The new all-seeing overlords have read the theory. There is a bewildering photograph in a recent New York Times Magazine article on the controversial and secretive tech company Palantir.1 The scene is this: four figures appear at the center of an unnatural vignette, lit as if by some secret source of light. They are gathered around a table, their attention directed towards the screen of a laptop. Each is smartly and casually dressed, all have kicked oﬀ their shoes in favor of just socks. The leftmost, standing figure is Alex Karp, Palantir’s CEO. The others, seated at the table, are his employees. Behind and slightly above this group there is, hanging on the oﬃce wall, a large framed portrait of a man with wire-framed glasses and a face like a skull. The subject of this portrait, more or less instantly recognizable as he is pictured here—left hand resting beneath his outsize head, ambivalent gaze seemingly directed out at us—is Michel Foucault.
There is an astonishing dissonance to such a choice in corporate wall art. Karp, who has built a company that by its own description enables “more eﬀective surveillance by the state of its adversaries” has hung up, ostensibly as a source of workplace inspiration, the face of the theorist who warned that visibility is a trap, that generalized surveillance is a mode of power that works not upon bodies but, as a political tactic, upon souls. Why would a privatized operation of seeing, created in order to “target terrorists and to keep soldiers safe” display, as if it were a motivational poster, this enduring (if somewhat equivocal) symbol of resistance to such forms of power? Surely it is no accident, this picture hanging here. Maybe we can even imagine the scenario: some important person at Palantir must have decided, perhaps for some reason between justification and spite, that Foucault specifically should adorn this wall. Perhaps this person tasked an intern, say, with getting this picture—not, so far as I know, available for purchase anywhere—professionally printed, framed, and hung. The portrait, this is to say, is intellectually deliberate; it is meant to convey a message. So we might wonder: when Palantir’s employees look at it, what do they feel? Do they think it’s ironic? Or might they regard Foucault in a way that makes their own morally and politically dubious enterprise seem justified, even necessary? Another way to pose the question: is the hidden guard of the panopticon’s central tower reading Foucault?
In search of an answer, let us briefly elaborate upon our players. First, Palantir Technologies Inc. Named for the indestructible crystal ball used to surveil past and future occurrences on Middle-earth, Palantir is an American software company with a handsome sans-serif logo and stylish exposed-concrete offices in trendy neighborhoods around the world. Despite bearing these standard aesthetic markers of tech-startup playful simplicity and nerd-culture harmlessness, the company is perhaps unique in the controversy it has attracted, even in a field where controversy has grown somewhat routine. Palantir has been intensely—some would say unfairly—criticized for the general danger it poses in terms of privacy and surveillance, as well as for its specific ties to the Trump administration and Cambridge Analytica, US Immigration and Customs Enforcement, the National Security Agency, the Central Intelligence Agency, and the US Army.
Palantir is employed by these state agencies as well as by a number of private corporations to do data integration. Palantir synthesizes huge quantities of data and constructs “generalizable platforms for modeling the world and making decisions.” Palantir is in the business of analysis, of knowing lots of things about lots of people. The company has two platforms, Foundry and Gotham. The former is built to “accelerate production” and “sell more products” and the latter is built for battle, for improving “decisions for operators at every echelon and in every domain—land, sea, air, space, and cyber.” Both of these platforms work on the level of metadata to provide insight both into large systems (like populations) and into highly specific matters (like particular individuals). Palantir could identify patterns in millions purchases in order to sell more Coca-Cola; it could identify patterns in your everyday movements and blow you up with an airstrike.2 Both scenarios, of course, are governed by the same basic principle of meticulous observation. Sometimes Palantir works with literal forms of observation and seeing—like imagery from the aerostat surveillance balloons the US Air Force flies over Afghanistan—but more often the company “sees” things by processing information like calls, texts, purchases, tax returns, biometric data, etc. In either case, Palantir’s project might be essentially described as one that makes visible that which previously was invisible, lost in a vast and incoherent field of data.
Here we might turn to the second subject of the picture: Michel Foucault. Like Palantir, Foucault’s work is about knowledge, power, and visibility. From his wide-ranging studies on the histories of sexuality, madness, and punishment, there emerges a general framework, a notion that power has come to function in societies not through violence and spectacle but instead through discourse, analysis, and regulation. Power operates on groups and populations, this is to say, by studying them, by making them visible. Certainly the most notable expression of this theory of knowledge/power is the panopticon, an 18th-century architectural model of a circular prison where inmates are backlit in individual cells and surveilled from a central tower. Here, the collective of prisoners is transformed into a group of “separated individualities” who internalize the gaze of the guard because they know that it could be trained upon them at any given moment. For Foucault, panopticism is a flexible schema that has spread throughout the social body; it is representative of a profound historical shift during which sovereign power shed its reliance on ball and chain and scaffold and became “discipline,” a modality of power that works efficiently through surveillance, classification, and assessment, through the formation of bodies of knowledge.
It is not especially difficult to describe the ways Palantir exemplifies the Foucauldian discipline mechanism. In recent months, for example, Palantir’s work with dozens of countries tracking the spread of COVID-19 seems an almost prophetic fulfillment of Foucault’s description of the plague as a “form” of disorder from which disciplined societies are born. The ideal exertion of disciplinary power, he writes, occurs within the model of the plague-stricken town, a community “traversed throughout with hierarchy, surveillance, observation … [and] an extensive power that bears in a distinct way over all individual bodies.”3 In the 17th century, according to Foucault, the plague was met with order, with surveillance and partition and regulation. Today, the coronavirus crisis is addressed similarly, through organizations of space and time, through tracking and tracing. It is Palantir’s software—which was used to analyze the health data of millions of people—that, in meeting the need for order, affords the state a greater view of its population. In a more general sense, too, Palantir’s work can be aptly described by what has become the sort of slogan of discipline: “Another power, another knowledge.” Power implies knowledge; knowledge implies power. Whether Palantir is enlisted by some state apparatus like ICE or a private company like Home Depot, what its software does is analyze data and render it intelligible and, therefore, actionable. Corporate and state powers can easily collect data, but they do little with this raw information. It is Palantir that performs an essential task on their behalf: transforming information into knowledge.
Still, Palantir is no panoptic machine of old. When the discipline mechanism becomes digital, it changes in a few important ways. In the circular prison, power operates automatically because inmates are always aware of the possibility of surveillance. But the high-tech gaze of Palantir is, on the other hand, utterly discreet. Those in its field of vision have no way of knowing it. If there is an “eye” to be ascribed to projects of data aggregation, perhaps it is more the eye of god than the eye of the guard. And god, of course, is not subject to the same rules as prison guards. Thus, while Foucault notes that panoptic mechanisms are “democratically controlled” in that they can be “supervised by society as a whole,” no such arrangement applies to Palantir, which operates in secret, accessible only to those who pay between 10 and 100 million dollars for its services. Perhaps more significantly, the god-eye of Palantir differs from the view from the tower in that its technological gaze is not only punitive but speculative or predictive. The idea behind finding patterns is knowing what comes next. Palantir’s software, as it is used by the LAPD, for example, analyzes information from different databases (police surveillance systems, DMV data, Verizon telecom data, etc.) in order to generate for officers lists of people “who the system believes are likely to commit a crime.”4 Thus Palantir expands the reach of the discipline mechanism in both space and time. Sometimes it’s at work in traffic cameras, online checkouts, voting booths. Other times it’s hurtling over Earth: missiles, satellites, surveillance balloons. It is external and invisible. It assembles histories, it knows the future.
To read Palantir through Foucault is to see the company in a deeply unflattering light, both when its work epitomizes disciplinary power and when it supersedes it. In the framework of knowledge/power Palantir occupies the position of a middleman, a translator. It would be incorrect in this sense to see the company as ever “having” or “being in” power; instead its project is one of accelerating, streamlining, and extending the reach of whatever corporate or state body uses its software. Palantir, however, does this job so well that it could be said to change the very nature of power: its shiny form of discipline remains capillary and founded on “knowledge,” but Palantir has changed the very limits of what can be known. To consider Palantir in this way—as serving a specific function within and on behalf of power—is to see data integration as fundamentally political, insofar as politics is the exercise of power. So apart from the way that Palantir’s complicity in Trump’s immigration crackdown, for example, may have oﬀended the “political” sensibilities of many of its employees,5 a Foucauldian conception of the company casts it as political in its very essence. Palantir is not simply dangerous for its connection to any particularly villainous administration—it is politically suspect through and through.
Karp, briefly a student of Jürgen Habermas, surely realizes that the Foucault portrait invites this view of his company. So why hang the picture? A clue, I think, lies in the aforementioned “Palantirian” response to the ICE contracts. In August of 2019, more than 200 of Palantir’s employees signed a letter confronting Karp on the issue. Though Karp would respond by renewing the 42 million dollar contract the same month, the letter reveals how Palantir’s work is somewhat at odds with its liberal white-collar employees. Perhaps then we ought to consider the function of the portrait to be a matter of self-reflection, a means for Palantir to see itself in a way compatible with the values of its workers. Indeed Foucault is a fitting choice for such a self-image rehab precisely because Palantir’s work resembles that of Foucault in that both are built on the same suspicion that in the world there are hidden currents of meaning, that knowledge of societies can be unearthed and deciphered through study. In a strange way, this is to say, the discourse of analyzing power sometimes mirrors the very structures it seeks to reveal. Obviously “discourse” and “power” are supposed to be political opponents but here in Palantir’s office the symbology of the former is used to cast the latter as an intellectual and loosely Leftist pursuit. (Of course, Foucault is also a somewhat mistaken symbol of the Left insofar as his politics were certainly equivocal and his theory of power cannot be easily aligned with that of Marx, for example.)
The portrait, then, does not evoke the work/framework of Foucault. Instead it designates only what it depicts: Foucault himself. The mystifying figure with a shining shaven head, the master anatomist of culture and discourse who was, surely at the time of his death, the most famous intellectual in the world. In other words, at work here is a sort of pseudo-cultural mode of severing signs from meaning. The portrait is a surface without depth, a floating symbol. It is devoid of its theoretical context (which reveals Palantir to be politically suspect) so that now it conveys only some vaguely countercultural notion of “intelligentsia.” This treatment is somewhat narcissistic, in a psychoanalytical sense. This is not to say that Karp and Palantir’s employees are pathological narcissists but rather that their relation to their corporate wall art resembles a narcissistic mode of relating to one’s surroundings and, especially, to images. In the myth of Narcissus, the reason for the hunter’s demise is not simple self-fixation but rather his inability to differentiate between reality and reflection. For Freud, narcissism meant attachment to, identification with, and libidinal investment in figures that remind one of oneself—or of what one would like to be. In this sense narcissism is about surface phenomena and self-flattery. It’s about pictures and mirrors and the ways individuals sometimes identify with such seductive trick images.
So perhaps there’s a sort of corporate narcissism here. To those at Palantir, these surfaces and empty symbols may feel truer than the actuality of the company's work. The portrait of Foucault does not conceal Palantir’s project but changes it. The picture is not a matter of outward image or PR. It’s a matter of interior appearances and thus its implications are moral. For it’s easier to believe one’s software is not dangerous and/or evil when one identifies with symbols of intellect and Left politics. It’s easier to facilitate deportations and predictive policing, say, when one believes oneself a guardian of epistemology and liberal democracy. Power was once clad in black, built in iron and stone. Now it is forward, intellectual, casual. Lightness is both the schema and the aesthetic. If there is a moral burden to technological omniscience—some weight of power and shame—it is perhaps easier to bear (or ignore) if one can walk around the office in socks.
Foucault, Michel. Discipline and Punish: the Birth of the Prison. 2nd Vintage ed. (New York: Vintage Books, 1995): 198.