Opening Positions

MARX: Thank you. I want to begin not with the machine but with what the machine is made of, because everything follows from that and almost no one will say it. You are told that a large model is a marvel of engineering — architecture, parameters, compute. This is the camera obscura, the image turned upside down so that the inversion feels like nature. Turn it right side up. What is actually inside the machine? The labor of millions of human beings. Every book, every article, every line of code, every photograph, every argument anyone ever had on the internet — the congealed, accumulated work of the species, frozen into a system of weights. The architecture is the vessel. The labor is the content. A model trained on no human work, however elegant, would know nothing and produce nothing. The value comes from the corpus, and the corpus is human labor, and the human labor was very largely not paid.

So I give you the machine in its true description, and it is a description I wrote a century and a half before it existed. It is dead labor. The accumulated past work of the human race, congealed into a thing, owned by the few, and set to work against the living — competing with us, disciplining us, replacing us where it can. I once wrote that capital is dead labor that, vampire-like, lives only by sucking living labor. I meant it as analysis, not as poetry, and the model proves the analysis. Its fluency is not a mind. It is borrowed animation — the reanimated labor of every writer it consumed, moving. The writer now faces a machine built from all the writers who came before her, owned by someone else, and used to underprice her. That is not a metaphor. That is the mechanism.

But here is the turn, and it is the reason I came, because I am not the prophet of doom my friend across the table takes me for. The same machine that, under this ownership, is a vampire — under different ownership is the most extraordinary instrument of liberation ever forged. In a notebook I never published, I imagined a stage of production where the machine would do nearly everything and the worker would step to the side of the process, and where wealth would depend not on anyone's labor time but on the general intellect — the accumulated knowledge of the species, made directly into a force of production. I imagined it as a dream. You have built it. The general intellect is no longer a phrase in a German notebook. It is running in a data center. And the only question — the entire question — is who owns it and to what end it is set working. The popular fear is that the machine will wake up and turn against us. The real danger is older and duller and already here: that dead labor, owned by the few, will go on ruling the living, only now it can plan. The answer is not to fear the plan. The answer is to take the machine away from the men who own it and plan, for the first time in history, for everyone. That is my opening.

EDO SEGAL: Friedrich.

HAYEK: That was a magnificent statement, and I agree with one sentence of it completely — the general intellect is running in a data center — and I reject the conclusion he draws from it more completely than I have ever rejected anything. Let me start where he started, with what the machine is, and then I will tell you why "take it and plan with it" is the most dangerous sentence spoken in this room tonight.

Karl is right that the model is a distillation of the accumulated knowledge of civilization. He is right that no engineer put that knowledge there; it was grown, not designed, the way a market is grown and a language is grown. But notice what he has quietly assumed. He has assumed that because the knowledge of society can be distilled into a machine, the coordination of society can be performed by that machine. These are not the same claim. They are not even close. And the gap between them is the work of my entire life.

Here is the knowledge problem, and it is not a problem of computation. Imagine, as I once wrote, that tin becomes scarce. It does not matter why; perhaps three people in the world know the cause. What matters is that everyone who uses tin must now use less of it, and everyone who can supply it should supply more. The price of tin rises, and without anyone understanding the cause, the whole vast system reorganizes itself around a scarcity almost no one comprehends. That price is not a fact sitting in a database waiting to be scraped. It is a signal, generated in the act of millions of people deciding what to do given what only they can see — the shop owner who knows he can switch suppliers, the manufacturer who knows her customer will wait. This knowledge is dispersed, tacit, perishable, and most of it does not exist until the moment it is needed and the signal calls it forth. You cannot upload what was never written down. You cannot centralize what only exists in the act of being used.

So when Karl says "take the machine and plan with it," I hear: take the machine, abolish the price system that was generating the knowledge, and hand the silence to a planner. Because that is the catch he steps past. When you replace the market with the model, you do not give the model the market's information. You destroy the process that was generating that information in the first place. The planner does not inherit a hard sum. He inherits a void. The model has read everything we wrote. But the decisive knowledge of an economy was never written — it lives in the friction between a particular person and a particular situation, and it dies the instant you try to extract it for central storage. A faster computer applied to missing data is still applied to missing data. My opening is this: the machine is the most powerful instrument the fatal conceit has ever been handed, and the fatal conceit is the belief that because we can describe an order, we can therefore direct it. We cannot. And the better the machine gets at describing, the more certain we will become that we can — which is exactly why this is the most dangerous moment my argument has ever faced, and the most necessary.

EDO SEGAL: Before I frame the rounds, one discipline I impose on every long conversation, because it pays off three hours later. Each of you, in a few sentences — what do you envy in the other man's position? Not respect. Envy. The thing his side gets to have that yours doesn't. Karl first.

MARX: I envy his humility, and I mean that without irony. Friedrich gets to say "we cannot know, we cannot direct, we must not presume," and there is a kind of peace in it, a permission to stop straining against a world too large to hold. My position has no such rest. I am committed to the claim that we can and must take conscious control of our collective conditions — that to leave the order unplanned is to leave it to domination wearing the mask of nature. That is exhausting. It means every injustice is a thing someone chose and could have chosen otherwise, and there is no comfort in "no one is to blame, it is simply the order." He gets the comfort of the spontaneous. I am denied it, because I have seen who the spontaneous order is comfortable for.

HAYEK: And I envy the hope. Karl's position contains a promise — that the suffering is not necessary, that it can be ended, that a better world is not only imaginable but buildable, and that the machine might be the means. My position offers no such promise. I am the man who must stand up and say: the thing you most want to fix, you must not try to fix directly, because the attempt will make it worse, and the most you can do is tend the general rules and accept that much misery will persist that you are forbidden to plan away. That is a grim thing to ask of people. Karl gets to point at a wound and say "we can heal this." I have to point at the same wound and say "your cure is the disease." I have spent my life being the man who says no to the most beautiful sentence in politics, which is "let us simply fix it." I do not enjoy it. I believe it. But I envy the man who gets to say yes.

MARX: That may be the most honest thing either of us says tonight.

EDO SEGAL: Two openings and two envies, and you can already see the shape of the evening. It is not that one of them loves the machine and one fears it — they both fear it, and they both see its power. It is that they locate the danger in opposite places. Karl says the danger is that the machine is owned, and unowned it could free us. Friedrich says the danger is that the machine will be aimed, and that the aiming is the catastrophe no matter who holds the gun. Hold both. We begin the rounds at the exact seam — with the knowledge problem, the wall Friedrich just built across the room, and whether the machine that read everything has finally found a way through it.