Subject: Re: FW: Why would I pay for Ximian software?
From: "Perry E. Metzger" <perry@wasabisystems.com>
Date: 02 Jan 2002 17:22:19 -0500


Nick Jennings <nick@namodn.com> writes:
> On Wed, Jan 02, 2002 at 03:27:57PM -0500, Perry E. Metzger wrote:
> > In general, the open source software universe has this problem, which
> > is how to assure that smart people can make a living doing free/open
> > source software. Writing good large programs is a full time job, and
> > people need to have ways to paying for that.
> > 
> > Having started a company dedicated to an open source product (NetBSD
> > in my instance), let me say that this is a hard problem.
> 
>  Agreed. It's alot easier for someone to make a living of selling drugs,
>  or prostituting themseleves, than it is for someone to make a living
>  of doing good for the community. That's one of the things I hate about
>  Capitalism. It's designed around the dollar, not around progress, or 
>  freedom. 

Not true, I'd say. In general, the free market model lets people vote
pretty directly (with their dollars) for what their true resource
priorities are. It is hard to end up with a Pareto optimal resource
allocation in any other way -- as attractive in the abstract as
various socialist schemes are, in practice they end up misallocating
resources far worse than free markets will. So, as much as I'd like to
say my own personal business problems might have something to do with
the "evils of capitalism", I'm afraid the econ student in me has to
pretty much accept that we need to make open source work within the
context of the free market model.

(One can also argue that the moral side of "free software" and "free
markets", that is, not attempting to impose anything on people except
that they can't impose things on other people, is fairly similar in
both cases.)

However, lets skip the musing about a better world for now.

The issue in the world we've got is one of our not understanding very
well how to deal with the business issues.

The problem with the open source world is that we end up constructing
a public goods problem. Software -- especially open source software --
is very classically under the definition -- non-rivalrous consumption
(you and I can both run a program without impacting each other's use),
and (in an open source world) non-excludability. In fact, the open
source definition more or less defines open source software in those
terms (though as a lawyer would understand them rather than as an
economist.)

The classical theory of public goods is that you end up with commons
problems -- the thing will tend to be overconsumed and underproduced
because there isn't a sufficient mechanism to reward the producer for
their efforts given the amount of economic benefit derived by the
consumer -- i.e. we get a "market failure". The fact that a producer's
efforts help the world more than he is rewarded is called an
"externality", and externalities obsess public goods minded
economists.

However, as both a libertarian and a a strong believer in open source,
I'd like to think there is good reason to believe that the public
goods/market failure theories that would traditionally be applied here
aren't true. Note that I say "would like to think". However, there is
some strong indicative evidence in this direction.

The "solution" to the "market failure" we've traditionally used in
software is copyrights, but of course, everyone is here on this list
precisely because they believe that many indirect problems make
copyright not worth it. In particular, software that is closed doesn't
have a developer community, and does not well meet the needs of many
users who require source. Even if source is available to closed
software, there is no incentive to contribute to the community if it
merely benefits the bottom line of the copyright owner.

Thus, the evidence against the traditional arguments on externalities
and market failure: the fact that we can have entire operating systems
like Linux or NetBSD in existence without the traditional closed
source model shows that the open source model certainly provides
dramatic incentives of some sort to production beyond the model of
dollars going to a copyright holder. At least part of the
externalities are being handled in some manner, however mysterious.

We therefore know that the traditional "public goods" model is
flawed. It is obviously not the case that there will be little or no
incentive to production without exclusion and direct
compensation. Something outside the traditional theory is happening.

However, that is in fact a problem in itself. The economics of all of
this are as yet very poorly understood. We've got lots of theory that
explains why this all shouldn't work, but very little that well
characterizes why it at least partially does work anyway. In an
excellent book on the public goods problem that I once read ("The
Theory of Market Failure" edited by Tyler Cowan), there were a number
of case studies described by famous economists in which classical
examples of public goods and market failure are examined and
debunked. (It turns out for example that there were privately run
lighthouses at one time, and that there is a market in beekeepers
providing bees for crop pollination, both classic examples of public
goods in econ 101 texts which turn out to be mythical.)  However,
although we well understand that we've mischaracterized the public
goods problem, exactly how to examine these sorts of situations in a
way that *does* properly characterize them is still an open question.

All this is of very immediate interest to me. I've got a company that
I started with a bunch of engineers working for me doing NetBSD work
for a living, and we confront the issue of paying our salaries every
day of the week. Exactly what sort of models can work that don't
involve exclusion so that we can continue to make a living AND have
everyone reap the benefits of having the software be open is a very
important issue to us -- as I assume it is to many people here.

--
Perry E. Metzger		perry@wasabisystems.com
--
NetBSD Development, Support & CDs. http://www.wasabisystems.com/