Arlene McCarthy's software patent "factsheet"

Posted Sep 4, 2003 13:01 UTC (Thu) by forthy (guest, #1525) [Link]

Actually, you can see from her fact sheet that she things common practice of the EPO
is more or less ok, but that her proporsal is more restrictive (which is in fact the case).
Arlene for sure has a non-technical education, however I don't know what exactly she
studied (her bio is a bit too terse for a politician - if I were to employ here, I suspect
she's a drop-out).

The main point against software patents is that the technical invention and the
non-patentability of algorithms as such are contradictionary: Algorithm selection (for a
particular technical purpose) is the trivial task of a computer scientist. Any ordinary
computer scientist should be able to research and select algorithms for tasks at hand -
which then solve technical problems. The "inventive" part of CS is to create new
algorithms. Application is the non-inventive part.

As an example, you can use the GIF patent. This would still be legitimate, since it has
a technical purpose: reduce network transfer time for images. The algorithm in question
is LZW, an algorithm that can compress almost anything. The mere selection of LZW
and application on images dedicated for network transfer is an "invention". As
McCarthy says in her "factsheet", other uses of LZW still are free. However, application
of a compression algorithm in order to reduce transfer time *is* something that's trivial.
Which compression algorithm gets chosen depens on how good the CS engineer can do
his research.

We brought that one up when a EPO representative told us (at the university) about
the possibility of software patents - something the entire audience violently rejected
(people did only come in to listen because they wanted to have their concerns heard).
We asked him if he thought it was a clever idea to apply a standard compression
method on an image in order to reduce transfer time, and he said, yes, if it's the first
one who does that, it's a clever idea. We told him, no, it's not a clever idea, even if it's
the first one to do, because it's obvious. The only non-obvious part of the GIF patent
actually is the LZW algorithm itself, which is not patented as such.

IMHO the "technical innovation" part does put a restriction on software patents. It
restricts them to those software patents that can't be done without computers. That's
a good thing, but far from being sufficient. Software patents (despite there are so
many) don't cause much harm, because most companies use them defensively only
(like IBM vs. SCO). Sometimes, a company digs out such a defensive patent, and
starts to sue (example again: GIF patent). Mostly when they got that patent form a
merger.

A good legislation for patents (to follow the spirit of the patent system):

Algorithms may not be patented.

Mere applications of algorithms are considdered "obvious".

A patent involving an algorithm must be dominated by non-algorithmic inventions. This is
the case if the innovation does need an algorithm to work. Only the non-algorithmic
parts of the invention actually are protected by patent law. The disclosure of the
algorithm in question is necessary to understand how the invention works.

We won't get anything like that anytime soon.