From Bernard.Lang@inria.fr  Mon Oct 16 13:30:30 2000
Date: Mon, 16 Oct 2000 15:30:30 +0200
From: Bernard Lang Bernard.Lang@inria.fr
Subject: [Patents] Counterexamples to John, Hartmut, Edward

On Sun, Oct 15, 2000 at 02:26:15PM -0400, Gregory Aharonian wrote:
> 
> John Tobey writes:
> 
> >
> >This fallacy has been used too often in defense of software patents.
> >The program and the circuit did *not* take equal amounts of
> >brainpower, time, and cost.  Let me explain.
> >
> >The effort that goes into a software implementation like Amazon's web
> >site is quite different from the construction of a factory:  It
> >involves no risk, and its completion gives no benefit to society.
> >Only the patent-bearing *idea* of the software deserves to be
> >considered when comparing research/development costs.  As PILCH
> >Hartmut has observed, the core idea seldom requires more than an
> >afternoon on a train.
> 
> You explain nothing here.  Amazon's one click is a lousy example to use
> to justify software patents.  Consider instead, natural language
> translation, a software problem that has yet to be solved after forty
> years of research.  People have invested (i.e. risked) lots of time
> and money, with few useful results.  So if someone new decided to take
> another crack at the problem, and spend lots of money and time, it is
> no less a risky and time consuming effort than any hardware design
> problem.  So stop attacking all software patents by pointing to some of
> the more obvious and simple patents such as Amazon's 1-click, or my
> favorite, a 1996 US software patent for nothing more than 3D pie charts.
> There are many software problems as intellectually difficult and time
> consuming to solve as hardware problems.  And there are many stupid
> mechanical and hardware patents being issued as well - should all
> mechanical and hardware patents be rejected because of the stupid ones?


as it turns out, natural language processing (NLP) is one of my areas
of expertise.  And if there is one thing I can state about it, it is
that progress in that field has been considerably slowed down by
excessive use of IP protection (because people were aware of the
potential benefits), whether througth patenting, copyright or
industrial secret.  It is a very complex field, with difficult
problems, and the issue is not so much inventing as developping a
proper science of what needs be done.  NLP is a realm of complex
systems, involving combinations of many techniques, and probably one
area where the economic theories of Bessen and Maskin apply well.
   Doing research in NLP is difficult because one needs complete
systems to experiment on a component of the whole processing chain,
and systems which are available and open to freely test a separate
components are not easily found.
   Furthermore, in that specific area, there is one concept that we
developed in my group, which took years and sweat to be clearly
identified, even though many people, in other groups, were also
working on it... and even though it seems obvious after the fact.
This is of course not patentable.  But once the concept is known,
there are plenty of ways to use it to implement a variety of processing
techniques in a more natural way, and some companies are rushing to
patent that, even though the inventiveness is negligeable compared to
what it took to identify the concept.
   What matters in NLP is not patentable. The trivial applications
are.

What is my conclusion:
  there is no such thing as a god given right to a monopoly on a
creation.  The only criterion is the social utility of granting it.
  In the above case, I believe that research in NLP is more muddled
than anything else by the expected benefits from IP protection.  And
social utility is hence minimized.


>                               ====================
> 
> Edward Avis writes:
> 
> >So I could protect my program, which is an implementation in C of a
> >particular sorting algorithm.  But I couldn't stop others from making
> >their own inventions based on the same algorithm.  This is exactly what
> >copyright on computer programs achieves.
> 
> And that is why people are increasingly abandoning copyright protection
> for software - because it gives no protection, and turning to patent
> protection.  If I invent a new sorting algorithm, and can't control how
> others exploit it with copyright protection, I will either get a patent
> or not disclose my sorting algorithm (except for those many programmers
> who voluntarily disclose new algorithms for everyone to use, which is
> nice but their choice).  Increasingly copyright is becoming absolutely
> useless to protect software - it is not an alternative to patent
> protection.  This is part of the reason for the explosion of software
> patents in the US, and for the coming explosion in Europe - that people
> realize that copyright protection for software is increasingly useless.
> Better to fix the patent examination process, a solvable problem, that
> trying to convince people to give up patents.

The question is not whether people can get protection, but whether
society has any interest in granting it.

  Past experience gives no indication that better algorithms are
produced when protection is granted.

> So if you reject software patents because some of them are logically
> equivalent to a mathemtical proof, then you have to reject mechanical
> and hardware patents as well because some of them are logically equivalent
> to a mathematical proof as well.
> 
> Which I suspect some of you want to do - not to get rid of just software
> patents, but all patents.

non sequitur ... we may (or may not) want to get rid of mechanical
patents which are mechanically produced ... on the above basis,
without necessarily wanting to get rid of all patents.

Bernard


-- 
         Non aux Brevets Logiciels  -  No to Software Patents
           SIGNEZ    http://petition.eurolinux.org/    SIGN

Bernard.Lang@inria.fr             ,_  /\o    \o/    Tel  +33 1 3963 5644
http://pauillac.inria.fr/~lang/  ^^^^^^^^^^^^^^^^^  Fax  +33 1 3963 5469
            INRIA / B.P. 105 / 78153 Le Chesnay CEDEX / France
         Je n'exprime que mon opinion - I express only my opinion
                 CAGED BEHIND WINDOWS or FREE WITH LINUX




