Subject: Re: Exploring the limits of free software: Cygnus, and GPL
From: "Stephen J. Turnbull" <>
Date: Thu, 27 May 1999 03:18:11 +0900 (JST)

>>>>> "Jean" == Jean Camp <> writes:

    Jean> Don't be a snob, economics is not brain surgery.

I disagree with Jean's implicit evaluation of the difficulty of
economics, but I'm taking that offline.  More important, the model
Jean has in mind (I don't know exactly what it is) cannot possibly be
the one I'm talking about.  Here's why:

    Jean> Anyway by defintion FS is a public good, but the reasonably
    Jean> predicted lack of supply does not occur. To me this does not
    Jean> imply that economic models will never fail. ;->

As far as I know no serious economist has thought carefully about the
issue.  Not published, anyway.  Certainly, the obvious analogies (the
semimythical lighthouse, for example) suggest market failure here.
However, production in all the classical examples involves work; in
the case of software, it is play.  (Forgive a little poetry.)

Normally we economists ignore this play motive because of exclusivity
of consumption; as you point out, this fails for software, and thus
the play motive need not fizzle out within a few hundred meters as it
does for hobbies like gardening, or even running a lighthouse.  A
software hobby can (and regularly does) reach and benefit the whole

About one in three of my colleagues saw the point immediately at a
recent workshop.  None of them think it's more than a curiosity at
this point, but that does prove that economists can be open-minded.

The issues that remain are, well, everything: how does the play motive
interact with the profit motive?  Do viral terms generate more total
value in the long run than non-viral terms?  Do viral terms
_necessarily_ generate more free software (value) than non-viral terms
in the long run?  What other "tiny" motives in addition to pure "play"
might be leveraged by the public good nature of free software?  Etc,
etc; all of these will have bearing on the design of mechanisms
(licensing terms, employment contracts, and marketplaces) to encourage
production and improvement of free software.

    Jean> Economics would predict that free software, and therefore
    Jean> the resulting liberties, are underprovided,

Not as a theoretical certainty for a useful definition of optimal
provision, although I suspect that would be the case for most
plausible values of the key parameters.  I think it's downright
obvious that free software is vastly underprovided compared to the
first-best; in that best of all physically possible worlds, all
software would be free and we'd have more of it to boot.  But that
world isn't going to happen, period.  So the standard of comparison
must be a lot more subtle---and upon reflection, I'm not sure I have a
clue how to define an appropriate standard right offhand.

    Jean> that equivalent proprietary software would dominate.

The relevant terms need definition.  All I can do is make some general 

Equivalent, I won't touch.  But in aggregate terms, it's quite clear
that proprietary software dominates in any quantity measure you care
to mention, unless you restrict the definition of "software" in a way
I would find unacceptable.  (More revenue, more copies, more users,
more bugs.... ;)  The simple-minded "Econ 101" model is consistent
with this, but it is quite useless (especially now that Jean's
finished shooting holes in it) for further analysis.

While I don't think "more is better" is the end-all of social
analysis, it's a convenient, easily understood zero-th order
approximation, and it is a non-negligible component of both profit and
social welfare.  So in your definition of "dominate", you'll have to
be careful to account for it (or why you ignore it).

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."