Freeman Dyson wins the contest, and then says the contest is stupid

Institute for Advanced Study “The Prisoner’s Dilemma”

One of my favorite skateboarders when I was young was Natas Kaupas, an innovative skater who pioneered a lot of modern streetstyle. Natas was one of those skaters who could do things that no other skateboarders could, but he was not particularly successful in one arena that was important at the time: contests. Kaupas was once quoted as saying something like I just want to win a contest so that I can say that they are stupid. Compared to the spontaneous, groundbreaking skating that was taking place on California streets, skateboard contests were pretty stupid. But if you disrespect something you consistently fail at, you look like a sore loser. Once you stand on the podium, you can say that standing on the podium is meaningless.

Freeman Dyson, a famous mathematician, is doing something akin to what Natas Kaupas dreamed of. Dyson, along with William Press, recently published a paper in the Proceedings of the National Academy of Sciences that overturned our understanding of the Iterated Prisoner’s Dilemma (IPD). While IPD pioneer Robert Axelrod reached the conclusion that winning strategies are nice, retributive, and forgiving, Press and Dyson showed that strategies that are mean and manipulative can in fact “dominate” all other strategies. In the three-decades-old contest to come up with the best strategy for the IPD, Press and Dyson had won.

And shortly thereafter, Dyson started doing something interesting. Rather than taking credit for solving the universal mystery of human behavior, he instead decided to say that the whole construction of the IPD was stupid. Dyson has called the IPD a “toy”, seriously questioning the validity of game theory as a means of understanding the evolution of human cooperation. Given the very specific and unrealistic assumptions of games like the IPD, he makes the claim that they should not be applied to actual human evolution.

What is perhaps even more interesting is what Dyson believes can account for human cooperation: group selection. The IPD only allows for selection at the individual level — cooperation emerges from reciprocity, which can be optimal for individuals under the right conditions — but cannot represent selection on groups. In this article Dyson gives a very brief and unconvincing argument in favor of group selection as the driving force behind human cooperation. Drawing a comparison with the dodos of Mauritius, Dyson suggests that group properties rather than individual properties determine evolutionary success. This overly-simplistic explanation suggests that Dyson is unaware of decades of theoretical scholarship on group selection, much of which has rejected “naive” explanations of the kind he offers.

But whatever you may think of Dyson’s eloquence on the subject of group selection, you have to hand it to him: he has shown that one of the defining paradigms of individual selection is stupid. While pro-competition commentators went to town lauding Press and Dyson for showing that selfishness, manipulation, and extortion win at the IPD, most people overlooked one key fact: human populations are not overrun by selfishness, manipulation, and extortion. By showing that the IPD can be won by a purely selfish player, what Press and Dyson truly showed was that the IPD cannot possibly account for the level of cooperation we observe among humans. In other words, reciprocity and individual selection are inadequate explanations of cooperation. Something else must account for human cooperation; that something else seems like it could be group selection.