Beneath the Information, the Operation: Harun Farocki and the Work of Images

1. Against the Creepage Current

During a panel discussion at the third Berlin Documentary Forum in 2014, Harun Farocki addressed the moment in the early twentieth century when film eventually became “entertainment, education, and so on,” ceasing its vital function in producing objective knowledge. The filmmaker compared it to a loss of employment and the recognition that accompanies work: “That’s not good!” he declared indignantly. “It’s better to have work to do. We know that liberation is very often linked to having a job.”1

In a sense, Farocki had rediscovered a version of that missing “work” in the 1990s with his “operational images.” However, the distinction underlying his 2014 statement on the “work” of images had been there for some time. In 1970, outlining the theoretical context of The Division of All Days (Die Teilung aller Tage), a Marxist Lehrfilm introducing the concept of surplus value, he and his collaborator Hartmut Bitomsky wrote:

What is called “film” comprises a whole bundle of functions. Roughly speaking: Film exists to reproduce the commodity of labor power (vulgar entertainment). Film exists to convey information (productive force) and to provide it (science as productive force). Film exists to confer qualifications. It turns out that these functions interfere with and cancel each other out if they are not first separated.2

For the young Farocki, the actual “work” of film went beyond information and entertainment. Though it could be found in the early history of film’s scientific applications, Farocki saw a need for it to be updated and mobilized for agitational purposes in a potentially revolutionary present following 1968. This preference for images at work—especially for photography and film in the fields of photogrammetry, microcinema, industrial motion studies, chronophotography, and other forms of visual research and production using images—was to remain an integral part of Farocki's approach to filmmaking and his critical engagement with visual culture. The conceptualization and theorization of operational images, Farocki’s most influential intervention into the semantics of visual and media studies, was largely prompted by the agonizing abundance of images that signify, represent, and metaphorize rather than “work.”3 For Farocki, the way out of this predicament inevitably meant creating a counterweight to a corrupted capitalist visuality. It was, however, a different kind of exit than was hinted at in the German title of his most comprehensive text on operational images, “Der Krieg findet immer einen Ausweg” (“War Always Finds a Way,” quoting directly from Brecht’s Mother Courage and Her Children).

Originally drafted for a lecture delivered at the HfG Karlsruhe on October 15, 2003, and published thereafter in English and French translations in 2004 (prior to being made available in German in 2005), the essay provides a theoretical and historical backbone for his three-part installation Eye/Machine (2000–03) and the film War at a Distance (2003), his influential engagement with images no longer primarily intended for the human eye to see, but rather for machinic entities to read, scan, decode, and process further. Farocki complained of being “weary of the everyday practice of remythologizing everyday life and of the multiple and multi-channel program of images that are designed to mean something to us.”4 Operational images, on the other hand, do not usually participate in such a re-mythologization, which is caused by a type of semiotization that tends to have no connection to its material referent. They only do so when taken out of their operational context and into the realm of representation for promotional or explanatory purposes. Stripped of the functions of meaning, reference, representation, or fictionality, they actually offer an antidote, if not a pharmakon—which Farocki clearly understood as inherently ambiguous, i.e., both healing and toxic. Despite his investment in the political and epistemic operativity of images, Farocki’s discourse also introduced a therapeutic-pharmacological knowledge in a Derridean sense, i.e., affirming as well as opposing the technical knowledge embodied by operational images.

More important for Farocki, however, was Paul Virilio’s theorization of the “industrialization of vision” from his 1988 The Vision Machine. Virilio summarized much of the time’s knowledge surrounding machine vision and AI when he wrote of “sightless vision,” “synthetic perception,” and the computer’s responsibility “for the machine’s—rather than the televiewer’s—capacity to analyze the ambient environment and automatically interpret the meaning of events.” For Virilio, this technology necessitated that “the question of ‘artificial intelligence’ … would be used in industrial production and stock control; in military robotics, too, perhaps,” while “delegating the analysis of objective reality to a machine,” contending that it would rely on “the formation of optical imagery with no apparent base, no permanency beyond that of mental or instrumental visual memory.”5

Virilio’s thinking on the transitory “nature of the virtual image” was inspiring for Farocki, who tried to combine a critique of the “industrialization of thought” with his fascination with an ontology of film based on a notion of work, operativity, and knowledge production.6 Fighting what he once called the “creepage current of consciousness” produced by mass media, Farocki revisited and updated an earlier notion of operativity in relation to images.7 He put the audiovisual “creepage” that escapes reflection and critical attention against the “invisuality” of operational images.8 But the respective visibilities and invisibilities differ greatly: While operational images only become visible to the human eye under certain conditions, the invisibility of “industrialized” television images (or the “visual” in Serge Daney’s sense) results from excessive visibility.9 Farocki’s interest in images “that are part of an operation” stemmed from his critique of an especially pernicious kind of operation he named “re-mythologization” or “industrialization.” In many respects, these tendencies are also symptomatic of a devaluation of form in favor of “content” (as in “content provider” or “content moderation”), a process of mass deskilling that targets the very ability to read not only the supposed “meaning” of media, but also their grammar, syntax, structure, infrastructure, rhetoric, and so forth.

2. In Defense of “Formalism”

Influenced early on by Roland Barthes’s conceptualization of the mythologizing function of metalanguage and his notion of a transitive, “operational” language organizing the interaction of humans and nonhuman entities (the notorious example being Barthes’s woodcutter who “speaks” the tree), Farocki embraced heightened attention to structure in a Marxist, rather than a postmodernist, understanding of a politics of form.10 In his crucial 1956 text “Myth Today,” Barthes established semiology as a science of forms that examines meanings independently of their content.11 Rejecting the Stalinist ban on “formalism” as well as the “structuralism” that distained orthodox historicism and dialectical materialism, the young Brechtian proto-structuralist advocated the compatibility of formal analysis and historical criticism: “To parody a well-known expression, I would say that a little formalism leads away from history, while a lot of formalism leads back to it.”12 It did not take long for the anti-substantialist, anti-patriarchal, and anti-colonial reputation of structuralism to disappear almost completely, despite the attempts by Louis Althusser and his students in the 1960s to fuse Marxism and structuralism. By the end of the decade, structuralism had become routinely equated with the technocratic ideology of the emerging information society.

Harun Farocki, War at a Distance, 2003. Copyright: Harun Farocki GbR.

In 2003, almost forty years after he had drafted two short reviews of Mythologies,13 Farocki reflects on Barthes’s defense of formalism and formal analysis:

In this text, Barthes was concerned with asserting his own practice. He wanted to belong to the revolutionary left and yet not embroider Stalinist doilies, as the CPF demanded of intellectuals. Moscow pursued the semiotics advocated by Barthes with particular hatred, as it was not based on something foreign, but on formalism, the ostracized Russian avant-garde theory; the only theoretically new thing, according to Foucault, that emerged from communism.14

Farocki, obviously aware that semiotics and structuralist thought had grown embattled amidst the politics of the Cold War, not only mentions Barthes’s demonization in the Soviet Union, but also—probably just as importantly—Michel Foucault’s insistence on the origins of postwar structuralism in the Soviet formalism of the 1920s. For Foucault,

Formalism was probably in general one of the most powerful and complex forces in 20th century Europe. Moreover, Formalism was associated very often with conditions and even political movements, which were certainly equally stimulating each time. The relationship between Russian Formalism and the Russian Revolution should definitely be investigated precisely anew.15

Rarely, in a French context, has the connection between formalist abstraction and revolutionary practice been rendered so clearly.

Farocki, who himself had a history of applying formalist methodology on political action, was obviously interested in this Nietzschean brand of postwar anti-Stalinist-yet-Marxist formalism. Such a position is particularly evident in Barthes’s “Myth Today,” where he discusses the discourse of the “Leftist Myth” (“Le mythe, à gauche”). Here, Barthes is concerned with the assertion of an “object language” (“langage-objet”) that contradicts the depoliticizing and de-naming logic of a mythical metalanguage that always threatens to seize the “object language” parasitically. The distinction between “object language” and “metalanguage” proved central. Barthes introduced it to position the specific formalism of the semiology he was proposing—aimed at “total critique” as a form of political speech or “langage réel”—against the metalanguage of myth (i.e., ideology) that empties discourse of the political. In this context, the term “langage … opératoire” (operational language) is introduced, very much in the spirit of the “militant monism” (Devin Fore) of Sergei Tretyakov and his fellow operativists (despite the absence of any direct reference to the Soviet factographers).16 However, it is only mentioned once, as Barthes does not return to this term in his essay or anywhere else in his writings. Yet it must have stuck with the young Farocki, who, while not mentioning it in his 1965 Spandauer Volksblatt review of Mythologies, extensively quotes the passage about the woodcutter and the tree for a 1965 program on Sender Freies Berlin (Radio Free Berlin), before taking it up again in the 1990s and early 2000s.17

Barthes defines operational language as language that “acts the object” (“par lequel j’agis l’objet”) in the absence of a meaning (through representation, metaphor, or imaging) beyond action. Access to this register of expression is restricted to those who, like the woodcutter, form a unity with the object of their craft; all others are condemned to remain outside the realm of objecthood (or thingness, materiality): “My language is no longer the instrument of an ‘acted-upon tree,’ it is the ‘tree celebrated’ which becomes the instrument of my language. I no longer have anything more than an intransitive relationship with the tree; the tree is no longer the meaning of reality as a human action, it’s an image-at-one’s-disposal.”18 A relationship structured in this way, which stabilizes the subject-object dyad as an asymmetrical power differential, must be understood as an expression of class struggle and thus politicized.

It is striking how intricately Barthes’s semiological formalism is linked to a political demand for imagelessness. Even if he doesn’t refer to the visual image in any strict sense, and even if the “operational” language does not yet comprise the “operational image” as Farocki will conceive it, it nevertheless shows how Barthes and Farocki are connected by a shared “weariness of metalanguage.” Replacing a pragmatic, transitive langage-objet completely absorbed in the action with an intransitive metalanguage, the revolutionary, reality-changing discourse of political action is undermined—by a preserving, static, and apolitical “image.” The revolution, according to Barthes, “makes the world, and its language, its entire language, is functionally absorbed in this making,” whereas the “image” is the axe at the revolution’s tree.19

Why did Barthes dub this revolutionary language “operational”? In addition to a productivist emphasis on “the language of man as a producer,” one could interpret this as suggesting (for those who may recall this tradition, long suppressed by Stalinism) a connection to the program that Tretyakov and his fellow contributors to Lef and Novyi Lef, such as Arvatov, Chuzhak, Eikhenbaum, Gastev, Pertsov, and Vinokur, postulated and developed starting in 1925. For Farocki, however, this fragile connection between Barthes’s semiological concept of “operative” object language and the factographic method of “operating” and thus “producing” in a revolutionary context had more than just coincidental significance.

For these ideas to take hold, Farocki might have had to ignore a problematic aspect of Barthes’s essay and Barthes’s relation to the left in general. Toward the end of the section on the “Left Myth,” Barthes reflects on the “nature of the ‘Left’” (probably targeting a particular left militant group close to the French Communist Party) and operational language. Since the left “always defines itself in relation to the oppressed, whether proletarian or colonized,” it must also grapple with “the speech of the oppressed.” The latter is a type of discourse that “can only be poor, monotonous, immediate,” because it is inevitably the barren “speech of action,” and thus the result of language’s “destitution.”20

Without the “luxury” of metalanguage, the “speech of the oppressed is real, like that of the woodcutter; it is a transitive type of speech: it is quasi-unable to lie; lying is a richness, a lie presupposes property, truths, and forms to spare.”21 If semiotics, according to Umberto Eco’s aphoristic definition, is “the discipline studying everything which can be used in order to lie,” Barthes’s (operational, actionist) “parole de l’opprimé” could be called pre-semiotic.22 But Barthes also essentializes, in thinly veiled primitivist terms, the speech of the oppressed. As Charles J. Stivale has rightly noted, his portrayal of “the dryness and awkwardness of the ‘Leftist myth’” conveys his “own awkwardness in regards to the ‘speech of the oppressed,’” and thus “proves the poverty or indiscretion of the Left in talking about its own myths.”23 The status of operational language in relation to the “masks” (Barthes) of the left’s largely unacknowledged self-mythologization remains somewhat unclear, since the parole de l’opprimé is torn between nonrepresentational, unmediated operativity and myth-generating discourse. How then does this ambiguous status of langage opératoire, no matter how casual and ephemeral Barthes’s use of the term, resonate with Tretyakov and Farocki?

3. Beneath the Information, the Operation

In 1969, several years after reviewing Mythologies, Farocki penned a series of four articles for film, a short-lived, left-leaning West German film magazine, on questions of “agitational” film, didactic strategies, behaviorism, teaching machines, and cybernetic pedagogy. In doing so, he continued from the short agitprop films made as a student at the German Film and Television Academy (dffb) in 1967 and 1968.24 He also laid the theoretical groundwork for two Marxist Lehrfilme that he would make with Hartmut Bitomsky: The Division of All Days and Something Self-Explanatory (15 x) (Eine Sache, die sich versteht (15 x), 1971).25 After displaying some basic knowledge of the Tretyakovian distinction between “informative” and “operational” literature in an article on class struggle in the Third World for the German weekly Die Zeit the previous year,26 he emphasized and further developed his interest at the outset of the article “to scientize agitation and politicize science” (see the translation by Ted Fendt in this issue). Now focusing on the operativity of agitational films, Farocki declared that in contrast to “informative films” such as the traditional documentary or audiovisual reportage, which are not limited in the nature of their content, “operative films” have no subject beyond the operations they are an integral part of. Whereas the “informative film” can be used by all interested parties, an “operative film” is not about but with the political action, i.e., it “can be judged by the operation, by its outcome, and by the significance of the operation in class strategy. the theory of the operative film and the theory of the operation are intertwined.”27

Harun Farocki. Copyright: Harun Farocki GbR.

It is worth noting that Farocki’s references to Tretyakov somewhat anticipated the latter’s widespread rediscovery among the West German left.28 Until then, Tretyakov’s rediscovery had to prevail against the focus on the Brecht-Lukács debate. As a representative of Antigestaltungstheorie (anti-design theory), according to Georg Lukács’s dictum from the early 1930s, Tretyakov was long considered problematic for more Moscow-leaning Marxists, meaning that for some time “there was no reception of Tretyakov in the Federal Republic.”29 It was only the republication of Walter Benjamin’s “The Author as Producer” in 1966 that brought his name back to mind (though Benjamin’s discussion of Tretyakov in his famous 1934 lecture at the Institute for the Study of Fascism in Paris was shorter than its long-term impact would suggest).

All the more astonishing is Farocki’s determination to apply Tretyakov, or rather the “operativism” associated with his name, to a post-1968 need to learn from the problematic reestablishment of the left-wing myth. To this end, Farocki—and he was not alone—flirted with cybernetics-inspired techno-rationalism and collaborated with students of Helmar Frank, promulgator of Kybernetische Pädagogik. Inadvertently, albeit partly strategically, he thus aligned with (a post-“operativism”) psycho-physical “operationism” and military/management related “operations research” (Operationsforschung in German) on both sides of the Cold War constellation.30 These epistemics of control and planning were believed to offer a way out of the dogmatic or metalanguage aberrations of the New Left’s Kulturrevolution, as a kind of sobering-up that relied on “operative” efficiency in the mediation and addressing of knowledge through communication. This new approach was designed to converge seamlessly with political action. However, this was only possible if a traditional notion of information were abolished and the last remnants of a regime of representation dismantled. In another article for film from 1969, Farocki argued that even if information is successfully conveyed to a target audience, it still remains ineffective: “It only has an effect when the addressee is simultaneously given the opportunity to put the information into practice; in other words, when it is organizing, operational information.”31

4. The Utopia of a Materialist Semiosis?

Two decades on, Farocki would begin to document the formatting and technologization of education, schooling, job interview training, etc., in companies and government agencies in films such as Indoctrination (1987), How to Live in the FRG (1990), and The Interview (1997). The operations of an increasingly post-Fordist workplace as social factory became the subject of his films. The “discretion” of industrial work, as practiced and researched by Soviet Taylorists and factographers in the 1920s, then finally became a topic around 2014 with preliminary research for a project on “motion study” that would have reestablished contact with Tretyakov and Gastev, had Farocki not died in the summer of 2014.32

Harun Farocki, The Interview, 1996-1997. Copyright: Harun Farocki GbR.

At that time, the focus on operational images had receded into the background of his practice. Or, more accurately, operational images could be taken for granted as a firmly established “standard”—a favorite term of Farocki’s at the time—in increasingly automated, digitized, and virtualized environments of work and visuality. However, engagement with the seemingly ubiquitous, universal nature of such contemporary “operativism,” as theorized by Aud Sissel Hoel, Jussi Parikka, and Roland Meyer, among others, could benefit from a closer look at the multiple origins of the trope of operational images.

Returning to the aforementioned pharmakonic ambiguity, bear in mind that Farocki began to operationalize filmmaking in the 1960s by discarding older notions (and protocols) of “information”—i.e., documentation, representation, and metalanguage. In many ways, he shared Barthes’s fondness for the “denoted image,” which Barthes discussed at length in his 1964 essay “Rhétorique de l’image,” his early foray into photo theory. With regard to photography in advertising, Barthes intervened in the “denotation” of the relations and means of production of audiovisual images and the economy in general. He distinguished between “symbolic” and “literal” images, the latter being first of all “privative,” i.e., “constituted by what remains in the image when we (mentally) erase the signs of connotation … Utopianly rid of its connotations, the image would become radically objective, i.e., ultimately innocent.”33

For Barthes, the “utopian character of denotation” hinges on the “paradox” that it is the photograph’s analogical nature that defines it as being “without a code,” since it “possesses the power to transmit (literal) information without forming it with the help of discontinuous signs and rules of transformation.” He then compares the “utopian” denotation of the photograph with the denotation of a drawing whose “coded nature” is based on a historical “set of regulated transformations,” on its selectiveness (“the drawing does not reproduce everything, and often very little”), and its inability to be “without style.”34 Barthes thus differentiated between different types or levels of denotation, from the utopian to the less utopian, from the photograph to the drawing. He did not discuss the encounter of these different types in hybrid images with a diagrammatic function though, such as in “the image as a schema, as a plan, qualitative and quantitative analysis through comparison,” which Tretyakov contrasted with the impressionism of “artistic imaginism” in 1928, in the absolute desire “to purify our texts by orienting ourselves towards rough types of report and technical prose such as report, command, and lecture.”35

5. “political beings who talk to each other and criticize images”

Farocki, who was always interested in diagrams and the boundaries between different technologies and modes of inscription and representation, turned to precisely such hybrid imagery at a central point of his engagement with operational images: “Many operational images are crisscrossed with colored auxiliary lines,” he wrote in connection with his video Eye/Machine 3 (2003), referring to the so-called bounding boxes of digital object detection: “The lines convey emphatically what is important in the images and just as emphatically what should not be important under any circumstances. The excess of reality is denied—a constant denial with a counter-effect.”36 As if alluding to Barthes’s discussion of the visual semiotics of advertising, Farocki established a connection between the codes of computer vision and advertising:

We know from advertising that there is a real image to be seen, and that the brand name and price are inscribed in the image. The auxiliary lines that the image-processing computer adds to the real images go beyond this copresence. They are intended to express the essence of the image, to suspend the real image. These images thus constantly repeat that they are no longer important, only what can be read from them. 

And as in several other texts from around 2004, the paragraph concludes with the concept of a “counter-effect” (or “counter-action”) to be expected from such a denial.37

Harun Farocki, Serious Games III,film still, 2009. Copyright: Harun Farocki GbR.

Terms that could be used here as synonyms for “denial” include “bias,” “strategic ignorance,” “epistemic violence,” and “racial profiling.” According to Farocki’s understanding, the machine vision of operational images is based on the principle of selections programmed by algorithmic decisionism. We should not forget that it was the weaponization of machine vision in the Second Gulf War that led to the terminology and the exploration of operational images in Farocki’s Eye/Machine and War at a Distance, as well as subsequent works such as Serious Games I–IV (2009–10). As much as he may have been interested in pursuing his version of a post-representational, denotative, and thus operational image in documentary—sometimes referred to as “observational film” in the tradition of Direct Cinema, and certainly in dialogue with a factographic past—operational images were first and foremost the subject of “political beings who talk to each other and criticize images, and who are well aware of the difference between the first war, in which Kuwait was invaded and annexed by Iraq, and the second Iraq war.”38

Instead of allowing himself to be lured by the gamification of war into becoming an apolitical actor in the theater of operations—the position of military technicians and (drone) operators—Farocki insisted that it was necessary to observe the operative denial in operational images. This meant reckoning with the “counter-effect,” the boomerang of the violence exerted via “clean” warfare, smart bombs, targeted killings, and—in present and future wars—by AI, robots, drones, and other nonhuman agents, usually on distant people and territories rendered invisible, objectified, occupiable.

At the same time, he was fascinated by the efforts and contradictions of image processing performed by these Arbeitsbilder (“working images,” rather than “images of work”).39 In the early 2000s, as something of a human annotator avant la lettre, Farocki was hardly able to imagine the drudgery of today’s extraction and exploitation in an AI industry that had not yet arisen. He couldn’t yet relate it to the living labor—rendered invisible, though increasingly organized—in the archipelago of sweatshops and the globally distributed “computer vision lab” where click workers and content moderators toil in the Global South.40 Neither did he envisage the ecological consequences brought about by such excessive energy consumption and the waste of material resources on which object detection is based.

6. Missing a Complex Image?

Nevertheless, Farocki clearly recognized how images used in operations function as effective means of exclusion, destruction, and the rendering invisible of entire populations. An important source of inspiration for conceptualizing the significance of the invisibility of victims of military violence was Serge Daney. The brilliant film and television critic—the same age as Farocki—provided crucial answers early on to the urgent need for a theory of images appropriate to the conditions of operational visuality.41 In 1991 Daney, having written several op-eds for Libération on the US military’s deployment of images in the Second Gulf War, was invited by the Revue d’études palestiniennes to synthesize these reflections and extend them to the question of Palestine. The result was an essay in which Daney elaborated not only on his distinction between “image” and “the visual”—“image” as something that “still relies upon an experience of vision,” and “the visual” as no more than the “optical verification of a procedure of power, whatever this may be (technological, political, advertising, military)”—but also, toward the end, on the “problem of the image of Palestinians.” With regard to the latter, Daney contended that, in the course of history, the signifier “Palestinians” had won over what it was supposed to signify. And with a touch of awkwardness and nonchalance with regard to the speech of the oppressed—not unlike that of Barthes—he not only asserted that it is “easier to die for a word than to work at the image of a thing,” but also lamented the lack of a “complex image of the Palestinian reality.”42

There is certainly much to be said about how Daney grafts his distinction between “image” and “the visual” onto the question of Palestine—especially considering an alternative, less simplistic conception of who is capable and entitled to lay claim to the “complex image” of a reality or to claim its absence. But for the purposes of this essay, the evocative phrase “to work at the image of a thing” (de travailler à l’image d’une chose) establishes a—potentially generative—connection to Farocki’s emphasis on the work of images that precedes, undergirds, and critiques the variant (or deviant) of such work in the operational. The worked (or working) image must be wrested not only from the clichés produced by the media industry and the contemporary historiography associated with it, but also from the inadvertently political performativity of those images that are operative but neither recognize nor acknowledge the work on which they continue to be based. In many respects, Farocki’s investigation into the semiotics of operational images was also motivated by the need to counteract the catastrophic loss of images that are not clichés, that aim at the kind of complexity Daney was demanding, and that elicit a kind of “counter-effect” of possibility different from the exclusionary and all-too-often dehumanizing images that have long since ceased to work and merely operate.