Subject: Re: BSD, GPL and macroeconomics
From: "Jonathan S. Shapiro" <shap@eros-os.org>
Date: Thu, 14 Feb 2002 12:15:31 -0500

[MarkM, MaryAnn: the following question arose on the "Free Software
Business" mailing list. For different reasons, I thought each of you might
possibly be interested in my response.]

> Going back to business topics, do the people on this list believe there is
> an economic advantage in having publicly funded research under BSD-like
> usage conditionsas opposed to the GPL?

What follows is merely a personal opinion, but since some of my work is
funded this way I have given the matter some thought. I apologize in advance
for its length. My answer will be U.S. centric, but I think that the line of
reasoning could be applied equally well to other types of publicly funded
work.

At the start, I should clarify that I do not propose that others should
adopt my opinions, nor do I suggest that my opinions are always right or
should serve as a guide to others. I offer them only because they reflect
*some* degree of thought on the question from someone who has been on
several sides of the issues.

In the U.S., the government historically owned all the rights for research
they funded to dispense as they pleased -- generally by making patents
available for free. This research was viewed as a public trust.

In the 1950s, it was realized that this was not generating value: most of
the work was not getting used. In part, it's just too difficult to do
business with the US Government -- little things like they don't pay their
bills (a client of mine now has USG receivables 9 months overdue, and this
isn't uncommon). Basically, the government wasn't very good at
entrepreneurship. At the time, the ideas of open patent licensing and open
source had not taken hold, so it was decided to let the universities keep
the rights and attempt to leverage them. While universities may not be the
best place to do this type of entrepreneurship, the results to date do tend
to confirm that they are better at it than the government.

In the process, the educational mission and the "public trust research"
mission of the American university system is being eroded; the long term
effects of this on the system are only just starting to be visible. One
effect is that our faculties are spinning out companies. This is great for
national technology policy, but horrible for training the next generation of
technologists (at lots of levels). TANSTAAFL.

My View as a Taxpayer:

As a taxpayer, I find current practices objectionable. What is going on
today is that my money goes to the government, and the government then does
not allow me to use what I have paid for. In fact, I have to pay more for it
because if it *does* get returned to me, it frequently happens in
proprietary form. My test here is the "reasonable man" test. If someone
offered me a choice on this deal, would I take it? The answer is no. The
deal is only possible if you can impose the coercive power of the government
on the taxpayer.

In the area of national defense, I probably feel that the current state of
affairs is an acceptable tradeoff up to a point. I am receiving benefit in
the form of the preservation and protection of the society in which I reside
and from which I benefit. There is considerable room for debate about what
means and how much money should be put into "the common defense,"  but in
abstract I feel that this is a legitemate function of government.

When matters shift to a national technology policy front, I am less
sanguine. It is not proper that my funds should be taken from me without my
consent to benefit a private company exclusively. I believe that America
collectively and Americans individually have benefitted greatly from the
results of government funded research -- benefitted with ample return our
investment (in the form of taxes). We eat better, we are safer, we have a
better medical system, etc. etc. We cannot, of course, judge the road not
taken, but having spent years in startups, I am not at all convinced that
the private sector can successfully marshal this kind of resource in this
way without a monopoly. As supporting evidence, take note that the only
remaining research labs of substance in the US are Microsoft and the
government labs.

The part I am not happy about is the exclusivity part. I believe that
inventions enable industries, but that in the end they are only a small part
of the resulting value proposition. In many cases ideas produced from
taxpayer funding had many applications and were balkanized by ineffectual
technology management at universities and/or the business failures of the
licensees. It is my opinion that the results of government funded research
are a public trust, and should without exception be made available for
public use. For software, this means BSD license at a minimum.

Then we arrive (in my mind) at the dicey question: if the public supplies
the critical activation energy, shouldn't they be entitled to long-term
rewards? The funding strategy does not provide equity, so some other
compensation must be found. I suggest that "continuing benefit" is a
reasonable alternative. To achieve this, some other licensing mechanism --
perhaps GPL -- must be applied. After careful thought, I believe that the
"continuing benefit" test should be be applied to my work, and this is why
the EROS work at Hopkins has been done under GPL. The issue, to me, is as
follows: the failure or shortsightedness of an exclusive licensee should not
allow the investment of the American public to be squandered. GPL is not a
significant impediment to a successful business. There are lots and lots of
ways to combine proprietary and GPL value creatively if a business is
inclined to do so, and RedHat and others have amply demonstrated that *pure*
open strategies can be used to build successful companies.

My View as an Academic:

I chose to go into academia for two reasons: (1) I wanted to help train the
next generation of technologists. God knows somebody has to do it. (2) I see
my work is a worldwide public trust. It must be commercialized to succeed,
and we are about to do so, but I feel strongly that it's roots should be
open.

As an academic, I have the authority (at least for now) to decide that my
work will be GPL'd. As university technology transfer departments begin to
turn their attention to software, it is not clear how long this will remain
true.

This type of work is only possible if it is publicly funded. Industry has
demonstrated repeatedly that it does not have and mostly does not want to
fund a long term outlook. Nowhere is this more painfully clear than in the
arena of software security, which is the arena I work in. It's partly an
activation energy problem, and partly a problem of combining a technically
difficult area with a realistic deployment and transition strategy; few
people have or know how to assemble all of the requisite skills and
outlooks.

The public funds my work, and the plain fact is that they *aren't* given a
choice about it. The government has funneled nearly $1m into my lab during
my first 30 months, and it is likely that many more $millions will follow
over time. For DARPA, which is my primary funding source, I am pleased to be
able to say that (at least in academia) DARPA funds based on reasonably
effective judgement and that they do not re-fund consistently failing
providers. Their results have been indisputably impressive. Where academic
funding is concerned, they try hard to fund in a fashion that respects the
value of the taxpayer's money, and I believe that for the most part (in a
statistical sense) they succeed.

Also, DARPA has been experimenting quietly with GPL vs conventional types of
agreements. In due course they will build up a portfolio of experience from
which to  judge whether one method is better than the other. I'm a great
believer in empirical testing before you overturn something in public
policy.

However, the entire DARPA system this rests on a fabric of respect for
reputation and the ethics and personal commitment of the network of people
involved. Since 1958, DARPA has done well at maintaining this fabric. When
it eventually fails, as any social fabric must, long-horizon technology
funding in our field will fail with it.

As a publicly funded researcher, I am very conscious of an ethical
obligation to provide value back to the public. I am further conscious that
it is simply not cost effective to educate the public about each of these
esoteric areas, and that the "let the investor judge the value" model simply
does not work for some of the things that DARPA funds. I am one of a very
limited pool of people in the world who can judge what technical success
means in this area. Once we have technical success, others will be able to
see the impact, but the event horizon is rather far away. In this type of
endeavour, I do not know any method of evaluation other than "try trusting
the experts, evaluate their results, and iterate, recognizing that a large
percentage will fail."

As I say, it all rests on the social fabric, and that fabric has in fact
held up very well.

Alternatives:

I will add one further thought:

Mark Miller, a dear friend, a libertarian, and a firm believer in market
mechanisms, would argue that this type of activity *can* be funded directly
by the public. He would argue the merit of "idea futures" and a variety of
other very interesting ideas. I would like to see those ideas tested, and if
they are successful, I think they are a better means of funding the type of
work that I do.

In the end, however, I am very skeptical. Investors are conservative. They
generally do not fund "blue sky" work. Further, the specialized judgment
that it takes to make successful technology investments is not widely
available to investors. Reputation systems could be used to solve that, but
a viable alternative funding system for long-term work would take real work
and many years to bootstrap. Indeed, I think the practical complications of
bootstrapping such a system have been grossly underestimated in the
economics literature.

As I say, I think it's worth a try, but in the meantime I prefer not to
undermine the few mechanisms that we have in place for public trust funding.


So those are some of my thoughts, offered for whatever value they may
provide.


Jonathan S. Shapiro
Assistant Professor, Department of Computer Science
Johns Hopkins University