After the second homework assignment, faurik commented on the question referring to an iterated prisoner’s dilemma. Specifically, faurik explained how in the one-shot game, the Nash Equilibrium of both players confessing does not necessarily hold in this new game. The observation arises because players realize that the game repeats for an unknown number of times, so working together might payoff. In this iterated version, then, does “playing nice,” contrary to the one-shot Nash Equilibrium, actually payoff?
The director of Harvard’s evolutionary dynamics lab, Martin Nowak, examined that question in his research project, published in tomorrow’s edition of Nature. Seth Borenstein provides his own take on the study in this article from The Boston Globe.
For his study, Nowak enlisted 100 Boston-area college students to play a prisoner’s dilemma game repeatedly with dimes, following the payoff matrix below:
He then added a twist, a potential punishment to uncooperative players: Any player could punish a player who chose to defect by 40 cents, but this punishment cost the player a dime to enact. Interesting. What were the results? “The players who punished their opponents the least made the most money.” So, perhaps the common proverb that nice guys finish last isn’t quite correct. To see if these findings play out in real-life scenarios, Nowak plans to study chief executives’ behaviors.
I think the article misses a key point here. The paper published is a negative result (and a repeat of prior negative results).
One of the big problems in game theory now is trying to understand punishment. It’s obvious that punishment occurs. However, being the punisher does not seem to be a NE strategy for an individual. So why do people do it? This, like many other games, shows that individuals shouldn’t punish– still leaving us with the question of why it happens.
(I’m rather surprised this made Nature, since this is pretty well known already).