Subject: Re: New ESR paper: The Magic Cauldron
From: "Stephen J. Turnbull" <>
Date: Fri, 25 Jun 1999 17:28:34 +0900 (JST)

>>>>> "Craig" == Craig Brozefsky <> writes:

    Craig> "Stephen J. Turnbull" <> writes:

    >> But the "free rider problem" objection is everywhere, and you
    >> get that one wrong.

    Craig> Certainly there are users of any given peice of software
    Craig> which do not submit bug fixes or reports.  So, yes, there

I'm sorry, I should have been more careful.  A person is a free rider
only if the aggregate (over all users) value of their potential
contribution is greater than the personal cost of making it.  It's
probably necessary that they _be_ a user; in the context of a pure
public good that point is moot (if everybody isn't a user by
definition, it's not a pure public good by definition), so it's never
been made explicit that I know of.

    Craig> are "free riders" here.  We are all "free riders" to some

That's my point.  "All have sinned and fall short of the glory of
God."  Unlike the Apostle, however, I consider this not to be human
perversity but an inefficiently designed system.  I have no bad words
for "free riders," only for the system that achieves less than it
could were it to harness their energies voluntarily.

    Craig> extent.  Having never submitted a bug report for emacs
    Craig> myself, or gcc for that matter.  But I fail to see how
    Craig> there is a "free rider problem." because noone suffers when
    Craig> I do not contribute to every project I use, or any project
    Craig> for that matter.

If you can report and/or fix a bug before it bites others, and those
others in aggregate would be willing to pay you enough to make it
worth your while to do so, and you do not, for whatever reason, there
is a free rider problem.  "Free rider" is not a problem of malice or
ethical incompetence, it is a failure of coordination.

    Craig> Noone suffers

Depends on how you define that.  I define "suffering" as "foregone
opportunities to exchange values to mutual advantage."  There is
definitely a lot of suffering going on in today's software market by
that definition, and much of the reason for that is that noone has
figured out how to make free software work on MS-Windows scale.  The
flip side of the inefficiency due to monopoly via IP is the free rider
inefficiency when there is no IP.

    Craig> It seems that it is made a problem only when you introduce
    Craig> a "total value" for the project, which it seems you derive
    Craig> as some possible maximumization of a set of equations, a
    Craig> game-theoretic acme of productivity.  But this in itself

Yup.  What's wrong with that, may I ask?  I don't demand perfection
(in the sense of your "acme"), but if I have the opportunity to
contribute to getting closer to it, I will do so.

    Craig> does not explain why the user who does not report a bug is
    Craig> counted as a deduction here, but the Chinese teenager who
    Craig> never heard of the project is not counted as a possible
    Craig> contributor when settling up the karmic software debt.

The Chinese teenager _is_ counted as a deduction; whatever gave you
the idea that she isn't?  However, in the case of that Chinese
teenager the immediate problem is the lack of coordination that keeps
her from hearing about the project, not the lack of coordination that
keeps you from selling bugfixes.  Since she makes no decisions, she is
not a free rider.  So I didn't mention her.

    Craig> The determining factor of what is counted as a deduction
    Craig> from the possible "total value" appears to be that the
    Craig> "free rider" is a problem because they benefit from the
    Craig> software, at the imaginary expense of all of the
    Craig> contributors.  The Chinese teenager does not benefit, thus
    Craig> we do not deduct their enjoyment in our ledger of
    Craig> suffering.

I didn't write any such thing, and it's not true.  The problem of free
rider _in economics_ is not, at root, an ethical one of benefit at
others' expense.  It is a problem of making a contribution less than
"one's fair share" where we can all agree that "fair contribution" is
non-zero if benefit is non-zero.  The point is that if you can find a
way of collecting and aggregating those fair shares, the project's
benefits to all users will increase, typically very large multiples of
the aggregate contribution in many realistic free rider situations.

It's not a problem of deduction, it's failure to add.

BTW, if you want to deny that the fair share is greater than zero,
that's fine by me.  Do you plan to demonstrate how you have repealed
the Laws of Thermodynamics, or to explain why some people ethically
must bear a positive burden while others' ethical burden is zero?
I'll happily take the con side of either debate.

    Craig> So, it appears the the problematic "free rider" is the
    Craig> phantasmatic other who enjoys at our expense, who denies us
    Craig> access to a completely imaginary state of perfect
    Craig> productivity.  Sucking out our bodily fluids even.

No, the problematic "free rider" is one who having killed a rat won't
throw the rat's ass onto a farmer's field where it serves as fertilizer.

    Craig> It's not a problem.  RedHat is growing and large and SuSe
    Craig> is just as big, and Debian gets bigger by the day as well,
    Craig> and all of them have their CDs reproduced by CheapBytes.
    Craig> The companies are supporting 270+ employees, helping them
    Craig> pay their mortgages or pay of their college debt, and maybe
    Craig> eat sushi now and then.

I see.  You consider that Red Hat, SuSe, and Debian are obviously
_exactly_ big enough now?  Suppose they were given the resources to
create WIWAOL (WIWAOL is Windows apps on Linux, pronounced "we wail"),
the Windose equivalent of GNU?  Given the resources necessary, I bet
they could, and I bet they would.  Isn't that a better world?

I don't know that the necessary resources are out there, but then
neither did the dry land ecologists on Dune.  Isn't it worth doing
exploratory analysis to try to find them?  When we find them, we can
decide whether the policies necessary to assemble them can be
justified in the context of free software; it's not clear to me that
they can't.  That latter question is the core topic of FSB, of course.

I don't know that they're not available, either.  Free rider problems
can lead to big losses.

A quick estimate would be the value of Microsoft == the value of IP in
Windows and Word == the resources forgone because the open source
process has not produced Windows and Word (but surely can easily
produce something of equal value) == resources forgone due to free
rider problems in OSS.[1]  And that's just a start, because we haven't 
yet factored in the net contributions that would be elicited by the
opening of the source, or the additional benefits due to somewhat
lower prices of computer systems.

Note: Microsoft provides a lot of services to buyers (please, no
MS-bashing responses, I know about the gross failures too) not
accounted for above, which must be debited from the above estimate.
But which of you wouldn't jump at an ethical chance for a sum of
investment even two orders of magnitude less than the number above?

    Craig> Can you explain why the 'free rider' is a problem and not a
    Craig> benefit of digital reproduction and Internet distribution
    Craig> scales?

Because those factors haven't repealed scarcity, they've just pushed
it back to the development stage.  Or are you convinced that all the
useful software has already been written?  Programmers don't live by
bread alone, but they still have to eat.

[1]  How does real Mentat computation grab you?  I'm not sure that
that computation is correct, but it's worth checking.  May take a
while, sorry.  Bug reports welcome!

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."