The argument for broad ownership is usually filed on the left, among the languages of redistribution and the welfare state. The People's Share files it elsewhere too. The case for democratic abundance need not rest on the premise that the powerful owe the rest of us a share of what is rightfully theirs. It can rest on the older and harder premise that a great deal of what they hold was never legitimately theirs to begin with. The question the AI economy forces is not how to divide a pie more fairly. It is who fenced the field, and whether they left enough, and as good, for everyone else.
That phrase is Locke's, and Locke is the right place to begin — not because conservatives can be coaxed into liking redistribution, but because the most rigorous account we have of how property legitimately comes to exist turns out to indict the way the AI economy was built.
Enclosure, not scarcity
Locke's Second Treatise grounds property in labor: a person owns what they take from the common store and mix their labor with. But Chapter V attaches two limits to that claim, and they are distinct. The first is the spoilage limit — one may take only what one can use before it perishes. The second is the sufficiency condition — appropriation is just only where there is "enough, and as good, left in common for others."
The standard libertarian reading holds that money dissolves both limits. Locke does say, in sections 46 through 50, that the invention of durable money lets people store value past the point of spoilage, by tacit consent, and so accumulate without waste. But notice precisely what money is permitted to do there. It answers the spoilage problem. Gold does not rot, so hoarding it wrongs no one through waste. Locke never says, and the text will not bear, that money voids the sufficiency condition. Perishability and sufficiency are different objections, and money speaks only to the first.
Nozick — no one's socialist — reads it the same way. His version of the proviso is not about apples rotting in a barn; it is about whether an act of appropriation leaves others worse off than they were before it. He keeps it as a live constraint on acquisition, and he allows that where appropriation does worsen the baseline, it can require compensation. The honest Lockean position, then, is this: money lets you pile up coins. It does not let you fence the commons in a way that leaves others unable to mix their labor at all.
This is why the AI case cannot be waved away as ordinary accumulation. It is not the story of someone who traded nuts for gold and grew rich by consent. It is first appropriation — the enclosure of things that were not private property before anyone fenced them.
The first is authored work. Your writing, your code, your images, your music: these are the clearest possible case of labor mixed with the world to produce a thing, and under Locke's own theory the thing belongs to the one who made it. When a model is trained on that corpus, no labor of the trainer's was mixed with unowned nature. What was mixed was your labor with their machines. Behavioral traces — where you went, what you clicked — make a weaker labor claim, but they carry a second one in its place: no one consented to trade them into a training set. Tacit consent to the use of gold is not tacit consent to surveillance. The strong claim rests on labor; the remainder rests on the absence of consent; together they cover nearly everything the models ingest.
The second enclosure is the knowledge commons. A frontier model absorbs publicly funded research, the open web, and centuries of culture in the public domain, and returns it as a proprietary output that a new entrant cannot reproduce without a hundred-million-dollar training run. That is the sufficiency condition failing in plain sight. The appropriation does not leave enough, and as good, for the next person who would build.
A necessary precision, because it is where the argument is strongest: the compute itself is not the stolen commons. The GPUs, the datacenters, the capital that assembled them — those were genuinely built, and no Lockean disputes that built capital belongs to its builders. The enclosure is of the corpus, not the hardware. Which means the trained model is a joint product: private capital applied to a commons that private capital did not create. And joint products generate shared claims. That is the hinge of the whole argument.
From here the remedy follows without a single appeal to redistribution. A government that recognizes the labor-mixer's interest in the trained model is not seizing wealth to reallocate it. It is doing the one thing Locke says government exists to do — adjudicating a property dispute and protecting the prior claim. The instruments are property instruments: data dividends, opt-in licensing, local trusts that return a share to the people whose labor and culture made the model possible. Not a tax on profit after the fact. Recognition of co-ownership at the source.
Jefferson: ownership as capacity, not nostalgia
Jefferson gives the argument its political stakes.
"Dependence begets subservience and venality, suffocates the germ of virtue, and prepares fit tools for the designs of ambition."
— Notes on the State of Virginia
It is tempting to read this as agrarian sentiment, a fondness for wheat and yeomen. It is not. Writing to Madison in 1785, he called small landholders the most precious part of a state — not because farming is virtuous, but because owning the means of one's livelihood is what frees a person's judgment from the people they would otherwise have to please.
His mechanisms were mechanisms, not metaphors: land grants that put property directly in citizens' hands rather than under federal management; opposition to perpetual public debt, because debt makes a citizenry dependent on its creditors; schooling funded by the sale of land rather than by transfers. The through-line is ownership distributed at the source.
Translated to this decade, the Jeffersonian citizen needs a capital stake in the AI economy — a share in the models, a seat in the data cooperative, an actual claim on the automated firm — and not a monthly check administered from Washington. The distinction is not stylistic. Ownership without control is only a stipend with better branding. A dividend you cannot govern leaves you exactly as dependent as a wage; it merely changes the name of the patron.
The Anti-Federalists: the capture problem
There is an obvious objection from the other direction: if the models owe the public a share, collect it federally and pay it out — an American sovereign wealth fund for the age of AI. The Anti-Federalists supply the warning. Brutus argued that a government stretched over too vast a territory, armed with the powers to tax and to do whatever it judged necessary, would draw power inexorably toward the center and hollow out the states beneath it.
Apply that to a single federal fund holding the nation's AI dividends, and the prediction is not inefficiency. It is capture. One pot in one capital is the perfect object for precisely the concentrated interests the fund was meant to constrain; it would be lobbied, captured, and turned within a decade. The Anti-Federalist remedy is subsidiarity — hold the share at the smallest competent level. State and municipal data trusts that license local data. Compute cooperatives at the county scale. Professional guilds and mutuals that bargain collectively for model access. The Anti-Federalists were never against shared abundance. They were against its distant administration.
Guardrails for the design
The twentieth century is most useful here not as a roster of endorsements but as a set of constraints on how the thing gets built.
Hayek's knowledge problem cuts against anyone — including us — who imagines a bureau could correctly price access to a model or calculate what "enough, and as good" requires. The lesson is to make the shared interest tradable rather than rationed: shares that markets can value, not allocations an office hands down. Friedman's negative income tax was an argument about delivery — that cash preserves choice where in-kind welfare does not — and the dividend borrows that logic, though the funding is ours and not his: it comes from enclosure fees, the royalties and licensing charges owed for use of the commons, not from raising the income tax. Tocqueville and Nisbet argue for the bodies that should hold the shares — the unions, congregations, cooperatives, and mutual-aid societies that stand between the isolated individual and the paternal state and keep both in check. Ostrom shows what makes such commons endure: clear boundaries, benefits proportional to contribution, and enforcement close to the ground. The Alaska Permanent Fund works because Alaskans understand the oil to be theirs, hold it at the level of the state, and receive the dividend directly. Burke supplies the tempo — update homesteading doctrine for data, extend the tax treatment that already favors employee ownership, establish in law that training on public and authored material creates a compensable interest. Reform the property code that exists; do not erect a federal department of abundance.
Re-homesteading the frontier
Put together, the conservative case for the People's Share is not redistribution from above. It is the re-homesteading of an enclosed digital frontier, and it asks four things:
- Recognize the new property. Individuals hold a Lockean interest in the work their labor created, and a sufficiency-based claim wherever the enclosure of compute and corpus leaves nothing as good for those who come next.
- Distribute ownership at the source. Where firms license public data or public spectrum, require broad-based equity — employee ownership, customer ownership, local trusts — rather than taxing the profit afterward.
- Govern it close to home. Hold those shares in state trusts, municipal cooperatives, and voluntary associations, at the Anti-Federalist scale, where they can be watched.
- Pay cash where enclosure cannot be undone. Charge a royalty for use of the commons and return it as a dividend, preserving the choice of the people who receive it.
Each step keeps government inside its oldest and most defensible role. Defining and defending the boundaries of property is Lockean. Multiplying the number of independent owners is Jeffersonian. Keeping the power close enough to watch is Anti-Federalist. None of it requires a new theory of what the state owes us. It requires only that we take the old theory seriously enough to notice who has been violating it.