Subject: Re: Novel anti-software-patent article
From: "Stephen J. Turnbull" <turnbull@sk.tsukuba.ac.jp>
Date: Tue, 11 Jan 2000 11:02:53 +0900 (JST)

Points taken out of order, so that the less contentious material comes
first.

>>>>> "Bernard" == Bernard Lang <Bernard.Lang@inria.fr> writes:

    Bernard> even though the actual "discovery" (not invention, mind
    Bernard> you) may be located.

I don't understand the difference.  I don't see how it can be made
operational in terms relevant to this debate.  The social costs and
benefits are the same, and the patent office can't inspect mental
processes to see if they were creative or mere "discovery."  Nor can
they tell by introspection; if they could make that judgement, surely
they could do a better job on "obvious to any practitioner"!

    Bernard>   As to the nature of software, there is a well know
    Bernard> result by Curry and Howard, the Curry-Howard isomorphism,
    Bernard> that states that a program is of the same nature as a
    Bernard> mathematical proof. Hence patents on programs are patents
    Bernard> on proofs.

How does this differ from any process innovation (vs. a product such
as a drug)?  Ie, consider the following isomorphism: a (process)
patent is granted for a description of the process, a (software)
simulation is a description of the process, and therefore any process
that can be patented is isomorphic to a program, which is isomorphic
to a mathematical proof.

Note that the leap from physical process to description is a
difference in nature.  A physical object is not isomorphic to a
mathematical one; the relation is a model.  Is it then your argument
that something that has a physical reality separate from what the
patent inspector makes his decision on can be patented, but that
anything that the inspector can actually directly apprehend as a
mental construct cannot?

Or, a related proposition, can analog computations be patented but
digital ones not?

    Bernard> 3- (and this is a more recent view) encourage investors
    Bernard> to bring inventions to market by making sure they recoup
    Bernard> their investment.

Please also keep in mind that "passive" marketing (bringing an
existing product to market) is a creative activity, and that "active"
marketing (specifying the product to be attractive to a certain class
of customers) cannot in practice be distinguished from invention.  In
the following sense:

When I was a graduate student, my advisor wrote a bunch of claims
down, handed them to one of my colleagues (ie, another of his
advisees), and said "prove these."  He then gave some detailed advice
on methods of proof for some of the claims.  The graduate student was
given a special footnote stating that "most of the detailed proofs are
due to" him, but was not a coauthor.

I challenged my advisor on this.  His response was "These proofs are
similar in kind to the ones I assigned you (pl.) in the game theory
class; any student who passed that class could have done them.  They
are similar to those in his own research---they should be easier for
him and also help him.  Had he come up with a counterexample or a
significant extension, he would have been a coauthor or primary author
depending on the importance of the new contribution."  The graduate
student agreed (not in front of our advisor) with this reasoning, and
so have all the academics I have talked to about it.  The harshest
criticism I've heard was "many advisors would have been more
generous."  (FWIW, my classmate's opinion was that incident was one of
the two most positive influences on his development in graduate study,
and that it would have been far less valuable as a coauthor.)

The analogy to software is clear, I think, although not pleasant to
contemplate.

    Bernard>   Since investment is low for software, but patent
    Bernard> protection is outrageously expensive -- not just the cost
    Bernard> of applying, but all the legal and technical work, plus
    Bernard> going to court when it really matters -- patenting
    Bernard> increases considerably the cost of going to market, and
    Bernard> hence is counter-productive.

This is a one-sided computation.  Unless you can measure the incentive
effects---and you haven't---the conclusion is invalid.  Also, the
relevant comparison is not between the cost of getting a patent and
other costs (in particular, the cost of development), it is between
the cost of patenting and the revenues lost by not doing so.

There are two reasons why small firms and individuals don't get
patents (actually AFAIK a corporation cannot get a patent; the deal is
that the employee gets the patent but immediately assigns it to the
company).  One is that the risk is high, so risk aversion requires
them to get much higher returns than an entity with a large portfolio
of development projects.  (Note that one way to deal with this problem
is venture capital---and it's the venture capitalist that gets the
large share.  But this requires a relatively high-quality invention.
Another way to deal with this, if the inventor has less confidence
that he can strike it rich, is to go to work for one of the big
patent-factory firms.  Looks a lot like reality, doesn't it?[1])

But the second is that to really profit from a given patent,
substantial other costs must be paid: acquiring rights to related
technology, marketing costs, and so on.

In other words, _the net social benefit to granting a patent to an
small entity is less than the social benefit to granting it to a
larger one._

    Bernard>   As for many sciences, the investment is spread over the
    Bernard> intellectual community. Actually bringing the (live)
    Bernard> brain a native of country Y in company or country X, in
    Bernard> order to have him "invent" something with little actual
    Bernard> investment other than his salary, which then becomes the
    Bernard> property of X, is in my opinion stealing the investment
    Bernard> of Y.

I see.  The inventions of von Braun, Einstein, and others who left
Germany for the U.S. were "stolen?"

How about the inventor him or herself?  Why should the inventor be
forced to stay in a place that is unwilling to recognize her abilities
with a competitive salary or environment conducive to her work?


Footnotes: 
[1]  I'm well aware that my reasoning is partly driven by explaining
the existing institutions.  The point is that these purely economic
motivations are as plausible as the power-grabbing rent-seeking
explanations that underlie most descriptions of the "patent problem."
And, given that I dislike software patents, despite my stance in this
forum as Devil's advocate, it is less self-serving.

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
/* What's the big deal about the millennium? .............................
.... There are still 360 shopping days left until the millennial epoch! */