Subject: Re: Open letter to those who believe in a right to free software
From: "Stephen J. Turnbull" <>
Date: Thu, 28 Oct 1999 13:11:04 +0900 (JST)

>>>>> "Ian" == Ian Lance Taylor <> writes:

    Ian> I'll just note that I personally disagree with that ethical
    Ian> assumption as stated.

Stated where and when?  You mean the "elephants are not community
members" assumption?  Or the "activities which tend to destroy
communities in the long run, although they can't do it quickly, are
immoral" assumption?  Or the "`the community' is an emergent
phenomenon of the web of interpersonal community bonds" assumption?
Or the "preservation of `the community' is a positive ethical value"

I'm not being tendentious; I'm trying to show directly why it is
useful to make ethical assumptions explicit.

    Ian> Perhaps I made a poor choice of words.  By ``economic
    Ian> arguments'' I meant to include the assumptions which underlie
    Ian> the model which supports the logic.

    >>    This is extremely unfair to people like myself....

    Ian> I don't see why it is unfair.

Simply because people reject my conclusions which are often based on
ethical assumptions reasonably similar to their own, because the
conclusions look like those derived by people with rather different
ethical assumptions.  Not because they understand my assumptions and
logic and disagree with one or the other.  A combination of "post hoc"
and "ad hominem" fallacies, I guess.

    Ian> I only entered this conversation because you appeared to be
    Ian> arguing that it is possible to build an economic model which
    Ian> is ethically neutral.

    Ian> I don't believe that that is possible.

Can mathematical analysis be ethically neutral?  I believe it can.  An
economic model is nothing but a mathematical analysis.  The science
comes in in tying the measurable variables to the real world.  An
economic model can be correct analysis but terrible science.  (Most
game theorists successfully avoid thinking about this.)  The ethics
comes in in tying the objective functions to ethical values; an
economic model can be good science, in a value-neutral sense, but
ethically disastrous.  (Many economists successfully avoid thinking
about this.)

For example, some historian once argued that because England's laws
prohibiting theft became more and more strict, England was evidently
becoming more law-abiding over time.  This is very bad economics, as
science.  An accurate economic model of thief behavior would balance
the thief's perceived gain from successful theft against the possible
loss from getting caught.  Thus the point of more strict punishment is
evidently more effective deterrent.  It would seem then, that
according to economic analysis England was in greater and greater need
of a deterrent, possibly because theft was becoming more common.  In
fact economic history of the period showed that the bottom of the
income distribution was becoming poorer and that further such laws
were correlated with recession periods, making that possibility the
more likely conclusion.

I don't see where ethical content enters this model of theft, although
I have heard people state without serious justification that doing
economic analysis of theft is immoral (usually on a security-through-
obscurity-like rationale).

Now the question of what deterrent is appropriate would arise.  One
way to compute this economically would be to determine the deterrent
which has maximum capability to deter.  Evidently (there being no
appeal-all-the-way-to-the-Supreme-Court costs at that time), capital
punishment is maximal.  There still is no ethical content; nobody has
advocated anything.  This is merely a possibly useful lemma.

It is only when we cross over to advocating capital punishment that
ethical content enters.  The assumption that maximizing deterrence is
an appropriate goal, implicitly that the thief has no rights and the
thief's welfare should not be considered is essential to the inference
that capital punishment for a misdemeanor is good policy.  This is
simply evil.  However, the analysis itself was value-free, a simple
set of answers to a series of related what-if questions.  It is only
in combination that the implicit assumption, the analysis, and the
inferred policy recommendation are ethically horrible.

Evidently this example contains implicit assumptions that make it bad
science as well.  But without removing the assumption that the thief's
welfare is not to be considered, the qualitative conclusion,
advocating punishment that under most people's ethical values will be
excessively cruel, would be upheld.

    Ian> Since the reasons I work on free software have nothing to do
    Ian> with economics, an economic model will not tell me whether or
    Ian> not to work on free software.  I've already decided to do so
    Ian> in any case.

You are misconceiving economics, then.  Do you work on free software
because it helps to achieve your personal goals, more so than working
on proprietary software would?  If so, you are ruled by economics; a
scientifically accurate economic model of your behavior would predict
that you will work on free software.  If you work on free software
randomly, for reasons you do not understand, then no economic model
would predict your behavior.  But I doubt you would claim that.

    Ian> I think I can understand why my position would be frustrating
    Ian> to you.

Your position is not.  I have no desire to convince anybody to not
work on free software; despite my beliefs about the social results, I
believe every individual, everybody else doing as they do currently,
would be better off working on free software than not.

My personal beliefs aside, no economist in his right mind advocates
that individual humans use economic models for decision-making; "just
do it" is _much_ more accurate.  That economists do so advocate is a
layman's misconception.  Some economists are foolish enough to
recommend economic models to firms, but the operations researchers do
such modeling so much better it's pointless.

I do think that there are tradeoffs between personal spiritual health
and social material wealth, and that people who have very low degrees
of hacker-consciousness probably would not benefit from working on
free software, and that bringing material benefits to them, if
sufficiently large, is an important goal that justifies non-zero loss
of hacker spiritual health, if sufficiently small.  The point of
modeling individual behavior is to make such analysis possible, and to 
determine how to indirectly measure things like individual welfare.

    Ian> I have certainly seen the dynamic you describe.  I hope that
    Ian> it's clear that I personally am saying something different.

It is clear in your case.  It is also clear in certain other cases
that the dynamic is operative, and that frustrates and pains me.

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."