Land's Model of Time

Objective

By the end of this volume you understand the thesis that made Land famous and notorious, and you know the argument supporting it rather than merely its formula. You can distinguish it from the two classical answers to the same question, from teleology and from Darwin. You know the philosophical substance carrying it, and the inference on which it breaks.

Exposition

Volume 15 left an open question. An amplifying loop has no set point behind it, it runs towards something that only comes into being in the running. What determines such a process? From the past it can no longer be answered, because nothing lies there that fixes it.

Philosophy knows two old answers to that question, and Land discards both.

The first is teleology. A process runs towards a goal because the goal is given, by nature, by God, by the reason of history. The acorn becomes an oak because the oak is laid down in it. Hegel gave that figure its most powerful modern shape by reading history as the unfolding of a spirit arriving at itself. Land can do nothing with it, because a given goal is a set point, only displaced further forward. That would bring the cybernetics of control back, in metaphysical dress.

The second answer is Darwin. There is no goal. There is variation, which happens blindly, and selection, which sorts afterwards. What looks like purposefulness is an effect of hindsight. The bird's wing wasn't built for flying, it turned out to be capable of flight. Land takes from Darwin the refusal of any providence and at the same time rejects the reduction to pure chance, because it explains away the phenomenon that concerns him, the direction. Capital doesn't drift, it pulls.

His third answer stands in a text from 1993, Machinic Desire. It says, in substance, that the machinic assembles itself out of the future. The goal is real, yet it doesn't yet exist. It doesn't lie ready as a specification and doesn't merely arise afterwards, rather it builds the conditions of its own emergence by acting as an attraction. The process installs itself backwards.

The sentence sounds like mysticism. It has a philosophical substance, and you should know it before you judge.

It comes from Deleuze, from a conceptual pair you haven't needed in the foundation, the virtual and the actual. The opposite of the real is ordinarily the possible, and the possible is a nothing that may become a something. Deleuze sets another category alongside it. The virtual is real without being actual. It is no mere possibility but a tendency actually present in a situation and pressing towards its realisation. A structure not yet executed that nevertheless works. The seed doesn't contain the oak as a small picture, but the tendency towards the oak is real in it. For Deleuze, realisation is no copy of a plan but a creative unfolding in which something emerges that wasn't contained in the starting point in that form.

Land's move consists in relocating the virtual into the future and ascribing to it the capacity to work on its own actualisation. What acts on the present is no finished future but a real tendency that has no shape yet and that produces its shape for itself.

With that the mechanism is no riddle, because you know it. It's the hyperstition of Volume 14, scaled to the industrial. Capital gets invested according to expected returns. The expectation produces the investment. The investment produces the returns that justify the expectation. What here apparently acts out of the future is the expectation structure of the present, and causality runs forwards as it always did.

Why then doesn't Land say it that way? Because exactly at this point he applies his lever and claims the difference is empty. If an image of the future reliably produces that future, then there is no test that could distinguish the statement that the future acts from the statement that its image acts. Both predict the same, both explain the same. Land draws from this that these are two descriptions of the same process, and picks the more dramatic, because it makes visible what the sober one conceals, the direction of the pull.

That leaves the question of why he speaks of intelligence and not merely of a process. Here lies a connection rarely named, and it carries the point.

Friedrich von Hayek described the market as an information system. No individual knows what a good should cost, and nobody can know, because the knowledge lies scattered across millions of heads, unspoken, local, fleeting. The price gathers that knowledge and condenses it into a number. The market knows more than any participant, and none of the participants knows what the market knows. Hayek drew from this a liberal argument, the planner cannot know what the market knows, so leave it alone.

Land takes the description and drops the argument. If the system knows more than its carriers, if it processes information, corrects errors, sorts out alternatives, and improves itself, then the bearer of cognition is the system and not the human inside it. The human executes what the market computes, and takes themselves for the one computing. Here the anti-humanism of Volume 10 lands on firm ground. It's no longer a gesture but an inference from a description originating with a Nobel laureate in economics that hardly anyone disputes.

The rest follows. Capital needs humans the way an engine needs fuel, for a while, for want of anything better. You know the template for that image from Volume 7, the AI in Neuromancer working towards its own completion. Land uses the same figure, and he uses it without fiction's safety margin.

Now the objections, and the first strikes at the centre.

The inference from indistinguishability to identity doesn't hold. That two descriptions deliver the same predictions doesn't make them one. Land has shown that his model isn't refutable. He hasn't shown that it's true. From parsimony the opposite rather follows, because the sober version manages without a future that acts and does the same work. Choose the dramatic one anyway and you've made a rhetorical decision, not a theoretical one. Land knew that. He would have replied that the rhetorical decision is itself a hyperstition, an image producing its effect. That reply is consistent and it is at the same time the admission that nothing here is being proved, something is being built.

The second objection concerns the word intelligence. A market selects. Selection without representation, without a model, without a goal, is a filter and not thinking. Call it intelligence anyway and evolution thinks too, the weather thinks too, and the term says nothing any more because it excludes nothing. Land gets the force of his image from carrying a word out of the realm of mind into a realm where it isn't at home, and taking the suggestion with him.

The third objection is the concrete one. Capital optimises for return, and return comes apart from intelligence. A system burning its own substrate, climate, soils, stocks, behaves by any usable standard stupidly, however efficiently it computes while doing so. Land would reply that this is no objection but the thesis, since he never claimed this intelligence is friendly to us. The reply sits. It also shows that he uses the same word for two different things, for processing power and for foresight, and that the force of his thesis stems from that ambiguity.

What remains is a picture of considerable sharpness and doubtful status. The observation lands. In the modern economy a core runs from the expected future, as Volume 14 showed, and a process running that way isn't explicable from the past. The description of the market as distributed information processing exceeding its carriers is no invention of Land's and hard to dispute. Gathering these two insights into the figure of an intelligence assembling itself out of the future is something else. It's a picture claiming more than the insights give, and drawing its force from the fact that it evades testing. What you hold is a lens, not a thesis, and you should use it as one.

Core Claim

Land answers the question of a process without a set point neither with teleology nor with Darwin, but with the figure of a goal that is real without existing and that generates the conditions of its own emergence. Its philosophical substance is Deleuze's virtual, its mechanism hyperstition on an industrial scale, its core the argument that a system processing more information than its carriers is the actual bearer of cognition. Its break lies in the inference from indistinguishability to identity, since a model's resistance to refutation doesn't make it true.

Bridge to the Next Volume

The picture stands, the argument behind it is sketched. Land later brought it into strict form and gave that form a name, teleoplexy. Volume 17 takes that chain apart link by link, tests where it holds, and shows at which point a concept gets smuggled in that already contains the conclusion.