Prisoner's Dilemma

John K Clark (johnkc@well.com)
Wed, 2 Jul 1997 20:08:09 -0700 (PDT)


-----BEGIN PGP SIGNED MESSAGE-----

On Mon, 30 Jun 1997 Michael Butler <mmb@best.com> Wrote:

>I maintain that being able to determine the counterpart's past
>dealings with you is pretty darned close to "recognizing" the
>counterpart.

You must know how to contact me, but you don't need to know my physical
location or know what I look like. You must know one of my names, but I could
have others. You might have dealings with 2 names with radically different
reputations and not know and not need to know that both are me. I might make
part of my money honestly and it's important that I have a good reputation
for that name so that people will know by examining my history that if make a
deal they will treated well. I might also make money as a flim-flam artist
selling perpetual motion machines, but I use a different name for that.
As long as you make a deal with the first name you're safe because it's
to my interest to maintain a good reputation for it.

On Tue, 1 Jul 1997 Kennita Watson <kwatson@netcom.com> Wrote:

>[OVERSIMPLIFICATION ALERT]
>In a population of 10 people, if 9 play tit-for-tat and one always
>defects, the 9 will be defected against 10% of the time, and the one
>will be defected against 100% of the time. This isn't useful for
>producing descendants.


The tit-for-tat strategy is to always cooperate on the first transaction and
then do what your counterpart does. If the population is large then most
transactions will be the first encounter with an individual. This too is an
oversimplification of course.

John K Clark johnkc@well.com

-----BEGIN PGP SIGNATURE-----
Version: 2.6.i

iQCzAgUBM7sV3n03wfSpid95AQFW1QTvVLnuWzsaNZ+EVNHyEq72PjczIsv8E743
t3PXZlTJUCRnMxWrO95Hl06dUpdyUyY5SprUfAqLbA919Y00xIOu68mE8fw3nJLE
Buwue/e92R8zXpaQfclbV7XKw+LtdV/UaO8d0ZLw6CdwI4FWW3GmL3fbDw3npQyA
mhRLdlDgw3xSUM1SSs5SEnstT/E7ZiCYz50V+WoeBgns1ERGEgg=
=RVdE
-----END PGP SIGNATURE-----