craig@usinset.inset.com (Craig Hubley) (02/07/90)
(somehow a prior copy of this escaped without a body - don't worry, I doubt this group will win any 'newsgroup volume bandwidth pig awards' anytime soon!) I am interested in cellular automata where the cells engage in exchanges with each other - and the rewards/survival reflect their degree of success at dealing with each other (sort of a neighborhood simulation). Axelrod wrote about this sort of a game in "The Evolution of Cooperation" using the Prisoner's dilemma inequality (reward for defection > reward for dual cooperation > reward for dual-defection > reward for cooperation when the other defects (sucker's payoff)). There were some rules for cells to adopt a neighboring strategy when it was successful - there was a distinction between the rules that scored best, and those that best convinced their neighbors - for instance Tit for Tat (cooperate first turn, thereafter do whatever the other did on the previous turn) was the best overall strategy, but since it can never score better than any individual opponent, it is never perceived as 'winning' by an individual (which was required by the conversion/growth rule). I am wondering if anyone else has done anything with cellular automata that hav such detailed interactions with each other, and if their work is in print anywhere. Thanks, Craig Hubley