Talk:Prisoner's dilemma/Archive 1

Archive 1

Archive 2

Archive 3

→

Archive 5

Would it be worthwhile to mention some of the simulations/games (like Core Wars) that tended to show that tit-for-tat strategies usually were more successful in the long run? --loh (2001-06-19)

---

The title of Axelrod's book serves as a link to a description of his simulations that will be filled in at some point. --LDC

---

Or how about the movie WarGames, which showed (rather pedantically) that sometimes they aren't? ;-) --KQ

---

Could somebody add about how iterative PD depends on exact values of game matrix ?

One-time PD work for every matrix like this:

   - + 
-  a 1
+ -1 b
Where -1<a<b<1

Iterative PD requires additional assumption that a<0, because if a>0 then switching-cooperation is the best algorithm (you achieve more with not cooperating than opponent loss from it, so it's best if you both switch, not cooperate)

Are there any other dependencies ? --Taw

---

I don't understand the reference to Core Wars. That is a competitive game, where the objective is to destroy the opposing program. To paraphrase a popular movie, "two programs enter, one program leaves".

Also, it should be noted that in the original Prisoner's Dilemma, communication between the prisoners doesn't really help much. Even after agreeing to cooperate, each individual may still want to defect and get the highest possible outcome (going free). We should probably start up a whole new topic on Game Theory. We also need to work in references to rationality and superrationality. -- ansible