Societal Iterated Prisoner's Dilemma
The Prisoner's Dilemma is a situation in which two people must choose to either 'cooperate' or 'defect' on each other. Regardless of the other person's choice, choosing to cooperate increases the combined benefit and choosing to defect increases one's personal benefit. Such situations thus have the unfortunate property of making it in each player's own interest to defect and thus produce the lowest possible communal well being.
The Prisoner's Dilemma has been extended to look at sequences of such interactions (the so-called Iterated Prisoner's Dilemma). In this case, there are many more strategies available, because you can take the other player's past actions into account when deciding what to do. A particularly important consequence of this is that cooperation can be in a player's own interest, since you have to meet the other player multiple times and suffer the consequences of your actions. The Prisoner's Dilemma has also been extended to situations with multiple players and communal interactions. Extensive details on these and other variants, can be found in the Stanford Encyclopedia of Philosophy.
In 2001 I further extended this model to include public knowledge of the other players' past interactions. This gives a very interesting model in which players have detailed reputations. It also leads to many interesting implications for the development of cooperation in such groups. For instance, players can use this information to defect on 'soft targets' (those that have not retaliated when others have defected upon them). Alternatively players could use the information to punish those who defect on players that have always cooperated — a kind of 'peace-keeping'. There are thus ways in which this public reputation encourages defection and ways in which it encourages cooperation.
I have named this model the Societal Iterated Prisoner's Dilemma (SIPD) and wrote a paper on it called 'Exploitation and Peacekeeping: introducing more sophisticated interactions to the iterated prisoner's dilemma'. I presented this paper at the World Congress on Computational Intelligence in 2002 and it was well received. It is quite readable for the non specialist while containing interesting ideas for those studying game theory, ethics or economics. The research was supervised by Alan Blair who provided much valuable advice and encouragement.
There have been several important developments since this paper was written and I have summarised them in an article on my unpolished ideas page. This summary also includes a link to a zipfile containing the source code of my program and all the data used in the paper.
This work has been quite exploratory in nature and is not the most precise work I have done in ethics, but I think the model is natural and important and so hope that others will take it further.