Fwd: Re: [Patents] Re: EPO/McCarthy logic

[Carsten Svaneborg <svanebor@mpip-mainz.mpg.de>]

The continued discussion. I agree with issue regarding
public discussion.

Subject: Re: [Patents] Re: EPO/McCarthy logic
Date: Tue, 11 Mar 2003 11:30:57 +0100
From: Carsten Svaneborg <svanebor@mpip-mainz.mpg.de>
To: svanebor@mpip-mainz.mpg.de

On Tuesday 11 March 2003 09:36, Bernard Lang wrote:
> PS
>   what I meant is that these properties (speed, size, ...) are
> properties of object code ...  not of source code. For a computer
> scientist source code defines functionality.

Naturally you can let the computer optimise code, either
by known optimisation techniques, by trial and error using
geneting algorithms, or by partial evaluation.

However, optimisation only works well the more information
you can supply the optimiser with (which is the whole idea
behind partial evaluation). Within computational physics you
would not expect the optimisation routines of the compiler,
to know clever ways of rewriting the equations of motion, or
to simulate interactions between charged particles with
periodic boundaries, which is a known and notoriously hard
problem because the potential is only conditionally convergent.
This requires pen and paper optimisation.


Still, I we all agree (I expect) that getting a patent granted
simply because it has "further technical effect" is completely
nonsense, because any program written by a not completely
novice will have lots of sources for "further technical effects".

The root issue of this thread is the threat of MacCarthys
amendments are accepted with a restrictive definition
of technical,  then that will the "further technical effect"
concept (as Xavi argues in [15]). HOWEVER, without such
a restriction, the MacCarthy amendments will instead
endorse the concept, and will require all EU nations also
to do so.

Which is ultimative badness, because then non-technical
inventions with trivial "further technical effect" known from
prior art can be patented.

--
  Mvh. Carsten