[net.math] Iterated prisoners dilemma

rlh@cvl.UUCP (Ralph L. Hartley) (09/24/84)

There is a strategy that I think can do better than tit-for-tat
(tit-for-tat cooperates int the first round and in all other rounds
does whatever its partner did in the last round).

I call my strategy parting-shot. It plays as follows.

All moves but the first and last
	- do what the opponent did in the last move
First move
	- cooperate
Last move
	- defect

Regardless of the opponent, parting-shot always gets a better score
than tit-for-tat.

				Ralph Hartley
				rlh@cvl
				siesmo!rlgvax!cvl!rlh

bae@fisher.UUCP (The Master of Sinanju) (09/26/84)

>	There is a strategy that I think can do better than tit-for-tat
>	(tit-for-tat cooperates int the first round and in all other rounds
>	does whatever its partner did in the last round).
>
>	I call my strategy parting-shot. It plays as follows.
>
>	All moves but the first and last
>		- do what the opponent did in the last move
>	First move
>		- cooperate
>	Last move
>		- defect
>
>	Regardless of the opponent, parting-shot always gets a better score
>	than tit-for-tat.
>
>					Ralph Hartley

Unfortunately, the rules of the game specify that competition shall continue
for a random number of rounds, so it is not possible for your program to 
determine when the 'last move' will occur.

-- 
            Brian A. Ehrmantraut

    {ihnp4, twg, decvax, ucbvax}!allegra!fisher!bae