The Civic Mill

The ant and the map

No ant knows where the food is. Each sets out randomly, and the one that stumbles on food turns for home, laying a faint trail of scent as it goes. The next ant along is more inclined to follow that trail than to wander off, and if it finds food the scent is deepened on the way back. Given time, short routes are strengthened while detours fade, and within an afternoon the colony settles on very nearly the optimal path from nest to dinner. No ant could draw the map. There is no map — only the ground, which remembers what has been done upon it, and a few thousand simple creatures each doing one small thing in response.

Biologists call this stigmergy, combining the Greek words for mark and work: coordination through marks left by work done. Termites use this method to raise their cathedrals without 3D printers or construction cranes, and wasps build paper cities; the whole vast enterprise of a colony gets done without a single meeting ever being called. Such self-organization through information processing seems to me one of the more hopeful facts about the world. Intelligence need not live in anybody’s head; it can live in the arrangement of things, which can evolve to multiply the incremental additions and subtractions left by whoever is passing by.

What would it be like to live in a society that decided some of its questions that way?

A Tuesday, some years from now

On your way to work there is a board in the bus shelter, and this month it holds a question from the water authority. The reservoir is low and something has to give, and rather than announce the answer the authority has put the question to everyone who waits for a bus. You have three stops’ worth of time. So you take a card and write down what a hosepipe ban would mean in your house, or a price rise, or a summer of brown lawns — not what you think should happen, only what it would cost you — and you pin it up with the others. Then you write down your own idea about what to do, and you do this before you read anybody else’s, so that the board gets your fresh thought rather than an echo of one already there. Only then do you read what the people before you have left. You file your idea beside the ones it most resembles, and when you see that one of the ideas on the board would plainly hurt the family whose card is two along from yours, you draw a line through it on their behalf. The bus comes. You have been four minutes.

You will never meet the people who were there before you, or those who come after. But by the end of the month several thousand of them have passed through that shelter, each spending a few minutes and each leaving a mark, and the answer the board has arrived at is one the town can live with, because the town wrote it. The water authority is bound to publish that answer beside its own and to explain, in public, wherever the two differ.

At lunch your phone buzzes. Your name has come out of a lottery, exactly as a juror’s does, and you are asked to spend twenty minutes this week on the hospital’s visiting-hours question with eleven strangers whose names came out of the same hat. Nobody volunteered and nobody was picked for being clever, and that is precisely the point. The people at the centre of the process are whoever chance put there, so there is no seat for anyone to buy.

When you get home, the fridge door is at war. Too tired to cook, says one note. Not pasta again, says another. By six o’clock Tacos has two notes stacked upon it, Tacos but vegetarian has appeared beside it and is propping it up, and somebody has put two sullen strokes through Leftovers. Nobody in the house has seen anybody else all day. It is tacos, and there is nothing to argue about, because everyone has had their say and everyone can see the count.

None of this is science fiction. Every piece of it has been tried somewhere, by someone. What has not happened yet is the habit of it — the day when a formal democratic decision board in a bus shelter is no more remarkable than the timetable beside it.

Every group has a process

When I worked at QinetiQ, fresh out of university, one of the first things I was taught was that any group of people working together has a process, whether or not anyone designed it. A meeting with no agenda still has one; it is called whoever talks first and loudest. A legislature has one. A family has one. A queue has one. The only questions worth asking are whether the process was chosen on purpose, and whether it is any good.

Computers taught us this long ago. Ask a hundred programmers to sort a list and you will get dozens of different programs, each arriving at the same sorted list by a different route, and nobody finds this strange, because nobody imagines there is only one way to sort. Yet when it comes to people deciding things together we use perhaps four procedures — the vote, the committee, the boss and the shouting match — and we spend an astonishing amount of energy arguing over which of the four is least bad, and almost none inventing a fifth.

So here is the thought at the heart of this piece. Democracy is not really a ballot. It is what it costs to take part. A ballot is a great deal of attention spent very rarely on a question that somebody else wrote. The fridge door is a little attention spent often on a question you are living inside. Between those two lies almost everything a society has to decide — how the water is shared, when the hospital opens its doors, what gets built and where — and almost none of it is ever asked of us. Not because anybody is hiding it, but because we have never built the machinery that would make asking cheap. That is the whole project. Clever processes, so that the price of a say comes down to a few minutes, and a few minutes is a price anyone can afford.

Saturday on the Square

Years ago I had an idea I never got to try. What if a conference were organised not around talks, with questions at the end for whoever could reach the microphone, but around an algorithm — a procedure each person stepped through in turn, so that the thinking was done by the whole event rather than by its loudest members? I filed it away with the other things I meant to get round to. It took a move to Madison, Wisconsin, and a farmers’ market, to bring it out again.

The Dane County Farmers’ Market is held every Saturday on Capitol Square, and if you have never seen it you should picture this. The Wisconsin State Capitol stands in the middle of a broad green square; the market runs around all four sides of it; and the crowd — thousands of people, with dogs and pushchairs and paper bags of sweetcorn — walks the square in one direction, round and round the building in the middle, the whole morning long. It is a lovely thing to be part of. It is also, if you have ants on the mind, a slightly uncanny thing to watch. I have stood on the Capitol steps more than once and thought: there is the colony, and there is the nest, and every one of these people is walking in a circle around the place where the decisions are made.

So we put a board on the Square. The Civic Mill is a noticeboard at a market stall, and the offer is simple: give us five minutes, and remake a decision the building in the middle has already made, without being told what the building chose. There is no facilitator and no discussion to run. There are only the rules, and the board, which remembers.

The rules are the ones you met in the bus shelter, and they are worth setting out once, because together they are the whole algorithm. Each visitor begins by writing down what the question costs them — the worry, not the opinion — so that the problem on the board is described by the people who actually have it, before a single cure is proposed. Each visitor then writes their own idea before reading anyone else’s, and files it beside the ideas it most resembles, so that a hundred independent guesses go onto the board rather than one guess copied a hundred times. And each visitor may strike out one idea, but only on somebody else’s behalf: the stroke must point at a worry already on the wall, and the strokes are counted. Write, file, strike. Each act is small, each leaves a mark for the next person to work with, and out of the marks an answer condenses that nobody wrote. How the count is done, and everything else you would need to run a board of your own, is on the project site, starting with the field guide.

The first board we ran was given the Foxconn decision, which most people in Wisconsin will remember: a famous company offered to build a factory in the state in exchange for several billion dollars in tax breaks. About twenty shoppers, a few minutes each. What they worried about, overwhelmingly, was whether the deal was honest. And the answer they settled on was neither yes nor no but a citizens’ assembly — ordinary people drawn by lot, with the company in the room to answer their questions, weighing the costs against the benefits in the open. Given a process, in other words, the crowd asked for a better process. The full report is published, beside what the legislature actually did.

What I did not expect, and what I now think matters more than any single answer, is what the board does to the people who use it. When you stand at it, the question in your head is not look what that idiot wrote. It is I hope I’ve put mine in the right place. You are not judging the other notes; you are working out how yours relates to them, and that small mechanical task is quietly disarming. When you pick up the pen to strike something out, it is to protect a stranger’s worry that you will never feel yourself. And losing feels different too. When you can see your own worry on the wall and your own idea in its stack, being outvoted is no longer being overruled. It is losing fairly, in daylight, to your neighbours. People who get used to that start to expect it, and people who expect it start to ask why the building in the middle of the square does not work that way.

The mill

Every process has a way of failing, and it would be dishonest to sell you this one without showing you where it breaks. The ants, after all, have their own.

When a column of army ants loses the scent of anything new, each ant simply follows the one in front, and the column closes into a circle that turns, and turns, until the ants drop from exhaustion. It is called an ant mill: perfect coordination, going nowhere. I chose the name for the board with that in mind, and not only for the grinding of grain. A crowd that only reads what is already on the board will circle it in exactly the same way, which is why every visitor writes before they read. A board can be flooded by an organised few. A steward who leans on the answer can bend it. And the people who stop at a market stall have chosen themselves, so a market crowd is not a random sample of anyone, and a board is not yet a lottery, though it is a step towards one.

None of this is a reason to distrust processes. It is a reason to be careful what a process asks of you. A process that costs a great deal — an evening in a draughty hall, a campaign, a career — demands your trust, because you cannot afford to try another. A process that costs four minutes at a bus stop asks for no trust at all. If it goes wrong you have lost four minutes, and the board can come down on Saturday night and a better one go up on Monday. That is the deepest reason for keeping the cost of participation low: cheap processes can be doubted, tested and replaced, while expensive ones have to be believed in. The very first sketch of all this, long before the Square, was a democracy tree — a tree beside a footpath on which walkers tied a tag as they passed, each adding to what the last had left, none of them staying long enough to argue. The board is that tree with rules. It will be broken and mended a good many times before it is done well, and I find that cheering rather than otherwise. We are allowed to keep trying, and trying in public with a stall and a pen is about as cheap as trying gets.

Whoever passes

The Civic Mill is a project of Sortition USA, a member-run nonprofit with chapters across the country that makes the case for democratic lotteries — citizens’ assemblies, impartial commissions, oversight boards, legislative juries — on the conviction that ordinary people, chosen by chance and given a fair process, can decide as wisely as those elected to decide for them. Its Madison chapter runs the board every Saturday and publishes every result. The Deliberation Gateway is where we try to work out what such processes are, why the good ones work, and how to build the next one.

The Madison board is below, as data. Load it, compute the answer, then drag a single note somewhere else and compute again. Watch the answer move — not because anyone changed what they wrote, but because of where they chose to put it.

No ant knows the way. The colony finds it anyway, a little work each, by whoever passes. There are as many ways for people to work together as there are numbers, and this is one of them. Go and invent another.

Further reading

• The Civic Mill — the project site: the field guide, the steward’s guide, the Madison docket, and every board’s report and data.
• The Saturday Board: The Foxconn Decision — the first board, 5 September 2026.
• Sortition USA — our partner, and the case for democratic lotteries.
• Liberating Structures — some three dozen small patterns for how groups work together, among them 1-2-4-All, 25/10 Crowd Sourcing, TRIZ and Troika Consulting.
• The Co-Intelligence Institute — Tom Atlee’s work on collective wisdom, including the Wise Democracy Pattern Language and the Group Works deck.
• The Democracy Tree — DG 001, where the idea was first sketched.