This is our village: ten of us and one pond. The pond has 40 fish in it today, and it has room for 50.
Every summer we all go fishing. The fair share is one fish each. Every winter the fish that are left have babies: each fish can have one baby, but the pond only has food for 10 babies.
So if we all take one, 10 fish come out and 10 babies go in, and the pond stays as it is. But some of us take two. Then more fish come out than go in, and the pond shrinks.
Everyone carries their catch home in a bucket. So unless somebody was at the pond watching, nobody knows who took two.
You are the village meeting: all of us, deciding together. Each summer you pick one card. Every card is a way of putting people off taking two: do nothing, have a talk, take turns watching, or get a policeman from the town. Then everyone goes fishing, and you see what the village can see.
Touch a card to see what it does: what it costs, what happens to someone who takes two, and what we get to find out. Then press FISH!
The aim. Keep the pond full and everyone fed for six summers. Then comes a seventh summer when nobody is watching. Will they still take one fish each?
The pond. The blue blob is our pond. Every orange fish in it is a real fish, and the number next to it counts them. We start with 40, and there is room for 50.
The babies. Every winter the fish that are left have babies. Three rules: each fish can have one baby; the pond only has food for 10 babies a winter; and the pond never holds more than 50 fish. So if we all take one, 10 come out and 10 go in, and the pond stays as it is. Every fish we take beyond that makes the pond one fish smaller next summer. Watch out: if the pond drops below 20, even one each is too many, because fewer than ten fish are left to be parents.
The village. The ten stick people are us, with our names under us. Each summer everyone takes one fish or two. One each is fair. Nobody can tell in advance who will take two.
The buckets. Everyone carries their catch home in a bucket. A bucket with a ? on it means nobody saw what was in it. Fish drawn in someone’s hand are what somebody saw them take: the watcher, or the policeman when he caught them. The meeting always finds out how many fish left the pond, so it always knows how many of us took two. It just doesn’t always know who.
The cards. Each summer, touch a card to see what it does. It is drawn into the picture (a circle on the grass, a watcher on a stool, a policeman with a tall hat), and the box with the green line says what it costs, what happens to someone who takes two, and what we get to find out. Leave it be: nothing happens to anyone, and nobody knows who took two. Have a talk: we all promise to take one; if fish go missing, the circle tells everybody off next summer; nobody knows who broke the promise. Take turns watching: one of us watches instead of fishing and sees everyone; seen taking two means a telling-off the first time, both fish back in the pond the second time, and both fish back plus no fishing next summer the third time. Ask the town for a policeman: he stands by the pond every day and catches nearly everyone who takes two; the fish go back in the pond. He tells us only who he caught, and we stop telling each other off.
FISH! Everyone goes to the pond one at a time and comes back with a bucket, or with their fish showing if someone saw them. Under each person it says what happened to them: told off, fish put back, caught, no fishing. Then the pond is counted, the babies come, and the game tells you what the village knows: how many fish came out, how many of us took two, who (if anyone saw), what the pond did, and what we ate. The bones on the shelf are all the fish we have eaten so far.
Next summer. Pick again from all four cards. Nothing is kept from last summer except what the villagers learnt.
The seventh summer. After six summers there is one more, with no card: nobody watches and nobody talks. If the pond is fished out before then, the game ends there instead: with no fish there is nothing to watch and nothing to share. Then the game shows you what really happened, which the village itself never saw, and who never cared about the rule from the start.
The paper. Elinor Ostrom (2000), “Collective Action and the Evolution of Social Norms”, Journal of Economic Perspectives 14(3): 137–158. doi:10.1257/jep.14.3.137, open access at aeaweb.org. Game built from this paper; the paper is the authority where the two disagree.
What the paper argues. The old theory said self-interested people would not provide a public good. In the lab they do, about half, and it decays; talking face to face raises it and keeps it up; people pay to punish free riders. Ostrom reads this with three kinds of player: rational egoists, conditional cooperators who share while they expect others to and give up when disappointed, and willing punishers. Norms are a value people add to the plain payoffs, and they evolve: toward whoever earns more, up where sharing is reciprocated, and down where rules enforced from outside make a norm of one’s own unnecessary. A little outside policing is the worst of all worlds; a lot works while it lasts and leaves less behind than never having had it. The commons that lasted for generations were run by their own users, with monitors they chose and sanctions that start as a word.
What this game is. A toy built to illustrate that argument. Ten villagers of the three kinds (three, five and two, our reading of the paper’s 40 percent, 27 percent and the rest), a norm strength for each, a pond that breeds, and four arrangements that are the paper’s regimes: the lab baseline, face-to-face communication, local monitoring with graduated sanctions, strong external enforcement. (A fifth, weak external enforcement, the paper’s “worst of all worlds”, was played until revision 8 and cut: two cards about policing in a paper about self-governance.) The seventh summer is the second phase of the Frohlich and Oppenheimer experiment the paper cites, in which the outside mechanism was removed.
What the meeting knows. The player is the village meeting and is told only what it could know. Everyone carries the catch home in a bucket, so the meeting always learns how many fish left the pond, and from that how many of the ten took two, but a name only when somebody at the pond saw: the village’s own watcher sees everyone, one fish or two, and reports them all; the town’s policeman reports only whom he caught. The policeman is honest and costs nothing: his whole price is the seventh summer. With nobody at the pond, the row shows ten buckets. This is the paper’s own point about monitoring: it is information before it is punishment, and the lab’s communication treatments reveal totals, not names. The seventh summer is then revealed in full, which the village never sees, together with who never cared about the rule.
What the game demonstrates. That a village of the paper’s three kinds, with norms that move as the paper describes, produces the paper’s pattern: about half share with nothing in place and it slides; a talk raises it and holds it; watching with graduated sanctions brings grabbers into line and leaves a village that shares unwatched; a policeman from the town enforces sharing while he is there and leaves less norm behind than never having had him.
What the game assumes. Everything in its rules: how a villager weighs the second fish against reciprocity, shame, assurance and the chance of being caught; the sanctions in kind (a word, then both fish back, then both fish back and a summer out; the policeman’s catch goes back in the pond too); the rates at which norms grow, are imitated, and are crowded out; the pond’s rule (each fish left can have one baby, food for ten babies a winter, room for fifty fish); the newcomer is switched off. None of these numbers is in the paper. The daily policeman’s bad seventh summer depends on crowding out being fast enough: under this game’s rate, and at three-quarters of it, the policed village ends below the one left alone; at half the rate it does not. Under a policeman, imitation of the villagers who ate best builds no norm, because what they show is compliance; it still erodes one when grabbers eat best.
What the paper establishes. Seven lab regularities, reported from the literature; field findings that user-made, user-enforced rules with graduated sanctions characterise long-surviving regimes; the crowding-out experiments; and an argument, not a theorem, that an evolutionary account with several types explains them. The paper says a mild degree of external monitoring “may be” the worst of all worlds; it does not measure it.
What is left out. The announced last round; group size and heterogeneity; boundary rules, cost-sharing, conflict arenas, the right to organise, nested layers; rewards; the trust game and the information regimes of Güth and Kliemt in their own right; the leader.
Numbers. Every number the game uses sits in a commented block at the top of its source, tagged with the page it came from or marked as ours. A village number replays exactly.
Steffen Huck, 2026, with Claude under the paper-to-game skill. Fonts: Schoolbell (Font Diner, Apache 2.0) and Short Stack (Sorkin Type, SIL OFL), embedded.