This is an interview by DIFFRACTIONS member Patrick Leftwich [Kitsumute] with Alexander R. Galloway. Alexander R. Galloway is an American media theorist, philosopher, and computer programmer known for his work at the intersection of critical theory, digital media, and continental philosophy. He holds a Bachelor’s degree in Modern Culture and Media from Brown University (1996) and a Ph.D. in Literature from Duke University (2001).
Since 2002, he has been a professor in the Department of Media, Culture, and Communication at New York University, where he is currently Professor of Media, Culture, and Communication. Galloway is also a founding member of the Radical Software Group (RSG), an art collective founded in 2000, through which he developed influential software art projects such as Carnivore (which won a Golden Nica at the Prix Ars Electronica in 2002) and Kriegspiel. His major book publications include Protocol: How Control Exists After Decentralization (2004), Gaming: Essays on Algorithmic Culture (2006), The Exploit: A Theory of Networks (2007, co-authored with Eugene Thacker), The Interface Effect (2012), and Uncomputable: Play and Politics in the Long Digital Age (2021).
Galloway’s work has been recognized with numerous prestigious awards, including a Guggenheim Fellowship in 2019, the Berlin Prize from the American Academy in Berlin, and a Golden Nica from the Prix Ars Electronica; his writings have been translated into a dozen languages. He has held visiting professorships at the University of Pennsylvania (Spring 2012) and Harvard University (Fall 2016), and has been a member at the Institute for Advanced Study and a fellow in the Cinepoetics group at the Freie Universität Berlin. Most recently, he co-authored Digital Theory (University of Minnesota Press, 2025/2026) with M. Beatrice Fazi, Matthew Handelman, and Leif Weatherby.
Patrick Leftwich: I would like to begin with a fairly general question that can lay the groundwork for the rest of our conversation as well as serve as a point of entry for readers unfamiliar with your work. Already in 2004 in Protocol: How Control Exists After Decentralization you challenged the widespread belief in academia and public discourse that decentralization—a defining feature of network technologies such as the World Wide Web—stands apart from, or even opposes, state and corporate control. How do you understand the relationship between media, power, and thought today? Has your view changed since Protocol, and if so, in what ways?
Alexander R. Galloway: Studying the protocols was a lesson in deterritorialization. Not all protocols, of course: some establish rhythms of coherency and correction, structure and hierarchy. But certain aspects of the Internet protocols are a study in flexibility, heterogeneity, contingency, and complexity. And yet these deterritorialized networks don’t fall apart; they hold together astonishingly well. My 2004 book was an attempt to show how control and organization continue to thrive within a seemingly anarchic communications infrastructure.
Ironically, protocol itself was vanishing and transforming during those same years, confirming the old maxim that certain phenomena become legible at the moment of their disappearance. The rise of platforms and cloud computing in the mid to late 2000s marked a turn away from protocol, a reactionary return to older models of centralization and proprietary control. In psychoanalytic language, the Oedipal structure has been reestablished. Rhizomes don’t matter so much today; we’re back to roots and trees.
PL: It seems to me that your notion of the relation between the analogue and the digital has shifted over the course of your work, a shift that appears reflective of your engagement with the three great French metaphysicians of the second half of the 20th century: Deleuze, Laruelle, Badiou. How do you currently conceive of the analogue-digital relation? How has your position evolved, and to what extent was that evolution driven by internal conceptual tensions as opposed to transformations in the technical and the social themselves?
AG: Some key moments mark this conceptual development. The first moment for me was to insist on the digital as a mode of mediation using discrete units, rather than a synonym for today’s consumer electronics. The first moment was the easiest and most obvious, because, in fact, that’s the dictionary definition of the digital. It has been challenging, however, to pursue every possible consequence. One consequence is to focus on discretization itself. What does it mean to discretize? What is a cut? What is a unit? How do monadic units become arranged in regular series? These are hard questions. Interestingly, there’s already an established science for this, the science of arithmetic. And that was the second moment: to acknowledge arithmetic as the elemental science of the digital, which simultaneously reveals geometry as the elemental science of the analog.
As you can see, it is important to open the conceptual window when studying the digital and the analog. And it’s important to open the historical window as well. It’s just not sufficient to say that the digital means contemporary consumer electronics. Computers can be analog, after all, and at the same time the digital is very old. In fact, discretization is one of the oldest topics in philosophy, going back to the pre-Socratics. This led me to the Pythagoreans and ultimately to Euclid, who already furnished a clear and rigorous definition of the analog and the digital (via the old Greek concepts of analogos and logos), as well as the related concept of the non-digital or irrational (alogos). This disrupted a few preconceptions: the analog and the digital are not binary opposites or contraries; in fact, the analog is a kind of reduplication or echo of the digital, just as the word ana-logos contains an echo of logos.
Looking back to the ancients also invites a reappraisal of digital studies in the present. Some of the most sophisticated digital theory happened in the middle of the twentieth century during the heyday of structuralism, well prior to the advent of the personal computer. In other words, if a return to arithmetic is necessary, digital studies will also need to return to structuralism, for those years were a high-water mark for investigations into symbols, logics, language, grammars, codes, and binary relations. Just as Lacan characterized his seminar as a “return to Freud,” I suspect digital studies will need to define itself through a return to structuralism. Mind you, I’m not necessarily a fan of structuralism; my claim is about how these two fields belong together, not an endorsement (or dismissal) of either field. And at the same time there can be no return to post-structuralism within digital studies. I insist (against some vocal protests from Derrideans and others) that post-structuralism has very little to offer digital studies.
In sum, Laruelle opened an avenue of inquiry—he would have called it a discovery—in which the digital is a generic operation within philosophy rather than the digital being understood through consumer electronics. I thought I had explored that avenue sufficiently in 2014 after finishing a book-length study of Laruelle. It soon became obvious, however, that I had only scratched the surface. And so I have been struggling for the last several years to elaborate a general theory of the digital and the analog. Filling out the trilogy would indeed be attractive, with Deleuze as a chaperone through the analog and Badiou through the digital, but I doubt I will ever be able to accomplish something so ambitious in scope.
PL: Could you elaborate on why you think post-structuralism has very little to offer digital studies? Is this connected to the questions of method: can the digital be abstracted from the mixed digital–analog architectures and/or practices by philosophies grounded in an essentially analog conception of media? If the analog is not the opposite of the digital, but rather stands in a tangential relation to it, what exactly makes the post-structuralist approach misleading or inadequate?
AG: Post-structuralism was too closely wedded to the jargon of analogicity. Recall the old terms from late-’60s and 1970s theory: slippage, residue, play, otherness, aporia, lack, trace, assemblage, decentering, desire, abjection, the Real. Digital technology can’t adequately account for these terms, and the phenomena they describe, except through clunky approximation. Note that I tend to sideline anthropological accounts of “digital culture,” which often project human traits onto machines, enacting a version of the pathetic fallacy. In other words, if we bracket the human user, I’m not aware of any actually existing digital machine where the integers are deterritorialized, or where procedures exhibit indeterminate play, or where variables follow the logic of desire and lack. On the contrary, those sorts of phenomena are militantly policed and actively excluded from digital tech. There’s a lot of magical thinking within theory circles that wants to say otherwise. But not much evidence exists to support it. So perhaps post-structuralism will supply the negative space, a ground for the digital’s figure, where the limits and shortcomings of the digital are revealed via silhouette, by way of what the digital can’t do.
Of course digital-analog mixtures exist. Keep in mind, however, that hybrid and mixture are analog technologies, so they already tilt in favor of the analog. Heterogeneous interfaces are the very lifeblood of the analog. (Which is one reason why Deleuze was such a virtuosic analog philosopher; he understood so much of the world in terms of heterogeneous interfaces.)
My comment about the analog as a twin or echo is helpful to undercut conventional descriptions of the digital and the analog as contraries or opposites. More metaphysical accounts will even privilege the analog as a kind of natural or “real” ground, apart from which the digital appears as pure formalization or abstraction (and hence the digital is somehow false or inauthentic). That’s an old story, and rather boring, if not also incorrect. I like to stress the analog as a twin or echo because it cuts against people’s intuition by inverting the priority, where the logos ratio (a/b) is now somehow “first,” while analogos doubles or re-articulates two logos ratios into an equivalence of likeness (a/b ≡ c/d). To push the point to the extreme: the qualitative analog “real” is merely the equivalence of two “false” digital signifiers; and if that doesn’t satisfy the Derrideans, I don’t know what will!
PL: Three of your books are organized around key figures of the digital that are related to systems of power based on algorithmic computation: protocol, exploit, and interface. These function both as technical terms for constructing the architectures of these systems and as conceptual tools affording critical access to them. I keep returning to the conclusion of your last book, Uncomputable, where you discuss the black box—another technology that is simultaneously a metaphor and an abstract machine specific to our post-cybernetic era. There you suggest that the black box might point towards an “antiweb” form of resistance (in a way that previous figures cannot). This seems connected to your longstanding engagement with the anarchist tradition, with asymmetry, blackness, and becoming-imperceptible. Could you elaborate on this? What is it about the black box as an idea and technology that distinguishes it from other figures of the digital, and how might it translate into forms of political organization or ethical practice?
AG: The black box is an enormously influential technology today. Simply put, a black box is an apparatus with a strict boundary between inside and outside, where the inside is illegible and unknown, while the outside is defined by an encoded, legible interface. And yet the inside and outside remain linked: manipulate the interface and a corresponding event will happen inside the kernel. In this way, the black box is a device for regulating flows of access and understanding.
All sorts of things follow the logic of the black box. So, phones and laptops are black boxes because their insides are inscrutable (barring deep forensic intervention), even as we peck and poke them. But a bicycle, say, isn’t a black box because the gears and brakes are readily available for inspection. Interestingly, the black box is a strictly formal technology. This means we can find black boxes that aren’t simply physical devices. The best example of this is the function. Functions are a mainstay of computer programming, where functions are “encapsulated” into an inside (the guts of the function, what it actually does) and an outside (the function signature or interface). In other words, even the mathematical shorthand for a function that we all learned in school, f(x), is a perfect illustration of the black box.
The argument at the end of Uncomputable has to do with historical change. In the nineteenth century, Marx described the commodity as having both a “rational kernel” and a “mystical shell.” But today the dynamic has inverted. Today the kernel is mystical, but the shell is rational, often quite literally. Laptops, for instance, are covered with alphanumeric symbols (keyboard), rigidly arranged pixels (screen), and explicitly defined ports of various kinds both physical and within the radio spectrum. So while I remain a Marxist, we need to recalibrate the concept of critique to accommodate the specific architecture of the black box. Today the problem is not so much “the rational kernel within the mystical shell,” as Marx put it in 1873, but somehow the inverse: a rational shell surrounding a mystical kernel.
PL: Do you have a proposition, or at least a sense of direction, as to how critique should be transformed—both as a theory and as a practice—in order to adapt to the ubiquitous black box and the inverted logic of commodity?
AG: Critique has historically been understood as an experience of revelation through unveiling or demystifying, a kind of anagnorisis, if you like, or what Paul de Man once characterized as “blindness and insight.” So what should critique look like today? One difficulty in answering the question comes from hermeneutics itself. A long and rich history of hermeneutic interpretation exists for scholastic and literary texts, but we don’t have very good examples yet for applying hermeneutics to data, source code, and symbolic interfaces (these being the primary “texts” in question when it comes to digital computers).
The black box has fully automated the logic of blindness and insight, at least in a superficial sense. For example, I personally don’t know how a GPU multiplies floating-point numbers. That knowledge is black-boxed for me. But I could gain this knowledge if I were to study electrical engineering and machine language. It’s not so much a question of misrecognition or fetishism (as critique was for Marx); the critical difference is strictly pragmatic and functional.
So if critique is to remain viable, and I hope it will, given how unappetizing the post-critical proposals of the 2000s were, then critique will need to deemphasize the logic of blindness and insight, in favor of something more like decoding or reprogramming. Twenty years ago I proposed the term “allegorithm” as a portmanteau of “allegory” and “algorithm.” It’s clunky, I know. But my ambition was to port existing forms of figuration (metaphor, metonymy, allegory, etc.) over to computation, monitoring which of these techniques still work and which don’t. Of course, this has been tried before in other contexts; recall how Lacan associated metaphor with symptom and metonymy with desire.
PL: Is the paradigm of network and protocol still driving contemporary power, or have synthetic media and generative agents added a new layer to the socio-technical stack that transforms its topology? If exploit—through hacking or computer viruses—once functioned as the paradigmatic form of protocological resistance, what might occupy that role today?
AG: I mentioned already that the status of protocols has diminished over the last twenty years, as open distributed protocols ceded territory to closed proprietary platforms. Agentic AI only accelerates this trend. Today we have “intelligence as a service,” with corporations jealously guarding their proprietary models and data. This is ironic, given that training data is often expropriated from the public commons (consider LLMs), in a process similar to what Marx once termed “primitive accumulation.”
The main counter-trend has been cryptocurrency and the blockchain; they remain largely protocological within a sea of proprietary tech, or perhaps they’re a hybrid of the two. To be clear, I reject cryptocurrency due to its dubious politics. On this point I agree completely with the late David Golumbia, who argued persuasively that cryptocurrencies are right-wing technologies. This is not to mention the blockchain’s profligate carbon footprint and relative lack of convincing use cases. But blockchain technologies do tend to resemble protocol in ways not evident in conventional proprietary tech.
PL: After decades of pushing critical philosophy to its limits, one cannot help but notice that antihumanism, at least its diluted version, has been appropriated by ideologues of contemporary capitalism and vectoralist power. This has prompted a rebound of humanism in philosophy, accompanied by various sorts of human-exceptionalist sentiments. How do you situate yourself within this shift? Do you assign valence to the concept of the human? Or should intellectual liberation continue to pursue a stripping away of the human from thought?
AG: I lament the antihumanism-to-computationalism pipeline. As you suggest, so much antihumanist discourse has fed directly into the kind of techno-centric philosophy that sustains the vectoralist class (to borrow a term from McKenzie Wark). Forty years ago it felt important to undermine the essence of the human, as Donna Haraway did in her famous essay on the cyborg. But today’s cyborg is an autonomous hunter-killer drone operating in the hinterlands of the American military command. Today’s cyborg is a trillionaire’s ex-lover crooning “I wanna be software” while trying to sound ironic. Honestly, I’d rather be a goddess than a cyborg.
This is, however, a live debate. Scholars like Melinda Cooper, for instance, insist that capitalism doesn’t degrade the human altogether, so much as subsidize specific kinds of persons at the expense of others. A related argument is evident in the work of Sylvia Wynter, where the category “human” is inherently anti-Black. Similar postures are evident in certain kinds of queer, trans, and feminist theory. In one sense, I suspect that antihumanism can no longer be categorized as an unalloyed good, as it was during certain phases in the history of critical theory (broadly conceived). And yet if antihumanism has no role as a widely scoped truth, it still might function well as a tactical intervention, useful in specific ways for specific reasons. Overall I don’t see humanism as the monster it once was construed to be. In 1946, Sartre insisted that “existentialism is a humanism,” albeit with a newly redefined human. And we know that Marxism is, and was, a humanism. Non-standard philosophy is a humanism (according to Laruelle). And if justice is still the key question of our age, from Palestine to climate catastrophe to class crimes and global wealth inequality, I expect humanism will rightly return to center stage and remain there for the foreseeable future.
PL: I would also like to ask about your position in the ongoing debate concerning computation, language, cognition, and reason. Some thinkers, for instance, Leif Weatherby, argue that LLMs simulate language without actually thinking, while others, including Katherine Hayles and Luciana Parisi, have suggested that even pre-transformer neural networks exhibited rudimentary, albeit nonhuman, forms of cognition. Where do you situate yourself between these two polar perspectives? Or does the debate itself rest on a false opposition that calls for a different conceptual framework altogether?
AG: Thinking has been better studied by others; it’s not something I’ve written on or considered deeply, except to recall that old line from Heidegger that we are still not thinking. If one defines cognition broadly enough, many natural phenomena could be categorized loosely under the heading of nonhuman cognition. At such a diffuse scale, however, the term begins to lose its own specificity. I think of expanded cognition as a desire more than a theory: the desire to confer cognition on all creatures great and small, like Epimetheus dispensing various powers and qualities to all the animals. But in the end, no, I’m not on the octopus-mushroom-slime mold side of the cognition debates. Old materialism is superior to new materialism.
As for computers, there’s no evidence that they are even remotely conscious, and indeed mountains of evidence point in the other direction. Alan Turing’s 1950 paper, “Computing Machinery and Intelligence,” bears a lot of the blame for this misconception. The Turing Test is one of the worst technologies in computing. It’s so fraudulent. For instance, it promises intelligence and “thinking,” and yet never defines thinking or positively confirms its existence. Meanwhile, the Test requires a human arbiter (the questioner), and thus ironically requires human intelligence in order to circumvent the question of intelligence. And it promises something like truth, but uses mimesis, which since Plato has been defined as an anti-truth technology. So the Turing Test has many problems. Yet, alas, the Test sits today as the very bedrock supporting so much of AI.
So we need a better account of computation, better than asking, “Is this instance of mimesis successful or unsuccessful?” Meanwhile, I agree with Yuk Hui that contemporary AI is, essentially, a David Hume Machine. And Hume is a dead end when it comes to explaining thinking. Skepticism, nominalism, empiricism, rejecting causality—Hume’s techniques are all ways to undermine and defeat consciousness, not to define it or characterize it in any positive sense. In other words, the Hume Machine (for AI) is an admission of AI’s failure, not its success. I’m rather conventional on this point: to understand thinking, we should no doubt rely on the tradition of rationalism (within which theology and metaphysics are often the most highly evolved forms), but even more so on the two most important sciences of the psyche, namely phenomenology and psychoanalysis. That’s where we will best understand thinking.