Fishing boats work the same lake, farmers pump from the same aquifer, and commuters share the same road. Each additional user lowers the return of all the others, and no individual user bears that cost.
The game tests whether a group restrains its use of a shared resource when restraint is in the interest of the group and not of any member. It shows the dissipation of the rent that the resource could yield, and, with communication switched on, whether a group can govern itself without an outside authority. The structure mirrors the public goods game, with the difference that members take from a common stock where in the other game they give to one.
Gordon (1954) showed that under open access a fishery attracts effort until the average return equals the return available elsewhere, at which point the resource yields no surplus. The argument received its familiar name from Hardin (1968), and with a fixed number of users its conclusion is less extreme. Each member increases effort until the member's own marginal return equals the outside wage, ignoring the reduction imposed on the others, so the group works the resource more than is best for it.
In the game here, an hour on the lake yields a catch that falls in proportion to the hours of the whole group. The group does best when the total hours equal the difference between the initial catch and the outside wage, divided by twice the rate at which the catch falls. In equilibrium the total is higher by the factor of two times the group size over the group size plus one. With the default settings each of four members fishes for 8 hours in equilibrium, against 5 hours at the group optimum, and the group keeps 64 percent of what the lake could add to its income. The share kept falls toward zero as the group grows. Ostrom (1990) documents the institutions by which communities have escaped this outcome.
In the experiments of Walker, Gardner and Ostrom (1990), groups of eight faced an equilibrium that left 39 percent of the maximum rent. Groups whose members had a small endowment earned 37 percent on average. Groups with a large endowment, which gave each member the capacity to do more damage, earned about minus 3 percent, with heavy losses in the early rounds and a gradual approach to the equilibrium afterwards.
Communication changes the result considerably. Ostrom, Walker and Gardner (1992) report that groups with the large endowment earned 21 percent of the maximum without communication, 55 percent after a single opportunity to talk, and 73 percent when they could talk repeatedly. Sanctions without communication achieved little once their cost was counted, at 9 percent, while groups that talked and chose their own system of sanctions reached 90 percent.