A note on order: this is the second of three pieces that go together. If you are starting here, read "After Privacy, a People" first, then this one, then "The Missing Variable." Each stands alone, but they were meant to be read in that order.

There is a tradition of thought - running from mechanism design through quadratic voting and funding to the book Plurality - that has, to my mind, the best diagnosis of institutional failure now on offer. It says, roughly: power and ownership concentrate faster than societies learn to cooperate; our two great aggregators, the market and the ballot, each pick up only one signal - the market hears money, the ballot hears headcount, and neither registers how much people care or how differently they see; and technology, which could have connected people across difference, has mostly driven them apart. If you accept that diagnosis, the response follows: redesign the instruments. Make votes express intensity. Make funding match the breadth of support rather than the depth of a pocket.

I accept the diagnosis almost entirely, and I have learned more from this tradition than from most of what passes for democratic theory. What follows is not a rebuttal. It is an observation about a missing layer, offered to people I expect to keep learning from.

The observation is this: every one of these mechanisms presupposes something it cannot create. Before a vote can be weighted, split, or spent, there must be an answer to a prior question - who is voting? Not their name; their singularity. Is this participant one human being, or a thousand identities run by one person? Will this participant exist tomorrow, carrying the consequences of today's choice, or dissolve and re-enter fresh, owing nothing for what they did? A mechanism is a rule for combining voices. It has no way, from inside itself, to establish that the voices are real, that they are distinct, and that they persist. That is not a parameter of the mechanism. It is the ground under it.

The tradition knows this. Sybil resistance appears in the papers as an acknowledged assumption, deferred to later. And the deferral has a structural consequence: the mechanisms live as guests. A participatory budgeting poll hosted by a city council member. A deliberation pilot run with a state's civic organizations. Each experiment borrows someone else's population, someone else's registry, someone else's trust that the participants are who they seem. The instruments travel from institution to institution, showing what they can do - and nowhere do they stand on ground built for them. Mechanisms everywhere, a demos nowhere.

I want to be precise about what a demos is, in this sense, because the word usually gets used loosely. I mean something minimal and almost technical: a set of participants each of whom is verifiably one real, singular, continuous human being - countable without being surveilled, accountable without being exposed. Not a community of shared opinion. Not a network of wallets. A body that can be counted once and holds still while you count it.

Why does this matter so much? Because the two failures that make collective life break down are not, at bottom, mechanical.

The first is the absence of standing - of real freedom in the structural sense. The decisions that shape a life are made at scales where the person has no channel: not the price of their labor, not the algorithm that curates their world, not the wars and debts made in their name. You can hand such a person the most finely tuned voting instrument ever designed, and they remain what they were: someone whose voice is heard, at best, inside a sandbox that power built and power can close. Refining the vote does not create the standing to vote on what matters.

The second is the absence of trust. Whatever is left of shared life runs on a dwindling reserve of it, and the digital layer - where more and more of that life now happens - runs on almost none, for a structural reason rather than a moral one: online, you cannot know that the other is one person, or a person at all. Where singularity cannot be established, accountability cannot attach; and without accountability, the rational stance toward strangers is defense. No weighting formula repairs this, because the formula has to assume the honesty of its own inputs. Trust is not an output of mechanisms. It is what they run on.

So the tradition and I are working on different problems - and this is complementarity, not superiority. Mechanism design refines how people decide together. It does not, and does not claim to, establish who the people are to one another, or whether they are free to decide anything of consequence. Those two questions - trust and standing - are answered, if they are answered at all, by something older than an algorithm: by constituting a body. A demos is not designed the way an auction is designed. It is founded, joined, verified, and counted - and only then does it become the place where refined mechanisms can finally run on their own ground.

The previous essay in this series described how such a body has, in fact, been constituted: verified personhood without disclosed identity - a check that establishes a living, singular human and then discards what it saw; membership by free choice rather than birth; one verified person, one vote, with no path by which money, reputation, or contribution converts into voting weight. On the question of weighting, this project and the plural tradition genuinely diverge - we load decision quality onto process rather than onto the ballot - and I have made that argument elsewhere and will not repeat it. The point here is different and friendlier: whatever one believes about how votes should be counted, the counting has to happen somewhere, on ground that can bear it. That ground now exists. It is small - I will not pretend otherwise; its members number far from the millions its argument imagines, and a demos this early proves a class of possibility, not yet a scale. But it is real, it runs, and its whole design answers precisely the two prerequisites the mechanisms defer: singularity you can verify, and accountability that does not cost anyone their privacy.

Which suggests an experiment that, until recently, could not have been run by anyone.

Every comparative claim about collective choice - that weighted votes reveal preferences better, that unweighted votes resist capture better, that deliberation before voting beats deliberation after - has had to be tested on soft ground: unverified crowds, borrowed populations, participants who may be one person or fifty. A verified demos changes what is testable. Run the mechanisms - plural, quadratic, filtered, anything - as advisory instruments on a population where every participant is verifiably one continuous human, and for the first time the results measure the mechanism rather than the noise. The constitutional floor does not move: binding decisions remain one person, one vote, because for us that floor is the point. But above the floor there is room - sandboxes, polls, consultations - where instruments from a rival tradition could be tried against our own, honestly, in the open, on ground built for exactly this. The two bets I described in the previous essay - refine the tools inside existing institutions, or constitute a subject outside them - are usually framed as alternatives. Here is the place where they stop being rhetorical positions and become a shared experiment.

I do not know which bet wins. It would be convenient to believe they converge; they may not. A demos without refined instruments stays coarse - a blunt majority that cannot register intensity. Instruments without a demos stay guests, homeless however good they are. What I know is that these two failures are not the same failure, that each tradition holds the cure for the other's, and that this is now demonstrable rather than arguable.

The invitation is simple: put the plural instruments and ours on the same verified population, in the open, and see which one holds up. That experiment could not have been run before now.