Subject: Re: patents
From: "Stephen J. Turnbull" <>
Date: Tue, 02 May 2006 16:40:51 +0900

>>>>> "Seth" == Seth Gordon <> writes:

    Seth> I wonder if a great deal of chaff could be cut out of the
    Seth> patent system by simply requiring that a patent must involve
    Seth> a combination of *more than two* previously known methods in
    Seth> order to qualify as "not obvious", so that

Good-bye Diffie-Hellman and RSA patents, which everybody admits clear
any reasonable bars for originality and unobviousness.  Nope, can't
work.  And OTOH, as you point out, almost certainly if that is
required, all that will happen is that patent applicants will rephrase
their claims to make it appear that multiple methods are involved.
That should rarely be very difficult.

The reason for the approval of frivolous claims is two-fold: patent
examiners have _by law_[1] perverse incentives (to grant, rather than
deny, patents, and to do so quickly) and they generally lack both
training and access to relevant information.  An adversarial approval
process solves both problems at one stroke.  The USPTO examiners will
still have those incentives, but they are by and large honest civil
servants, and will pay attention to the "amicus" arguments, especially
presentation of relevant prior art.

Another thing to note about this is that prior art introduced in this
way can be added to the USPTO's databases, and that should be a
requirement in any reform bill.

[1]  Remember, the USPTO originally opposed granting software patents
in general, until they were directed to do so by a court.  Then, being
ethical bureaucrats, they did what they were told, as best they could
with the resources they had.

Graduate School of Systems and Information Engineering   University of Tsukuba        Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
        Economics of Information Communication and Computation Systems
          Experimental Economics, Microeconomic Theory, Game Theory