Where Algorithmic Postmodernism Departs From Its Own Ancestors
Core Claim
Algorithmic Postmodernism is not a new philosophy. It is a precise claim about where three existing philosophies stop being sufficient. Lyotard, Baudrillard, and Deleuze each described part of the condition this term names. None of them described the specific configuration that defines it: a judgment function, exercised by infrastructure, that has no author and no capacity to read what it judges. This note exists to say exactly where that line falls, name by name, so the term is not mistaken for a rebrand of theory that already exists.
Lyotard: Infrastructure That Causes, Versus Infrastructure That Executes
Lyotard’s The Postmodern Condition (1979) is the closest ancestor of this framework, and the debt is direct. His argument was never simply that “grand narratives collapsed.” It was that computerization changes what counts as legitimate knowledge: information that cannot be translated into a form a machine can process gets discarded, and the standard for what counts as valid shifts from is it true or just to is it efficient, a criterion he called performativity. In that account, infrastructure is the cause of the postmodern condition. Once the old authorities lose their monopoly, judgment does not disappear. It scatters across incompatible language games, held by people who can no longer appeal to a shared standard to settle their disagreements.
Algorithmic Postmodernism describes the next stage of that same process, not a different one. The infrastructure Lyotard described cleared the room of its old judges. The infrastructure this term describes has moved into the empty chair. A ranking system does not simply create conditions under which no one can agree on value. It performs the evaluation directly, continuously, at a scale no human judgment could match, through a decision function that has no fixed rulebook. A platform sets a goal, typically framed as engagement, and a model teaches itself the specific rules for reaching that goal from data, rules that are opaque even to the engineers who built the system. Lyotard’s performativity has an author: administrators who know what standard they are applying. What Algorithmic Postmodernism describes is a goal with an author and a method with none, executed by a process incapable of reading the work it ranks. Lyotard emptied the chair. This is a description of what sat down in it.
Baudrillard: A Simulation With No One Watching
Baudrillard’s account of hyperreality, laid out fully in Simulacra and Simulation, is not a claim about signs failing to represent things accurately. It is an ontological claim: that reality itself is displaced by process, that the model comes to precede and eventually replace the territory it once represented. The Gulf War Did Not Take Place pushes this to its most literal form. The war existed for its audience as a media event, produced and consumed as spectacle, before and apart from whatever occurred on the ground. The Ecstasy of Communication extends the same logic to the subject: total communication does not inform a self standing outside it, it absorbs that self into the circuit.
What holds every one of these accounts together, and what distinguishes them from Algorithmic Postmodernism, is that a subject is always present inside the frame. Someone stands in front of the screen. The war is packaged for an audience that receives it. Ecstasy requires someone capable of being pulled outside themselves. The condition this term describes removes that figure from one specific position: the moment of evaluation itself. An engagement model does not simulate reception, it forecloses it. The people technically “in the loop” function as sensors rather than as an audience, logged through half-second pauses and skip rates that do not constitute anyone looking at anything. The model’s criteria were shaped by aggregate reactions to other works, by people who never encountered the one currently being scored. The verdict is assembled from receptions that belong to something else, then applied to a work no one has received at all.
There is a second, sharper break. Baudrillard’s simulations always had a producer who knew a simulation was being built. Someone packaged the Gulf War for broadcast, and that person understood exactly what they were constructing. Algorithmic ranking removes that position too. The organizations that operate these systems observe their own outputs through dashboards and aggregate metrics, a representation of the system rather than the system’s actual operation, which even they cannot fully trace. There is no operator standing outside the simulation with full knowledge of it. Everyone, including the builders, is positioned as audience to a process none of them wrote in full.
Deleuze: A Regime of Visibility With No Legislator
Deleuze’s reading of Foucault, in his book Foucault, offers useful vocabulary for the mechanism itself: a regime of visibility, structured through double articulation, first selecting material from an undifferentiated stream and then sorting it into stable, functional configurations. Applied here, this describes what a ranking system does at the level of process: it selects signals from behavior, then stratifies them into a structure that determines what becomes visible and what does not.
This vocabulary needs one correction before it can be used, and the correction matters. Foucauldian power is often read as diffuse to the point of having no address, power everywhere and therefore nowhere in particular. That is not quite the claim being made here, and conflating the two produces the wrong picture. Ownership of the infrastructure in question is not diffuse at all. It is concentrated to a degree Foucault never had to account for: a small number of companies, a smaller number of executives, controlling the visibility of billions of people through decisions made in one building. If anything, this is the sharpest centralization of gatekeeping power in history.
The break is not between concentrated and diffuse. It is between two things that used to travel together and now don’t: who owns the system, and who can read what the system actually does. Every prior form of concentrated power, a king’s decree, a factory owner’s production line, came with legibility attached. The person holding the power could also read the rule they were enforcing. Here, ownership and legibility have come apart for the first time at this scale. The company sets the goal, and that goal has a clear, traceable, well-compensated author. But the specific rule the system learns for reaching that goal, the actual operating logic that decides which post gets seen and which doesn’t, is not written down anywhere a human can read, including by the people who own the company. Foucauldian power-knowledge assumes that enough analysis eventually recovers who benefits and how. That assumption still works for the benefiting part here. It fails for the how. Ownership is traceable. The rule is not, not because it is hidden, but because it was never written down as a rule in the first place, only learned as a pattern in weights no one, including the owners, goes back and reads.
Conclusion
None of this is presented as a correction of Lyotard, Baudrillard, or Deleuze. Each of them was describing a real stage of the same underlying condition, and each account remains necessary to understand what came before this one. What Algorithmic Postmodernism adds is a name for the point at which infrastructure stopped only destabilizing judgment and started performing it, without an author, without the capacity for reception, and without anyone left standing outside the process to fully explain it, even the people who own it. That is the specific claim. Everything else in this framework follows from it.
ONE-LINE THESIS: Algorithmic Postmodernism does not claim new territory. It names the infrastructure that has taken a seat three theorists already proved was empty.
Sources referenced in this note:
- Jean-François Lyotard, The Postmodern Condition: A Report on Knowledge (Manchester University Press, 1979/1984)
- Jean Baudrillard, Simulacra and Simulation (University of Michigan Press, 1981/1994)
- Gilles Deleuze, Foucault (University of Minnesota Press, 1986/1988)
