TPS zoologist Bryce Ryan sends along this bit, it's from Science but has quite a bit of economics in it.
- One thing that's missing is that cooperation is an expected outcome if the discount rate and difference in payoffs are structured properly. Punishment need not play a role. Repeated games + Realization that cooperation can help you = Cooperation, even in a selfish world. That's the folk theorem, basically, and it answers the question as to why games that look like typical prisoner's dilemmas don't cause the problems you'd expect. It's a function of repeated dealings.
- They test between the punishment and non-punishment groups and find more long-term income for the punishment groups...well, as a public choice guy, tell me about the collective action aspect of punishing the group?! They are groups of three...could you just buddy up with one other person in the group and blindly punish the third member for no reason? Also, realize that the structure of the punishment is such that the personal choice is "give to the common good or have it confiscated." Sure sounds Lenin-esque, huh? Not sure that approximates the choice for individuals in this society, but that's always the default criticism for experimental work. Also, one big problem with common pool problems are not that there's shirking (of course, that's a problem) but that it's oftentimes tough to tell who's shirking. If there's a can to chip in money for the local park-- how can you tell who put money in? Yes, it can be overcome, but it's still a problem-- it's not like common pool problems can't be overcome either, so it's valid to consider.
- By the way, this is a very common experiment to show the detrimental effects of poorly defined property rights.