Subject: Re: Open letter to those who believe in a right to free software
From: "Stephen J. Turnbull" <>
Date: Wed, 17 Nov 1999 15:46:02 +0900 (JST)

>>>>> "rms" == Richard Stallman <> writes:

    rms> I think we are failing to communicate, since you start with
    rms> "no", but the following details seem to agree with what I
    rms> said: the model includes parameters describing what people in
    rms> society do value.  I called these "other people" because they
    rms> are *other than the reader of your article*.

OK.  That was not clear.  But what I was denying was that I was making
value assumptions (yet, inevitably they will creep in), at least not
value assumptions that you would object to.

So what is the value assumption I am making?  I've thought about it
rather carefully, informed by the thinking of both economists and
their critics.  The only value assumptions that I can see that I have
made are (1) that what individuals want should matter in evaluating
what is good for society, and (2) that some aggregation of what
individuals _could_ want is sufficient to characterize what is good
for society.  This rules out Nazism, the "l'etat c'est moi"-ism of
Louis XIV[1], and theocracy, but it doesn't rule out the kinds of
things you want considered ("freedom", "domination") as far as I can

Do you have a problem with those assumptions?

Measurement and model implementation difficulties for values that are
not among those conventionally counted as economic are important, as
you recognize.  Those are not assumptions, though.

    rms> But I'm calling attention to a different question: [...] 
    rms> "What should I value?"  A question the reader would, I hope,
    rms> ask perself.

Sure.  Me too.

    rms> That latter question is not, I believe, the one you seek to
    rms> address.

No.  I explicitly avoid it.  I do have that job, as a parent and as an
academic advisor.  But not as a research economist.

As an economist my primary job is attempt to construct a model which
flexibly reflects (1) the values of people in society and (2) their
modes of interaction, for the purpose of describing the likely effects
of changes in the modes of interaction on the outcomes.  As auxiliary
tools I will provide "social welfare functions" and other summary
statistics, but they are auxiliary.  Sophisticated readers can (and
do) construct alternatives for themselves, and the unsophisticated can 
ask for second opinions or look at more disaggregated results.

The point of the the model itself is to attempt to summarize the
infinite complexity of society in a way that the reader can grasp, and
so decide for him or herself whether the results of a given change are
good or bad.  Thus, a secondary job is to present the model in such a
way that the assumptions, whether values, factual, or analytical, are
exposed so that the reader can judge their accuracy, and whether the
inevitable sacrifice of accuracy is justified by the difficulty of
formulating a better model.  And perhaps, for readers with a certain
degree of sophistication, giving them the ability to change the
parameters of the model for themselves and see what comes out.

Ah yes, there's another value assumption.  This one about science: an
unbiased but inaccurate model is better than no explicit model.

    rms> But your article will convey some message about it
    rms> implicitly--it cannot help doing so.  Just what that message
    rms> will be will depend on what the article actually says; but it
    rms> seems likely that a paper described as "economic" will
    rms> encourage people to (without conscious consideration) adopt
    rms> more "practical" values, values that they associate with the
    rms> word "economic".

It's unkind of you to rub my nose in that.  I live every day of my
professional life knowing that no matter what I do sleazy politically
motivated advocates will pick from the range of results (sensitivity
analysis) that I publish those that are most favorable to their
position, and claim that I support their cause.

The nature of economic results being what they are (theoretically
ambiguous and empirically imprecise), the only way to prevent that is
not to publish.  But in every profession there are liars-for-hire; the
demand for biased results _will_ be satisfied.  At least I can publish
my own, removing those biases I can recognize, discussing how my
methods remedy the biases in others' work, and acknowledging that as a
human being I have probably unwittingly introduced yet other biases.

I don't see why I'm obliged to cater to the ignorance of those who
will not learn, and especially not to abandon a set of useful tools
simply because they are easiest to apply to the marketized portion of
the economy.

    rms> Especially if that is what the author in fact values; the
    rms> author's views would tend to come through in various implicit
    rms> ways.

See my .sig.  What do you think I value?

It is true that I do value freedom of choice for others, that
therefore for them I value their values, and that for 99.44% of
humankind those values are (outside of a small circle of family and
friends) first and foremost materialistic.  Breaks my heart, but
that's what valuing freedom is about---allowing others to make their
own mistakes.

    rms> So the article could discourage people from thinking about
    rms> their values.  That would be detrimental to the success of
    rms> free software.

Not my job as an economist to worry about that.  That's your job.  And
mine, of course; I'm an advocate in my not-for-professional-journal-
publication time.  But _as an economist_ going beyond trying to expose
my assumptions to criticism, and trying to remove biases, risks going
in the other direction toward creating new biases.

Furthermore, it is important to me that my model be specified in a way 
that makes it useful to people who don't think like me.  Adding a
specific spin to it would reduce that usefulness.

    rms> If the article discusses various alternative worlds with
    rms> different proportions of people valuing this or that, they
    rms> might help reinforce for the reader that one's values are not
    rms> fixed, they are something one can think about.

Economists have been doing precisely this at least since the
Bernoullis, and I plan to continue that grand tradition, although in
somewhat broader ways than most economists do.  But it doesn't seem to
have helped much.  So this is not my cross to bear, not this week.
(Again, I'm talking about as a publishing research to a faceless and
(I hope!) large audience; one-to-one or one-to-few, as in parenting,
advising, upper-class and graduate seminars, and general bull sessions
it _does_ help and it _is_ my job.)

    rms> This, for me, is the main point about the article, the point
    rms> that seems most important to me.

My tenure and promotion committee does not agree.

Nor do I.  Providing reliable tools by which people can conduct the
relatively straightforward and reliable computations that deal with
economic valuation is more important.  I want to show people that the
economic costs of using free software that they _fear_ are not
realistic, that the economic benefits are rather higher than they
_thought possible_.

I don't think that will change their values.  I do think it may make
it easier for them to acknowledge values that they suppress, such as
freedom and "fun", if they can place confidence in a model that says
"free software works like this in the economy," rather than devoting
energy to fearing the worst outcomes for the unknown processes by
which it operates.  (Or worse, using overly simplistic analogies to
non-software industries that predict disaster.)  This puts the burden
on _them_[2] to evaluate the tradeoff between the economic values
which can be (relatively) reliably and accurately modeled, and their
other values which are much less susceptible to economic (or any
science that I am aware of) modeling.  And that's where it should be.

I also think that such a model is of much greater interest to FSB than 
an advocacy tool without accurate reflection of the economic realities 
would be.

I have written and published advocacy stuff, I will write and publish
more.  But to everything there is a time and place.

[1]  Well, under the usual non-dictatorship postulate that more than
one member of society matters.  ;-)

[2]  Yes, I said that I _could_ model the not-conventionally-economic
values.  And I can.  It's even useful, given an audience of
professionals used to that convention.  But it would be misleading of
me to present such a model to non-specialists, who would not be able
to interpret a single index with some abstract "weighting" parameter
incorporating both kinds of values as easily as they can interpret the
choice between "freedom and no MS Word" and "no freedom to share with
MS Word".

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."