Subject: Re: the Free Software Movement in Industry
From: Ian Lance Taylor <ian@airs.com>
Date: 24 Aug 2001 13:44:10 -0700

"Stephen J. Turnbull" <turnbull@sk.tsukuba.ac.jp> writes:

> All that's going on here is the ancient argument about the prisoners'
> dilemma.  A says, "The dominant strategy for all players is to fink,
> therefore the solution is `all fink'."  And B says, "But that sucks
> socially, so it can't be the answer."  Life's a bitch, Mr. B, face it.
> 
> The fact is, that _is_ the answer.  "So, Doctor, what do we do about
> it?"  "Well, if it hurts when you do that, stop doing that."  We have
> to stop playing the prisoners' dilemma.  The question is, "what
> alternative game can we agree to?"  And, "how can the rules we set be
> enforced?"  That's what Seth is asking.

A one-shot prisoner's dilemma is actually resolvable if you assume
that everybody will think pretty much the way you do.  As I recall,
Hofstadter described this as ``super-rational'' (essay collected in
his book ``Metamagical Themas'').

In any case, a one-shot prisoner's dilemma is rare.  For one thing, in
the real world people are permitted to discuss their decision.  Also,
companies normally play a repeated prisoner's dilemma.  And the
results for a repeated dilemma are quite different.  See Robert
Axelrod's ``The Evolution of Cooperation:''
    http://ishop.wordsworth.com/bin/bkdetails.asp?isbn=0465021212&sessionID=ww2475191175616

For example, nuclear war during the cold war period is easily
understood as a repeated prisoner's dilemma, and there was some
literature analyzing it in those terms.  And, in fact, we never did
have a nuclear war.


Anyhow, I don't think writing everything as free software is a
prisoner's dilemma, because it's not obvious that, e.g., everybody is
better off if telco billing system is free software.

Ian