Talk:Introduction to Game Theory/Prisoner's Dilemma

The classic dilemma is not a confession, but rather blaming the other prisoner

The numbers used are poor, clearly they work to create the same outcome, but both confessing, and only one confessing both result in 10 man-years being served. In order to make the outcome counterintuitive, the price of a double confession should be 6 or more, but less than 10 years each. This way, the Nash equilbrium is actually the worst possible outcome in terms of total man-years served. Further more, the idea is not to pick based on your opponent's choice being unknown and therefore using an average case, but to assume that your opponent is also selfish, and figure that he is for sure going pick assuming your selfishness as well.

I have gone through and changed Person A to Andy and Person B to Bob to make it easier for someone to read. I've also followed the suggestion above and increased the jail term for a double confession to 7 years apiece. I suggest that the part on expected jailterm is removed, as it isn't required at all, and I think confuses the matter. On the confession matter, from an aesthetic point of view, the problem would 'sound' better if it was plead guilty or not guilty.--MarkyParky 23:47, 25 August 2006 (UTC)

I agree that it should definitely not be a confession. This doesn't make any sense, if one confesses to a crime one is then charged with the crime and served a correctional sentence. However, it is traditional for members of the judicial/correctional system to offer a reduced or rescinded sentence for a crime committed if he testifies against his co-conspirator codefendent. I believe that instead of person A confessing, person A should "rat" on person B. Perhaps narc may be another possible word.