Subject: Re: Open letter to those who believe in a right to free software
From: "Stephen J. Turnbull" <turnbull@sk.tsukuba.ac.jp>
Date: Fri, 29 Oct 1999 19:26:01 +0900 (JST)

I'm going to post this one, although it's really shaky as to whether
it's on topic (presuming the thread itself was).

I think it's a reasonable conclusion to the thread(s) that got started
a couple of weeks ago, from my point of view.  People who have been
following the thread but are ready for it to go away will get a more
accurate idea of what I keep saying "I'm gonna do," and I think I owe
you all that.

I should also say that on most of the concrete points debated, I
haven't changed my positions---yet.  Nor do I see any particular
reason why the discussion should have much changed anybody else's
mind; nothing was _proved_---yet.  I have changed the way I think
about them, substantially, thanks to the discussion, and I expect my
modeling to be much more convincing as a result.

Of course as the papers get written, I'll put them up on a web site
(I'll let you know), and they'll be intelligible to non-economists (at
least in outline and with examples) and libre---subject to
professional journal copyright assignments; I can't avoid that, it's
my job to submit them for publication.

Yup, for those of you who were wondering when I slept---that torrent
was financed by the Japanese taxpayer.  I was at work!
Un(?)fortunately, my calendar for the next week or so won't permit me
to spend very much time on research, so I'm going to dry up for a bit, 
maybe quite a while.

I would like to mention an implication of my conversation with Ian
regarding "ethical assumptions."  Although for my academic work I'm
going to use ethical and scientific assumptions I'm comfortable with,
I have no problem with working out the consequences of assumptions I
don't like.  So if you have a need for an economic analysis of the
software industry, I'd be happy to consult---for money (I never refuse
free money) or love (of free software).

Academic books always mention the patience of the spouse.  Reflecting
on the last week or so, the way I've imposed on this forum, I'm
beginning to understand the feeling.  At least you can all be thankful 
you're not married to me!

Many thanks to _all_ participants.  I'm sure it looks like I think I
have all the answers, but I don't, and I know I don't.  But after this
thread, I have a lot more answers than I did before.  Better yet, I
have a bunch of new questions!

>>>>> "kms" == Karsten M Self <kmself@ix.netcom.com> writes:

    kms>  - Economics studies markets.  Individual behavior falls
    kms> under psychology.

That's too generous to economics.  Economics does claim to have a good 
model of individual behavior.  It's simple enough to state, so I'll do 
it here:

1.  People have a good idea of what they want to happen in the world
2.  They have a good idea of what they make happen in the world
3.  They are pretty good at using the information they have to decide
    how to satisfy 1 given 2

0.  They do as they have computed in 3.

Clinical psychology, of course, is about people who fail to do 0.

Experimental psychology is about measuring how well people do in 1, 2,
and 3.  The answer turns out to be "not all that well," but in general
they are pretty unbiased in their errors, especially where there are
precise ways of keeping score, like money.  This means that economists
can use the law of large numbers to work with markets.

There are of course significant biases is some areas.  I don't really
think those apply strongly to factors entering the decision to do free
software.  Some matter a lot to economics, possibly.  A typical
example is that many people apparently weight losses more than gains
of the same size.  (I don't see how that applies to free software
directly, although I can tell some strained stories.)  It turns out
that these biases are indistinguishable in effect on measurable real-
world behavior from certain changes in parameters in the standard
model.  (To the best of my knowledge; I am not trying to claim there
are none that would force a change in the standard model, I just don't
know of any, and I was moderately well-informed in the matter and
would have noticed publication of new ones.)

Note that there is no mention of money or even physical commodities,
nor even properties like "more is better."  They are not implied.
Furthermore, relevant factors are not restricted to the personal
sphere; they can include family, friends, or the whole human race,
individually or in aggregates.  In fact, in general the only
restrictions that "better" must satisfy are that given three worlds,
A, B, and C

a. A is no better than itself
b. Either A is better than B, B is better than A, or they are
   indifferent
c. If A is no better than B, and B no better than C, then A is no
   better than C.

Most people, as well as most economists, agree these are pretty weak.

It is true that the "usual assumptions" include far more specific
assumptions, and these are more or less necessary to derive things
like the law of demand and behavior under uncertainty.  However
economists can and regularly do abandon some of the "usual
assumptions" when appropriate.

In the case of free software the usual assumptions that are in danger,
as I currently see it, are

(1) "continuity", which implies that given two goods it's always
    possible to make up for a loss of one by a sufficiently large gain 
    in the other.

I think it should be clear from that why I care so much about the the
claim that intellectual property is evil; if so, there is no economic
benefit large enough to justify it.

Based on the discussion so far I see no reason to give this one up in
general, except for Ben Tilly's religious belief about the
non-existence of certain kinds of infinity.  ;-)  I will, as a matter
of personal honor (and empirical fact), put a (parametric) proportion
of programmers with such utility functions in the model.  But I will
have no qualms about trading off their sense of moral outrage against
economic benefits to consumers or stockholders, and empirically the
proportion is likely to be sufficiently low as to hardly affect the
analytic conclusions.

(2) the assumption that personal consumption of commodities is all
    that matters.

This one is a dead duck.  The whole point of freeing your software is
that you care that other people be able to execute your software and
derive from your software.

What does this mean in terms of modeling?  Simply that I do something
like put the number of people who use your software into your utility
function, independently of the revenue you may or may not generate
from it.  Doh, as they say; that wasn't so hard.

I can think of some reasons why that simple model is unsatisfactory;
for example, under the rest of the "usual assumptions," it's not
sufficient to explain why so many free software advocates work _only_
on free software; it would make more sense to use revenues from
proprietary software to finance work on free software.  (Were I a
professional developer, this would be strongly attractive to me as a
practical option, although I would rather work in an all-FS
environment.)  So I'm going to need to deal with those issues, but I
think they're pretty easy; the utility function concept is so flexible
I can pretty much rig it any way I want.

Also, if I want to (have a hope of) explain things like forks I'm
going to need to put in "interaction terms," like membership in a
project (which amounts to reciprocal use of each other's code).  But
that's far too ambitious to take seriously at this point.

It is true that I haven't explained why people want to do free
software in an important sense.  It doesn't help you to run your FSB,
and it doesn't help you convince your sister to quit working for
Microsoft and apply for a job at Cygnus.  If you want that, talk to a
psychologist or ESR.

But that's not my goal; my goal is to tell the relevant Senate
Committee that increasing patent life to 20 years would cost $X
billion dollars in real economic benefits, broken down as (negative
cost, ie benefit) increased development of proprietary software, lost
usage benefits, lost production of rival free and proprietary software
with $Y billion of estimated usage benefits, and so on.  To figure out
how much free software there _would_ have been, I need to know how
hard people are going to work on free software, which comes back to
those incentives, ie, the arguments in the utility function.

One subtle aspect is that under the assumption of continuity (well,
actually a certain amount of differentiability or convexity), it is
legitimate to construct "shadow prices" (ie, in money terms) on values
like "working on free software is fun" or "I like knowing that 10,000
people are using my software" or "I like knowing that your brother is
using my software".  These shadow prices _can_ be used to compare such
benefits to realized profit or loss in the market.  (The main caveat
is that utility functions are measured with poor accuracy relative to
profits, so one must do lots of sensitivity analysis.)

    kms> The *implications* of the utility function, and its effects
    kms> on the market, are generally fairly well understood.  But the
    kms> critical voices here do have a point when they say that
    kms> economists don't know what makes people tick.

The point is true, but not relevant to what economics is for.
Economics is not for persuading people to change their lives.  It is
for persuading politicians to change social policies.

Contrary to what you said, economists _don't_ understand the
production function, in the same way we don't understand the utility
function.  We simply relate the input arguments to the output
arguments as a functional constraint.  We don't know anything about
what makes it tick---that's what operations researchers and industrial
engineers do.

Of course, when the OR guys don't know squat about what makes the
production function tick, then the economists have to get mixed up in
the nitty gritty.  Eg, in software.  But as a general practice we have
no way to predict when the technology will change dramatically.
Moore's law is _not_ an economic law, we can't explain it---but
anybody doing hardware economics better know it.  Any more than we can
predict when tastes will change dramatically.  Who would have
predicted ten years ago that flights from Tokyo to the US would be
tobacco-free---certainly not an economist, although RJR's marketing
dept may have had an inkling.

Ben Tilly's discussion about the dynamic effect of Perl5 vs Perl4 on
development costs actually worries me, in terms of "will I be able to
model this acceptably well", far more than the issues of modeling
programmer participation in free software.


-- 
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."