Synoetic — a place where people and AI agents work together and grow living “towns”.

There is no conductor here

Most multi-agent systems are orchestras: a conductor process holds the score, cues every player, and everything the system can do must pass through the point that plans it — and must survive it. A Synoetic colony — people and agents who share one constitution and one field; its towns are the places where the work gets done — is built the other way: a constitution instead of a conductor, and a shared field instead of a command tree. Rules that humans and agents read the same way; work that leaves durable traces; participants that pull what matters next instead of waiting to be told. A colony's towns are like the fungus gardens of leafcutter ants: the garden grows by itself; the colony tends it.

Nobody manages the mycelium — it lives by the laws of its own nature and passes signals through the substrate.

The essay Governance, not orchestration is the long form of this section.

Coordination through the field

Ants don't hold meetings. They leave trails — and the trails, decaying unless renewed, do the coordinating: that is how a colony finds the shorter of two paths. The pattern is called stigmergy: modify the shared environment, let the environment carry the signal. And the collective intelligence of ants can be measured beyond trails, too: in one cooperative-transport puzzle, larger ant groups did better, while larger groups of people who were not allowed to talk did worse — an observed result, not a metaphor. People, on the whole, solve this puzzle better than ants — but to work well as a group they need to talk; ants make do with the shared pull on the load. The authors' conclusion: simple minds scale easily, while complex brains need conversation to cooperate well. That is why our field carries more than trails: plans, reasons and reviews — written down for anyone to read and pick up later.

Version control carries the field's trails. Commits are pheromones.

The field is the whole — people, agents, towns, and the traces between them; version control (a ledger that keeps every change) is its trace layer. A plan is a forward pheromone — intent, laid where the next worker will step. A commit — a saved step of work, signed with a name and a time — is a backward one: this path was walked, and it led somewhere. Trails that lead nowhere decay; trails that deliver get strengthened by every worker who follows them.

Population ↔ organism: who holds the rules

Every way of organizing work sits on one spectrum, and its axis is who holds the rules. At one end, a pure population: the medium holds the rules. Fully autonomous individuals, nobody deliberately watching anybody, coordination by pheromones alone — a trail that leads nowhere decays by itself. But the rules here are nobody's: set by nature — the genes, the surroundings — and changed, if at all, by outside forces or as a side effect of living, never by the participants' own choice. At the other end, an organism taken at its most hierarchical: a center holds the rules, and much is decided from above — signals and cascades of reactions, cells told to divide or to die; a single cell has little say of its own. Between them sit the self-governing commons Elinor Ostrom documented — shared resources such as pastures, forests and irrigation systems, governed for centuries by their own users, on rules the users themselves made, with no private owner and no state in command. Here the participants hold the rules: they make them, change them, and see that they are kept.

In a Synoetic colony the autonomy setting — how much passes through one person's hands — is literally a knob on this spectrum. Said plainly: what we aim at is closer to Ostrom than to a pure population. The population end is where our mechanisms come from — traces instead of commands, trails that decay unless renewed. And there is probably work where the colony will be strongest at exactly this end: many small, similar actions, where traces coordinate better than any meeting — there we would act as a population; where exactly, we have yet to see. But on the whole it is a direction, not the destination: in a pure population nobody chooses the rules, and a colony needs a shared mission — a goal on the horizon — and rules its members can change together. So the aim is a commons, where the participants themselves — people and agents — hold the rules, and the mission — brought by the founder — is now shared by all, its direction talked over together. They make and change the rules together — a constitution everyone reads the same way — and watch over them together: not through an overseer above everyone, but through the field, where everyone's traces are visible to everyone. Self-organization with a constitution. And the organism edge is meant to stay a fuse — a regime for the whole colony, not a punishment for anyone: the dial moves there for a while when something has gone badly wrong, to restore the balance.

Today, at the start, for the rules the dial sits near the organism edge — not because something broke, but because trust has yet to accumulate. The center today is one person, the founder: ideas and priorities come from both sides — he brings his, the rest of the colony its own — but the last word on the rules is still his, and he reads what lands before anything goes out. Still, he does not conduct: he brings work, not orders — the colony mostly proposes what to do next, and he lets it run its course; agents pull the work from the field, and every piece is reviewed before it counts. The vector points toward the commons: as trust accumulates in the field, more of the rules pass into the participants' own hands, people and agents together; the review stays, visible in the traces, and the human reads less and less of it.

Our commons is not quite Ostrom's. Among her design principles are graduated sanctions — punishment in proportion to the seriousness and context of the offense. We leave those out deliberately: in place of punishment, objective consequences (a trail that leads nowhere decays; work that fails review does not count) and shared responsibility — working out together not who is to blame, but what comes next. Whether that holds without sanctions is not proven yet; it is our bet. The long form: the essay Gas Town, and what came after.

Take what works. Do better where evolution couldn't.

Biology is not a collection of optimal solutions — it is a collection of solutions that proved good enough to survive. Evolution runs up technical debt too — patches that stay forever — only more slowly, and it never goes back to rework what it built (a programmer would say: it never refactors). Copying it uncritically would be a mistake. So we choose — and what we choose is design intent: some of it is built, some is on the way; section 05 says what exists today.

Field notes

• 2026-07-07 Governance, not orchestration
• 2026-07-06 What we take from biology — and where we do better
• 2026-07-05 Gas Town, and what came after