study (New York: Minor Compositions/Autonomedia, 2013), 93. See also Ian Baucom, Specters of the Atlantic Finance Capital, Slavery, and the Philosophy of History (Durham, N.C.: Duke University Press, 2005).

236   Notes to Chapter 1

67. Anna Lowenhaupt Tsing, The Mushroom at the End of the World: On the Possibility of Life in Capitalist Ruins (Princeton, N.J.: Princeton University Press, 2015), 39. 68. Mezzadra and Neilson, Politics of Operations, 3. 69. Mezzadra and Neilson, 3. 70. See Alberto Toscano and Jeff Kinkle, Cartographies of the Absolute (Winchester: Zero Books, 2015). See also chapter 4 of this work. 71. Mezzadra and Neilson, Politics of Operations, 5. 72. Mezzadra and Neilson, 5. 73. Mezzadra and Neilson, 162. 74. Mezzadra and Neilson, 67. 75. See, e.g., Ned Rossiter, Software, Infrastructure, Labor: A Media Theory of Logistical Nightmares (London: Routledge, 2016). 76. Rossiter, 142. 77. Rossiter. 78. Deborah Cowen, The Deadly Life of Logistics: Mapping Violence in Global Trade (Minneapolis: University of Minnesota Press, 2014), 3. 79. See, for example, Melinda Cooper, “Turbulent Worlds: Financial Markets and Environmental Crisis,” Theory, Culture, & Society 27, nos. 2–­3 (2010): 167–­90. 80. Gökçe Önal, “Media Ecologies of the ‘Extractive View’: Image Operations of Material Exchange,” Footprint, Autumn/Winter 2020, 31–­48. 81. These are topics left with less attention in this book only because there are so many excellent studies on them already. See, for example, Caren Kaplan, Aerial Aftermaths: Wartime from Above (Durham, N.C.: Duke University Press, 2018); Lisa Parks, Cultures in Orbit: Satellites and the Televisual (Durham, N.C.: Duke University Press, 2018); Antoine Bousquet, The Eye of War: Military Perception from the Telescope to the Drone (Minneapolis: University of Minnesota Press, 2018); Paul K. Saint-­A mour, “‘Applied Modernism’: Military and Civilian Uses of the Aerial Photomosaic,” Theory, Culture, and Society 28, no. 7 (December 2011), 241–­69; and Jeremy Packer and Joshua Reeves, Killer Apps: War, Media, Machine (Durham, N.C.: Duke University Press, 2020). 82. Brian Massumi, Ontopower: War, Powers, and the State of Perception (Durham, N.C.: Duke University Press, 2015). 83. These arguments continue to take shape in the next chapter, focusing on operational images in the context of AI and data. The topic of the “invisible image” has featured prominently as a point of reference in many of the approaches, due in no small part to Trevor Paglen’s mobilization of Farocki’s work. The chapter tackles the question of “invisibility.” It proposes a shift from the centrality of the invisible—­a s it emerged as a key onto-­ epistemological hinge for scientific imaging of unseen worlds—­to the invisual, a term by Mackenzie and Munster. This term takes a more central place in tying operations of images into questions of platformed data while outlining aspects of machine learning. The operational shift from image to

Notes to Chapter 2   237 nonimage becomes a central thread of tension throughout the next chapter as images become more like platforms, datasets, and models. 84. Siegert, “The Map Is the Territory,” 15.

2. What Is Not an Image? 1. Georges Didi-­Huberman, Images in Spite of All: Four Photographs from Auschwitz (Chicago: University of Chicago Press, 2008), 133. See also Tomas Dvořák, “Beyond Human Measure: Eccentric Metrics in Visual Culture,” in Photography Off the Scale, ed. Tomas Dvořák and Jussi Parikka (Edinburgh: Edinburgh University Press, 2021), 41–­60. 2. Roland Barthes, Camera Lucida, trans. Richard Howard (New York: Hill and Wang, 1981), 6. Peter Geimer builds this case, referring to Barthes, in Inadvertent Images: A History of Photographic Apparitions, trans. Gerrit Jackson (Chicago: University of Chicago Press, 2018), 1. 3. Erwin Panofsky, “Perspective as Symbolic Form” (1927), in Picture Industry: A Provisional History of the Technical Image, 1844–­2018, ed. Walead Beshty (Arles: Luma Foundation, 2018), 274. 4. Panofsky, 21. 5. Louise Amoore, Cloud Ethics: Algorithms and the Attributes of Ourselves and Others (Durham, N.C.: Duke University Press, 2020), 43. 6. Wendy Hui Kyong Chun, Programmed Visions: Software and Memory (Cambridge, Mass.: MIT Press, 2011), 18. 7. On transparency, gender, and labor in photography, see Michelle Henning on the “rendering invisible of reproductive work” in the case of Lucia Moholy. Henning, Photography: The Unfettered Image (London: Routledge, 2018), 44–­62. 8. Michel Foucault, Discipline and Punish: The Birth of the Prison, trans. Alan Sheridan (New York: Vintage Books, 1977), 200. Foucault’s note was specifically on the reversal of the principle of the dungeon (lack of light) to the excess of light and visibility of the panopticon. My aim is not to bring the discussion in this chapter to the military-­surveillance complex or the panopticon, but the notion of representational visibility can still be problematized. On current racial politics of exposure in data and AI systems, see Ruha Benjamin, Race after Technology: Abolitionist Tools for the New Jim Code (Cambridge: Polity, 2019). 9. Trevor Paglen, “Invisible Images (Your Pictures Are Looking at You),” New Inquiry, December 8, 2016, https://thenewinquiry.com/invisible -images-your-pictures-are-looking-at-you/. 10. Zach Blas, https://www.joaap.org/issue9/zachblas.htm. 11. See, for example, Gregor Schneider’s Invisible City installation in Athens (Onassis Stegi: 2017) https://www.onassis.org/whats-on/fast-forward -festival-4/fff4-invisible-city. 12. Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021).

238   Notes to Chapter 2

13. Quoted in Geimer, Inadvertent Images, 140. 14. Peter Geimer, “Was ist kein Bild? Zur ‘Störung der Verweisung,’” in Ordnung der Sichtbarkeit: Fotographie in Wissenschaft, Kunst und Technologie, ed. Peter Geimer (Frankfurt: Suhrkamp, 2002), 313–­41. 15. Geimer, 327–­28. 16. See Vilém Flusser, Into the Universe of Technical Images, trans. Nancy Ann Roth (Minneapolis: University of Minnesota Press, 2011), 16. The reference to the impossible comes through Rosa Menkman’s artistic research project on impossible images and resolution studies. See https:// beyondresolution.info/IM-POSSIBLE-BLOB. 17. Paul Virilio, The Vision Machine, trans. Julie Rose (Bloomington: Indiana University Press, 1994), 59. 18. Geimer, Inadvertent Images, 151. 19. A special thanks to Ryan Bishop for the dialogues on the prefix force of the concept. 20. Allan Sekula, “The Traffic in Photographs,” Art Journal 40, no. 1 (1981): 15–­25. See also Peter Szendy, The Supermarket of the Visible: Toward a General Theory of Images, trans. Jan Plug (New York: Fordham University Press, 2019). 21. Adrian Mackenzie and Anna Munster, “Platform Seeing: Image Ensembles and Their Invisualities,” Theory, Culture & Society 36, no. 5 (2019): 5. 22. Adrian Mackenzie, Machine Learners: Archaeology of a Data Practice (Cambridge, Mass.: MIT Press, 2017), 92–­102. It is here that Mackenzie develops Deleuze and Guattari’s notion of the partial observer when writing about machine learning: “Every observer in this domain is partial because the humans cannot see lines or curves in the multidimensional data, the functions that underpin models such as logistic regression or linear regression can transform data in the vector space but can’t show how well they see it, and the processes of optimization only see the results of the model and errors, not anything in its referential functioning” (100). 23. Sean Cubitt, “Mass Image, Anthropocene Image, Image-­Commons,” in Dvořák and Parikka, Photography Off the Scale, 25–­26. 24. Ingrid Hoelzl and Rémi Marie, Softimage: Towards a New Theory of the Digital Image (Bristol, UK: Intellect Books, 2015), 5. 25. Cubitt coins these examples, including Playfair’s, as “proto data media.” Cubitt, “Image + After I: Photography as Print and as Scientific Instrument,” Fotomuseum Winterthur, September 1, 2017, https://www.fotomuseum.ch /en/series/processing/. 26. Mackenzie and Munster, “Platform Seeing,” 6. 27. See Rebecca Uliasz, “Seeing Like an Algorithm: Operative Images and Emergent Subjects,” AI & Society 36 (2021), https://doi.org/10.1007/s00146 -020-01067-y. 28. Artist-­scholar Rosemary Lee works with similar questions, investigating machine l­earning techniques as the site of an expansion of (the meaning of) images. Lee also poses a question that is relevant for pattern recognition vis-à-vis representation: “Could such an image err on the side of hyper-­

Notes to Chapter 2   239 visibility, in which viewers may perceive all an image has to offer on its surface, yet be unable to grasp any meaning from it?” Lee, “Machine Learning and Notions of the Image” (PhD diss., University of Copenhagen, 2020, 147). 29. Vladan Joler and Matteo Pasquinelli, “The Nooscope Manifested. AI as Instrument of Knowledge Extractivism,” online essay, 2020, https:// nooscope.ai/. 30. Joler and Pasquinelli. 31. Joler and Pasquinelli, 12. 32. Benjamin Bratton, The Stack: Software and Sovereignty (Cambridge, Mass.: MIT Press, 2015), 374. 33. A term that I have, with Abelardo Gil-­Fournier, used in our work on environmental surfaces and images. See, for example, the video piece Seed, Image, Ground (2020) at https://www.fotomuseum.ch/en/situations-post/seed -image-ground/. 34. Throughout this discussion of platforms and invisual observations, there’s a strong connection to N. Katherine Hayles’s work on nonconscious cognition as a key parallel theoretical strand. The notion of invisuality speaks to this bundle of sensors and recursive loops of operations: “Nonconscious cognitions are increasingly embedded in complex systems in which low-­level interpretative processes are connected to a wide variety of sensors, and these processes in turn are integrated with higher-­level systems that use recursive loops to perform more sophisticated cognitive activities such as drawing inferences, developing proclivities, and making decisions that feed forward into actuators, which perform actions in the world.” N. Katherine Hayles, Unthought: The Power of the Cognitive Nonconscious (Chicago: University of Chicago Press, 2017), 24. 35. Bratton, Stack, 374. 36. Alexander R. Galloway, Gaming: Essays on Algorithmic Culture (Minneapolis: University of Minnesota Press, 2006). 37. “Although extreme, Baudrillard’s conflation of information (and thus computation) with transparency resonates widely in popular and scholarly circles, from fears over and propaganda behind national databases to examinations of ‘surveillance society.’ This conflation is remarkably at odds with the actual operations of computation: for computers to become transparency machines, the fact that they compute—­that they generate text and images rather than merely represent or reproduce what exists elsewhere—­must be forgotten.” Wendy Hui Kyong Chun, “On Software, or the Persistence of Visual Knowledge,” Grey Room 19 (Winter 2004): 26. 38. See, for example, Lev Manovich, Cultural Analytics (Cambridge, Mass.: MIT Press, 2020). 39. Nick Srnicek, Platform Capitalism (Cambridge: Polity, 2017); Bratton, Stack. 40. In Bratton’s definition: “A platform may be defined as a standards-­ based technical-­economic system that may simultaneously distribute interfaces into that system through their remote coordination and centralizes

240   Notes to Chapter 2

their integrated control through that same coordination. Platform logic refers first to the abstracted systems logic of platforms and the tendency on the part of some systems and social processes to transform themselves according to the needs of the platforms that may serve and support them, both in advance of their participation with that platform and as a result of that participation.” Stack, 374. 41. Ruggero Eugeni and Patricia Pisters, “The Artificial Intelligence of a Machine: Moving Images in the Age of Algorithms” NECSUS, July 6, 2020, https://necsus-ejms.org/the-artificial-intelligence-of-a-machine-movingimages-in-the-age-of-algorithms/. 42. Mark Graham, Rob Kitchin, Shannon Mattern, and Joe Shaw, eds., How to Run a City Like Amazon and Other Fables (London: Meatspace Press, 2019). 43. Instead of “visible/invisible,” Pasquinelli’s term for machine learning operations is another visual metaphor that refers to renaissance and seventeenth-­century painting: “Machine learning’s view of the world is also anamorphic: Even if it respects the shape, or topology, of the world, it distorts its proportions.” Matteo Pasquinelli, “How a Machine Learns and Fails—­ A Grammar of Error for Artificial Intelligence,” Spectres, November 20, 2019, https://spheres-journal.org/. 44. Jannice Käll, Posthuman Property and Law: Commodification and Control through Information, Smart Spaces and Artificial Intelligence (London: Routledge, 2022). 45. Sarah Keenan, Subversive Property: Law and the Production of Spaces of Belonging (Abingdon: Routledge, 2015), 17. 46. My aim here is not to venture into questions that experts in contemporary critical legal scholarship already discuss in detail. 47. Éric Alliez and Maurizio Lazzarato, Wars and Capital, trans. Ames Hodges (Los Angeles: Semiotext(e), 2018). 48. Hito Steyerl, Duty Free Art: Art in the Age of Planetary Civil War (London: Verso, 2017), 47. 49. Approaching these issues from a slightly different angle, more focused on the subject, see Shane Denson, Discorrelated Images (Durham, N.C.: Duke University Press, 2020). 50. Sean Cubitt on the politics of the mass image: “A database hoard of unregarded X-­rays or unwatched CCTV footage is an affirmation of the world’s existence. It states the facts. It is they rather than the tiny number of photographs we actually pay attention to that constitute the record of reality. The reality they record becomes reality through these records.” Cubitt, Anecdotal Evidence: Ecocritique from Hollywood to the Mass Image (Oxford: Oxford University Press, 2020), 225. On cloud computing and data as sites of pattern recognition, see also Amoore, Cloud Ethics, 49. 51. See Andrew Fisher, “On the Scales of Photographic Abstraction,” Photographies 9, no. 2 (2016): 203–­15.

Notes to Chapter 2   241 52. Actionrecognition.net online at http://actionrecognition.net/files/dset .php. 53. Marcel Mauss, “Techniques of the Body,” Economy and Society 2, no. 1 (1973): 72. 54. On cinema and cybernetics, see Ute Holl, Cinema, Trance and Cybernetics, trans. Daniel Hendrickson (Amsterdam: Amsterdam University Press, 2017). 55. While formatting pertains to “organisation of data into a dataset” (Joler and Pasquinelli, “Nooscope Manifested”), the produced statistical model itself becomes an operation of formatting the world according to its calculated, weighted prescriptions. 56. Steyerl, Duty Free Art, 47. 57. Daniel Rubinstein and Katrina Sluis, “A Life More Photographic: Mapping the Networked Image,” Photographies 1, no. 1 (2008): 9–­28. See also Andrew Dewdney and Katrina Sluis, eds., The Networked Image in Post-­digital Culture (London: Routledge, 2022). 58. For a complementary perspective on the visual culture of computing, including on infrastructures of rendering, see Ranjodh Singh Dhaliwal’s work including his PhD dissertation (forthcoming as a book): Dhaliwal, “Rendering the Computer: A Political Diagrammatology of Technology,” PhD diss., UC Davis, 2021, Publication # 28540442 on ProQuest. 59. For an interesting discussion of this history, see Leonardo Impett, “The Image-­ T heories behind Computer Vision,” #DHNord2020—­ T he Measurement of Images, November 19, 2020, https://www.meshs.fr/page /theimaget201021155656. 60. See, for example, “Excavating AI: The Politics of Images in Machine Learning Training Sets,” Excavating AI, September 19, 2019, https:// excavating.ai/. 61. Nicolas Malevé, “On the Data Set’s Ruins,” AI & Society 36 (2021), https://doi.org/10.1007/s00146-020-01093-w. 62. See Uliasz, “Seeing Like an Algorithm.” 63. Malevé, “On the Data Set’s Ruins.” 64. This is a candid reference to Deleuze and Guattari’s notion of the haecceity as a singularity consisting of intensities, not merely a divisible instance of a general concept or unit. “A cloud of locusts carried in by the wind at five in the evening; a vampire who goes out at night, a werewolf at full moon. It should not be thought that a haecceity consists simply of a decor or backdrop that situates subjects, or of appendages that hold things and people to the ground. It is the entire assemblage in its individuated aggregate that is a haecceity; it is this assemblage that is defined by a longitude and a latitude, by speeds and affects, independently of forms and subjects, which belong to another plane.” Gilles Deleuze and Félix Guattari, A Thousand Plateaus, trans. Brian Massumi (Minneapolis: University of Minnesota Press, 1987), 262. 65. Amoore, Cloud Ethics, 163.

242   Notes to Chapter 2

66. Referring to the JPEG, Hoelzl and Marie write: “The ‘Photographic’ in Joint Photographic Experts Group stands for ‘continuous-­tone still images,’ which are images with smooth transitions such as paintings, photographs or drawings, in contrast to ‘discrete tone images’ such as line drawings. For the JPEG group, the ‘photographic’ is no longer tied to a specific recording and printing technology; instead, the term designates an array of digital images that can be compressed in the same manner. Put differently, ‘photographic’ designates a particular aesthetic distribution (continuous-­tone) of pixels that can be correlated with each other during the compression process.” Hoelzl and Marie, Softimage, 69. 67. This is on par with Kember and Zylinska, who argue that new media should not be seen as a collection of “discrete objects” but as processes of mediation and remediation. Sarah Kember and Joanna Zylinska, Life after New Media: Mediation as a Vital Process (Cambridge, Mass.: MIT Press, 2012). 68. Impett, “Image-­T heories.” 69. Relevantly, Louise Amoore’s use of the term “aperture” carries forward a camera-­based technique into the question of a machine learning and cloud analytic, but as she points out, it is actually about cultural techniques of “dividing, selecting, and narrowing the focus of attention.” Amoore, Cloud Ethics, 162. Bootstrapping optical terms is done both to establish a continuity from digital data to earlier regimes of distribution of visuality and to ensure those are not seen only in terms of seeing or optics of vision: they are, fundamentally, about openings and ethicopolitical closures, to paraphrase Amoore. 70. Daniel Chávez Heras and Tobias Blanke, “On Machine Vision and Photographic Imagination,” AI & Society 36 (2021), https://www.springerprofessional .de/en/on-machine-vision-and-photographic-imagination/18595710 ?fulltextView=true. 71. Allan Sekula, “The Body and the Archive,” October 39 (Winter 1986): 6. See also Jane Birkin, Archive, Photography and the Language of Administration (Amsterdam: Amsterdam University Press, 2021). 72. Sekula, 5. 73. Sekula, 58. 74. Trevor Paglen, “Invisible Images.” The phrasing was already used by Paul Virilio in The Vision Machine: “Once we are definitely removed from the realm of direct or indirect observation of synthetic images created by the machine for the machine, instrumental virtual images will be for us the equivalent of what a foreigner’s mental pictures already represent: an enigma.” (60). 75. On the chemistry of visual culture, see Esther Leslie, Synthetic Worlds: Nature, Art and the Chemical Industry (London: Reaktion, 2005). 76. Malevé, “On the Data Set’s Ruins.” 77. Benjamin, Race after Technology. 78. “PPB consists of 1,270 individuals from three African countries (Rwanda, Senegal, and South Africa) and three European countries (Iceland,

Notes to Chapter 2   243 Finland, and Sweden), selected for gender parity in the national parliaments.” The Gender Shades project site is at https://www.media.mit.edu/projects /gender-shades/overview/. 79. Pasquinelli, “How a Machine Learns and Fails.” 80. Pasquinelli. 81. Friedrich Kittler, Optical Media, trans. Anthony Enns (Cambridge: Polity, 2010), 38. 82. Matthew Fuller and Andrew Goffey, Evil Media (Cambridge, Mass.: MIT Press, 2012). 83. This is a point articulated in more detail by Amoore in Cloud Ethics. Furthermore, to note: the notion of bias has its technical meaning in deep learning and this is in danger of getting confused with the more social, everyday meaning of bias. See also Käll, Posthuman Property and Law. 84. See Noopur Raval, “Interrupting Invisibility in a Global World,” ACM Interactions, July–­ August 2021, https://interactions.acm.org/archive/view /july-august-2021/interrupting-invisibility-in-a-global-world. 85. Here, the work of ethnographic studies of various aspects of digital labor has been instrumental in elaborating the embodied aspects and real-­ world situations of, for example, Turkers, but also the other labor of content moderation, among other things. See, for example, Lilly Irani, “Difference and Dependence among Digital Workers: The Case of Amazon Mechanical Turk,” South Atlantic Quarterly 114, no. 1 (2015): 225–­3 4. See also Ben Burbridge, Photography after Capitalism (London: Goldsmiths Press, 2020). 86. Malevé, “On the Data Set’s Ruins.” 87. On topos and media archaeology, see Erkki Huhtamo, “Dismantling the Fairy Engine. Media Archaeology as Topos Study,” in Media Archaeology: Approaches, Applications, and Implications, ed. Erkki Huhtamo and Jussi Parikka (Berkeley: University of California Press, 2011), 27–­47. 88. Joanna Zylinska, “Undigital Photography: Image-­M aking beyond Computation and AI,” in Dvořák and Parikka, Photography Off the Scale, 231–­52. 89. As Rosemary Lee points out in her PhD dissertation about AI and images, there are interesting earlier examples of experimental processes of algorithmization of the operator as a way to create images. As Lee argues, avant-­garde examples range from early twentieth-­century automatism and procedural practices as well as post–­World War II avant-­garde techniques of regularity, chance, and procedure, not least in the works of John Cage, Fluxus, etc. But also in early computer art, the idea of the extended site of execution of algorithmic instructions takes place. Quoting Lee: “Vera Molnár, an early proponent of computer art, explored the artistic and aesthetic potential of computational processes in paintings created according to what the artist referred to as an imaginary machine, or ‘machine imaginare’ (1960). Following algorithmic instructions to execute the paintings, the artist took on the conceptual role of a computer—­one that (or whom) computes. The

244   Notes to Chapter 2

computational dimension of these works is attributable, thus, not to a digital computer, but to computation as a procedure enacted in order to produce the work.” Lee, “Machine Learning,” 54. 90. Quoted in Alex Soojung-­Kim Pang, “‘Stars Should Henceforth Register Themselves’: Astrophotography at the Early Lick Observatory,” British Journal for the History of Science 30, no. 2 (June 1997): 188. On the standardization of images as labor, he writes: “As historians have come to realize, the assumption that because scientific practices are social they are nothing but social misses what is most interesting: how these practices come to produce things—­standardized animals, theories, experimental data, photographs—­ that are reasonably trustworthy” (202). 91. Lindsay Smith Zrull, “Women in Glass: Women at the Harvard Observatory during the Era of Astronomical Glass Plate Photography, 1875–­1975,” Journal for the History of Astronomy 52, no. 2 (2021): 115–­46. 92. “Labour in a single shot” workshops’ website at https://www.labour-in -a-single-shot.net/en/films/. 93. I will return later to the topic through a discussion of logistics of the image as it pertains to capitalism and labor through Alberto Toscano and Jeff Kinkle, Cartographies of the Absolute (Winchester: Zero Books, 2015). See also Thomas Stubblefield, “Database Labour: Supply Chains, Logistics, and Flow,” in Labour in a Single Shot: Critical Perspectives on Antje Ehmann and Harun Farocki’s Global Video Project, ed. Roy Grundmann, Peter J. Schwartz, and Gregory H. Williams (Amsterdam: Amsterdam University Press, 2022), 329–­45. 94. On a slightly different level, but related to data and work, Aaron Koblin’s sheep work is one such proxy of labor: 10,000 drawings of a sheep commissioned through the Amazon Mechanical Turk platform, http://www .aaronkoblin.com/work/thesheepmarket/. 95. “Image-­model” is the architecture theorist John May’s term, referring to the contemporary form of statistical images and statistical seeing that relates to possible futures: “Image-­models contain simulations of all possible futures.” John May, Signal, Image, Architecture (New York: Columbia Books on Architecture and the City, 2019), 97. 96. Estelle Blaschke, “From Microform to the Drawing Bot: The Photographic Image as Data,” Grey Room 75 (Spring 2019): 65. On questions of active, operational memory in computer history with implications for operational archives, see Orit Halpern, Beautiful Data: A History of Vision and Reason since 1945 (Durham, N.C.: Duke University Press, 2014), 67. Halpern writes of memory as “an active site for the management and execution of . . . operations.” See also Amoore, Cloud Ethics, 50. 97. Here, I would not go so far as to claim that diagrams are the opposite of images, but that the operative diagrammatic—­as well as the cultural techniques such as tables and charts—­features as the persistent distorted doppelgänger variations of the operational image. Wolfgang Ernst’s term “operative diagrammatics” develops C. S. Peirce’s notion of a graphic visualization of

Notes to Chapter 2   245 a scheme that Ernst, in his own words, extends from the observer to the material circuitry. Instead of reading this as opposite to an image as Ernst implies, I see these as different aspects of the image, actualized in the operational image that can also feature diagrams. See Wolfgang Ernst, Technológos in Being: Radical Media Archaeology and the Computational Machine (London: Bloomsbury, 2021). 98. Janet Vertesi’s Seeing Like a Rover is an excellent case of operational images in scientific remote sensing (such as the Mars Rover). The analysis of digital images demonstrates the different practices of composition at play. This ranges from the centrality of the community of scientists and operators to the described importance of the digital CCD (Charge-­Coupled Device) sensor as the basis for photon detection that can be operationalized in relation to diagnostics of material composition: “When paired with optical filters, pixel values reveal information about an imaged object’s ability to reflect light in a particular wavelength. This can be used as a diagnostic tool to identify mineralogical composition” (75). Similarly, an example like the use of false color images is fitting for addressing the invisuality of epistemic imaging. To quote Vertesi again: “False color, to the Rover scientists, does not imply a false image; nor is the image artificially painted to produce spectacular views. Rather, the colors arise from a mathematical relation between pixels across the included image frames, enabling the viewer to see when objects in the scene reflect light in different wavelengths. Thus the distribution of colors in a false color image demarcates, highlights, or otherwise identifies invisible features of the imaged terrain” (80). From Janet Vertesi, Seeing Like a Rover: How Robots, Teams, and Images Craft Knowledge of Mars (Chicago: University of Chicago Press, 2015). 99. Joseph Vogl, “Becoming-­Media: Galileo’s Telescope,” trans. Brian Hanrahan, Grey Room 29 (Winter 2008): 14–­25, 17. 100. Vogl, 22. See also Dvořák, “Beyond Human Measure.” 101. See, for example, the special issue of Log journal on “Model Behavior” (Fall 2020) for an overview of discussions in architecture. 102. Geimer, Inadvertent Images, 146. 103. Lorraine Daston and Peter Galison, Objectivity (New York: Zone Books, 2007), 256. 104. See Amoore, Cloud Ethics, 47–­55. 105. See Joel Snyder, “Visualization and Visibility,” in Picturing Science, Producing Art, ed. Caroline A. Jones and Peter Galison (New York: Routledge, 1998), 379–­99. 106. Pasquinelli, “How a Machine Learns and Fails.” 107. See Birgit Schneider, “Climate Model Simulation Visualization from a Visual Studies Perspective,” WIREs Clim Change 3 (2012):185–­93, https://doi .org/10.1002/wcc.162. 108. Joler and Pasquinelli, “Nooscope Manifested.” 109. Amoore’s notion of the cloud analytic is here again a relevant reference point. Amoore, Cloud Ethics.

246   Notes to Chapter 2

110. Amoore. 111. See, for example, Bernhard Siegert, Passage des Digitalen (Berlin: Brinkman & Bose, 2003); John Bender and Michael Marrinan, The Culture of the Diagram (Stanford, Calif.: Stanford University Press, 2010); Sybille Krämer, Figuration, Anschauung, Erkenntnis: Grundlinien einer Diagrammatologie (Berlin: Suhrkamp, 2016). 112. The focus on practice, on praxis, is here indebted to McKenzie Wark’s discussion on the problematic mode of contemplation and erasure of praxis in Object-­Oriented Ontology. See Wark, General Intellects: Twenty-­One Thinkers for the Twenty-­First Century (London: Verso, 2017), 269–­85 113. Gilles Deleuze, Foucault, trans. Seán Hand (Minneapolis: University of Minnesota Press, 1988). 114. W. J. T. Mitchell, “What Is an Image?” New Literary History 15, no. 3 (Spring 1984): 503–­4. 115. On the politics of AI, see Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (New Haven: Yale University Press, 2021). 116. Daston and Galison, Objectivity, 383. See also Nina Samuel, “Images as Tools: On Visual Epistemic Practices in the Biological Sciences,” Studies in History and Philosophy of Biological and Biomedical Sciences 44, no. 2 (June 2013): 225–­36 117. See Dhaliwal, “Rendering the Computer.” See also NVIDIA, “GPU-­Based Deep Learning Inference: A Performance and Power Analysis,” White Paper, November 2015, https://www.nvidia.com/content/tegra/embedded-systems /pdf/jetson_tx1_whitepaper.pdf; and Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Advances in Neural Information Processing Systems 25 (2012): 1–­9, https:// proceedings.neurips.cc/paper/2012/file/c399862d3b9d6b76c8436e924a68c45b -Paper.pdf. 118. Gaboury, Image Objects, 27–­5 4.

3. The Measurement-­Image 1. Matthew C. Edney, Mapping an Empire: The Geographical Construction of British India, 1765–­1843 (Chicago: University of Chicago Press, 1997). See also Jill H. Casid, Sowing Empire: Landscape and Colonization (Minneapolis: University of Minnesota Press, 2005). 2. Mackenzie and Munster scaffold their analysis of platform seeing and invisuality through Henri Bergson’s early twentieth-­century ontology that described the universe and matter as images. Adrian Mackenzie and Anna Munster, “Platform Seeing: Image Ensembles and Their Invisualities,” Theory, Culture & Society 36, no. 5 (2019): 3–­5. 3. Donna Haraway, “Persistence of Vision,” in The Visual Culture Reader, ed. Nicholas Mirzoeff (New York: Routledge, 2002), 679. 4. Vilém Flusser, Into the Universe of Technical Images, trans. Nancy Ann Roth (Minneapolis: University of Minnesota Press, 2011), 16.

Notes to Chapter 3   247 5. Flusser, 16. 6. See Xiaowei Wang, Blockchain Chicken Farm (New York: FSG Originals/ Logic, 2020). Technical images might work by statistical abstractions, but this does not lessen their material force. It only demonstrates that abstractions also organize worlds—­for better or for worse. Many cultural theorists still say “for worse” due to an aversion to abstractions seen as disembodied and alienating. Still, there might be more nuanced accounts available about the politics of abstractions that can be developed through discussions of operational images. For example, research on the politics of climate images and models is a good example of awareness of the connection between concrete (in)visual practices and calculated abstractions such as predictions and forecasting. See Birgit Schneider, Klimabilder: Eine Genealogie globaler Bildpolitiken von Klima und Klimawandel (Berlin: Matthes & Seitz, 2018); and Paul N. Edwards, A Vast Machine: Computer Models, Climate Data, and the Politics of Global Warming (Cambridge, Mass.: MIT Press, 2010). 7. Volker Pantenburg, “Working Images: Harun Farocki and the Operational Image,” in Image Operations: Visual Media and Political Conflict, ed. Jens Eder and Charlotte Klonk (Manchester: Manchester University Press, 2016), 55. 8. Nora Alter, “Two or Three Things I Know about Harun Farocki,” October 151 (Winter 2015): 151–­58. 9. Mary Terrall, “Representing the Earth’s Shape: The Polemics Surrounding Maupertuis’s Expedition to Lapland,” Isis 83, no. 2 (June 1992): 218. 10. Valérie November, Eduardo Camacho-­Hübner, and Bruno Latour, “Entering a Risky Territory: Space in the Age of Digital Navigation,” Environment and Planning D: Society and Space 28, no. 4 (August 2010): 582. 11. This topic is also discussed in our recent book Photography Off the Scale, which is one reference point for my chapter here and my colleague Tomas Dvořák’s recent writing on photographic measure. Tomas Dvořák, “Beyond Human Measure: Eccentric Metrics in Visual Culture,” in Photography Off the Scale, ed. Tomas Dvořák and Jussi Parikka (Edinburgh: Edinburgh University Press, 2021), 41–­60; Tomas Dvořák and Jussi Parikka, “Measuring Photographs,” Photographies 14, no. 3 (2021): 443–­57. 12. Harun Farocki, “Commentary from ‘Bilder der Welt und Inschrift des Krieges,’” Discourse 15, no. 3 (Spring 1993): 79, 78–­92. 13. Albrecht Meydenbauer, “Das Denkmäler-­A rchiv und seine Herstellung durch das Messbild-­Verfahren,” 1896, online scanned at Landesbibliothek Coburg, http://digital.bib-bvb.de/view/bvbmets/viewer.0.6.4.jsp?folder _id=0&dvs=1662325927377~5&pid=10981138&locale=en_US&usePid1 =true&usePid2=true. 14. Arago quoted in Patrick Maynard, The Engine of Visualization: Thinking through Photography (Ithaca: Cornell University Press, 1997), 205. 15. Here the terminology of “immobile mobiles” (Latour) is apt in describing this process of transporting measurements and images as data back to centers of calculation (centers of viewing, seeing, analyzing). Bruno Latour,

248   Notes to Chapter 3

“Visualisation and Cognition: Drawing Things Together,” Knowledge and Society: Studies in the Sociology of Culture Past and Present 6 (1986): 32. 16. On instrumental images and climatology, see Birgit Schneider, Klimabilder: Eine Genealogie globaler Bildpolitiken von Klima und Klimawandel (Berlin: Matthes & Seitz, 2018). 17. Joel McKim, “Into the Universe of Rendered Architectural Images,” Unthinking Photography, June 2019, https://unthinking.photography/articles /into-the-universe-of-rendered-architectural-images. To quote McKim in full: “Other developments in rendering software, by companies like Lumion, are making the translation of CAD drawings to photo-­realistic still and animated visualisations virtually automated and instantaneous through the use of image libraries of objects, scenery and materials. Architectural renders, in other words, are increasingly becoming crucial tools within real-­ time design practices—­we could call them ‘operative’ or ‘operational’ images, to use a term employed by both the filmmaker Harun Farocki and the philosopher Sybille Krämer. As operational images, renders need perhaps to be interpreted and questioned less in terms of what they represent, and more in terms of what they are capable of doing, what sorts of effects they bring about, and what kinds of futures they project.” 18. See Raymond Bellour, “The Photo-­Diagram,” in Against What? Against Whom? ed. Antje Ehmann and Kodwo Eshun (London: Koenig Books and Raven Row, 2009), 147. 19. Sybille Krämer, Figuration, Anschauung, Erkenntnis: Grundlinien einer Diagrammatologie (Berlin: Suhrkamp, 2016), 14–­17, 131. See also again Latour, “Visualisation and Cognition.” 20. Joerg Albertz, “140 years of ‘Photogrammetry’: Some Remarks on the History of Photogrammetry,” asprs.org, May 2007, 504, https://www.asprs .org/wp-content/uploads/pers/2007journal/may/lookback.pdf. 21. Albertz, 504. 22. For the price list, see Meydenbauer, “Das Denkmäler-­A rchiv.” 23. Lacan in “On the Gaze as Objet Petit a,” quoted in Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021), 38. 24. On the concept of the navigational image, see the e-­flux special issue (June 2019) Navigation beyond Vision at https://www.e-flux.com/journal/101/. 25. Eyal Weizman, Forensic Architecture: Violence at the Threshold of Detectability (Cambridge, Mass.: MIT Press, 2017), 17–­21. 26. Weizman, 20. 27. For several insights into the history of science of technical, instrumental images, see Horst Bredekamp, Vera Dünkel, and Birgit Schneider, eds., The Technical Image: A History of Styles in Scientific Imagery (Chicago: University of Chicago Press, 2015). 28. See, for example, Aimé Laussedat, Leçons sur l’art de lever les plans : Comprenant les levers de terrain et de bâtiment, la pratique du nivellement ordi-

Notes to Chapter 3   249 naire, et le lever des courbes horizontales à l’aide des instruments les plus simples (Paris: Impr. Impériale, 1861). 29. Samuel Y. Edgerton, The Mirror, The Window and the Telescope: How Renaissance Linear Perspective Changed Our Vision of the Universe (Ithaca, N.Y.: Cornell University Press, 2009), 168–­69. 30. A more detailed media archaeology would pay attention to such key operations as Girard Desargues’s perspective calculation from the mid-­ seventeenth century. Here, perspectival drawing and modeling moved to a methodology that allowed us to model objects in hypothetical geometrical space. Thank you to Pasi Väliaho for pointing to Desargues’s fascinating work. For a link between perspective and phantom space (as well as architecture), see Iman Fayyad, “Phantom Space,” Log 43 (Summer 2018): 139–­52. 31. Mario Carpo, The Second Digital Turn: Design beyond Intelligence (Cambridge, Mass.: MIT Press, 2017), 101–­14. For a detailed analysis on why computer graphics is not about linear perspective as visuality but as instructions for rendering, see Gaboury, Image Objects. 32. Also, Aimé Laussedat seems to refer to it in such terms at the back of the history of the linear perspective and descriptive geometry. Laussedat, Recherches sur les instruments, les méthodes et le dessin topographiques: Tome 2 (Paris, 1898–­1903), 14, Bibliothèque nationale de France, département Sciences et techniques, 8-­V-­27824 (2). 33. Mary Ann Doane, “Temporality, Storage, Legibility: Freud, Marey, and the Cinema,” Critical Inquiry 22, no. 2 (Winter 1996), 328. 34. Apophenia in these terms appears in Benjamin Bratton’s work as well as Hito Steyerl’s. See, for example, chapter 5 in Steyerl, Duty Free Art: Art in the Age of Planetary Civil War (London: Verso, 2017). 35. Alphonse Bertillon, “Selections from Theoretical Study of Signalment (1896),” in Picture Industry: A Provisional History of the Technical Image, 1844–­2018, ed. Walead Beshty (Zürich: Luma and CCS Bard, 2019), 181; Allan Sekula, “The Body and the Archive,” October 39 (Winter 1986): 3–­6 4. 36. See also Erika Balsom, “Moving Bodies: Captured Life in the Late Works of Harun Farocki,” Journal of Visual Culture 18, no. 3 (2019): 358–­7 7. 37. Balsom, 363. 38. Pasi Väliaho, Mapping the Moving Image: Gesture, Thought and Cinema circa 1900 (Amsterdam: Amsterdam University Press, 2010), 40–­51. 39. Marta Braun, Picturing Time: The Work of Etienne-­Jules Marey (Chicago: University of Chicago Press, 1992), 66 40. Balsom, “Moving Bodies,” 363. As a sidenote, I point the curious reader to look up a particular measuring device that Alberti proposed in the sixteenth century, which itself was to reproduce proportions and shapes of the human body / sculpture with precision. The device, called “the definitor,” features in Alberti’s De Statua (at least in the first Italian translation of 1568) and is also described in Carpo, Second Digital Turn, 117–­19. 41. “In Western societies, straight lines are ubiquitous. We see them

250   Notes to Chapter 3

everywhere, even when they do not really exist. Indeed, the straight line has emerged as a virtual icon of modernity, an index of the triumph of rational, purposeful design over the vicissitudes of the natural world.” Tim Ingold, Lines: A Brief History (Abingdon: Routledge, 2007), 152 42. William Rankin, After the Map: Cartography, Navigation, and the Transformation of Territory in the Twentieth Century (Chicago: University of Chicago Press, 2016); Sean Cubitt, The Practice of Light: A Genealogy of Visual Technologies from Prints to Pixels (Cambridge, Mass.: MIT Press, 2014), 45–­7 7. 43. Friedrich Kittler, Optical Media: Berlin Lectures 1999 trans. Anthony Enns (Cambridge: Polity, 2010), 94. 44. Carpo, Second Digital Turn, 115. 45. Carpo, 115–­17. 46. Gaspard Monge, Géométrie descriptive: 4e édition augmentée d’une théorie des ombres et de la perspective (Paris: Vve Courcier, 1820), 1. For an overview on Mongean geometry as well as an insight into the British alternative context of using isometric perspectives, see Peter J. Booker, A History of Engineering Drawing (London: Northgate, 1979). 47. Monge, 1. 48. Kittler, Optical Media, 94. 49. Jean-­Baptiste-­A ntoine-­A imé Sanson de Pongerville, Gaspard Monge et l’expédition d’Égypte (Paris, 1860). 50. John Bender and Michael Marrinan, The Culture of the Diagram (Stanford, Calif.: Stanford University Press 2010), 156; Booker, History of Engineering Drawing, 92. 51. Arago, quoted in Ariella Aïsha Azoulay, Potential History: Unlearning Imperialism (London: Verso, 2019), 39 (epub version). 52. Laussedat, Recherches sur les instruments, 4–­6, 14. 53. Laussedat, 14. 54. Thank you to Xinyi Wen for pointing me to the case of the graphical telescope by Varley. See Xinyi Wen, “The Quest for ‘Perfect Truth’ Embodiment, Objectivity, and Cornelius Varley’s Graphic Telescope” (MPhil essay, Cambridge University, 2020). Wen is referring to the questions of truth-­to-­ nature and mechanical objectivity as articulated by Lorraine Daston and Peter Galison, Objectivity (New York: Zone Books, 2010). On camera lucida and Laussedat, see also Stuart I. Granshaw, “Laussedat Bicentenary: Origins of Photogrammetry,” Photogrammetric Record 34, no. 166 (June 2019): 131–­32. 55. Kittler, Optical Media, 118. 56. See Matthew C. Hunter, Wicked Intelligence Visual Art and the Science of Experiment in Restoration (Chicago: University of Chicago Press, 2013). 57. Larrie D. Ferreiro, Measure of the Earth (New York: Basic Books, 2011), 252. Furthermore, territorial expansion meant in many cases property expansion, including land taxation rights for the East Indian Company. Thanks for this note goes to Pasi Väliaho. 58. Schneider, Klimabilder, 146–­49.

Notes to Chapter 3   251 59. Josef Maria Eder, History of Photography, 4th ed., trans. Edward Epstean (New York: Dover, 1978), 398 (Orig. pub. 1932). 60. Granshaw, “Laussedat Bicentenary,” 128. 61. John Law, “On the Methods of Long-­Distance Control: Vessels, Navigation and the Portuguese Route to India,” Sociological Review 32, no. 1 (1984): 234–­63. See also Peter Sloterdijk, In the Interior World of Capital, trans. Wieland Hoban (Cambridge: Polity, 2013), 98–­101. 62. Richard Copeland’s preface to Beautemps-­Beaupré’s An Introduction to the Practice of Nautical Surveying and the Construction of Sea-­Charts, trans. Richard Copeland (London: R. H. Laurie, 1823), vi. 63. Cubitt, Practice of Light, 53. 64. Laussedat, Recherches sur les instruments, 12n2, 23–­24. 65. Anthony Vidler, “Photourbanism: Planning the City from Above and from Below,” in A Companion to the City, ed. Gary Bridge and Sophie Watson (Malden, Mass.: Blackwell, 2000), 35–­45. 66. Cubitt, Practice of Light, 70. 67. Schneider, Klimabilder, 163–­68. 68. Ingold, Lines. 69. Ingold, 155–­56. 70. Ingold, 155–­56. 71. See Weizman, Forensic Architecture; Matthew Fuller and Eyal Weizman, Investigative Aesthetics: Conflicts and Commons in the Politics of Truth (London: Verso, 2021). 72. Sloterdijk, In the Interior World of Capital, 100. 73. November, Camacho-­Hübner, and Latour, “Entering a Risky Territory,” 585. 74. November, Camacho-­Hübner, and Latour, 586–­87. 75. Ariella Aïsha Azoulay, Potential History, 40 (epub version). 76. For a detailed view that is more complex than what I mention, see Michael Kershaw, “The ‘Nec Plus Ultra’ of Precision Measurement: Geodesy and the Forgotten Purpose of the Metre Convention,” Studies in History and Philosophy of Science 43 (2012): 563–­76. 77. Rankin, After the Map, 280. 78. Rankin, 281. 79. Law, “Methods of Long-­Distance Control.” 80. Mary Louise Pratt, Imperial Eyes: Travel Writing and Transculturation, 2nd ed. (London: Routledge, 2008), 15. 81. In what is now northern Finland, including locations such as Tornio and Kittilä, or “Kittis.” 82. In what is now Ecuador. 83. Pierre-­Louis Moreau de Maupertuis, La figure de la terre: Déterminée par les observations de Mm de Maupertuis (Paris, 1738), v. 84. Ferreiro, Measure of the Earth, 28. For travel descriptions of the Northern expedition, see de Maupertuis, La figure de la terre.

252   Notes to Chapter 3

85. Terrall, “Representing the Earth’s Shape,” 219. 86. On differences between photogrammetry and remote sensing, see also Armin Gruen, “Everything Moves: The Rapid Changes in Photogrammetry and Remote Sensing,” Geo-­spatial Information Science 24, no. 1 (2021): 33–­49, https://doi.org/10.1080/10095020.2020.1868275. 87. Latour, “Visualisation and Cognition.” 88. Latour, 220. 89. Isaac Todhunter, A History of the Mathematical Theories of Attraction and the Figure of the Earth: From the Time of Newton to That of Laplace, vol. 1. (London, 1873), 97. 90. Terrall, “Representing the Earth’s Shape,” 227–­37. In addition, gravity measurements with the help of a pendulum were meant to correlate results about the shape of the earth. “Important pendulum experiments were made at Pello, which is close to Kittis. The result is that a pendulum which oscillates in a second at Paris will make 59 more oscillations in 24 hours at Pello than at Paris” (99). 91. Terrall, 223. 92. Terrall, 223. 93. Terrall, 223. 94. Ferreiro, Measure of the Earth, 150. The measurement unit of “toise” is about 6.39 feet, or 1.949 meters. 95. See also Denis Cosgrove, A Cartographic Genealogy of the Earth in Western Imagination (Baltimore: Johns Hopkins University Press, 2001). 96. On the point of simulation of the earth from Albrecht Dürer onward, see also Leon Gurevitch, “Google Warming: Google Earth as Eco-­ machinima,” Convergence: The International Journal of Research into New Media Technologies 20, no. 1 (2014): 85–­107. 97. Ferreiro, Measure of the Earth, 133. 98. Azoulay, Potential History, 5–­6. 99. Gurevitch, “Google Warming.” 100. Carpo, Second Digital Turn, 120–­29. 101. Gaboury, Image Objects. See also Steve F. Anderson, Technologies of Vision: The War between Data and Images (Cambridge, Mass.: MIT Press, 2017).

4. Operational Aesthetic 1. Leon Gurevitch, “Google Warming: Google Earth as Eco-­machinima,” Convergence: The International Journal of Research into New Media Technologies 20, no. 1 (2014): 103. See also Denis Cosgrove, A Cartographic Genealogy of the Earth in Western Imagination (Baltimore: Johns Hopkins University Press, 2001). 2. Roland Meyer, “Formatting Faces: Standards of Production, Networks of Circulation, and the Operationalization of the Photographic Portrait,” in Format Matters, ed. Marek Jancovic, Axel Volmar, and Alexandra Schneider (Lüneburg: Meson Press, 2020), 163. On formats more broadly, see also

Notes to Chapter 4   253 Jonathan Sterne, MP3: The Meaning of a Format (Durham, N.C.: Duke University Press, 2012). 3. Thomas Elsaesser, “Simulation and Labour of Invisibility: Harun Farocki’s Life Manuals,” Animation: An Interdsciplinary Journal 12, no. 3 (2017): 214–­29. 4. Rosa Menkman, Beyond Resolution (2020) project page, https:// beyondresolution.info/Beyond-Resolution-1. 5. A reminder that the notion of territory is, itself, already a calculative category as Stuart Elden notes: “It comprises techniques for measuring land and controlling terrain, and measure and control—­the technical and the legal—­must be thought alongside the economic and strategic.” Elden, “Land, Terrain, Territory,” Progress in Human Geography 34, no. 6 (December 2010): 799. This approach underpins the relation of the operational image to territory—­or territory already itself full of operational schemas of the sort that Elden lists. 6. Birgit Schneider, Klimabilder: Eine Genealogie globaler Bildpolitiken von Klima und Klimawandel (Berlin: Matthes & Seitz, 2018), 115. 7. Anders Engberg-­Pedersen also addresses “operational aesthetics,” and it is the title of one of the chapters in his forthcoming book, Martial Aesthetics: How War Became an Art Form (Stanford, Calif.: Stanford University Press, 2023). 8. See Cormac Deane, “The Control Room: A Media Archaeology,” Culture Machine 16, https://culturemachine.net/vol-16-drone-cultures/the-control -room/. 9. Pamela M. Lee, Think Tank Aesthetics: Midcentury Modernism, the Cold War, and the Neoliberal Present (Cambridge, Mass: MIT Press, 2020). 10. Elsaesser, “Simulation,” 219. On aesthetics of the interface as “regulation,” see Jan Distelmayer’s argument about the graphical user interface as a site of operations: “This tradition presents the aesthetics of regulation as an ‘order of selectivity’—­offering options and reassuring usability as a freedom of choice in the form of menus, buttons, lists, and the like. This ‘freedom as control’ is a question of strictly defined and prepared choices.” Distelmayer, “Carrying Computerization: Interfaces, Operations, Depresentations,” in Image—­Action—­Space: Situating the Screen in Visual Practice, ed. Luisa Feier­ singer, Kathrin Friedrich, and Moritz Queisner (Berlin: De Gruyter, 2018), 64. 11. Ilka Brasch, Film Serials and the American Cinema, 1910–­1940 (Amsterdam: Amsterdam University Press, 2018); Tom Gunning, “Crazy Machines in the Garden of Forking Paths: Mischief Gags and the Origins of American Film Comedy,” in Classical Hollywood Comedy, ed. Kristine Brunovska Karnick and Henry Jenkins (New York: Routledge:1995), 7–­105; Lisa Trahair, “The Narrative-­Machine: Buster Keaton’s Cinematic Comedy, Deleuze’s Recursion Function and the Operational Aesthetic,” Senses of Cinema, October 2004, https://www.sensesofcinema.com/2004/comedy-and-perception /keaton_deleuze.

254   Notes to Chapter 4

12. Harris, quoted in Brasch, 44. 13. Neil Harris, “The Operational Aesthetic,” in Popular Fiction: Technology, Ideology, Production, Reading, ed. Tony Bennett (London: Routledge, 1990), 402. 14. Wanda Strauven, “The Observer’s Dilemma: To Touch or Not to Touch,” in Media Archaeology: Approaches, Applications, and Implications, ed. Erkki Huhtamo and Jussi Parikka (Berkeley: University of California Press, 2011), 148–­63. See also Seth Giddings, Toy Theory (Cambridge, Mass: MIT Press, forthcoming). 15. Gunning, “Crazy Machines,” 100. See also Trahair, “Narrative-­Machine.” 16. Marianne van den Boomen, quoted in Distelmeyr, “Carrying Computerization,” 61. 17. On visibility and software, source code, and more, see also Wendy Hui Kyong Chun, Programmed Visions: Software and Memory (Cambridge, Mass: MIT Press, 2011). 18. See also Ryan Bishop on “autotechnological” education in Bishop, Comedy and Cultural Critique in American Film (Edinburgh: Edinburgh University Press, 2013). 19. Brasch, Film Serials, 58 20. Brasch, 59. 21. Brasch, 61. See also Paul DeMarinis, “Erased Dots and Rotten Dashes, or How to Wire Your Head for a Preservation,” in Media Archaeology: Approaches, Applications, and Implications, ed. Erkki Huhtamo and Jussi Parikka (Berkeley: University of California Press, 2011), 211–­38. 22. See Stephen Kern, The Culture of Time and Space, 1880–­1918 (Cambridge, Mass.: Harvard University Press, 2003). 23. Geocinema, The Making of Earths (2020), ZKM Critical Zones exhibition, https://geocinema-zkm.netlify.app/. 24. See for example Blair Thornton, “Robotics and Autonomy: Exploring Earth’s Inner Space,” lecture given at the 2019 Centre for Machine Intelligence Symposium, University of Southampton, October 2019, https://www .youtube.com/watch?v=MNjVh0VvMNk&feature=youtu.be. 25. Recursions is also a theoretical figure picked up by Shaoling Ma in her analysis of Chinese state sensing projects as they recursively operate on different scales from policy discourses to questions of surveillance of ethnic minorities. Ma, “Big Earths of China: Remotely Sensing Xinjiang along the Belt and Road,” Critical Inquiry 49, no. 1 (Autumn 2022): 77–­101. 26. See Ma, “Big Earths of China.” 27. Stefanie Hessler, Prospecting Ocean (Cambridge, Mass: MIT Press, 2019). 28. The Anthropocene Observatory was funded and produced by the Haus der Kulturen der Welt (HKW, Berlin) and executed by the Territorial Agency (John Palmesino and Ann-­Sofi Rönnskog), Armin Linke, and Anselm Franke. 29. “Across a number of specific international agencies and organisations, information about scientific research is acquired, registered, evaluated, pro-

Notes to Chapter 4   255 cessed, stored, archived, organised and re-­d istributed. These behind-­t he-­ scenes processes and practices, that lead to the equally complex decision making procedures, form new discourses and figures of shift. The Anthropocene Observatory documents these practices in a series of short films, interviews and documentary materials: aim of the project is to illustrate in detail the unfolding of the thesis of the Anthropocene in its many streams of influence.” The Anthropocene Observatory project, https://www.territorialagency.com /anthropocene. 30. For an extended discussion of critical zones, observatories, and environmental imaging and art practices, see Bruno Latour and Peter Weibel, eds., Critical Zones: The Science and Politics of Landing on Earth (Cambridge, Mass: MIT Press, 2020). 31. Digital Earth as a name is not related to Digital Earth as the term used in the DBAR context. 32. Geocinema Network home page, https://geocinema.network/. 33. Geocinema Network home page. 34. H. D. Guo, Z. Liu, and L. W. Zhu, “Digital Earth: Decadal Experiences and Some Thoughts,” International Journal of Digital Earth, 3, no. 1 (2010): 31–­46, https://doi.org/10.1080/17538941003622602. 35. “2009 Beijing Declaration on Digital Earth,” International Journal of Digi­ tal Earth, 2, no. 4 (2009): 397–­99, https://doi.org/10.1080/17538940903444380. 36. Z. Liu, T. Foresman, J. van Genderen, and L. Wang, “Understanding Digital Earth,” in Manual of Digital Earth, ed. H. Guo, M. F. Goodchild, and A. Annoni (Singapore: Springer, 2020), https://doi.org/10.1007/978-981-32 -9915-3_1. 37. Liu et al. 38. Ma, “Big Earths of China.” See also Gabriele de Seta, “Gateways, Sieves and Domes: On the Infrastructural Topology of the Chinese Stack,” International Journal of Communication 15 (2021):2669–­92. 39. Andrew Chatzky and James McBride, “China’s Massive Belt and Road Initiative,” Council on Foreign Relations, January 28, 2020, https://www.cfr .org/backgrounder/chinas-massive-belt-and-road-initiative. 40. Chatzky and McBride. 41. Elise Misao Hunchuck, Marco Ferrari, and Jingru (Cyan) Cheng, “Prologue to the Sky River,” Avery Review 53 (2021), http://averyreview.com /issues/53/prologue-to-the-sky-river. 42. Asia Bazdyrieva and Solveig Suess, “The Future Forecast,” e-flux, February 1, 2020, https://www.e-flux.com/architecture/new-silk-roads/313108 /the-future-forecast/. 43. Bazdyrieva and Suess. 44. Huadong Guo, “Big Earth Data: A New Frontier in Earth and Information Sciences,” Big Earth Data 1, nos. 1–­2 (2017): 9. 45. Ma, “Big Earths of China,” 95. 46. Liu et al., “Understanding Digital Earth.” 47. Liu et al.

256   Notes to Chapter 4

48. Liu et al. 49. Liu et al. 50. John Tresch, “Cosmologies Materialized: History of Science and History of Ideas,” in Rethinking Modern European Intellectual History, ed. Darrin M. McMahon and Samuel Moyn (New York: Oxford University Press, 2014), 155. On the multitude of globes, see also Cosgrove, Cartographic Genealogy. 51. On the powers of administration, see also John Guillory on the information genre. Guillory, “The Memo and Modernity,” Critical Inquiry 31, no. 1 (Autumn 2004): 108–­32. 52. See Hai Ren, “Infrastructure as a Planetary Sculpture: The Future of the Belt and Road Initiative in the Anthropocene,” Belt & Road in Global Perspective, March 18, 2021, https://munkschool.utoronto.ca/beltandroad /article/infrastructure-as-a-planetary-sculpture-the-future-of-the-belt-and -road-initiative-in-the-anthropocene. 53. Bruno Latour, “Visualisation and Cognition: Drawing Things Together,” Knowledge and Society: Studies in the Sociology of Culture Past and Present 6 (1986): 32. 54. Geocinema, “Footnotes to the Making of Earths,” n.d., https:// geocinema-zkm.netlify.app/. 55. Thomas Elsaesser, “Early Film History and Multi-­Media: An Archaeology of Possible Futures?” in New Media, Old Media, ed. Wendy Hui Kyong Chun and Thomas Keenan (New York: Routledge, 2006), 17, 13–­25. 56. Ute Holl, Cinema, Trance and Cybernetics, trans. Daniel Hendrickson (Amsterdam: Amsterdam University Press, 2017). 57. Geocinema, “Footnotes.” 58. Bernard Dionysius Geoghegan, “An Ecology of Operations: Vigilance, Radar, and the Birth of the Computer Screen,” Representations 147, no. 1 (2019): 59–­95. 59. As Cormac Deane argues in “The Control Room” article, at the back of the Control and Command doctrine that materializes in those architectural spaces, a row of imagined spaces of interfaces of control also emerge that feature a peculiar variety of operational images—­namely Fantasy User Interfaces. Such fictional entities feature mostly in science fiction cinema, from The Bourne Identity (2002) to Children of Men (2006) and many others. Deane argues that “the FUIs on display in these films enable a scientific, or scientific-­seeming, gaze that visualizes features of an environment that are otherwise invisible (infra-­red, nano-­scale, informational, unrecorded, behind corners, etc.). The FUI often renders bodies, objects, and buildings transparent, producing schematic images of them that can be rotated on several axes.” But observation is often less exciting than prescribed in fantasy images, and diagrammatic invisuality is harder to demonstrate in a pictorial fashion. 60. Lorraine Daston and Elizabeth Lunbeck, eds., Histories of Scientific Observation (Chicago: University of Chicago Press, 2011). 61. Alberto Toscano and Jeff Kinkle, Cartographies of the Absolute (Winchester: Zero Books 2015), 208, emphasis in the original. See also Thomas

Notes to Chapter 4   257 Stubblefield, “Database Labour: Supply Chains, Logistics, and Flow,” in Labour in a Single Shot: Critical Perspectives on Antje Ehmann and Harun Farocki’s Global Video Project, ed. by Roy Grundmann, Peter J. Schwartz, and Gregory H. Williams (Amsterdam: Amsterdam University Press, 2022), 329–­45. 62. Toscano and Kinkle, Cartographies of the Absolute, 213–­14. 63. Bazdyrieva and Suess, “Future Forecast.” In this passage they quote Huadong Guo et al., “Big Earth Data: A New Challenge and Opportunity for Digital Earth’s Development,” International Journal of Digital Earth 10 (November 2017): 1–­1 2. 64. On this sort of double geography and data, see Louise Amoore, Cloud Ethics: Algorithms and the Attributes of Ourselves and Others (Durham, N.C.: Duke University Press, 2020), 29–­55. 65. Sean Cubitt, Finite Media: Environmental Implications of Digital Technologies (Durham, N.C.: Duke University Press, 2017), 159. 66. See Geoff Cox and Jacob Lund, The Contemporary Condition: Introductory Thoughts on Contemporaneity & Contemporary Art (Berlin: Sternberg, 2016). 67. For a critical reading, see also chapter 2, “The Ship: Transatlantic,” in Christina Sharpe, In the Wake: On Blackness and Being (Durham, N.C.: Duke University Press, 2016). 68. Bazdyrieva and Suess, “Future Forecast.” 69. See Jussi Parikka, A Geology of Media (Minneapolis: University of Minnesota Press, 2015). 70. For a longer discussion of ground truth and “fake” geographies, see Abelardo Gil-­Fournier and Jussi Parikka, “Ground Truth to Fake Geographies: Machine Vision and Learning in Visual Practices,” AI & Society 36, 1253–­62 (2020), https://doi.org/10.1007/s00146-020-01062-3. 71. Asunder project website at https://asunder.earth/. 72. On workshops and artistic methods, see Daphne Dragona, “What Is Left to Subvert? Artistic Methodologies for a Post-­digital World,” in Across and Beyond: A Transmediale Reader on Post-­digital Practices, Concepts, and Institutions, ed. Ryan Bishop et al. (Berlin: Sternberg, 2016), 186–­98. 73. Part of the transmediale 2018 theme Face Value and in collaboration with other colleagues from the Winchester School of Art, including myself, Miha Brebenel, and Ryan Bishop. https://2018.transmediale.de/program/event /surface-value-landscape-prediction-an-amt-workshop-day-1. 74. On visual culture and the politics of Earth observation platforms such as Global Forest Watch, see Birgit Schneider and Lynda Walsh, “The Politics of Zoom: Problems with Downscaling Climate Visualizations.” Geo: Geography and Environment 6, no. 1 (2019), https://doi.org/10.1002/geo2.70. See also Lisa Parks, “Signals and Oil: Satellite Footprints and Post-­communist Territories in Central Asia,” European Journal of Cultural Studies 12, no. 2 (2009): 137–­56. 75. Landscape Prediction, Gil-­Fournier’s GitHub page: https://github.com /abe-/landscape-prediction.

258   Notes to Chapter 4

76. I am here referring to the complex set of epistemic, design-­led, speculative work at Strelka Institute’s Terraforming think tank / education program that ran from 2020 to 2022. See https://theterraforming.strelka.com/. 77. “This specifically modern form is what we call firmative speculation, a firming (from the late Latin firmã) or solidifying of the possibilities of the future. It is a speculative mode that seeks to pin down, delimit, constrain, and enclose—­to make things definitive, firm. The ur-­image of such agency is the firm, a type of business house (emerging in Germany in 1744) that capitalizes on market conditions, working toward an optimal level of production that will ensure maximum profit and minimum cost while always on the lookout for fresh opportunities for expansion—­aggressively pushing its products through advertising, shaping new needs, and consuming publics.” Uncertain Commons, Speculate This (Durham, N.C.: Duke University Press, 2013), https://speculatethis.pressbooks.com/chapter/chapter-1/. 78. Melinda Cooper, “Turbulent Worlds: Financial Markets and Environmental Crisis,” Theory, Culture & Society 27, nos. 2–­3 (2010): 170. 79. Cubitt, Finite Media, 160. 80. Christina Vagt, “Physics and Aesthetics: Simulation as Action at a Distance,” in Action at a Distance, by John Durham Peters, Florian Sprenger, and Christina Vagt (Minneapolis: University of Minnesota Press, 2020), 52. 81. Jimena Canales, “Operational Art,” in Visibility Machines: Harun Farocki and Trevor Paglen (Baltimore: University of Maryland Press, 2014), 37–­5 4. 82. Canales, 45–­46. 83. On the centrality of the plot and tricks in design, see Benedict Singleton’s work. For an overview, see Benedict Singleton and Robin Mackay, “Yarncast: Plots, Platforms and Sinister Designs,” Urbanomic, Feb 2, 2015, https://www.urbanomic.com/podcast/yarncast-benedict-singleton-plots -platforms-and-sinister-designs/. 84. Lee, Think Tank Aesthetics.

5. The Post-­lenticular City 1. Viola Zhou, “Drones Light Up Shanghai’s Sky with a QR Code (That You Can Scan)” Vice, April 20, 2021, https://www.vice.com/en/article/88n9vb /shanghai-drone-show-qr-code. 2. Lawrence Grossberg, “The In-­Difference of Television,” Screen 28, no. 2 (Spring 1987): 28–­46. 3. Grossberg. 4. Katherine N. Hayles, “Addressing Space: Augmented, Enfolded, and Exfolded Hybrid Models and Their Implications,” lecture given at the University of Virginia School of Architecture, June 11, 2020, https://www.youtube .com/watch?v=ZCWI4_HonjI. On Shanghai urbanism, see Anna Greenspan, Shanghai Future: Modernity Remade (Oxford: Oxford University Press, 2014). 5. Christian Ulrik Andersen and Søren Bro Pold, The Metainterface: The Art of Platforms, Cities, and Clouds (Cambridge, Mass.: MIT Press, 2018),

Notes to Chapter 5   259 81–­1 20. See also Rob Kitchin and Martin Dodge, Code/Space: Software and Everyday Life (Cambridge, Mass.: MIT Press, 2011). 6. See Jannice Käll, Posthuman Property and Law (London: Routledge, 2022). 7. See Daniel Rubinstein and Katrina Sluis, “A Life More Photographic: Mapping the Networked Image,” Photographies 1, no. 1 (2008): 9–­28, https:// doi.org/10.1080/17540760701785842. 8. See Anthony McCosker and Rowan Wilken, Automating Vision: The Social Impact of the New Camera Consciousness (New York: Routledge, 2020), 55–­7 1. 9. N. Katherine Hayles, Unthought: The Power of the Cognitive Nonconscious (Chicago: University of Chicago Press, 2017). 10. Mark Andrejevic, Automated Media (New York: Routledge, 2020), 94–­112. 11. Siegfried Kracauer, The Mass Ornament: Weimar Essays, trans. and ed. Thomas Y. Levin (Cambridge, Mass.: Harvard University Press, 1995), 62. 12. Lidar imaging is divided into two kinds: “Topographic lidar typically uses a near-­infrared laser to map the land, while bathymetric lidar uses water-­penetrating green light to also measure seafloor and riverbed elevations.” National Ocean Service, “What Is Lidar?” https://oceanservice.noaa .gov/facts/lidar.html. 13. Siegfried Zielinski, Deep Time of the Media: Toward an Archaeology of Hearing and Seeing by Technical Means, trans. Gloria Custance (Cambridge, Mass.: MIT Press, 2006). See, e.g., Horst Bredekamp, Vera Dünkel, and Birgit Schneider, eds., The Technical Image: A History of Styles in Scientific Imagery (Chicago: University of Chicago Press, 2015). 14. On computer graphics, see Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021). 15. Some, like John May, go as far as to claim that photography and imaging are two radically separate things: “Unlike photographs, in which scenic light is made visible during chemical exposure, all imaging today is a process of detecting energy emitted by an environment and chopping it into discrete, measurable electrical charges called signals, which are stored, calculated, managed, and manipulated through various statistical methods (Bayesian, Gaussian, Poissonian).” May, Signal, Image, Architecture (New York: Columbia Books on Architecture and the City, 2019), 45. 16. Jennifer Gabrys, Program Earth: Environmental Sensing Technology and the Making of a Computational Planet (Minneapolis: University of Minnesota Press, 2016), 58. 17. Peter Galison, “Images Scatter into Data, Data Gather into Images,” in Iconoclash: Beyond the Image Wars in Science, Religion and Art, ed. Bruno Latour and Peter Weibel (Cambridge, Mass.: MIT Press, 2002), 319. 18. On modeling as following, see Matthew C. Hunter, “Modeling: A Secret History of Following,” in Design Technics: Archaeologies of Architectural

260   Notes to Chapter 5

Practice, ed. Zeynep Çelik Alexander and John May (Minneapolis: University of Minnesota Press, 2020), 45–­70. 19. McCosker and Wilken, Automating Vision, 65–­67. 20. See ARCore overview at https://developers.google.com/ar/discover. 21. Sean Cubitt, “Mass Image, Anthropocene Image, Image Commons,” in Photography Off the Scale, ed. Tomáš Dvořák and Jussi Parikka (Edinburgh: Edinburgh University Press, 2021), 25–­40. 22. Erika Balsom and Harun Farocki, “The New Constructivism: Erika Balsom and Harun Farocki Discuss Parallel I–­IV,” in Harun Farocki: Programando o Visível, ed. Jane De Almeida (São Paolo: CinUSP, 2017), 130. 23. Paul Virilio, The Vision Machine, trans. Julie Rose (Bloomington: Indiana University Press, 1994). 24. Some of these questions—­and some of the answers—­pair up with what Matthew Fuller and Eyal Weizman define as the “hyperaesthetic”: capacities of sensing beyond the photographic image that acknowledge that all matter can be read in relation to questions of aesthesis and thus become an element in this expanded notion of imaging. See Matthew Fuller and Eyal Weizman, Investigative Aesthetics: Conflicts and Commons in the Politics of Truth (London: Verso, 2022), 71–­82. Susan Schuppli’s work also speaks directly to this expanded notion of sense and imaging. For more, see Schuppli, Material Witness: Media, Forensics, Evidence (Cambridge, Mass.: MIT Press, 2020). 25. Patrick Maynard, The Engine of Visualization: Thinking through Photography (Ithaca: Cornell University Press, 1997). 26. “Thus, supposing the composition of the atmosphere to be unchanged at the heights investigated, its density at these heights will be known, after making various corrections, if we can project a sufficiently strong beam of light to allow for the light scattered in a given region being detected and measured by photo-­electric apparatus after collection by a large reflector.” Edward Hutchinson Synge, “A Method of Investigating the Higher Atmosphere,” London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 9, no. 60 (1930): 1015. 27. Friedrich Kittler, “A Short History of the Searchlight,” in Operation Valhalla: Writings on War, Weapons, and Media, ed. and trans. Ilinca Iurascu, Geoffrey Winthrop-­Young, and Michael Wutz (Durham, N.C.: Duke University Press, 2021), 62–­68. 28. Edward Hutchinson Synge, “A Suggested Method for Extending Microscopic Resolution into the Ultra-­microscopic Region,” London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 6, no. 35 (1928): 356–­62. 29. Jimena Canales, “Einstein’s Discourse Networks,” Zeitschrift für Medien-­und Kulturforschung 5, no. 1 (2014): 23. 30. Wolfgang Ernst, Chronopoetics: The Temporal Being and Operativity of Technological Media, trans. Anthony Enns (London: Rowman & Littlefield, 2016); Joanna Zylinska, Posthuman Photography (Cambridge, Mass.: MIT Press, 2017). 31. American Geosciences Institute, “What Is Lidar and What Is It Used

Notes to Chapter 5   261 For?” https://www.americangeosciences.org/critical-issues/faq/what-lidar -and-what-it-used. 32. William Rankin, After the Map: Cartography, Navigation, and the Transformation of Territory in the Twentieth Century (Chicago: University of Chicago Press, 2016), 15. 33. Rankin, 15. 34. See also Zeynep Çelik Alexander, “Scanning: A Technical History of Form,” in Design Technics: Archaeologies of Architectural Practices, ed. Zeynep Çelik Alexander and John May (Minneapolis: University of Minnesota Press, 2020), 71–­102. 35. Mario Carpo, The Second Digital Turn: Design beyond Intelligence (Cambridge, Mass.: MIT Press, 2017), 122. 36. Carpo, 122. As Sean Cubitt points out, the charge-­coupled device or the CCD chip is also an interesting technical object of what I call “invisuality” (after Mackenzie and Munster) in this book: it regulates how data is shifted “from spatial to temporal and back to spatial ordering” as it organizes the incoming (photon) data via a clocking mechanism. “The CCD imposes a very specific order on the light it gathers, governed by mask, column, and clock and characterized by whole-­number steps of equal duration and area. The result is an array of discrete, ordered units.” Cubitt, The Practice of Light: A Genealogy of Visual Technologies from Prints to Pixels (Cambridge, Mass.: MIT Press, 2014), 101. 37. The interconnected ecology of sensing is central to what Gabrys writes about planetary-­sensing networks too. These are less about navigation in the concrete sense of the term but part of a similar operational epistemology of sensing that also includes prediction/forecasting at different scales: “Sensors working together within a network establish a computational pattern of correspondences, where the physical sighting, sensor type, coding, and correlating of data coalesce into an environment of sensor data that in­form observations. When the ‘system is the sensor’ and the network operates as a sort of distributed instrument, it might be possible to create models and forecasts of ecological processes and, through these sensor systems, act upon environments.” Gabrys, Program Earth, 43. 38. Shane Denson, Discorrelated Images (Durham, N.C.: Duke University Press, 2020). 39. Kaja Silverman, “What Is a Camera? or: History in the Field of Vision,” Discourse 15, no. 3 (Spring 1993): 3–­56. 40. This operational aesthetic has some of its roots in the military history of the twentieth-­century in the form of operational research, but the expansion to new institutional situations of management has spread across a much wider societal field since the Cold War. 41. Moholy-­Nagy, quoted in Cubitt, Practice of Light, 95. 42. David Mattison, “Mountain Photography,” in Encyclopedia of NineteenthCentury Photography, vol. 1, ed. John Hannavy (London: Routledge, 2008), 949.

262   Notes to Chapter 5

43. Peter Ainsworth, “Between Real and Virtual, Map and Terrain: ScanLab Projects, Post-lenticular Landscapes,” Philosophy of Photography 10, no. 2 (2019): 269–­81. 44. See, for example, Mark David Spence, Dispossessing the Wilderness: Indian Removal and the Making of the National Parks (Oxford: Oxford University Press, 2000). 45. May, Signal, Image, Architecture, 45. 46. Ainsworth, “Between Real and Virtual,” 273. 47. ScanLAB, “Post-­lenticular Landscapes,” n.d., https://scanlabprojects.co .uk/work/post-lenticular-landscapes/. 48. ScanLAB. 49. Allan Sekula, “The Traffic in Photographs,” Art Journal 41, no. 1 (1981): 15–­25. 50. Geoff Manaugh, “The Dream Life of Driverless Cars,” New York Times, Nov 11, 2015, https://www.nytimes.com/2015/11/15/magazine/the-dream-life -of-driverless-cars.html. 51. McCosker and Wilken, Automating Vision, 65. 52. Ranjodh Singh Dhaliwal, “Rendering the Computer: A Political Diagrammatology of Technology” (PhD diss., University of California, Davis, 2021), ProQuest #28540442. 53. NVIDIA, “Solutions for Self-­Driving Cars & Autonomous Vehicles,” n.d., https://www.nvidia.com/en-gb/self-driving-cars/drive-platform. 54. NVIDIA, “NVIDIA Drive videos,” n.d. https://www.nvidia.com/en-gb /self-driving-cars/drive-labs/. 55. See also Benjamin Bratton, “The City Wears Us: Notes on the Scope of Distributed Sensing and Sensation,” Glass Bead, 2017, http://www.glass -bead.org/article/city-wears-us-notes-scope-distributed-sensing-sensation /?lang=enview. Bratton writes: “For example, driverless cars are emblematic of big heavy machines sensing/learning in the streets. Their proprioceptive sensors include wheel speed sensors, altimeters, gyroscopes, tachymeters, touch sensors, while their exteroceptive sensors include multiple visual light cameras, LiDAR range finding, short-­and long-­range RADAR, ultrasonic sensors on the wheels, global positioning satellite systems/geolocation aerials, etc. Several systems overlap between sensing and interpretation, such as road sign and feature detection and interpretation algorithms, model maps of upcoming roads, and inter-­car interaction behavior algorithms. Along the gradient from fully to partially autonomous, the humans inside provide another intelligent component that may be variously copilot or cargo, and together they form a composite User ambling through the City layer of The Stack.” 56. Directly quoted from the ScanLAB project page at https:// ScanLABprojects.co.uk/work/dreamlife-of-driverless-cars/. 57. This formulation is partly an implicit reference to and riffing with Brian Massumi’s ways of talking about movement. 58. Manaugh, “Dream Life of Driverless Cars.”

Notes to Chapter 5   263 59. Christian Doppler, “Über das farbige Licht der Doppelsterne und einiger anderer Gestirne des Himmels,” facsimile reproduced with a translation in Alec Eden, The Search for Christian Doppler (Vienna: Springer Verlag, 1992), orig. pub. 1842. 60. Ryan Bishop and John Phillips, Modernist Avant-­Garde Aesthetics and Contemporary Military Technology: Technicities of Perception (Edinburgh: Edinburgh University Press, 2010), 101. 61. Bernard Dionysius Geoghegan, “An Ecology of Operations: Vigilance, Radar, and the Birth of the Computer Screen,” Representations 147 (Summer 2019): 69. 62. Geoghegan, 69. 63. Photography thus also relates to the history of spatial computing and sensors: “We can divide sensors into two large groups, passive and active. Passive sensors detect and observe electromagnetic radiation that is absorbed, reflected, or scattered from objects, and they use an external energy source from the physical environment, such as the sun. An active sensor provides its own energy source, which is directed at the object. The sensor then measures the response to the object. When we use a camera without a flash, the camera is a passive sensor, observing and recording light from the sun reflected from the objects in the camera’s lens. A camera that uses a flash provides its own light for the object and then measures and records the light that is reflected off of the object.” Sashi Shekhar and Pamela Vold, Spatial Computing (Cambridge, Mass.: MIT Press, 2019), 69. 64. Jimena Canales, “Flash Force: A Visual History of Might, Right, and Light,” in Seeing with Eyes Closed, ed. Elena Agudio and Ivana Franke (Berlin: Association of Neuroesthetics, 2011), 34–­41. 65. As Geoff Manaugh writes: “Humans are not the only things now sensing and experiencing the modern landscape—­that something else is here, with an altogether different, and fundamentally inhuman, perspective on the built environment.” Manaugh, “Dream Life of Driverless Cars.” Humans of course never were the only ones, considering the various scales of sensing and experiencing from animals to bacteria, from organic surfaces to technical media even prior to recent computational infrastructures. See Shannon Mattern, A City Is Not a Computer: Other Urban Intelligences (New York: Princeton University Press, 2021). See also Jussi Parikka, Insect Media: An Archaeology of Animals and Technology (Minneapolis: University of Minnesota Press, 2010). 66. Manaugh, “Dream Life of Driverless Cars.” 67. See also Wendy Hui Kyong Chun, “Crisis, Crisis, Crisis, or Sovereignty and Networks,” Theory, Culture & Society 28, no. 6 (2011): 91–­112. 68. Chun. 69. Talbot, quoted in Maynard, Engine of Visualization, 152. Cf. Fuller and Weizman, Investigative Aesthetics, 79, on “the image-­being of all material surfaces” in this light too. 70. See Abelardo Gil-­Fournier and Jussi Parikka, “‘Visual Hallucination of Probable Events’: On Environments of Images, Data, and Machine Learning.”

264   Notes to Chapter 5

In Big Data: A New Medium? ed. Natasha Lushetisch (New York: Routledge, 2020). 71. Manaugh, “Dream Life of Driverless Cars.” 72. See Mattern, City Is Not a Computer. 73. Bratton, “City Wears Us.” 74. Steve F. Anderson, Technologies of Vision: The War between Data and Images (Cambridge, Mass.: MIT Press, 2017). 75. See, e.g., McCosker and Wilken, Automating Vision. 76. Adrian Mackenzie and Anna Munster, “Platform Seeing: Image Ensembles and Their Invisualities,” Theory, Culture & Society 36, no. 5 (2019): 7. 77. Mackenzie and Munster, 18. 78. Fei Wang, Sanping Zhou, Stanislav Panev, Jinsong Han, and Dong Huang, “Person-­in-­WiFi: Fine-­Grained Person Perception Using WiFi,” arXiv, March 30, 2019, 2, https://arxiv.org/abs/1904.00276. 79. Wang et al. 80. Adrian Mackenzie, Wirelessness: Radical Empiricism in Network Cultures (Cambridge: MIT Press, 2010). 81. Wang et al., “Person-­in-­WiFi,” 2. 82. See Andersen and Pold, Metainterface. 83. Quoted from the Abandon Normal Devices festival website (2016), https://www.andfestival.org.uk/city-cant-see/. Where the City Can’t See was commissioned by Abandon Normal Devices, St Helens Heart of Glass and University of Salford Art Collection. 84. A trailer of the film is available at https://vimeo.com/188626212. 85. Cf. Andrejevic, Automated Media, 94–­112. 86. David Meyer, “How Cloud Computing Contributes to Autonomous Driving—­A Thought Experiment,” MS&E 239 (blog), July 11, 2018, https:// mse238blog.stanford.edu/2018/07/davidmyr/ how-cloud-computing -contributes-to-autonomous-driving-a-thought-experiment/. 87. Mark Hansen, Feed Forward: On the Future of Twenty-­First Century Media (Chicago: University of Chicago Press, 2015); Erich Hörl, “The Environmentalitarian Situation: Reflections on the Becoming-­Environmental of Thinking, Power, and Capital,” Cultural Politics 14, no. 2(2018): 153–­73; Florian Sprenger, Epistemologien des Umgebens: Zur Geschichte, Ökologie und Biopolitik künstlicher environments (Bielefeld: Transcript, 2019); Gabrys, Program Earth. 88. Zylinska, Posthuman Photography. 89. Volker Pantenburg, “Manual: Harun Farocki’s Instructional Work,” in Against What? Against Whom? ed. Anthe Ehmann and Kodwo Eshun (Cologne: Walther König, 2009), 98, 93–­100. 90. Bratton, “City Wears Us.” 91. See Gabrys, Program Earth, 43. 92. As Otto von Gruber put it in the 1930s, “The problem of automatic plotting instruments concerns the representation on a map of an object shown on photograms, without the need for carrying out computations

Notes to Conclusion   265 point by point or for graphical constructions.” Von Gruber, Photogrammetry: Collected Lectures and Essays (London: Chapman & Hall, 1932), 276. 93. Sprenger, Epistemologien des Umgebens. 94. Parikka, Insect Media. 95. On navigation, see Doreen Mende, “The Navigation Principle: Slow Image,” e-flux lectures, November 29, 2017, https://www.e-flux.com/live /165096/e-f lux-lectures-doreen-mende-the-navigation-principle-slow -image/. 96. Following this navigational focus, it’s clear how the operational image relates to the instrumental image as in Allan Sekula’s use of the term: these sorts of images operate in automated systems and as part of large-­scale infrastructural technologies, where such images include those used in aerial imaging, military systems, automated navigation, or robotic object recognition guided movement (an example being the factory assembly reliance on automatism and action). 97. Andersen and Pold, Metainterface; Benjamin H. Bratton, The Stack: On Software and Sovereignty (Cambridge, Mass.: MIT Press, 2015): 220–­26. 98. Mattern, A City Is Not a Computer. 99. See Käll, Posthuman Property and Law; Benjamin Bratton, The Revenge of the Real: Politics for a Post-­pandemic World (London: Verso, 2021), 51–­59.

Conclusion 1. Thank you to Robert Pietrusko for clarification and discussions about the process of construction of the black hole image. Pietrusko was part of the team that participated in the film Black Holes: The Edge of All We Know (2020), directed by Peter Galison, that features the work of the Event Horizon Telescope team project. Any possible misunderstandings in the following elaboration are to be credited to me. 2. Peter Galison, “How Do You Photograph a Black Hole?” MoMA, May 17, 2021, moma.org/magazine/articles/563. 3. Lorraine Daston and Peter Galison, Objectivity (New York: Zone Books, 2010). 4. ESO, “First Image of a Black Hole,” April 10, 2019, https://www.eso.org /public/images/eso1907a/. 5. Farocki had referred to the phantom image and the phantom shot as a persistent feature of cinema and animation: a perspective impossible for a human to occupy, like in the early cinema of the 1920s of a “camera that had been hung under a train,” or then in the age of intelligent weaponry, “the film that takes up the perspective of the bomb,” or the impossible physical worlds conjured in animation, including computer images. Farocki, “Phantom Images,” Public 29 (2004): 13. 6. MIT Computer Science and Artificial Intelligence Lab, “A Method to Image Black Holes,” June 6, 2016, https://www.csail.mit.edu/news/method -image-black-holes; Katherine L. Bouman, Michael D. Johnson, Daniel Zoran,

266   Notes to Conclusion

Vincent L. Fish, Sheperd S. Doeleman, and William T. Freeman, “Computational Imaging for VLBI Image Reconstruction,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2016, 913–­22, https:// www.cv-foundation.org/openaccess/content_cvpr_2016/html/Bouman _Computational_Imaging_for_CVPR_2016_paper.html. 7. Galison, “Photograph a Black Hole?” 8. Benjamin Bratton, The Terraforming (Moscow: Strelka, 2019); Orit Halpern, “Planetary Intelligence,” in The Cultural Life of Machine Learning: An Incursion into Critical AI Studies, ed. Jonathan Roberge and Michael Castelle (Cham, Switzerland: Palgrave Macmillan, 2021), 227–­56. 9. Halpern, “Planetary Intelligence,” 232. 10. See Janet Vertesi, Seeing Like a Rover: How Robots, Teams, and Images Craft Knowledge of Mars (Chicago: University of Chicago Press, 2015), 215–­37. 11. On calibration, cleaning, and disciplining of images, see Vertesi, 53–­103. 12. On the nonexistence of the digital image, see, for example, Wolfgang Hagen, “Es gibt kein ‘digitales Bild’: Eine medienepistemologische Anmerkung,” Archiv für Mediengeschichte 2 (2002): 103–­1 2; Claus Pias, “Das digitale Bild gibt es nicht: Über das (nicht-­)Wissen der Bilder und die informatische Illusion,” Zeitenblicke 2, no. 1 (2003), https://doi.org/10.25969/mediarep/4082. See also Jacob Gaboury, Image Objects: An Archaeology of Computer Graphics (Cambridge, Mass.: MIT Press, 2021) as well as Ranjodh Singh Dhaliwal’s ongoing work on rendering, which already moves the question from primacy of ontology to operations. On questions of digital images, I thus also want to point to a quote by Sean Cubitt that resonates with the centrality of “operation” in the media theory of Wolfgang Ernst: “It is not that digital images ‘don’t exist’ when they are not being run by software: films and TV shows in the can could be said not to exist by the same token, as could a photograph kept in a drawer or a painting secreted away in a private cabinet. Rather, each rendering of an image to a display or printer requires a unique run of the code. Similar to running an otherwise inert filmstrip through a projector, running code is an action encouraging a potential medium to become itself. In cinema the sense of event, and in domestic television viewing the familiarity and comfort of real-­time flow, give this becoming a role in consumption. The on-­demand, packet-­switched Internet refers us to run time, the time during which software executes the instructions in the code. Even more than live broadcasts, this brings us to real time, even though, in every instance, code and software have to be written in advance of their run.” Sean Cubitt, The Practice of Light: A Genealogy of Visual Technologies from Print to Pixel (Cambridge, Mass.: MIT Press, 2014), 243. 13. In addition to existing scholarship on scientific photography, for recent work on the history and philosophy of photography articulated between light, materiality, and measurement, see the special issue Light Sensitive Material starting from the introduction. Michelle Henning and Junko Theresa Mikuriya, “Light Sensitive Material: An Introduction,” Photographies, 14,

Notes to Conclusion   267 no. 3 (2021): 381–­94. See also Tomáš Dvořák & Jussi Parikka, “Measuring Photographs,” Photographies 14, no. 3 (2021): 443–­57. 14. Quoted in Patrick Maynard, The Engine of Visualization: Thinking through Photography (Ithaca: Cornell University Press, 1997), 259. 15. Thomas Elsaesser, “Making the World Superfluous: An Interview with Harun Farocki,” in Harun Farocki: Working on the Sight-­Lines, ed. Thomas Elsaesser (Amsterdam: Amsterdam University Press, 2004), 184. 16. Thomas Elsaesser, “Early Film History and Multi-­media: An Archaeology of Possible Futures?” in New Media, Old Media: A History and Theory Reader, ed. Wendy Hui Kong Chun and Thomas Keenan (New York: Routledge, 2005), 13–­25. 17. Harun Farocki, “Controlling Observation,” trans. Laurent Faasch-­ Ibrahim, in Elsaesser, Harun Farocki, 294. On recent work on the invisual operations of carceral capitalism, see Jackie Wang, Carceral Capitalism (Los Angeles: Semiotext(e), 2018). 18. Thomas Elsaesser, “Harun Farocki: Filmmaker, Artist, Media Theorist,” in Elsaesser, Harun Farocki, 24. 19. Christa Blümlinger, “Slowly Forming a Thought While Working on Images” in Elsaesser, Harun Farocki, 168. 20. Elsaesser, “Harun Farocki,” 19. See also Nora Alter, “Two or Three Things I Know About Harun Farocki,” October 151 (Winter 2015): 151–­58.

This page intentionally left blank

INDEX

Page numbers in italic refer to illustrations. Abney, William de Wivelselie, 6 Adorno, Theodor, 17 agencement, 39 Ainsworth, Peter, 190–91 Alberti, Leon Battista, 43, 45, 108, 114 Albertz, Joerg, 106 Alembert, Jean Le Rond d’, 8 Amazon, 19 Amoore, Louise, 60, 247n69 Andersen, Christian Ulrik, 178 Andrejevic, Mark, 180 Anthropocene Observatory, 152 apparatuses, 40 Application Program Interfaces (APIs), 170 Arago, François, 5, 102, 115–16 architectural modeling, viii, 45–46, 185 artificial intelligence (AI; machine learning), 15, 18, 19, 79–81, 87, 90, 150, 167, 198, 213–17; activations and operations linked to, 21; fake geographies and, 169; for games, 60, 68; at institutional level, 13; invisible technical images in, 29, 91; modeling functions in, 77, 80, 86; regularities extracted in, 78; social impact of, 82; statistical models underlying, 76; video linked to, 171 astronomic images, 1, 62, 84–85 Asunder (Brain, Oliver, and Sjölén), 169, 171, 172 Atlas of Emotion (Bruno), 28

autonomous vehicles, 30, 49, 73, 191, 195 AVA (atomic visual actions) dataset, 74 Azoulay, Ariella Aïsha, 115, 128, 135–36 Bailey, Solon Irving, 1–2, 4, 7 Bal, Mieke, 28 Balsom, Erika, 17, 111 Barad, Karen, 33, 37, 38, 39–40, 55, 87 Barnard, E. E., 6 Barnum, P. T., 145, 173 Barthes, Roland, 58, 79, 81 Baudrillard, Jean, 12, 219, 239n37 Bauordnungslehre (Neufert), 45 Bayesian heuristics, 216 Bazdyrieva, Asia, 150, 156, 157, 166, 169 Beautemps-­Beaupré, Charles-­ François, 117–18, 119 Bellour, Raymond, 17 Belt and Road Initiative, 155–56 Bender, John, 7 Benjamin, Ruha, 50, 82 Benjamin, Walter, 148 Bergson, Henri, 245n2 Bertillon, Alphonse, 48, 109 bias, 82–83 Bilibili (Chinese company), 177 Bishop, Ryan, 196 Blackett, Patrick, 13 black holes, 30, 211–14, 218 Blanke, Tobias, 79–80 Blas, Zach, 61 269

270  Index

Blaschke, Estelle, 86 Bonnen, Marianne van den, 147 Bradotti, Rosi, 16 Brain, Tega, 169 Brasch, Ilka, 145–46, 148–49, 153 Bratton, Benjamin, 69, 199, 215, 239n40, 262n55 Braun, Marta, 110 Bredekamp, Horst, 34 Bridgman, Percy Williams, 31 Broca’s Brain (Sagan), ix Brookes (slave ship), 49–50 Browne, Simone, 33, 49–51 Brunelleschi, Battista, 108 Bruno, Giuliana, 28 Bouguer, Pierre, 133, 134, 135, 136 Bouman, Kate, 214 building information modeling (BIM), 185 Buolamwini, Joy, 82 Burch, Noël, 168 Busch, Emil, 106 camera obscura, 89, 116, 191 Canales, Jimena, 32, 173–74, 191 Cannon, Annie Jump, 3, 6 capitalism, 15, 29, 32–33, 44, 50–56, 83, 168; platforms linked to, 71, 72, 94, 209 Carpo, Mario, 108, 114, 186 Carte de Ciel, 3 Cassini, Jean-­Dominique, 129 Chaplin, Charlie, 147 Chávez Heras, Daniel, 79–80 China-­Pakistan Economic Corridor, 155 CHIRP (Continuous High-­resolution Image Reconstruction using Patch priors), 214 chronophotography, 84, 89, 109–10, 123 Chun, Wendy Hui Kyong, 60, 61, 71, 93 climate data, 48, 91, 155, 156, 162, 164, 172

Clinton, Bill, 154 Colson, René, 63 Commercial and Political Atlas and Statistical Breviary (Playfair), 67 Common Objects in Context (COCO), 76 Community Earth System Model (CESM), 169 computer graphics, 42, 94, 137, 234n53 computer vision, 58, 75–77 conceptual persuasion, 24, 25, 217 Cooper, Melinda, 172 Counter Music (Farocki), 180, 206, 207 Count of Monte Cristo, The (Dumas), 173–74 Cowen, Deborah, 54 Crawford, Kate, 79 Creators of Shopping Worlds (Farocki), 142, 180 Critical Zones (exhibition), 152 Crookes, William, 6 Cubitt, Sean, 48, 66, 124, 167, 172, 240n50, 261n36, 266n12 cybernetics, 32 Daguerre, L. J. M., 6 Dark Matters (Browne), 49 Darwin, Charles, 135 dashboards, 21 Daston, Lorraine, 89, 93 data visualization, 42, 47, 66, 89, 110, 111, 112, 113, 125 Deane, Cormac, 256n59 DeepMask, 21 de la Rue, Warren, 6 Deleuze, Gilles, 16–17, 27, 39, 43, 149, 218, 238n22, 241n64 Denison, Shane, 187 Derrida, Jacques, 89 Desargues, Girard, 249n30 Deville, Edouard, 189 DexNet (robotic tool), 69 Dhaliwal, Ranjodh Singh, 94, 193

Index  271 Diderot, Denis, 7–8 Didi-­Huberman, Georges, 57, 75 Digital Belt and Road Initiative (DBAR), 92, 143, 150–59, 161–63, 166, 171, 206 Digital Earth program, 153–55, 157, 166, 167 Disease Surveillance and Risk Mapping, 170 Distelmayer, Jan, 253n10 Doane, Mary Ann, 109 Dodge and Burn (Syjuco), 61 Doppler, Christian, 185, 195, 199 Doppler effect, 195–96 Draper, John, 6 Dream Life of Driverless Cars (ScanLAB project), 192, 194, 195, 199 Dumas, Alexandre, père, 173–74 Dürer, Albrecht, 124, 139, 140 East India Company, 117 Eastlake, Elizabeth, 216 Edgerton, Samuel, 108 Ehmann, Antje, 18, 85, 164 Elden, Stuart, 253n5 Elsaesser, Thomas, xi, 17–18, 141, 145, 149, 162, 217 embedded vision, 191–92 Engell, Lorenz, 34 Enlightenment, 103 Ernst, Wolfgang, 32, 244n97 Eshun, Kodwo, 18 Euclid, 114 Eugeni, Ruggero, 71–72 Euler, Leonhard, 114, 124 Event Horizon and Black Hole Imaging project, 30, 212–14, 215 extractivism, 44, 52, 54–56, 189 Eye/Machine I–III (Farocki), 11 Facebook, 21, 66, 145 facial recognition, 48, 49, 70, 72 Farocki, Harun, vii, viii, 21, 23–25, 41, 48, 73, 85, 96, 107, 118, 146, 164, 173, 206, 207, 217–19; city of

transport viewed by, 180; film installations by, 11; at IKKM, 34; on industrialization of thought, 14–15; invisible control systems and, 149; as media archaeologist, 17; measurement images in works of, 30, 98–99; military themes of, 12–13, 15, 33, 52, 103; montages of, 218; navigational themes of, 42; operational aesthetics and, 144; operational images defined by, 19; operative invisibilities and, 141–42; simulations viewed by, xi Fashion-­MNIST (dataset), 76 FieldSAFE (dataset), 76 figure de la terre, La (Maupertuis), 129–30, 132 First Gulf War (1990), vii Fitzpatrick Skin Type system, 82 flattening, 103, 106, 125, 170, 171 Fleming, Williamina, 3 Flickr, 77, 81 Flusser, Vilém, 96, 219 following, 42 forensics, 109 Forgotten Space, The (Sekula and Burch), 168 formatting, 140–41 Foucault, Michel, 16, 43, 60, 92, 142 Framing Territories (Geocinema), 26, 41, 150, 162, 166 Fraunhofer, Joseph von, ix–x, 62, 195 Friedrich, Kathrin, 32–33 Fuller, Matthew, 16, 260n24 Gaboury, Jacob, 94 Gabrys, Jennifer, 182, 232n32, 261n37 Galileo Galilei, 87–88, 108 Galison, Peter, 89, 93, 213, 215 Galloway, Alex, 71 Geimer, Peter, 63, 64, 88, 90 Gender Shades project, 82 Geocinema, 26, 30, 41, 92, 143, 149–62, 206

272  Index

geodesic measurement, 4, 20, 30, 98, 126–28, 131, 132, 135 Geoghegan, Bernard, 163, 196 Geographical Information System (GIS), 183 Géométrie descriptive (Monge), 114–15, 116 gig economy, 83 Gil-­Fournier, Abelardo, 26, 96, 143, 169, 170–71, 172 Global Forest Cover Change, 170 Go (board game), 68–69 Google, 183 Gore, Al, 154, 155 GPS (global positioning system), 46, 48, 128–29, 130–31 GPUs (graphics processing units), 94, 193 Gravity’s Rainbow (Pynchon), 13 grids, 43–56, 111, 196; in game of Go, 68–69; in mapping, 130 Grossberg, Lawrence, 177–78 Gruber, Otto von, 264n92 Guattari, Félix, 27, 39, 238n22, 241n64 Gunning, Tom, 145, 147 Gurevitch, Leon, 139 Halley, Edmund, 117 Halpern, Orit, 215 Haraway, Donna, 50, 95–96, 98 Harney, Stefano, 50 Harris, Neil, 146 Harvard Observatory, 1–2, 62, 85, 195 Hayles, N. Katherine, 178, 179, 239n34 Heidegger, Martin, 34–35, 232n20 Henning, Michelle, ix, 2, 8 Henry, Paul, 6 Henry, Prosper, 6 Herschel, Frederick William, 181 Hitch, Charlie, 14 Hoel, Aud Sissel, 20

Hoel, Aurora, 32–33 Hoelzl, Ingrid, 67, 242n66 Holl, Ute, 162 Horkheimer, Max, 17 How Not to Be Seen (Steyerl), 61, 142 Humboldt, Alexander von, 135 Hunter, Matthew C., 233n35 Ice Cores (Schuppli), 152 Imagenet (dataset), 76–77, 81, 86 Image of the City, The (Kinkle), 165 Images of the World and Inscription of War (Farocki), 11, 99, 100, 101, 102, 104, 105, 127 Impett, Leonardo, 79 “Impossible Images” (Menkman), 26 Inextinguishable Fire (Farocki), 15 Ingold, Tim, 125–26 intellectual property, 72 Intergovernmental Panel on Climate Change (IPCC), 48 Internationale Kolleg für Kultur­ technikforschung und Medienphilosophie (IKKM), 33–34 Introduction to the Practice of Nautical Surveying, An (Beautemps-­ Beaupré), 118, 119 iPad, 185 Jameson, Fredric, 164, 165 Janssen, Jules, 6 Joler, Vladan, 68 Joselit, David, 44–45 Kahn, Herman, 13 Käll, Jannice, 73 Keaton, Buster, 147 Keenan, Sarah, 73 Kinect (gaming device), 186 Kinetics (dataset), 74 Kinkle, Jeff, 164, 165 Kittler, Friedrich, 114, 219 Klee, Paul, 169 Kracauer, Siegfried, 180

Index  273 Krajewski, Markus, 34 Krämer, Sybille, 34, 35, 103, 125 Krone, Hermann, 6 labor, 82–86 Labour in a Single Shot (Farocki and Ehmann), 85, 86, 164 Lacan, Jacques, 107 La Condamine, Charles-­Marie de, 133, 134, 135, 136 Lambert, Johann Heinrich, 114, 127 Landsat Earth Observation (1972), 155 laser scanning, 190–91, 202 Latour, Bruno, 8–9, 37, 131, 152, 160 Laussedat, Aimé, 118–22, 249n32 Leavitt, Henrietta Swan, 3–4, 5, 6–7, 10 Lee, Pamela, 144 Lee, Rosemary, 238n28, 242n89 Lek, Lawrence, 17 Leroi-­Gourhan, André, 35 Li, Fei-­Fei, 77 LIDAR (light detection and ranging), 22, 67, 174, 181, 184–93, 197–208, 219 Li Guoqing, 160, 161 LILA BC (dataset), 76 Lindseth, Frank, 20 Linke, Armin, 143, 152 Linnaeus, Carl, 129 Litvintseva, Sasha, 26 logistics: of DBAR, 161, 163; extraction and finance linked to, 52, 54–55, 83; grids and diagrams in, 49, 50, 51; of image compilation and naming, 77; of image vs. space, 22, 48, 51; infrastructure and, 167, 168, 178, 191; invisual operations as, 165–66; military, 127, 163; slave trade linked to, 49, 50 Lumière, Louis, 85 Lury, Celia, 229n82 Lynch, Kevin, 165

Ma, Shaoling, 157 Mach, Ernst, 62 machine learning. See artificial intelligence machine vision, vii, 15, 22–24, 61, 69–70, 91, 93, 181, 183, 186, 193, 203, 205, 208; expanding definition of, 200; genealogy of, 182, 187; military applications of, 11, 12 Macho, Thomas, 34 Mackenzie, Adrian, 22–23, 47, 65–69, 73, 200 Magellanic clouds, 3–4, 5, 7, 10 Making of Earths, The (Geocinema), 149–53, 154, 158, 160–62, 163 Malevé, Nicolas, 77–78, 79, 81, 83 Manaugh, Geoff, 192, 193–94, 195, 198, 263n65 Marey, Etienne Jules, 84, 89, 109–10, 111, 112, 113, 123, 126 Marie, René, 67, 242n66 Marrinan, Michael, 7 Mars Rover, 18, 228n59, 245n98 Massumi, Brian, 16, 24, 32, 217 Mattern, Shannon, 45, 209 Maughan, Tim, 203 Maupertuis, Pierre Louis Moreau de, 129–30, 132, 133 Mauss, Marcel, 75, 78 May, John, 190, 244n95, 259n15 MbM (Made by Machine) dataset, 80 McCosker, Anthony, 192 McKim, Joel, 248n17 Mechanical Turk, 19, 77, 83, 86 Menkman, Rosa, 26, 47, 142, 213–14 metadata, 66–67 Métrophotographie (Saconney), 123–24 metrophotography, 113, 120, 123, 124, 127 Meydenbauer, Albrecht, 99, 102–3, 106, 109, 118, 122, 126–27, 180

274  Index

Meyer, Roland, 141 Mezzadra, Sandro, 32, 51–54 microphotography, 62 military operations, 11–15, 33, 52, 103, 127, 163, 173; descriptive geometry and, 136; machine vision and, 12, 24; measurement and, 99, 115 Mitchell, W. J. T., 93 model images, 86–92 Moholy-­Nagy, László, 188 Molnár, Vera, 242n89 Monge, Gaspard, 114–15, 116, 136 Mongkut, king of Siam, 151 Moro, Jeffrey, 234n52 Moten, Fred, 50 Movidius, Visual Processing Unit, 193 Munster, Anna, 22–23, 65–69, 73, 200 Muybridge, Eadweard, 189, 191, 197 Myriad 2 Visual Processing Unit, 193 Nancy, Jean-­Luc, 52 Napoleon Bonaparte, emperor of the French, 128 National Agriculture Imagery Program (NAIP), 76 Neilson, Brett, 32, 51–54 Neufert, Ernst, 45 new materialism, 37–39 Newton, Isaac, 129, 130 next frame prediction, 143, 168, 170, 171 Neyt, Adolphe, 6 nonbattle, 16 NVIDIA, 66, 94, 193 Oliver, Julian, 169 On Building (Alberti), 114 Open Images (dataset), 76 operationalism, defined, 31 operations research (OR), 13, 32, 140, 174 Order of Things, The (Foucault), 43

Paglen, Trevor, 17, 23, 25, 64, 79, 173, 174; Farocki viewed by, 21, 236n83; on invisible images, 29, 61, 81 Pang, Alex Soojung-­K im, 244n90 Panofsky, Erwin, 58, 79 Pantenburg, Volker, 17, 19, 96, 177, 206 Parallel I–IV (Farocki), 142 Pasquinelli, Matteo, 68, 72, 82 Pasteur, Louis, 62 pattern recognition, 18, 58, 68, 91, 109, 174–75, 183 Payne-­Gaposchkin, Cecilia, 6 Peirce, C. S., 244n97 Pencil of Nature, The (Talbot), 198 Perrine, Charles, 85 Perspectiva liber (Lambert), 127 perspective, 108, 110, 121 Peters, John Durham, 4 Phillips, John, 196 photo-­diagram, 17 photogrammetry, 30, 117, 118, 124, 126–27; architectural, 99; existing technology and, 113–14; instruction in, 122, 123; instrumentation and, 102, 106, 109–10; mathematization linked to, 3, 98, 114; uses of, 120–21 photography/photographs, 29, 48, 57–58, 98; aerial, 50, 55, 118, 124; as black box, 79; digital, 74; drawing compared to, 108; flash, 197; imperialism and, 136; as model images, 88–89; money linked to, 72; objective accuracy ascribed to, 83; in pattern recognition, 108–9 photometry, 2, 5–6, 184 Pickering, Edward, 1–2, 10 Pietrusko, Robert, 212 Pilot Parliaments Benchmark (PPB), 82 Pisters, Patricia, 71–72 platform seeing, 65

Index  275 plausibility, 171–72 Playfair, William, 67 Poe, Edgar Allan, 145, 146 Pold, Søren Bro, 178 politics of operations, 51–52 Post-­lenticular Landscapes (ScanLAB Projects), 189, 192 Pratt, Mary Louise, 129 Princes Connect! (role-­playing game), 177 Project Tango, 183 Pynchon, Thomas, 13 Pythagoras, 114 QR codes, 177, 178, 202 racism, 15, 23, 49–51 radar, 185, 196 Raman spectroscopy, 3 RAND Corporation, 13 Rankin, William, 185 relational databases, 46, 66–67, 68 remote sensing, 6–7, 30, 32, 50, 55, 97, 118, 149–54, 159–60, 166, 179; in China, 149–50, 154, 158; of extraction sites, 56; in Geocinema’s artworks, 92; of light spectra, 195; types of, 131 “Resolution Studies” (Menkman), 47, 142 Ritter, Johann Wilhelm, 181 robotics, 69, 87, 97, 208 Rotman, Brian, 232–33 Rutherfurd, Lewis Morris, 6 Ruttman, Walter, 180 Saconney, Jacques-­T héodore, 123–24 Sagan, Carl, ix Satz, Aura, 26 Scanlab, 30, 181, 189–98, 202–3, 206 Schneider, Birgit, 125, 143–44, 233n33 Schuppli, Susan, x, 152

science and technology studies (STS), 8, 9, 37, 38, 53, 154 Seeing Like a Rover (Vertesi), 245n98 Sekula, Allan, 25, 48–49, 66, 72, 80–81, 164, 168 Serres, Michel, 37, 38 Seven Bridges of Königsberg problem, 124 sewing machines, 148 Shekhar, Sashi, 263n63 shopping malls, viii Siegert, Bernhard, 35, 38, 41, 42; grids viewed by, 43–44; hybrid objects viewed by, 37, 43; at IKKM, 34 simulations, xi, 173, 205 Sjölén, Bengt, 169 slave trade, 50–51 Sloterdijk, Peter, 127 smart cities, 180 smartphone, 70, 183 Spacelink satellite, 4 spectrometer, ix Speer, Albert, 45 Sprenger, Florian, 208 Stabius, Johann, 139, 140 standardization, 45, 78, 152, 164, 189 statistical modeling, 91 Steyerl, Hito, 17, 23, 25, 41, 57, 61, 74, 76, 142 Strelka Institute, 26–27, 153 Strindberg, August, 62 “Surface Value” workshop, 169 surveillance, 49–50, 70, 72 surveying, 95, 108, 115, 116, 126–27, 129 Suess, Solveig, 150, 156, 157, 166, 169 Syjuco, Stephanie, 61 Synge, Edward Hutchinson, 184, 187, 260n26 Talbot, Frederick Henry Fox, 198 Taylor, Frederick Winslow, 84 telegraph, 148, 172, 173–74 telephone, 148

276  Index

telescope, 87–88, 116, 117, 131–32, 212–13 TensorFlow, 21 Terraforming program, 26–27 Terrall, Mary, 130, 131 Toscano, Alberto, 164, 165 traffic control, viii, 175 transparency, 4, 60–61, 71, 81, 93, 194 Tresch, John, 159 Tsing, Anna, 51 Uncertain Commons collective, 172 “Underweysung der Messung” (Dürer), 124 unimages, 63–64 Vagt, Christina, 173 Väliaho, Pasi, 17 Varley, Cornelius, 116 Vertesi, Janet, 228n57, 228n59, 245n98 Vertov, Dziga, 180 Virilio, Paul, 11, 15, 16, 64, 195, 219, 242n74 Vismann, Cornelia, 34, 36, 44 visual processing units (VPUs), 192 Vogel, Hermann Wilhelm, 6

Vogl, Joseph, 88, 90 Vold, Pamela, 263n63 Volkmer, Ottomar, 62 Voltaire, 130 Wagner, Beny, 26 Weibel, Peter, 152 Weizman, Eyal, 107, 260n24 Wen, Xinyi, 116 Where the City Can’t See (Young), 203–4 Whipple, John Adams, 6 White Out (Menkman), 142 WiFi, 188, 201–2 Wilder, Kelley, 3, 6 Wilker, Rowan, 192 Winthrop-­Young, Geoffrey, ix, 36, 231n14 Workers Leaving the Lumière Factory (Lumière), 85 Yosemite Valley, 189, 190, 191, 197 Young, Liam, 203–4, 206 ZKM (Zentrum für Kunst und Medien Karlsruhe), 152 Zylinska, Joanna, 84

This page intentionally left blank

This page intentionally left blank

Jussi Parikka is professor in digital aesthetics and culture at Aarhus University in Denmark. He is also visiting professor at the Winchester School of Art and at FAMU at the Academy of Performing Arts in Prague, where he leads the project Operational Images and Visual Culture. His books include Insect Media: An Archaeology of Animals and Technology (Minnesota, 2010), Digital Contagions: A Media Archaeology of Computer Viruses, and A Geology of Media (Minnesota, 2015), as well as the coauthored The Lab Book: Situated Practices in Media Studies (Minnesota, 2022). He is coeditor of Photography Off the Scale: Technologies and Theories of the Mass Image.

This page intentionally left blank

PLATE 1. Fraunhofer lines help to visualize and analyze the distribution of the solar spectrum that becomes operationalized for chemical analysis and forms one core part of the history of remote sensing. Image drawn based on the original.

PLATE 2. Four stills from Harun Farocki’s Eye Machine III (2003). Courtesy of Antje Ehmann. Copyright Harun Farocki GbR.

PLATE 3. While Neuferts’ Architects’ Data is one example among many in the history of territorial and architectural planning, gridded images have become a default environment for design way beyond specific manuals. One could build an entire archaeology of grids inside the design space of an Autodesk software suite. An image of Autocad 2022 Interface. Courtesy of Athina Stamatopoulou.

PLATE 4. A Landsat image. The Making of Earths, 2020, Geocinema. Still from a video. Courtesy of the authors.

PLATE 5. A series of predictions of the movements of the Ucayali River in Peru as part of Abelardo Gil-­Fournier’s experimental workshops that test how geographical datasets can become visual experiments. Some of these sequences display predictions that clearly “fail” as the river seems to dissolve itself. The failure is visually interesting: a visual accident becomes a reminder of aerial images of floodings and rivers out of control. Reprinted with permission.

PLATE 6. Dream Life of Driverless Cars, originally produced for the New York Times, 2015. ScanLAB Projects; reprinted with permission.