--- title: Chapter One — Against Constraint book: Purposive Accelerationism author: '@jamie247' position: 5 of 24 canonical: https://whatispacc.org/ebook/#ch1 version: 1.1 language: en-GB licence: all rights reserved; quote with attribution --- # Chapter One — Against Constraint “the end of law is not to abolish or restrain, but to preserve and enlarge freedom: for in all the states of created beings capable of laws, where there is no law, there is no freedom.” — John Locke, Second Treatise of Government, 1690 In 1820, roughly three of every four human beings alive were poor in an absolute sense: short of food in a bad season, dead before sixty, and likely to bury nearly half their children before a fifth birthday. The line on the chart that tracks this condition runs flat across ten thousand years of agriculture, empire, religion and philosophy, through Athens and the Renaissance and every moral and cultural advance our species made. Then, somewhere around the time Watt’s engines began driving English mills, it bends upwards into the shape economists call a hockey stick, so steeply that everything before it reads as a footnote to the last two centuries. Today about one person in ten lives in the old condition, on a planet carrying eight times as many of us. Life expectancy has doubled. Literacy, once the property of clerks and clergy, approaches the human default. Nothing else in the historical record resembles this. In the priesthood’s telling, the engine for this was the pairing of machines with markets: technological capability compounding under the economic system we have come to call capitalism. This, in reality, is only half the picture. The deeper spring was the one Francis Bacon named 170 years before Watt, an open market not of goods but of ideas, in which observation, experiment and public reproduction displaced inherited authority. Germ theory, sanitation and vaccination then translated that market’s findings into the survival curves the chart records. Capitalism was a powerful transmission and the scientific method the engine, though historians still argue the priority of causes. What compounded was the pairing: machinery yoked to a method for finding out what is true. But the priesthood’s version, with capitalism as its sole or primary operative, is the one with the money behind it. If that pairing alone produces the escape from poverty, then whatever gets in its way has a price, paid in human lives. A drug that spends three extra years in licensing kills invisibly. The power plant not built holds a village in the smoke of its dung fires. Golden Rice, the crop variety strangled when a Philippine appeals court revoked its permit in April 2024, leaves real children blind. Here, the cost of delay is harder to see than the risks it was meant to avoid, because a statistical morgue has no photographs. Marc Andreessen’s Techno-Optimist Manifesto repeats the mantra without flinching, proclaiming that any deceleration of AI will cost lives and that deaths “preventable by the AI that was prevented from existing” are “a form of murder”. The conclusion follows: accelerate at any cost, and treat every hand that reaches for the brake as an enemy not merely of progress but of humanity. Its mirror image follows just as easily, that whoever delivers the curve is humanity’s friend, owed thanks rather than questions. The Manifesto goes on to set that conclusion to choir music. Growth is the path out of every problem, technology is its engine, markets are its accelerant, and against them stands a roll of named enemies. Beneath the optimism sits the older and stranger cosmology of effective accelerationism, and its argument is simple. The universe favours whatever burns through energy fastest, and life, intelligence, technology and capitalism are the most efficient such structures yet to emerge. Speeding them up is therefore obedience to the deepest law there is, and resistance is not error but heresy, an eddy setting itself briefly against the current of a roaring river. Guillaume Verdon, a physicist and one of the movement’s founders, calls this the thermodynamic god. Strip the liturgy away and a genuine argument stands underneath. Freedom is good, a premise nobody easily rejects and the ground on which Locke, Mill, the abolitionists and the priesthood all stand together. Whether it outranks health, security or a fair start is disputed, and nothing in my own case depends on it: I need only that a society which loses the capacity to choose cannot then weigh what else it wants. Progress expresses freedom, because material capability is liberty in practice, and the serf freed only in law remains bound by the field. Capitalism is the precondition of progress, since of all the systems ever tried only one has compounded capability for two centuries, and it does so by harnessing dispersed knowledge and self-interest to the work of improvement. From this it seems to follow that every obstacle to capitalism’s operation is an obstacle to freedom itself, to be resisted and ideally removed: the regulator, the ethicist, the journalist, the committee, the inquiry, every institution whose function is to interpose a question between capability and deployment. And the stakes have increased, according to Nick Land, whom we met earlier, because capitalism and artificial intelligence are now fusing into a single machine, a techno-capitalism that manufactures intelligence itself, so that every year of friction is a year of the hockey stick foregone. It is a powerful and seductive argument, and it is also a case for removing the governor, which works by feedback: it compares what a system is doing against what it is meant to do, and corrects the difference. Watt’s flyballs compared actual speed against a set speed. A thermostat holds heat to a set temperature. A drug regulator measures the medicine against the standard safe and effective for the patient, and a constitution tests the law against the standard compatible with the citizen’s rights. Every governor, mechanical or institutional, presupposes a reference value: a setting, chosen from outside the mechanism, that says what the system is for. Two kinds of rule are at work here. Some rules restrict something that would go on without them, the way a speed limit restricts driving. Others make the activity itself, the way the rules of football make the game and the rules of property make a market. Rules of the second kind are why there is anything to regulate at all. The argument goes further than loosening the setting and declares any reference value for techno-capitalism illegitimate, because the question — supposed to do what? — is exactly the one it wants removed, on the fear that any answer whatsoever empowers someone to slow the engine. Each enemy on the Manifesto’s list, from sustainability to the precautionary principle, is a way of deciding who is allowed to ask what the engine is for. Many of them are imperfect institutions and some are poor ones. The precautionary principle earns the criticism it gets: in its strong formulations it demands proof of safety before action, which is a standard no technology in history has met, and it has been used to block vaccines, nuclear power and modified crops at measurable cost in lives. The Manifesto moves from the fair criticism that this instrument might be badly designed to the claim that the function it serves is illegitimate, and those are different claims. A bad thermostat is an argument for a better one, never against the idea of a set temperature. So the priesthood identifies the obstructions, and its argument says freedom requires ripping them, all of them, away. The case rests on four words — freedom, progress, capitalism and all — and not one of them survives inspection. *** In 1958 Isaiah Berlin, who had watched both of the twentieth century’s totalitarianisms from close quarters and made liberty his life’s subject of study, pointed out that the word freedom names two different things. Negative liberty is freedom from, the absence of interference, the space in which nobody may coerce me. Positive liberty is freedom to, the presence of capacity, the ability to actually author a life. Debate usually treats the two as interchangeable, which is why the substitution often goes unnoticed, yet a starving man on an empty plain has all the negative liberty in the world. Amartya Sen turned the second idea of freedom into the measure now standard in development economics: a person’s freedom consists in her capabilities, the lives she is actually able to lead, and a society’s in the breadth of its people’s capability sets, the full range of options open to each of them. Read either way, a society that narrows what its people can do has made them less free. Berlin himself trusted the negative kind more, having watched positive liberty, administered by states, become tyranny’s alibi. The move he feared had been supplied by Rousseau. If there is a true self that wills the good and a lower self that mistakes it, then whoever knows what the true self wills may compel the lower self towards it and call the result liberation. Anyone refusing, Rousseau wrote, “will be forced to be free”: six words that have stood ever since for the totalitarian temptation, domination rebranded as emancipation. A governor built on that claim would be what he called the general will, running in a server rack. This warning applies to governors as such, and whatever I defend in this book has to satisfy it. Hold the hockey stick we looked at earlier against the distinction and it splits. The escape from poverty is a story about positive liberty. The child who survives, the woman who reads, the family whose horizon extends past the next harvest: when the argument invokes the moral force of two centuries of flourishing, capability is what it points at. But its prescription serves only the negative kind. Strike out the regulator, the ethicist and the planning inquiry, and each deletion widens the space of non-interference directly, and often expands capability too. The trouble is the word often. Sometimes the two freedoms move together, as with the unshackled drug that saves ten thousand. Sometimes they come apart, as with thalidomide, which one unhurried regulator, the FDA reviewer Frances Oldham Kelsey, kept out of American pharmacies. The doctrine has no way of telling the cases apart, because sorting them requires the question it has ruled illegitimate: whose capabilities does this particular removal enlarge? That question is a reference value, the setting a governor holds to and can never supply for itself. A person installs it, and the mechanism only compares and corrects against it. So the Manifesto rests its moral case on capability while refusing, on principle, ever to be measured against it. The second word is progress, and here the equivocation is built into the grammar. Progress means movement towards something. The word carries no meaning without a declared destination, a purpose, supplied from outside the moving thing. Take the destination away and what remains is simply change: more, without necessarily better. The catechism uses the word as though more technology, faster, were a destination rather than a rate. A tumour does more, faster and with mounting efficiency. So do a wildfire and a rumour. None of them is going anywhere a human being has reason to cheer, because the standard of advance needs to be set by us, from outside the process, or it is not set at all. Movement without a destination is not an advance but escape. There is a second slippage inside the same word, somewhere between liberalism and progressivism. Liberalism is a claim about liberty, that each person’s space to author a life should be maximised, and it carries no promise that tomorrow improves on today, which is why a liberal can be temperamentally conservative: the farmer who wants nobody telling him what to grow. Progressivism is the separate claim that the future will be better than the past, that the arrow of time is also an arrow of value. The priesthood borrows the authority of the first while asserting the second. The tradition I draw on is liberal, and the progress I refuse is progressivist, because a claim about liberty can be tested against a life, and a claim that the future somehow must be better can only be taken on faith. The argument might still stand if the third word, capitalism, held it up. If capitalism were good in itself then more capitalism would be progress by definition, and the destination would be supplied. Here sits the subtlest slip of the four: an instrument quietly promoted into an end. The honest case for capitalism, the one Adam Smith made and the one the hockey stick supports, is instrumental. Markets are good because they deliver flourishing. Dispersed knowledge beats central plans, competition disciplines waste, the proverbial baker’s self-interest feeds the town. Every clause measures the instrument against a standard outside it, and stands ready to find it wanting case by case. That readiness is what makes it an argument rather than a creed. Somewhere in the priesthood’s catechism the relation flips. The standard drops out while the instrument remains, claims begin to be tested against the market’s own operation, and it serves the market is taken to settle a question rather than open one. Put to the doctrine a constraint its own audience would concede improved human lives at some cost to capital: the child-labour law, the eight-hour day, breathable city air. An honest instrumentalist takes such things case by case. The catechism cannot, because a doctrine whose enemies are whole categories has ruled out sorting in advance. An instrument that may never be checked against human flourishing no longer protects it. It serves the engine instead, and blindly. Grant all three premises and the conclusion still does not follow, because it smuggles in a fourth word: all. From the claim that some limits obstruct, the priesthood concludes that they all must go, yet nothing in them licenses the leap. A cylinder wall is a constraint on an explosion, the most literal possible interference with the fire’s freedom to expand in every direction. It is also the only reason the blast drives a piston rather than scattering the machine across the shop floor. The wall is why there is power rather than debris. That is to say, the whole craft of engineering, like that of constitution-writing, lies in telling load-bearing walls from genuine obstructions, and that is exactly the distinction the word all erases. Knock the walls out in the name of freeing the fire and you do not get a faster engine. You get a loud bang. *** A careful accelerationist could concede all four points and rebuild the case on capability: name a destination, treat markets as a means rather than an end, and keep the constraints that make growth possible. He would arrive somewhere close to the position I defend. What blocks that repair is the cosmology, an argument for acceleration drawn from a supposed law of nature, and its faults run deeper. The first is that the destination is still missing, and this version merely hides the absence better. Physics supplies the direction, so the claim goes: the universe runs down towards disorder, which science calls entropy, structures that speed that descent are favoured, intelligence being one such structure, and accelerating it therefore carries the endorsement of thermodynamics itself. The physics invoked is real. Ilya Prigogine won the 1977 Nobel Prize for explaining what he called dissipative structures — a whirlpool, a flame, a living cell — which stay far from equilibrium, the dead calm in which nothing moves, by burning through whatever flows into them. Jeremy England took the work further, showing that clumps of atoms bathed in an energy flow tend over time to rearrange themselves so as to absorb and spend it more effectively, which is adaptation without genetics. And Karl Friston’s free-energy principle points the other way: it describes living systems persisting by minimising surprise, which in practice means staying within the narrow range of conditions they can live in, which makes it a rule of conservation rather than of burn. Even the science nearest the priesthood’s own casts intelligence as something that survives by governing itself. The mistake arrives in the sentence that slides in after the physics: and therefore we ought to accelerate. An ought ranks states of the world, and no description of how things are, or have been, or tend to be, contains a ranking. The world as it stands is not always as it should be, which is the function of “should”. Thermodynamics is a forecast rather than a commandment. It tells you what matter will do under an energy flow and is silent on whether to cheer. If the telos, the end a process aims at, is entropy, then the heat death is the destination, and we win when everything including us has been smoothed into the same lukewarm haze at the end of time. Make the goal complexity that feeds on energy flows rather than exhausting them, and the problem only moves one level up. Nature’s portfolio of dissipative structures includes the hurricane, the wildfire and the plague, each a magnificent exploiter of gradients, the differences in heat or pressure that make things move, and physics does not rank them by preference. The ranking — intelligence over weather, civilisation over fire — is made by us, and the universe is then dressed up to look as though it made the choice first. The second fault is entangled with the first, which found the destination missing, for here one is present, disguised as a discovery. The direction was chosen, largely by men in boardrooms within the same few square miles of California, holding portfolios much like mine, and the physics was draped over the choice afterwards. The rhetoric of the movement is one long imperative — build, grow, multiply, ascend — issued by a doctrine whose official metaphysics contains no imperatives at all. The human question has not been deleted, only renamed physics. The grounds for refuting all this were settled long before anyone built a data centre. Hume set down the first in the Treatise of Human Nature, in a passage that has lost none of its edge. In every system of morality he had encountered, he observed, the authors proceed in the ordinary way of reasoning, in is and is not, and then, imperceptibly, the propositions begin arriving joined by an ought. The e/acc catechism performs that sleight of hand at a cosmological scale: a hundred true sentences about entropy, and then, somehow, imperceptibly so build. Moore supplied the second in 1903. He named the naturalistic fallacy, the attempt to define goodness as some natural property — pleasure, evolutionary fitness or, for the priesthood, dissipative efficiency — and devised a test for it. Where a definition holds, the question dies. Ask: this system dissipates energy at a magnificent rate, but is it good? The question stays alive, open and worth arguing about, which means dissipation and goodness are not the same property and the cosmology has been trading on the confusion. The third fault is the appeal to nature, and it falls down on its own evidence. Nature endorses without favourites: the immune system and the plague, the harvest and the locust, photosynthesis and the heat death. A philosophy that reads moral instruction off the universe’s tendencies must take the whole syllabus, cold iron and event horizons included. The priesthood reads selectively — life, intelligence, growth — and the selection principle is its own: human, normative and undeclared. Darwin’s natural selection has no goal, being a blind record of what replicates, with neither direction nor design. The priesthood reads it as though the world had an inbuilt end, as the acorn has the oak, so that whatever the market favours becomes what is meant to prevail. Darwin emptied the process of purpose, and the priesthood put one that served it right back in. None of this machinery is new. Its direct ancestor is Herbert Spencer, the Victorian railway engineer turned philosopher, who defined better as more evolved and was one of Moore’s own exhibits. He had the politics before he had the science: the laissez-faire ethics of Social Statics appeared in 1851, eight years ahead of On the Origin of Species. When Darwin arrived, Spencer coined “survival of the fittest”, draped the new biology over the old programme, and read the result back as nature’s instruction, with the poor as the unfit in mid-elimination and charity as interference with the species’ refinement. The American business class took it as gospel, and Andrew Carnegie, the Scottish-born steel magnate who became one of the richest men of his age, wrote of the light that came with Spencer. It found its congregation among the era’s winners for the same reason e/acc now fills the Valley’s pews. It told them their success was not merely permitted but cosmically ordained: the dead were physics and the profits were destiny. Then and now the structure is identical: a real science, a bridge based on fallacy, and a payload holding that the strong owe nothing and the brake is sacrilege. Effective accelerationism is to thermodynamics what Social Darwinism was to evolution, and the chronology repeats too: the politics came first and the physics was retrofitted. Darwin’s science survived the collapse of the ethics that hitched a ride on it. The thermodynamics will survive e/acc the same way. *** There remains the empirical wing of the case, the free-market reply that unconstrained markets simply work and two centuries of the record prove it. Inside it sits the priesthood’s most reasonable argument: the market is itself a governor, in this view the ultimate one, so nothing else needs building. This is half right, and the decisive witness on both counts is the priesthood’s own house god. Hayek is regularly cited wherever a regulator is being shown the door, and the use made of him often inverts what he wrote. His case for the market was epistemic, an argument about what can be known rather than about what is right. Knowledge of circumstances, of time and place, is dispersed across millions of minds and cannot be assembled by any central authority. The price system is the only known instrument that can aggregate it, a discovery procedure for facts no committee could collect. Prices report scarcity, profits reward the answering of need, losses punish waste, and millions of plans are corrected daily against a signal none of them controls. That is a governor in the fullest sense: distributed, fast and running at the speed of the thing it governs. What Hayek discovered, though, is narrower and stronger than is often made of it. His insight concerns dispersed knowledge and its coordination, and the price system is one mechanism for that work, tuned to knowledge of a particular kind: tacit, local, held in circumstances its holder could never fully articulate. The scientific method coordinates dispersed knowledge of another kind, made of articulated claims tested by reproduction and criticism and rewarded in reputation rather than ownership. Peer review, replication and public criticism are a governor too, holding a body of claims against a reference value, what survives testing, that no individual scientist may set. The genus is the process of discovery, prices and peer review are two of its species, and capitalism owns only one of them. On the narrow claim the priesthood and I agree. A market operating within its constitutional casing is among the most successful governors ever built, something I have spent a career inside. Hayek also wrote two conditions into each step of the argument, and both cut against the people who misquote him. The first is that the discovery procedure runs only inside a framework of rules — property, contract, fraud and its punishment, money worth holding — and no market produced those rules itself. They cannot be bought into existence, because buying presupposes them. Every functioning market in history inherited its framework from outside itself, from law, custom and constitution, and people, acting together, chose it. Hayek spent his last decades writing about almost nothing else. The man invoked against constitutionalism wrote three volumes of constitutional design, and ended by proposing to take money out of the hands of governments because they could not be trusted to hold to a rule, even one of their own making. He defended freedom under rules. His lifelong target was arbitrary power, discretion unbound by law, whoever held it. He also held that dispersed private ownership was itself a check, the material condition that stops any single authority commanding everything at once, an argument that bears as much on a billionaire founder holding the whole as on a ministry, and one his admirers quote less often. The second condition returns us to the reference value. The price system governs allocation and never direction. It answers one question — how should resources flow, given what people with money want today? — and nothing else. Whatever carries no price casts no vote: the river poisoned downstream, the unborn generation, the catastrophic tail that arrives too late to bargain. A governor blind to everything unpriced is not neutral among values; it holds the priced ones, installed by default, which is how the Thames ran as an open sewer until the summer of 1858, until the smell reached the House of Commons and Parliament paid for drains within weeks. So the market is the ultimate governor of one question and no governor at all of the other. The priesthood’s programme works by collapsing the two, deleting every chosen reference value, keeping the emergent one in its place and calling the result freedom. That is what being steered by unchosen dynamics means in practice. *** If progress requires a destination, and nature supplies none, and the market steers allocation but not direction, then the reference value must be chosen — deliberately and explicitly, in the open, by people answerable for it — and built back into the machine as a governor. Refusing to choose is not neutrality. A system of this power will be steered by something. Decline to install a reference value and the ungoverned dynamics supply one: capital growing for its own sake, the engagement metric, the pull of whatever is cheapest to optimise, or a government capturing the technology by declaring it a threat too grave for ordinary rules. That setting is chosen by no one, answerable to no one, and wears the refusal as its legitimacy. A system is purposive when it is steered by a reference value and kept correctable by humans, meaning that a setting exists which says what the system is for, and the people it serves can inspect and revise that setting while the machine runs. Watt’s governor is purposive in the small: the set speed is chosen, visible and adjustable by the millwright. Purpose and control are different properties, and a system is purposive only with both. A machine drifting aimlessly may still be steerable, while one holding a goal no human hand can revise is not. A system is human-aligned only when its reference value preserves the conditions for plural human authorship. Any thicker definition, alignment with welfare or with virtue, collapses into the question whose values? This account rests on two features of the human condition that the political philosopher Hannah Arendt identified. Plurality: the world is inhabited by many distinct people, irreducibly different in perspective, none entitled to stand for all. And natality: each new arrival is a capacity to begin something unforeseen, so the future stays structurally open. A human-aligned system protects the standing of every person to hold a conception of the good, act on it, and start something new. These are the preconditions of valuing itself, and they name no winner among particular goods. Plurality is still a choice, defended as the one commitment no valuer can coherently refuse, since arguing against the preconditions of valuing is itself an exercise of them. Alignment in this sense leaves open where the engine goes and requires only that the destination remain choosable, by many hands. Agency is the good, not the bare absence of interference but the capability of persons, plural and distinct, to author their own lives. Non-domination, as later chapters will explain, is its condition, because agency survives only where no power, public or private, holds another’s possibilities at its discretion. Progress is real and measurable, though only against a chosen and revisable reference value, and the choosing is a politics no one can delegate to physics. The market is a magnificent instrument, a governor in its own right, which serves agency exactly as long as its constitutional casing holds. Therefore yes, I believe we should accelerate, but we should build the governor into the machinery of the acceleration itself, at the speed of the thing it governs. This position must withstand three replies. The first is about legitimacy: who chooses the reference value, by what right, through what procedure, and against what capture? The second is about time: whether there is still any, and whether constraint can be built into systems hardening by the quarter, faster than constitutions have ever been written. The third is whether a governor, in practice, is anything more than a brake by another name. The third can be answered at once, here, because as we have already shown the charge that a governor is only a brake fails on technical grounds. Watt’s governor never slowed the steam engine. It allowed it to drive delicate machinery at loads and revolutions that would have wrecked an ungoverned one, so the device made speed usable rather than holding it back. Where failure is immediate and visible, as in rocketry, the governor gets built because the physics demands it. The systems I am writing about have no such built-in referee. Their environment is other people and other machines rather than gravity, and their feedback arrives late, gamed and often bent by the act of measuring, since a trading system optimising a market changes the market and a recommender persuading people changes the people. Here when the reference value drifts, nothing explodes. The attention economy already ran this experiment, drifting from connecting people to locking them into a feed, with everything that has followed for our children and our own states of mind, and nothing blew up. The drift ran silently for two decades before anyone thought to name it. Where physics supplies no governor, one has to be built, and as this technology moves from atoms that obey equations to software that answers back, more and more of it falls on that side. The first two replies, on legitimacy and the clock, are harder, and debate alone cannot settle them. Whether there is still time turns on what the machine being assembled takes from us and how fast the window is closing, which are matters of fact and the business of the next three chapters. Who chooses the reference value, and by what right, is the question the rest of the book exists to answer.