AI As A Time Machine

Great Thinkers, and the Past, Brought Back to Life

1

In many ways, AI can act as a time machine. We can take a snapshot of 1850, or 1789, by instructing an LLM to disregard everything from after that year. We can see something like what people really thought back then.

Maybe we can get concrete answers to concrete questions. Maybe we can travel back in time.

During the 1820s, what did white people really think about slavery? In the North? In the South? In the 1890s, what did people think would happen by, say, 1990? What did they foresee?

Or suppose that we want to know the original public meaning of the Fourteenth Amendment. Maybe we could use an LLM to find out.

A lot of research is now being undertaken in that vein. The basic goal is to give the past a voice by using data from the period that you want to understand. See

History LLMs: Giving the Past a Voice with Large Language Models

Here is something from the 1930s: https://www.nber.org/papers/w33373

There is much more. Can we see what Karl Marx would think of contemporary Russia? What Jane Austen would think about feminism? What Bentham would think about Peter Singer?

I am trying some of this. Here are some glimpses, from Claude.

2

Here is Hayek on Nudge. In my view, it’s both interesting and good.

A Review of Nudge: Improving Decisions About Health, Wealth, and Happiness

by Richard H. Thaler and Cass R. Sunstein

As it might have been written by F.A. Hayek

(An imagined review — Hayek died in 1992, sixteen years before this book appeared, so what follows is necessarily a work of pastiche, constructed from the positions he took across his actual writings.)

I must begin by conceding to the authors a point I have myself spent a career insisting upon, namely that the rational, calculating agent of the textbooks is a fiction, and a misleading one. I have never supposed that the individual mind possesses, ready at hand, a complete and consistent ordering of its own preferences, still less the computational power to apply that ordering correctly to every choice set before it. My own objection to the planned economy was never that the planner is less rational than the idealized consumer of the textbooks; it was that no one, planner or consumer, possesses the relevant knowledge in a form that permits its centralization. The authors’ catalogue of inertia, framing, and the misjudgment of small probabilities is therefore no embarrassment to me. It is, if anything, a vindication of the case I made long ago against what I called constructivist rationalism — the conceit that the deliberate, articulate reasoning of the mind is the sole or even the principal instrument by which men succeed in conducting their affairs. We function as well as we do not because we calculate well but because we are embedded in inherited rules, habits, and institutions which economize on the very calculation we are so poor at performing. On this diagnosis, the authors and I are in closer agreement than they perhaps realize.

It is at the point of remedy that I must part from them, and the parting is not a small one.

Having shown that the individual cannot be trusted to design his own choice environment unaided, the authors propose that this office be assumed by an architect — and the question I would put to them is the question I have put to every planner: by what means does this architect come to possess the knowledge that the individual lacks? The case for the market was never that the market participant is wise; it was that the price system aggregates, through the use of competition and trial and error, a quantity of dispersed and largely tacit knowledge — of particular circumstances of time and place, of personal value scales that the chooser himself may be unable to articulate — which no single mind, however well-intentioned or well-credentialed, can gather and hold. The authors’ choice architect faces exactly this difficulty in miniature. He must decide, for millions of savers of different ages, temperaments, family obligations, and private hopes, what the content of the helpful default shall be. He may do tolerably well on average. He cannot, by the nature of the problem, do well for the particular man whose circumstances diverge from the average, and it is precisely this man whom a rule imposed from above, however softly, will fail.

There is a second difficulty, which I would urge upon the authors with some insistence, regarding the distinction between the architecture of the corner shop and the architecture of the state. Where the designer of choice is one firm among many — a pension provider, a supermarket, a website — competing for custom against rival designers, his errors are subject to the correction that competition supplies: the man ill served by one default may take his business elsewhere, and the firms that nudge badly are, in time, disciplined by exit. This is choice architecture absorbed into the same spontaneous order that governs the market generally, and I find little in it to alarm me. But where the architect is the state, and the default is set once for an entire population by a single legislature or agency, the corrective of exit disappears. An error in design is no longer one competitor’s disadvantage; it is everyone’s burden, uniformly imposed and not easily escaped. The authors call their proposal libertarian because the formal liberty to opt out is everywhere preserved. I would ask them to consider that the absence of competitive alternatives to the designer himself is a deprivation of a different and, I think, more consequential kind — not a deprivation of the citizen’s liberty to choose, but of society’s liberty to discover, through the rivalry of many attempted solutions, which architecture actually serves men best.

I confess, further, to an old and not yet quieted anxiety about the rhetoric in which the proposal is dressed. I have spent a long portion of my life observing how the road to a comprehensively administered society is traveled not in great strides but in small, individually reasonable steps, each justified by the modesty and evident benevolence of its architects, none resembling tyranny, all together amounting to it. A measure that does not coerce, that wears the name “libertarian,” that asks only to arrange the furniture of choice rather than to forbid any door — such a measure is, for that very reason, less likely to summon the vigilance that a frank command would summon, and more likely to be extended, by degrees, from the pension form to provinces of life where the case for any uniform design at all is far weaker. I do not say the authors intend this extension. I say that the language of nudging supplies no principle by which its own boundaries are fixed, and a doctrine that relies on the continued good sense and restraint of those who hold the levers, rather than on a limit built into the structure of the levers themselves, is a doctrine I have learned, at some cost, to distrust.

I will close, then, where I began, in partial agreement. The authors have correctly diagnosed the poverty of the rational-actor model, and I credit them for it. Where the architecture they recommend remains the product of competing private designers, tested against one another and against the exit of the dissatisfied, I can accept it as a legitimate evolution within the market order, no more sinister than any other innovation in the arts of trade. Where it is proposed as an instrument of the state, set by a single hand for an entire population and insulated from the correction that rivalry alone supplies, I must withhold my approval, not because the architects are foolish or ill-intentioned — I have no doubt many are neither — but because no intention, however good, supplies the particular knowledge that only a free and competitive order can generate. The fatal conceit is not extinguished by calling it libertarian. It is only better mannered.

3

Here is John Stuart Mill on Nudge. In my view, it’s even better.

A Review of Nudge: Improving Decisions About Health, Wealth, and Happiness

by Richard H. Thaler and Cass R. Sunstein

As it might have been written by John Stuart Mill

(An imagined review — Mill died in 1873, well over a century before this book appeared, so what follows is necessarily a work of pastiche.)

The authors propose a phrase, “libertarian paternalism,” which I confess struck me on first encounter as very nearly a contradiction in terms, the sort of yoking of opposites that a clever rhetorician achieves by sleight of hand rather than argument. Having read the book through, I am prepared to withdraw the charge of contradiction, though not the charge of difficulty. The authors have identified something true and important; whether they have settled the question they raise, or only restated it more precisely, is the matter I wish to examine.

Their premise is this: that human beings choose badly, not from any want of freedom but from the structure of their own faculties — that the very organ by which we deliberate is subject to inertia, to the tyranny of the default, to the seduction of how a choice happens to be framed. And from this premise they draw a conclusion which I find, on reflection, difficult to resist: that since some arrangement of the choosing environment is unavoidable — a cafeteria must place the fruit somewhere, a pension scheme must enroll a worker by default into something — the only real question is whether that arrangement shall be left to accident, to the silent interest of whoever profits from our inertia, or shall be deliberately chosen with the chooser’s own welfare in view. Put thus, the authors are not proposing to circumscribe liberty at all; they are proposing to take responsibility for an architecture that already exists and is never neutral. This is a serious point, and I do not think my own principle, that the individual is sovereign over himself, supplies any ready answer to it, for the principle was framed against the open exercise of coercive power, and a default is precisely not that.

And yet I am not wholly easy, and the reasons for my unease bear directly on matters I addressed long ago.

I have argued that the value of liberty does not lie solely in arriving at the correct result, but in the exercise, by each person, of his own faculties of observation, reasoning, judgment, and even of feeling, in the work of forming that judgment. A man whose conclusions are right, but who has reached them by mere deference, has acted in a manner not greatly superior to that of an ape capable of imitation. My fear regarding the architecture the authors recommend is that it achieves its excellent results — and I do not doubt they are frequently excellent, measured in pensions saved and organs donated and debts unaccumulated — by going around the deliberative faculty rather than through it. The default is not an argument. It does not engage a man’s reason; it exploits his inertia. Where the authors are at pains to insist that the choice remains formally open, I would ask them to consider whether a liberty that is technically preserved but practically unexercised answers to the purpose for which I valued liberty at all. The capacity for choice, like a muscle, atrophies from disuse, and an architecture that quietly does our choosing for us, however benevolently, may produce a generation well provided for and poorly practiced in the office of self-government.

Against this I set their strongest reply, which I take seriously: that in the great majority of the cases they discuss, no deliberative faculty was in fact being exercised before the architects intervened. The worker who fails to enroll in the pension scheme is not engaged in a reflective rejection of saving; he has merely not gotten round to the form. Here I think the authors have the better of the argument, for my objection to paternalism was always an objection to the displacement of a man’s actual judgment by another’s, and where no judgment is being formed, there is none to displace. In matters of this kind — the dull, recurring, low-stakes arithmetic of daily provision, where most men never have and never will deliberate with care — I am content that the architecture be built well rather than badly, on the condition that it remain visible, contestable, and easy to refuse.

My remaining anxiety concerns not the cafeteria but the higher matters that the authors, to their credit, mostly leave alone, and that I would urge upon any government inclined to take their lessons further than the authors themselves intend. There is a difference between nudging a man toward the saving of his money and nudging him toward the saving of his soul, by which I mean those large questions of belief, vocation, and manner of life in which the very point of liberty is that each person work out the answer badly, if need be, by his own lights, rather than well by another’s. I would ask the architects of choice to remain forever modest about which domain they are in, to publish their defaults and their reasoning for them without concealment, for a benevolent design hidden from the designed-upon is a thing I cannot approve under any name, libertarian or otherwise, and to remember that the case for any nudge rests entirely on its surviving the light of day. On this last point I believe the authors and I are in complete agreement, and it is the condition under which I can recommend their book, with admiration for its cleverness and a standing reservation about how far its lessons should be allowed to travel.

— J.S.M.

4

And here is Mill on Hayek’s The Road to Serfdom:

A Review of Dr. F.A. von Hayek’s The Road to Serfdom

As it might have been written by John Stuart Mill

(An imagined review — Mill died in 1873, seventy years before Hayek’s book appeared, so this is necessarily a work of pastiche rather than a real or recovered document.)

---

There are books which instruct chiefly by the alarm they raise, and Dr. von Hayek’s is conspicuously of this kind. He writes as one who has watched, at close quarters, the degeneration of a civilized people into servility, and he writes with the urgency of a man determined that his adopted country shall not travel the same road by the same comfortable stages. For this urgency I have considerable sympathy. I have spent the greater part of my own intellectual life contending that the danger to liberty in a democratic age comes not chiefly from the tyrant but from the crowd — from custom, from the tyranny of prevailing opinion, from the gradual contraction of the space in which a man may think and live differently from his neighbors without penalty. Dr. von Hayek locates a kindred danger in the economic sphere, and shows how the machinery of central planning, however benevolently intended, tends to concentrate in the hands of the state a power over the minutiae of individual life that no Parliament should be trusted with and no citizen should be made to endure. On this point his warning deserves to be heard, and heard carefully, by precisely those reformers most inclined to dismiss it.

Yet I cannot accompany him the whole distance of his argument, and a reviewer owes his readers the grounds of his hesitation rather than a bare verdict.

My objection is to the rigidity of his road. Dr. von Hayek writes as though economic planning and personal freedom stood at opposite ends of a single track, such that a society which takes any step toward the former has committed itself, by a kind of mechanical necessity, to arrive eventually at the latter’s negation. I have myself argued, in my Principles, that the rules of distribution are a matter of social choice in a way that the laws of production are not — that a community may, after full and free deliberation, modify the institution of property, extend public provision, or experiment with cooperative enterprise, without thereby surrendering its soul to a bureaucracy. Whether such measures lead to servitude seems to me to depend entirely on whether the habits of free discussion, the independence of the press, and the individuality of citizens are maintained alongside them — not on the bare fact that the state has enlarged its functions. Dr. von Hayek’s argument too often proceeds as if the means determined the end irrespective of the temper of the people employing them, and this I think both an empirical overstatement and, in its way, a surrender of the very faith in human agency that animates the rest of his book.

There is, besides, a curious asymmetry in his treatment of coercion. He is alert, properly, to the coercion latent in a planning ministry; he is far less alert to the coercion latent in unregulated economic power, in the dependence of the laborer upon the employer, in the silent compulsion exercised by want upon a man who has no resource but to sell his labor on whatever terms are offered him. I have never believed that the removal of legal restraint is identical with the establishment of freedom, where one party to a bargain holds all the leverage and the other holds none. A doctrine of liberty that attends scrupulously to the despotism of the official and not at all to the despotism of circumstance is a doctrine only half built.

I would say, too, that Dr. von Hayek’s instincts are sounder than his architecture. His most persuasive pages are not those given over to prophecy — the inevitable descent from planning board to concentration camp — but those given over to a vindication of the individual conscience against the totalizing pretensions of any single design for society, however rational that design imagines itself. This is an argument I would have written myself, and have, in different language. Where he goes astray is in supposing that this argument belongs exclusively to the defenders of an unmodified market order, rather than to anyone — planner or liberal — who insists that the experiment of living remain open, that no class of expert be permitted to foreclose it on the public’s behalf, and that the test of any institution, public or private, be its tendency to enlarge or to diminish the free development of individuality.

In sum: a valuable corrective, oversold as a universal law. Dr. von Hayek has done a service in reminding an age enamored of administration that administration is not innocent of danger. He has done a disservice in suggesting that the danger admits of only one remedy, and that remedy a fixed and total abstention from collective contrivance. The friends of liberty would do better to take his warning and discard his determinism — to ask, of every proposed extension of public power, not “does this lead inexorably to serfdom” but “does this leave the individual freer, in the end, to think, to choose, and to err, than he was before.”

— J.S.M.

5

These are, of course, mere glimpses, a tip of an immensely large iceberg.