Re: Standard definition

From: Bill Spight (bspight@pacbell.net)
Date: Wed 30 Oct 2002 - 16:35:07 GMT

  • Next message: Grant Callaghan: "Re: Standard definition"

    Dear Derek,

    > Bill:
    > In the Iterated Prisoner's Dilemma the performance
    > is only cooperation
    > or defection. Neither TIT-FOR-TAT nor any other
    > strategy can be
    > described merely in terms of performance.
    >
    > Derek:
    > Yes it can, you just need enough performances to
    > deduce the the strategy. How else could you infer the
    > strategy of somebody who refuses to disclose it?

    In my just previous reply to this note, I brought up infinite sequences. Let me clarify what I had in mind.

    First, there are stategies that can be described in terms of one's own performance. For instance, alway defect. So I overstated the case.

    Second, I meant one's own performance, but I did not specify that, and moved away from that in my reply. Too early in the morning, I guess. My point is that a strategy is a condition-action pair. Performance is only in the action. If Wade intends to include conditions as well, fine. I think he should. But then there are better ways of describing the matter.

    Third, if we include the other's performance in the conditions, as TIT-FOR-TAT does, then we can apparently define the strategy finitely purely in terms of performance:

         {C}
         {cC}
         {dD}

    I.e., cooperate the first time, then do what the other did last time. But this definition depends upon interpretation, where the small letters indicate what the opponent did last time and the capitals indicate what to do *every* time the conditions are fulfilled. We can dispense with this interpretation by making the 'every' explicit, and that takes infinite sets.

    Best,

    Bill

    =============================================================== This was distributed via the memetics list associated with the Journal of Memetics - Evolutionary Models of Information Transmission For information about the journal and the list (e.g. unsubscribing) see: http://www.cpm.mmu.ac.uk/jom-emit



    This archive was generated by hypermail 2.1.5 : Wed 30 Oct 2002 - 16:38:12 GMT