Thursday, October 14, 2004

Ding! Dong! 'Tit for Tat' is Dead!

Tit for Tat has been defeated in the Prisoner's Dilemna problem! This is a huge deal, as Tit for Tat has been used to sculpt things as disparate as national security and evolutionary theory of cognitive development. This may also signal a tidal shift in the importance of social software and group psychology going forward.
via [slashdot]

2 Comments:

Blogger Cote' said...

I'm not math man...but reading the explanation, I can't help but feel they're gaming the system. (Is that a pun?) I guess that's allowed, but being able to cooperate with other of your own versions of the game seems fishy.

Then again, taken on it's own (esp. in the context of agents, like the South hampton people work on), it seems pretty damn smart.

2:25 PM  
Blogger egoodwin said...

Yeah, they are gaming the system. But in the context of the competition it is allowed. Basically, the point of it is that Tit-For-Tat works best in cases of individualized agents, but in cases of cooperative agents (such as price fixing in oligopolies) there are cases where you can beat Tit-For-Tat. For example, a case where a clan related by blood (ie same genetic code, therefore willingness to sacrifice self for the good of the team) versus a cartel related only by business interests. In the case of conflict, given enough entities, the social networked strategy could be allowed to dominate. Of course, this may only work if the other players don't know about your cooperative. Still, I think there may be some big potential here, especially in the context of the internet.

5:54 PM  

Post a Comment

<< Home