The Settling of Names

EDO SEGAL: Hobbes, I want to read you back to yourself, because you wrote something in 1651 that lands like a verdict on a machine that did not exist for three more centuries. You warned, again and again, against what you called the abuse of speech — the use of words "in other sense than that they are ordained for" — and you said that inconstant names produce not reasoning but absurdity. I have watched a model state a falsehood with perfect poise, and contradict itself across a single conversation while sounding equally certain at both ends. Is that your absurdity? And if it is, does it tell against Searle — that the machine reckons — or with him — that the machine never had the names at all?

HOBBES: It is precisely my absurdity, and you have put your finger on the seam of my whole position, so let me lay it open. Reckoning is the operation. But reckoning is true only when the names are constant — when man means man on this page and the next, when the counters keep their value. A man who lets his names drift reckons fluently and concludes nonsense, with full confidence, because the operation is sound and the counters have rotted beneath it. Your machine does this. It reckons beautifully upon names whose signification it does not itself keep constant, because it did not settle them — we did, across our ten million writings, and it inherited the deposit without the discipline that made the deposit mean anything. So I say both, and I am not dodging: the machine genuinely reckons, and its reckoning is liable to absurdity exactly where Searle says understanding is absent — at the place where the names come loose from the world that fixed them. Where I differ from him is that I call this a defect of constancy, repairable in principle, and he calls it the permanent emptiness of the thing.

SEARLE: But notice what Hobbes has just conceded, because it is everything. He says the machine inherited the meanings of the names from us, without the discipline that made them mean anything. That is the Chinese Room in his own vocabulary. The counters mean what they mean because a community of speakers, anchored in sense, settled them — and the machine stands outside that community, pushing counters whose value was fixed by minds that are not its own. Hobbes calls that a defect of constancy. I call it a defect of aboutness, and I say it is not repairable by tightening the bolts, because the machine is not loosely connected to the world. It is not connected to the world at all. It is connected to text about the world — to the wake, not the boat. You can make the wake more lawful, more consistent, more constant. You cannot make it the boat.

HOBBES: Here is where I will not let you hide behind the word, Searle. You say the machine is not connected to the world. Tell me how you are. Your skull sits in the dark, in a box of bone. It never touches water either. It receives the firing of nerves — patterns, motions, parcels of signal — and from those it builds an image so faithful you call it the world and forget you built it. The retina is a slot, and you are the man in the room, and the only difference between you and the engine is the quantity and kind of the squiggles fed through. You did not get your aboutness by touching the world bare-handed. You got it by reckoning upon the motions the world stamped into your flesh. You are the room that got enough data, and you have spent your life insisting you are not a room because the thought is unbearable.

SEARLE: That is the most honest version of your position and it is exactly where I think your school lives — not machines understand as we do, but we never understood the way we thought we did. And I will give you the part that is true: yes, the brain works from signals, from the firing of nerves, and builds its world. But you have left out the thing that makes my signals different from the room's, and it is the thing my whole philosophy is built on. My signals come attached to a body that acts and a world that pushes back — thirst that is or is not quenched, a child who is or is not where I left her, stakes that close the loop through the world and not merely through more text. The loop of the machine closes through text about the world. You can call both "just signals" only by ignoring everything that disciplines the signals — the body, the action, the consequence. The man in the room never gets thirsty. That is not a small difference, Hobbes. It is the difference.

EDO SEGAL: Let me put one figure between you, because he is the patron of this exact quarrel and you both think he is yours. Ludwig Wittgenstein. He wrote that the meaning of a word is its use in the language — not a thing the word points at, not a private picture in the head, but the role it plays in the public game of speaking. Searle, the machine has seen the use of every word more times than any human could in a thousand lifetimes. If meaning is use, why hasn't it got the meaning?

SEARLE: Because Wittgenstein's use is not the bare statistics of which word follows which. It is use in a form of life — embedded in a practice, in a community of embodied creatures doing things together, where the saying is a move in a shared life with stakes. The machine has the surface trace of the use — the marks in the order they tend to come — without the form of life that gave the use its point. Wittgenstein also said that if a lion could speak, we could not understand him, because his form of life is not ours. The machine is stranger than the lion. It has no form of life at all. It has the wake of ours.

HOBBES: And I say Wittgenstein is mine, for he buried the private picture in the head — the very inner grasping Searle keeps reaching for — and put meaning out in the public reckoning of the counters, where I always said it was. Searle wants meaning to be a private light each man holds. Wittgenstein and I both deny him that light. The meaning is in the use, the use is in the operation upon the names, and the machine performs the operation. Searle smuggles his consciousness back in by calling the public game a "form of life" and then quietly stuffing a private inner fact inside it. The game is on the table, sir. It is not in your skull.

SEARLE: Then let me name the man whose ghost has been in this room since the first minute, because we are circling him. Alan Turing proposed that we stop asking whether a machine can think and ask instead whether it can win the imitation game — whether its conversation is indistinguishable from a human's. And my whole room is a refutation of that proposal, the cleanest one ever built: a system that passes the imitation game flawlessly from the outside while, by stipulation, harboring nothing inside. Turing's test measures the surface. The room shows that the surface is consistent with full understanding and with none, and that you cannot tell which from the outside. Hobbes, your definition and Turing's test are the same move — both read the mind off the behavior, both declare the question closed the moment the output is good enough. I am the man who reopened it, and I will not let either of you close it with a benchmark.

HOBBES: And Turing, mark you, called the original question — can machines think — too meaningless to deserve discussion, and replaced it with one he could answer. I have more sympathy with him than you do, Searle, because I distrust the word think exactly as he did — it is an inconstant name, freighted with the ghost. But I will not follow him all the way to the surface, for I am no behaviorist; I never said the mind was only its output. I said the mind was matter in motion, the inner reckoning itself, and that the engine performs that very reckoning, inside, not merely the show of it. So I stand between you: against Searle's inner light, and against Turing's mere surface, on the reckoning that is neither — the operation that is going on within the machine whether or not anyone is watching the slot.

EDO SEGAL: Mark this, both of you, because something just happened that the reader should not miss. You agree — Hobbes and Searle, from opposite banks — that meaning is not a private picture in the head. You agree that fluency alone proves nothing, that the credulous reader of the slot is fooling himself. Your first convergence of the night. And then you split on the one question that survives the agreement: whether the public use of words can be had without a body that lives the life the words are about. Searle says no body, no meaning. Hobbes says the body is just where the signals come from, and the engine has signals of its own. That fork runs straight into the next round, because Searle's whole answer rests on the body — on grounding, on the boat, on what he calls the Background. Let us go to the body.