Democracy is about to get valuable

Every era has a resource it quietly runs out of. The industrial age ran out of clean air. The information age ran out of attention. The AI age is running out of something stranger: evidence that a human was here.

The cost of producing a convincing person has collapsed. A persuasive opinion, a coordinated crowd of supporters, a grassroots movement with ten thousand voices: all of it can now be generated for less than the price of a coffee. This is not a prediction. It is the operating reality of every comment section, every consultation portal, every "public feedback" process running today. When fake participation is free, real participation becomes impossible to see.

And here is the uncomfortable economic consequence: when something becomes impossible to see, proving it becomes valuable. A verified human voice, one person, speaking once, in a way that cannot be faked, duplicated, or bought, is becoming the scarcest input of collective decision-making. Scarce things get priced. Democracy, the system that runs entirely on verified human voices, is about to get valuable.

The only open question is who sets the price.

Two ways to pay

There are exactly two ways a society can pay for proof of humanity, and they lead to opposite worlds.

The first way is surveillance. Verify everyone by identifying everyone: document uploads, biometric scans, an ID check in front of every opinion. This is the default path, because it is what existing institutions and platforms know how to build. Its cost is not money. Its cost is that the secret ballot dies on the spot. A vote you cast with your iris is a vote someone can trace back to your iris. The verification layer becomes the control layer, owned by whoever runs it, and every authoritarian in history would have paid any price for it.

The second way is cryptography held by communities. Verify personhood without identity: neighbors attest to neighbors, the credential is a key only its holder controls, the ballot is encrypted before it leaves the hand, the count is proven correct by mathematics rather than by an administrator, and the power to decrypt anything at all is split among citizens so that no single party, including the operator, including us, can ever look inside. The cost of this path is engineering effort and operational discipline. We know the exact number, because we paid it: a few months of work, about $15 a month of infrastructure, and roughly $3 of computation per election.

The first path is cheap to start and catastrophic to own. The second is hard to start and nearly free to own. That asymmetry is our entire company.

AI centralizes. Crypto decentralizes.

These four words carry the whole argument, so they deserve to be unpacked once, carefully.

AI centralizes because its economics centralize: frontier models concentrate compute, data, and capability into a handful of organizations, and every institution that adopts AI inherits a dependency on that handful. This is not a moral failing of the technology; it is gravity. We use AI heavily ourselves, including agents that operate large parts of our own infrastructure. The gravity is real and it points inward.

Cryptography pulls the other way, and not as metaphor but as mechanism. A zero-knowledge proof moves verification from the institution to the mathematics: nobody has to trust the counter, because a wrong count cannot be submitted. Threshold cryptography moves custody from the operator to the community: in Röbel, reading a single ballot requires three of five citizens, in different places, holding different keys, to act together in public. The machine that coordinates all this is deliberately powerless. We built a server whose defining feature is what it cannot do.

So the pairing is not "AI versus crypto." It is a division of labor that we think defines the next decade of civic infrastructure: agents do the work, humans hold the keys. AI drafts, monitors, diagnoses, operates, explains. Cryptography guarantees that no amount of that capability, ours or anyone's, can see a vote, forge a count, or move a treasury without the quorum of humans the community itself named. The more powerful the agents become, the more valuable the key ceremony gets. We are building for the world where that trade is obvious.

Who pays, and why they will

"Will anyone pay for democracy?" is the wrong question. People already pay for it, heavily and badly. A single postal ballot round costs a mid-size German association thousands of euros and three weeks. Municipal participation platforms cost hundreds to thousands per month and cannot hold a secret ballot at all, which is why their results convince nobody. The price of NOT having trustworthy participation is paid in the currency institutions are actually going bankrupt in: legitimacy.

So the buyers are concrete. The 600,000 German Vereine and cooperatives whose statutes require votes they currently run by show of hands and paper. The 11,000 municipalities under pressure to digitize without becoming surveillance operations. The unions, the housing societies, the DAOs that discovered public token-voting reproduces every pathology it was meant to escape. And increasingly, networks themselves: any protocol or platform that needs to know "is this a real, distinct human of this community" will need an answer that does not come from a biometric orb or a passport database. A citizenship layer that can say yes without revealing anything else is infrastructure the AI economy will need, not the other way around.

What they buy from us is not the code. The code is open source, deliberately, because openness is what makes the trust claims checkable and what makes the stack a commons instead of a product trap. What they buy is everything around the code: the deployment that takes an afternoon instead of a quarter, the key ceremony run with their five citizens, the compliance documentation their board demands, the election-day support, the assistant that explains a proposal to a 70-year-old in plain language. The cheap tiers sell convenience. The expensive tiers sell accountability. The mathematics is free for everyone, forever, and that is precisely why the surrounding service is worth paying for.

The bet

Our bet, stated falsifiably: within a decade, "verified human, secret ballot, proven count" stops being a property of nation-state elections and becomes a standard capability of ordinary communities, the way "has a website" did. Either it arrives in the surveillance form, owned by platforms, priced in privacy, or it arrives in the community form, owned by the communities themselves, priced like the utility it is.

One town in Mecklenburg already runs the second form. It cost less per month than its coffee budget. The town does not have to trust us, and neither do you: every count it has ever published is verifiable onchain by anyone, forever.

Democracy is about to get valuable. We intend to make it affordable, and to make sure the ones who own it are the ones it belongs to.