Nathaniel Shar's Comments on the Press-Dyson Article on the Iterated Prisoner's Dilemma

While I agree that the article is interesting from a mathematical standpoint, I strongly disagree with the authors' use of the word "evolution", as well as their use of related words like "fitness landscape". I think it is badly misleading, and I hope it is not intentionally so.

It is clear that the authors are trying to create some connection between their findings on IPD and biological evolution (see, for example, the final sentence). But the connection is tenuous at best.

Evolution is the change in the frequencies of various heritable characteristics in a population over time. An individual cannot "evolve" in a biological sense. Yet there is no mention of populations, heritable characteristics, or frequencies of anything, in this paper; the only thing that changes over time is the behavior of an individual. The principal mechanism of biological evolution is natural selection (differential reproductive success), but neither reproduction nor selection is part of the model in this paper.

Instead, the authors use the word "evolutionary" to mean "changing in response to conditions", which is a nonstandard use of the word "evolutionary". Perhaps "adaptive" or "conditioned" would be a better term, but such a term would not so strongly imply a connection to biology.

Similarly, the term "fitness landscape" contains the biologically loaded word "fitness," which denotes the ability to survive and reproduce. There is no survival or reproduction in this paper, so the use of the word "fitness" is inappropriate.

While it is certainly no rarity for famous and well-known scientists, whether biologists or not, to demonstrate ignorance of evolutionary theory, I find it difficult to believe that widely knowledgeable and extremely smart people like Press and Dyson could really think that their use of the word "evolutionary" is properly descriptive. In particular, the authors cite Robert Axelrod's excellent book, "The Evolution of Cooperation", so we may presume that they have read it. Axelrod's computerized IPD experiments used a truly evolutionary model, involving populations, heritable characteristics, and so on, and he is therefore entitled to put the word "evolution" in the title of his book. After reading it, the authors have no excuse for not understanding what evolution is. And if the authors of this paper see their work as bearing upon evolution in the sense that Axelrod's work did, they are wrong.

I hope that they are merely wrong, because the alternative is to believe that they are deliberately misapplying the language of evolutionary theory in order to imply a connection with biology or with Axelrod's work. That would be much worse.

Experimental Math Spring 2013 class page