[Patents] Re: European arguing for patenting pure math formulas

[phm@a2e.de (PILCH Hartmut)]

Greg wrote something like:

> As an antidote to someone's pseudo-scientific posturing, what follows
> are the abstract, and one paragraph from the conclusion, of a journal
> article from a European philosopher that pure mathematical algorithms
> should be patentable, a proposal more ambitious than many made here in
> America.  It is good to see that some Europeans sticking to pure logic
> and semantics, strictly avoiding economic reasoning.

A few years ago Prof TAMAI Tetsuo of Tokyo University argued that either
should all mathematical abstraction be directly patentable, or software
should not be patentable.  What patent offices are currently doing is a
kind of halfway solution, indirectly claiming abstract logical principles
and pretending to be aiming only at a concrete physical realisation.

>From what I read below here it is not clear whether Paul B. Laat from
Groningen said anything else than Tamai.  The abstract does not suggest
that B. Laat is an economist who investigates the economics of patenting
abstract inventions.  Philosophers are concerned mainly with consistency.
And, yes, it is inconsistent to patent logical problems in relation to a
physical apparatus.  The physical apparatus is irrelevant to these
problems.

When Laat says that mathematics of course belongs to the useful arts, then
I wonder what kind of things do not.  Certainly artistic schemes such as
musical composition techniques, rhetorical devices or argument structures
are useful and can be formulated in mathematical terms.

By allowing unlimited patentability, the focus is shifted to another
question: the question of patent enforcability.  Let's say I have a patent
on the argument structure exposed above.  Should I now be allowed to
exclude everybody else from using this reasoning?  Or only from using it
for an "industrial application"?  Is a Disney film with a courtroom debate
in which this reasoning is used an "industrial application"?  What about a
film that uses a certain plot or narration technique?  Is the patent
lawyer's work "industrial" ?  The software author's work ?

This will lead into so many delimitation difficulties that you get right
back to the suggestion that there needs to be a material embodiment,
produced by an industrial apparatus, with marginal costs approximating a
constant above zero, and that an informational embodiment is not
acceptable, because information is equivalent in its economic and
communicative properties to thought itself.

That is again not very far away from the traditional limitation to applied
natural science, except that the average patent becomes much more trivial
and that you have huge unaddressed problems with prior art and with all
the well known negative economic effects of the patent system.  In the end
the conclusion can only be that of Tamai:  since it is inconsistent to
allow physical embodiments of abstract-logical innovation to be patented,
the best choice is not to allow such innovation to be patented at all.
Or, if you believe that the triviality problem can be overcome (as Greg
claims to believe): show us the running code before you try to conscript
us as alpha testers for something nobody believes will ever work.

Demanding a mathematical proof is quite a good idea.  I'd do it the other
way around:  a patent on an invention in the realm of applied natural
science is valid until the day when someone has mathematically reduced it
to prior knowledge.

>                                       ====
>
>     Patenting mathemtical algorithms: What's the harm?  A thought
>         experiment in algebra
>     International Review of Law and Economics, v20 (2000), 187-204
>     Paul B. de Laat
>     Faculty of Philosophy, University of Groningen, The Netherlands
>
>     Abstract
>         The patenting of software-related inventions is on the
>     increase, especially in the United States.  Mathematical formulas
>     and algorithms, though, are still sacrosanct.  Only under special
>     conditions may algorithms qualify as statutory matter: if they
>     are not solely a mathematical exercise, but if they are somehow
>     linked with physical reality.  In this article, the best results
>     are obtained if formulas and algorithms are only protected in
>     combination with a proof that supports them.  This argument is
>     developed by conducting a thought experiment.  After describing
>     the development of algebra from the 16th century up to the 20th
>     century (in particular, the solution of the cubic equation), the
>     likely effects on the development of mathematics as a science
>     are analyzed in the context of postulating a patent regime that
>     would actually have been in force protecting mathematical inventions.
>
>     Conclusion
>         What lessons can be drawn from this thought experiment?  If
>     the results are any guide to present-day developments, they would
>     argue for a BLANKET ACCEPTANCE of mathematical formula and
>     algorithms as statutory matter for patent applications.  Let us
>     finally acknowledge that mathematics belongs not only to the
>     'liberal arts' but also the 'useful arts'.  Let us abolish the
>     'pure-number condition'; even if it is all mathematical abstraction,
>     this is acceptable.  There is no need to specify links with
>     'physical elements' or 'process steps', with 'pre-computer' or
>     'post-computer process activity'.  Even if nothing of the sort is
>     specified, the claim is acceptable.  Let us not bother that the
>     claim 'wholly preempts' the use of the algorithm; let all possible
>     commercial applications effectively become protectable.  It should
>     be stressed that noncommercial, in particular academic, use is to
>     remain possible (experimental use exception).  Actually, it is
>     the development of both (pure) science and (applied) technology
>     that the system is designed to foster.  Such unconditional acceptance,
>     however, should be carefully formulated as to what is protected
>     and what is not.  While the protection of formulas tout court is
>     acceptal in societal terms, the best results seem to be obtained
>     when algorithms and formulas are only patentable in combination
>     with a supporting proof.  That arrangements keeps avenues open for
>     inventive mathematicians to develop alternative methods of proof.