[Patents] Re: Konno 1995: The Karmarkar patent and Software -- does math patenting work?

[phm@a2e.de (PILCH Hartmut)]

> Konno's book exaggerate the problems of the Karmarkar patent.  Sloppy
> research which question the rest of the book.

Did you read the book?

> For example, hartmut includes the following quote:

This was not a quote but my summary.

>         "However the Karmarkar patent did not create very significant
>         revenue for AT&T while causing significant inconvenience to
>         the industry and destroying much of the positive interaction
>         dydles that had previously characterised the cooperation
>         between linear programming research and industry in the US.

> Nonsense.  One of the reasons why ATT didn't make much money off the
> algorithm and its patent, was that mathematicians were quickly able to
> come up with alternative implementations of linear programming solving
> algorithms, while in the spirit of Karmarkar, did not infringe his
> patent.  In short, they engaged in the time-honored practice of
> working around a patent technique, some of which were even better than
> Karmarkar's approach.  Which is the very point of the patent system.

The patent was granted in 1988.  From 1984-88 Karmarkar and AT&T were very
secretive about the effectiveness of their algorithm and provoked academic
conferences by just bragging and not answering questions.  That is why the
academic community engaged in its own research and by 1988, when they
found out that Karmarkar/AT&T had obtained a patent, had already
significantly improved on the methods of Karmarkar.

> Companies didn't have to buy ATT's custom
> supercomputers that ran Karmarkar's algorithm (ATT set up a
> subsidiary I think to sell these machines), because they could run
> their own implementations on their own supercomputers, without fear
> of infringing the Karmarkar patent.

A program called OB1 did the job better than AT&T's system (KORBX), but
the mathematicians who wrote that program were nonetheless bullied by
AT&T's army of 1000 lawyers and settled for a royalty payment of 5%,
although they (probably) didn't infringe on the Karmarkar patent.

This could happen only, because there was a large math research community
that resented the Karmarkar/AT&T practise and was not keen on patents.

You can easily imagine a scenario where this is no longer the case or for
some other reasons no better method is found or only 1 or 2 are found
which are then also patented.  This can easily throw a whole range of
interesting fields of application back for 10-20 years.  With basic
mathematics, this range is infinite.

Konno also points to the fact that the first wordprocessors at the
beginning of the 80s were also sold for 10 million USD and marketed only
to a few large customers, similar to the Karmarkar custom computer.  This
of course is a problem that is not limited to abstract innovations, but
particularly exacerbated in this area.

> Challenging the patent was a waste of time because a) the patent was
> novel and unobvious and enabled and statutory, though b) of limited
> offensive use.

The program was
(a) not statutory:
    In the US the CAFC had voided that requirement, but not yet in Japan.
(b) not novel
    A russian mathematician had already found and published it in a
    russian math journal in the 60s, but nobody deemed it important at
    that time.  That surfaced in 1989 and is again very typical for
    the problems that the patent system will encounter especially in
    abstract areas like math which have a 2000 year history.

> So far from being some plague on the patent system, the Karmarkar
> patent was a good example of the system working.  To criticize
> software patents using Karmarkar is a really weak attack on the
> system.

To criticize a book you haven't read, just because you don't like the
conclusions, is ... well, err, Aharonians "scientific" style.

What is particularly interesting about Konno's book is that it builds its
argumentation around a much celebrated advance in mathematics and not
around some trivial patents.  So you get right to the core of where swpats
should be doing a good job for the promotion of science and the useful
art, and then find that they are failing badly even there.  Konno of
course also doesn't fail to mention that the patent applications for
Karmarkar's patent were badly written (by people with no understanding of
mathematics and lots of mistakes) and badly granted (again, much too
broad, by examiners with lack of understanding), thus showing clearly
that the patent examination system, even when it comes to true
achievments, does a lousy job compared to the referee system of even
third-class scientific journals.  Konno of course also doesn't fail to
mention that most software/math patents are, unlike Karmarkar's patent,
silly right from the start and are nevertheless granted due to the
inadequate rules and institutions of the patent system.

--
Hartmut Pilch
Federation for a Free Information Infrastructure        http://www.ffii.org/
Protecting Innovation against Patent Inflation        http://swpat.ffii.org/
For a software patent free Europe PLEASE SIGN http://petition.eurolinux.org/