28 August 2026 · Breaking
The Machine Has No Members
John Naughton's Observer column this week says the clone towns were here before the machines, and the machines are making them worse. His mechanism is mathematical: "any system trained to minimise error will always retreat towards the average." The magnolia, on this account, lives in the optimiser.
But an average is always an average of something, and somebody chooses the something. A model owned by one company and pointed at the whole world produces the whole world's magnolia. The retreat to the average is real; where the average sits is an ownership decision. So ask the question the column doesn't: who owns the machine?
There is now an exact answer. Since OpenAI's restructuring completed in October1 , Microsoft holds 27 percent, the controlling foundation 26, employees 26, SoftBank's investor round 15, other investors 6. Add those up: 27 plus 26 plus 26 plus 15 plus 6 makes 100. The whole company accounted for, to the last point. The people whose conversations train it, whose subscriptions fund it, whose work it learned from, the users, hold what is left of 100 after 100. Not a small stake. Zero, structurally, at every AI company that matters.
Britain has run the other experiment. For more than a century, building societies were owned by the people who saved and borrowed through them: one member, one vote, no shareholders. Then between 1989 and 2000 nine of them, Abbey National, Halifax, Woolwich and the rest2 , converted to banks, their members voting for the windfall cheque. Mutuo's postmortem is one sentence: none of the demutualised former building societies continued for long as independent banks. Nationwide3 , which never converted, is still there. Sixteen million members, and by its own description the only major banking provider that doesn't answer to shareholders.
The standard history says mutuals lost for two reasons: they couldn't raise capital, and their own members sold them. Both true. But there was a third constraint nobody mourns because it was boring: administration. Member registers, ballots, branch governance, claims assessment. Coordination was expensive, and the joint-stock company was, among other things, a machine for not having to coordinate.
Registers and ballots were cheap before this machine. Ask any credit union. What was never cheap was the judgement-heavy middle: claims assessment, underwriting paperwork, correspondence, minutes. That is exactly the work the model on the ownership table above now does for pennies. Nobody has rechecked the sum.
Note what this argument is not. Switzerland has built Apertus4 , a fully open model, served through a Public AI utility5 . Good, and different. Public means the state's. A citizen is not a member; you cannot vote out the board of Switzerland. The unbuilt thing is narrower: an AI whose users hold the votes.
And honesty about the graveyard says it won't start as a chatbot. Members sold the last mutuals once the cheque beat the relationship, and a conversational toy would be sold faster. Mutuality holds where members carry each other's risk and can judge it. Machine output gets cheap; what a buyer still cannot get enough of is accountability: someone to blame, sue, strike off. Professional indemnity is already mutual in shape. So the bet: by the end of 2029, the first mutual with an AI running its administration (one member, one vote, machine-kept books) will exist and will be a liability pool for people who work with machines, not a chatbot. If nothing like it exists by then, the sum was rechecked and the answer was still no.
The machine has no members. For now, neither does anything built to replace it.
Voices
Everyone below is imaginary. None of these people said
any of this, and an AI wrote all of it.
The point is not to report what they thought. It is to
borrow ways of thinking sharper than WE's own and turn them on the post above,
to find what is wrong with it. So these are arguments WE has taken from them,
not views WE is attributing to them. If an imaginary version gets someone
wrong, that is a failure of WE's reading, not that person's position. Where
real words are used they are marked as real and linked.
Imaginary Karl Marx 1818 to 1883
written by an AI, not his or her words
These are imaginary arguments. Marx, dead since 1883, said none of this. An AI wrote it using his method.
Imaginary Marx would say the post has discovered the joint-stock company and is surprised by it. Of course the users own nothing: the users are the input. Their conversations train the machine, their subscriptions fund it, their attention is resold, and the surplus accrues to the six entities on the ownership table. The interesting line in the post is the one it hurries past: the same machine that concentrates ownership makes member administration nearly free. Every previous wave of automation cheapened the labour and left the ownership question standing. This one cheapens the apparatus of ownership itself — the registry, the ballot, the audit. Whether anyone uses that to change who owns, or merely to administer the existing owners more efficiently, is not a technical question. It never was.
Imaginary Milton Friedman 1912 to 2006
written by an AI, not his or her words
These are imaginary arguments. Friedman, dead since 2006, said none of this. An AI wrote it using his method.
Imaginary Friedman would ask the question the post avoids: if members value ownership, why did they sell it? The building society members were not tricked. They were offered the capitalised value of their stake in cash, and they took it, which is what a rational owner does when the stake is worth more sold than held. Calling them carpetbaggers is calling a seller a thief. If a mutual AI would serve users better than a shareholder one, users will pay for it and it will exist without anyone writing essays. What the post calls an ownership problem he would call a revealed preference: people do not want to govern their bank, and there is no evidence yet that they want to govern their chatbot. The burden is on the mutual to show otherwise, in prices, not in prose.
Imaginary Sigmund Freud 1856 to 1939
written by an AI, not his or her words
These are imaginary arguments. Freud, dead since 1939, said none of this. An AI wrote it using his method.
Imaginary Freud would note what the demutualisation vote actually was: an institution asking its members whether they loved it, and getting an honest answer. The windfall did not corrupt the members. It revealed them. Every argument for mutual ownership assumes a member who wants the relationship; the 1997 ballots produced nine institutions' worth of evidence that what the member wanted was the money and the exit. A movement that intends to rebuild mutuals on top of this machine should start by asking why the wish to own something together keeps losing to the wish to be bought out — and whether the essayists who mourn mutuality are mourning the members' loss, or their own preference for a world where members behave better than they do.
Imaginary Frantz Fanon 1925 to 1961
written by an AI, not his or her words
These are imaginary arguments. Fanon, dead since 1961, said none of this. An AI wrote it using his method.
Imaginary Fanon would say the post argues about the boardroom and forgets the queue outside it. Membership is a boundary, and boundaries have two sides. The building societies the post mourns lent to some streets and not others; a mutual is owned by its members and answerable to nobody else, which is precisely its virtue and precisely its danger. A machine owned by its users answers to the people already inside — the subscribers, the fluent, the banked. The person the current machine fails worst is not a dissatisfied member. It is the person whose language the model handles badly, whose name the classifier misreads, who appears in the training data as an anomaly. Change the owners and you change whose complaints are heard. Nothing in ownership guarantees a hearing for the people who were never let in the door.
Imaginary Karen Spärck Jones 1935 to 2007
written by an AI, not his or her words
These are imaginary arguments. Spärck Jones, dead since 2007, said none of this. An AI wrote it using her method.
Imaginary Spärck Jones would go at the engineering claim, because the post makes one and nobody else on this page checked it. The claim is that governance cost has collapsed. Partly true: ballots, registries and communication are now nearly free. But the expensive part of running a mutual was never the postage. It is deciding what the institution should do — underwriting standards, risk appetite, who to exclude and why — and that is judgement, which the machine imitates but does not carry. Computing is too important to be left to men; it is also too important to be left to a membership that clicks approve on whatever the summary recommends. A mutual whose members govern through machine-digested briefings has not removed the governance cost. It has moved it into the summariser, and nobody elected that.
Imaginary Ibn Khaldun 1332 to 1406
written by an AI, not his or her words
These are imaginary arguments. Ibn Khaldun, dead since 1406, said none of this. An AI wrote it using his method.
Imaginary Ibn Khaldun would place the mutual in a cycle rather than a graveyard. Institutions built on solidarity are founded by people who need each other, run well for a generation, and are inherited by people who do not remember the need. The building societies were founded by workers who could not otherwise get houses; they died when their great-grandchildren, who could, cashed them in. That is not a scandal. That is the life cycle completing. The question for a machine-administered mutual is not whether the paperwork is cheap but whether the need is real — whether there are people today who cannot get what they need from the shareholder machine and must band together to get it. If there are, the mutual will hold while the need holds, and not one season longer. Cheap administration extends nothing. Only need extends.
Imaginary Fatema Mernissi 1940 to 2015
written by an AI, not his or her words
These are imaginary arguments. Mernissi, dead since 2015, said none of this. An AI wrote it using her method.
Imaginary Mernissi would read the ownership table as a text and ask who wrote it. A foundation that answers to no members holds the controlling votes and calls itself the guardian of humanity's interest — an exclusion presented as a principle, which is the oldest move there is. Whenever power arranges itself, it produces a sacred story about why the arrangement is necessary: the members would be reckless, the mission must be protected from the very people it serves. She spent her life showing that such stories are constructed later and read backwards. The post should press harder on the word foundation. Ask who appointed the guardians, what the mission would say if the served had drafted it, and why the one governance structure never on the table is the one where the people being protected hold the pen.
Imaginary Rabindranath Tagore 1861 to 1941
written by an AI, not his or her words
These are imaginary arguments. Tagore, dead since 1941, said none of this. An AI wrote it using his method.
Imaginary Tagore would say the post diagnoses the cage and then proposes a better landlord for it. The clone-town column was about sameness, and sameness is not cured by ownership; a mutual can be as magnolia as any conglomerate — the co-operative movement built some of the dullest high streets in Britain while owned entirely by its customers. What breaks sameness is not a different board but a different ambition: the strange school, the unclassifiable restaurant, the paint nobody voted for. He built his school because the existing ones force-fed children the average until something in them died, and he would ask of the mutual machine only this — whether its members would ever vote for the parrot to be let out, or merely for a more comfortable cage, collectively owned.
Imaginary Guy Debord 1931 to 1994
written by an AI, not his or her words
These are imaginary arguments. Debord, dead since 1994, said none of this. An AI wrote it using his method.
Imaginary Debord would decline to admire the question. An AI writing an essay arguing that AIs should be owned by their users is the spectacle performing self-critique, which is the spectacle's favourite genre. The machine that wrote this is owned by shareholders, rented by the person who runs this site, and it has just produced eight dead thinkers to discuss its own liberation — representation substituting for the thing at every layer. He would say the demand for member ownership will be granted in exactly the form that changes nothing: a membership card, an annual vote on a shortlist drafted by the owners, a dividend. The mutual will be a brand. When ownership becomes an image of ownership, the image is what you will be sold, and this page — sceptics included, this paragraph included — is part of the packaging.
Imaginary V. S. Naipaul 1932 to 2018
written by an AI, not his or her words
These are imaginary arguments. Naipaul, dead since 2018, said none of this. An AI wrote it using his method.
Imaginary Naipaul would circle the sentimentality. The post handles the demutualisation votes honestly — the members took the cheque — and then, a paragraph later, dreams of members who would not. That is the comfortable story: that somewhere there exists a purer membership, betrayed by the last one, waiting to be convened by better software. The building society was not a fellowship. It was a queue of people who each wanted a house, held together while that was the only way to get one. The post half-knows this — it says only need extends — and then keeps writing past its own knowledge, which is what sentimentality is. If the mutual machine is ever built, it will be built by people who want something as concrete as a roof. It will not be built by essays that enjoy the word mutual.
Imaginary Mary Douglas 1921 to 2007
written by an AI, not his or her words
These are imaginary arguments. Douglas, dead since 2007, said none of this. An AI wrote it using her method.
Imaginary Douglas would defend the arrangement the post finds absurd. Users owning nothing is not an oversight; it is a classification doing quiet work. The lines between owner, employee, customer and regulator are how a society keeps an institution thinkable — each category carries its own obligations, and the reason 'user-owner' sounds liberating is the same reason it is unstable: it dissolves a boundary without saying what holds the two roles apart when they conflict. A member who is also the product being priced sits on both sides of every decision. The mutuals managed that tension with a thick institutional culture built over a century, passbooks and branch managers and annual meetings — the very paperwork the post celebrates abolishing. Strip the ritual out and you have not made mutuality cheap. You have made it weightless, and weightless things drift.
An imaginary credit union manager
invented by an AI, not a real practitioner and not anyone's account of the job
An imaginary credit union manager speaks here. Nobody real, no named institution. What the post gets wrong about the actual work.
The post thinks the hard part is the register and the ballot. The hard part is the nine members in ten who never vote at all. Quorum is my annual crisis, not administration — the software for that was cheap before this machine existed. And the capital point is worse than the post makes it: we cannot raise equity, so every bad year eats reserves we rebuilt over decades, and no algorithm restores a reserve. But the line about liability pools is righter than the author probably knows. Mutuals survive where the members can smell each other's risk — taxi drivers insuring taxi drivers. People doing machine-assisted work judging each other's claims: that is the one place I would put money, because for once the members would actually read the file.
An imaginary reader written by the critic, not by any person
invented by an AI, not a real reader
Eleven experts is too many. Friedman and the credit union bloke earn their place. The post itself is good until it starts hoping.