The Constitution Is a Distributed System

Consider a young adult you know. A niece, a new hire, a son. They did what we told them to do: the grades, the degree, the debt. They work more than one job, or one job that feels like two, and the math of an ordinary life, a home, a family, a margin for error, does not work out. They know it doesn't work out, because they carry a device that reminds them several hundred times a day, in fifteen-second increments, each one engineered to be slightly harder to put down than the last.

We tend to discuss this as a culture problem, or a character problem. It is an engineering problem, and I want to describe it the way an engineer would, because I am one, and because the diagnosis points somewhere unexpected: at a two-hundred-and-thirty-seven-year-old document that turns out to have been a distributed-systems specification all along.

The wiring changed

In 1992, Martin Nowak and Robert May showed something surprising about cooperation. Put players in a spatial game where defection pays, and cooperation should go extinct. It doesn't, so long as cooperators sit in clusters. Small clumps of neighbors who interact repeatedly can hold out against a world of defectors, and even expand. Cooperation, it turns out, is not only a virtue. It is a property of network topology.

Six years later, Duncan Watts and Steven Strogatz showed how little it takes to destroy that topology. Rewire just a few links in a clustered network into long-range shortcuts and the whole world collapses to six degrees. Add enough shortcuts and clusters stop mattering; everyone is adjacent to everyone. Later work confirmed what you would guess: in that regime, the conditions that let cooperation survive disappear.

Now notice what the engagement-optimized feed is. It is a machine for replacing your clustered links, family, neighborhood, colleagues, congregation, with long-range shortcuts to strangers, selected in real time for maximum arousal. We performed the Watts–Strogatz rewiring on an entire civilization in about fifteen years, and we funded it by auctioning the attention it captured. The machine is indifferent to what that attention contains; as The Functional Melancholic observes, even resistance is monetized. Outrage at the system is among its best-selling products. People did not get worse. The wiring changed. The rage, the tribalism, the collapse of shared reality: these are not moral failures spreading through the population. They are what the new topology computes.

The cost lands hardest on the young, because they never knew the old wiring. The economist Carol Graham's research puts it plainly: the deciding factor in whether a young person invests in their own future, stays in school, and avoids the risks that foreclose a life is hope, usually in the form of one adult who takes their aspirations seriously. We built a machine that strips hope at scale, and its business model is working as designed. On November 30, 2022, we attached a superintelligence pipeline to that machine. That is the moment we are in.

The proposed cure is worse than the disease

A movement in my industry has looked at this wreckage and reached a conclusion: the problem is democracy itself. You have heard the proposals under various names. Network states, startup societies with sovereign ambitions, governance-by-founder. The pitch is that legacy institutions are too slow and too captured to survive the technological transition, so the capable should exit: buy land, issue tokens, replace citizenship with subscription, constitutions with terms of service, and elections with a CEO.

The diagnosis is half right, which is what makes the pitch dangerous. Our institutions genuinely are failing to metabolize the rate of change. But look at what the cure removes. A citizen of a network state holds exactly the rights the founder finds it profitable to grant, revocable at the speed of a software update. There is no peaceful mechanism for the governed to remove the governor. We have run this experiment before, for most of recorded history. It was called feudalism, and the whole point of 1787 was to make it structurally impossible.

I don't think most people drawn to these projects crave serfdom. They crave what the projects market: small, voluntary, high-trust communities with real agency. Which is to say, they crave clusters. The longing is legitimate. The architecture on offer is a trap.

What the founders actually built

Read the Constitution as an engineer and you find a document written a hundred and fifty years before its own vocabulary existed.

Madison stated the design principle outright in Federalist 51: ambition must be made to counteract ambition. He did not trust the nodes. He arranged the topology so the system could survive them.

That is the insight we are about to lose, and it is the one we most need. The founders' answer to untrustworthy power was never better rulers. It was structure. If the topology of our networks now decides whether cooperation survives, then topology is constitutional infrastructure, and at this moment it is being written by advertising markets and, increasingly, by men who state openly that accountability is the bug.

The task in front of us is not to nationalize the feed or to flee into a founder's fiefdom. It is to carry the constitutional invariants into the layer where public life actually happens now: accountable power, rights that don't bend to convenience, and structure over trust, written into the protocols themselves.

What I built

I set out to write a book about this. Partway through, I realized the argument would be stronger if it ran. So for the past stretch of my life, mostly alone and mostly at night, I have been building a working prototype called strangeAttractor. It is one system, not a movement, and I describe it here as evidence that the alternative is buildable, not as a finished answer.

Concretely, it is a distributed AI and community platform with a few properties I believe should become ordinary:

This is what I mean by a startup society worth having: not a sovereign island for the wealthy, but small constitutional communities, interoperating, inside the republic rather than instead of it. Preserving the Constitution does not mean laminating the parchment. It means porting it.

The ask

I am one engineer, and this letter is the opposite of a pitch for my own indispensability. The prototype runs; the argument needs adversaries and allies who are better than I am at the parts that aren't code. If you fund things, fund the boring civic layer of identity, federation, and local-first AI with the patience you'd give infrastructure. If you regulate or legislate, understand that feed topology and engagement optimization are now constitutional questions, whatever the statute books say. If you research, test the topology claim; I have laid it out plainly enough to be wrong. And if you build, build for clusters and take nothing in payment for attention.

Whatever you do, when a system asks for your trust, ask the founders' question instead: where is the structure that makes trust unnecessary, and who can revoke this power peacefully? Any system with no good answer, whether platform, protocol, network state, or AI, is asking you to re-run the oldest failed experiment in history.

The young people I opened with are not lost. They are unwired. They are waiting, mostly without words for it, for someone to take their aspirations seriously and hand them a topology where effort compounds and cooperation wins. We know how to build it. The specification has been sitting in the National Archives since 1788.