An experiment in game theory: Part I

So, last night I performed an experiment.  I wanted to test out whether Robert Axelrod's use of game theory in explaining the evolution of cooperation worked in the modern world of social media.

Axelrod has successfully shown that how we structure interactions with each other determine how effectively cooperation develops.  Cooperation is viewed pretty expansively.  I particularly like the description of the "Live and Let Live" phenomenon that existed in trench warfare during WWI.

The standard situation in game theory is the Prisoner's Dilemma.* The upshot of the game is that both parties benefit from mutual cooperation (e.g. live and let live) but one party could do better by exploiting or abusing the other.  Repeated experiments have shown that all parties do better in the long run by cooperating, but each individual has the short term incentive to exploit the other. Axelrod provides four "suggestions" for promoting and sustaining cooperation, or non-exploitative, behavior.

Don't be envious. Axelrod argues that you shouldn't compare how well you are doing to how well someone else is doing.  Instead, you should compare how well someone else could be doing in your shoes. The main lesson here is that you do not have to do better than others to do well for yourself.  

Be nice.  Here, Axelrod asserts that your default behavior should be to cooperate, especially if you are going to interact with that other person again. What's most important about this is that always starting off in the adversarial position creates an environment in which cooperation becomes a losing strategy.  If you start off fighting, you'll be fighting all of the time.

Reward good behavior and punish bad behavior.  The strategy that works best in the Prisoner's Dilemma is called Tit for Tat.  Basically, it means that when other people are nice, be nice right back at them.  But when they exploit or abuse you, you punish them.  When they start cooperating again, you forgive.  How much forgiveness you give depends on the environment. The willingness to forgive is itself exploitable unless you use a one to one reward/punishment strategy.

Don't try to outsmart other people.  People respond to your behavior. If good behavior is rewarded and bad behavior punished, as Axelrod says, "...your own behavior is likely to be echoed back to you."

I have become curious about whether the structure of social media promotes cooperation or exploitation. People have written that at the same time that social media allows people to connect across space (think about friends or family across the world that you regularly interact with on Facebook), it also engenders a form of social distance that allows people to say things in social media that they would not say in a face to face situation.

In the next posting, I will describe the experiment and my lessons learned.

*In this game, two burglars are caught at the scene of the crime and interrogated separately. If both burglars cooperate with each other and keep their mouths shut, they would each spend only 1 year in jail. However, if one decides not to cooperate and implicates the other, he will not go to jail at all.