[MUD-Dev] Star Wars Galaxies: 1 character per server
hart.s at attbi.com
Wed Feb 12 00:47:22 New Zealand Daylight Time 2003
> If, in the first game, Player 2 does not know that Player 1 is
> restricted in strategy, then Player 2's ideal strategy does not
> change when Player 1's restriction is lifted. And, in the second
> game, if both players can know the other's strategy, then the game
> reduces to the trivial.
I assume you mean "either in the first game 2 doesn't know about 1's
restraint, or he does and the game is trivial." My mistake, I meant
iterative prisoner's dilemma, in which case differentiating 'always
cooperate' from 'tit for tat' or 'always defect,' for example, only
takes a single 'defect' (though, of course, it is possible to have
all kinds of strategies which evade a single defect, as tit for 2
tats, and detecting those will take more sampling, etc. etc.)
> I am interested in how you see the ideal strategy shifting away
> from 'always defect.'
Clearly, 'always defect' is effective in a population of only-
defectors, only-cooperators. With tit for tats, the 'always defect'
player will consistently earn the lower mutual defection bonus
rather than the slightly higher cooperative bonus. Even if he
defects and the other cooperates once, the overall disadvantage,
depending on the makeup of the population and what is actually being
exchanged, may be great.
This is hair-splitting. I recognize that a MUD isn't prisoner's
dilemma, I was just reaching for a way to communicate an idea about
why it is not always better to give one person full protection from
the other in some way than not to protect either in the same way.
MUD-Dev mailing list
MUD-Dev at kanga.nu
More information about the MUD-Dev