Re: META: Memes and War [was: Tomorrow is a new day]

From: Keith Henson (
Date: Fri Nov 12 2004 - 06:40:16 MST

At 08:36 AM 12/11/04 +0000, you wrote:
> > This might work out, but I am not entirely comfortable with it.
> > sounds good, but as we all know, "rational" does not cut it with the
> > prisoner's dilemma game.
>As we all know? Considering you're the one ranting on evolution, haven't you
>heard of the "evolution of altruism" and such?

(Google is your friend.)

Google: Results 1 - 10 of about 63 for "Keith Henson" "evolution of
cooperation" OR Axelrod

Google Groups: Results 1 - 10 of about 47 for "Keith Henson" "evolution of
cooperation" OR Axelrod

And if you look, one of the articles where favorably I cite Axelrod's book
is my 1987 "Memetics and the Modular Mind article in _Analog_. So, yeah,
I've heard about it. :-)

>Prisoner's dilemmas are
>analyzable by game theory and rational tools.

"Distributed Rational Decision Making

"Sometimes efficiency goals and stability goals conflict. A simple example
of this is the Prisoner's Dilemma game where the unique welfare maximizing
and Pareto efficient strategy profile is the one where both agents
cooperate, Table 5.1. On the other hand, the only dominant strategy
equilibrium and Nash equilibrium is the one where both agents defect."

(Dawkin's from Selfish Gene, Second Edition, "Nice Guys Finish First" chapter.)

"So I have worked out by impeccable logic that, regardless of what you do,
I must defect. And you, with no less impeccable logic, will work out just
the same thing. So when two rational players meet, they will both defect,
and both will end up with a fine or a low payoff. Yet each knows perfectly
well that, if only they had both played COOPERATE, both would have obtained
the relatively high reward for mutual cooperation ($300 in our
example). That is why the game is called a dilemma, why it has even been
proposed that there ought to be a law against it."

Above, I didn't say "iterated prisoner's dilemma's game" and I didn't say
you could not analyze it with rational tools. It would take more time to
find than I want to put into it this morning, but one of the studies I
read reported the way naive people play the one-round game is deeply
influenced by having the "rational" way to play the game explained to
them. (I.e., they defect.) That's my meaning of "'rational' does not cut it."

>As an aside note, a better algorithm than the typical tit-for-tat has been
>found a while ago, but you probably know about it already, although IMO the
>new algorithm is against the spirit of the thing.

That's really interesting. Does anyone have a pointer?

Keith Henson

PS. If you have not read this classic, it ranks high among the canonical
AI books.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT