My interest in applying experimental methods to the study of public goods was inspired by my prepublication knowledge of the proposed Groves-Ledyard “solution to the free rider problem.” Consequently, we have the rare circumstance in which an economic theory and experimental tests of it are published fairly close together in time. Public goods was a natural for experimentation: one had a hypothesized problem in incentive failure and an incentive-compatible solution. It was necessary to establish, empirically, that there was a free rider problem in the first place and that, in the second place, the proposed solution improved matters. Ultimately, what emerged was: (1) there is a free rider problem but it is not as severe, empirically, as the strong version of the theory predicts; (2) the “solution” works quite well, but so do other “solutions” (such as the auction mechanism) with much less attractive static theoretical equilibrium properties (they have a blizzard of static equilibria); (3) all group-decision procedures require a stopping rule, and this has its own incentive effects whatever the nature of the static model; and (4) our theories are woefully weak on dynamic process analysis.

Why do not “free rider” mechanisms perform more poorly than they do? The answer, I think, may be related to the question of why opportunistic behavior is not more common in private contracting, why there is not more crime — at least petty crime — since it appears to pay, why property rights tend to be respected in the absence of continuing enforcement and why tipping is commonplace even in restaurants people are unlikely to visit again, and so on.