Re: help on strategy for a game????

Parth wrote:

> Wayne Rasmussen <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> > "Randolph M. Jones" wrote:
> >
> > > Parth wrote:
> > > > the game is as follows:
> > > >
> > > >                  Agent 2
> > > >
> > > >                 C       D
> > > >              ________________
> > > >             |       |       |
> > > >           C | 3, 3  |  1, 5 |
> > > >             |_______|_______|
> > > > Agent 1     |       |       |
> > > >           D | 5, 1  |  0, 0 |
> > > >             | ______|_______|
> > > >
> > > >
> > > > The game is to be played 1000 times against another agent... No fixed
> > > > strategy seems to work against all other fixed stategies... suggest
> > > > some cooperative mixed strategy for this game please.
> > >
> > > For extra credit, name this famous dilemma.
> > >
> > > Is this homework from your philosophy class or your AI class?
> >
> > Prisoner's dilemma.  I remember that the tit-for-tat algorithm was best
> > (back in the 80's) if you didn't know the exact number of turns.  IIRC,
> > someone had a slightly better than the tit-for-tat but can't recall the
> > details.   But the contest is fair when you play against all the other
> > agents (one at a time) for the 1000 times then total the scores for all the
> > runs.  If someone has a bad agent and you are stuck against him, you could
> > be screwed.  So everyone has to play everyone else.
> >
> > Now if the teacher has his own code for the other agent, he is likely using
> > tit-for-tat.  If you know for sure that exactly 1000 times is used, then on
> > the last one you play D instead of C.
>
> No the game is not prisoners dilemma... the payoff of 1 in the matrix
> cells - [0][1] and [0][1] make it different from prisoners dilemma.

It is the prisoners dilemma.  Show me a credible source, other than yourself, that 
this is not a problem of that
class.