A musing on The Evolution of Cooperation

[Mark Graber]

Robert Axelrod's THE EVOLUTION OF COOPERATION was one of my favorite books when I was in graduate school.  One point the book made was that "tit for tat" strategies are problematic.  The basic feature of a "tit for tat" strategy is that I cooperate in a prisoner's dilemma game until you defect (fail to cooperate).  i then punish you by defecting in the next round and continue to defect until you begin to cooperate.  The problem is as follows:  Suppose I misperceive your action as a defection (non-cooperation) when you really intended to cooperate.  I will defect the next round.  Of course, if you are also using a "tit for tat" strategy,you will respect by subsequently defecting and we will get into an endless defect cycle, which provides the worst payoffs for everyone.  May I suggest two good lessons this work might teach.

1. Where cooperation is needed, persons should strongly presume that others are cooperating and perceive their posts in that light.  A cooperative endeavor, such as a conlaw list serve, will function better if occasional defections are not punished, then if cooperative posts are accidentially punished.

2. Punishments should be short-lived.  if you believe this post violates list etiquette rules and must be punished, respond with a devastating "Graber is a rude idiot" that you believe will convince all reasonable readers.  I, of course, will unconvincingly respond that you are the rude idot.  Still, presume your first defection did all the damage and return to cooperative mode, not responding to my defection.

Axelrod shows, in my view, that society's function best when these rules are followed.  I think they are good ones.  At any rate, we might have a more interesting conversation on the structure of conversation and cooperation than more recent issues.

MAG