As a social scientist who conducts extensive field research, as well as doing experiments and theoretical work, I find the target article resonates well with my own research. Since one of the earliest experiments on punishment was conducted by myself, Roy Gardner, and James Walker (Ostrom et al. Reference Ostrom, Walker and Gardner1992), the origin of that article may be interesting to readers of this issue. As Guala indicates, we had earlier reviewed a very large number of in-depth case studies of settings where users organized their own governance system related to common-pool resources. In my 1990 book, I reported on a massive effort to synthesize findings from this large number of studies. One of the principles that I derived from this study was that long-lasting and robust institutions tended to use “graduated sanctions” (Ostrom Reference Ostrom1990). By graduated, I meant that users of successful common-property regimes would sanction one another for observed nonconformance to their own rules and would first gently remind one another of such infractions. The gentle reminders would be exercised one or two times, but after using “shame” to try to bring someone around to following rules, there would be other punishments that would be imposed in an ever-greater level of cost to the recipient. The final punishment might be rather severe.

I also observed punishments being administered in the field and was puzzled, because one could not explain such punishments by users themselves using game theory. Therefore, I asked Roy Gardner to develop a rigorous game-theoretic model and worked with James Walker to test that model in the lab.

When we gave the participants in an experiment an opportunity to pay a fee to fine someone else, they did indeed use it contra to the game theory prediction. In fact, we found they overused it. We did not quite know how to explain the overuse, but we finally used the term “blind revenge” to explain the fact that frequently the sanctioner would direct the fines against others whose computerized record showed that they were highly cooperative. We thought they probably figured that the cooperators were initially fining the non-cooperators and the non-cooperators then fined the cooperators in revenge.

In the massive case studies we had worked through, we did not find many cases of blind revenge. Hence, we decided to move to the next step in the lab and give participants an opportunity to communicate and decide on their own rules. Those who engaged in self-governance then did not use fines very often. They increased their levels of cooperation to the point that their net benefits were very close to optimal. Thus, the combination of agreement and discussion first and evidence that others were cooperating, led to a much better result than an externally designed sanctioning system without a set of rules that the participants had agreed to.

Although in the Janssen et al. (Reference Janssen, Holahan, Lee and Ostrom2010) study we did not give participants an opportunity to design their own rules, we gave them the opportunity to simultaneously engage in communication and use of fines, the use of fines alone, and the use of communication alone. As Guala comments, when communication and punishment opportunities were combined, the subjects did the very best in the lab.

There are several issues being debated by a variety of very distinguished scholars. There is no question that humans have the capability of engaging in serious punishment of each other; but that should not lead us to conclude that the way of achieving long-term sustainability is by enabling participants to punish each other without enabling them to engage in serious discourse about the rules they want to adopt and how they should be observed and sanctioned. When participants in a dilemma setting are able to engage in serious discussion, debate about their joint future, and agree on rules that limit strategies, they have much less need to use punishment against defectors. Monitoring each other and initially shaming those who do not comply with their rules is, however, an essential component for sustaining that cooperation over time. Stronger sanctions are not often needed, but their authorization backs up the use of mild sanctions when rule-breaking behavior is initially observed. Our recent research related to forestry institutions around the world demonstrates that when the users monitor each other's behavior in a forest, forest conditions are substantially enhanced (Coleman Reference Coleman2009; Coleman & Steed Reference Coleman and Steed2009; Chhatre & Agrawal Reference Chhatre and Agrawal2009).

I am glad to see these issues being raised in a way that makes it possible to move forward to a better understanding of the role of punishment in overcoming social dilemmas of various kinds.