Re: Strawmen and Urban Legends

[Stefaan A Eeckels]

On Thu, 28 Dec 2006 13:32:44 +0100
Merijn de Weerd <merijn+nospam@realemail.net> wrote:

> On 2006-12-28, Rui Miguel Silva Seabra <rms@1407.org> wrote:
> > google were from her press releases!) -- advocate that a machine
> > which contains software may be suitable for a patent if it
> > satisfies all the criteria. Nevertheless, the software component
> > can't be patented, as
> 
> I don't understand the practical difference between a patent on
> a machine which contains a software component, and a patent on 
> the software component of that machine. Why should you be able
> to get the one but not the other?

There should be no problem on getting protection for a software
component as long as that component is novel and non-obvious, and if
patent protection is sought for that component per se. One of the
considerations should be that a lot of algorithms have been published
before patent protection was extended to software, and that anything
that is part of an elementary CS curriculum should not be considered
for patenting. This means, inter alia, that the patent terminology
(which might have been suitable for the description of largely
mechanical or electromechanical devices) should be revamped for
software, and use approaches suitable for the description of software,
like UML or a real programming language. 

The current situation is that programmers cannot understand
software-related patents, meaning that patent examiners with a CS
education could be hard-pressed to link a known algorithm to its
description in patentese.

> Also doesn't this open a very big hole, in that people can now
> distribute the component separately from the machine and leave it
> up to the end user to assemble them? That's not legal anywhere else
> in patent law (contributory/indirect infringement), why should it
> be legal for software components?

Obviously, if a device is assembled from standard or non-patented parts,
then an individual can assemble the device for personal use, based on
the public description available as part of the patent grant, without
the knowledge of the patent holder. It is also allowed to build another
device out of these parts. If a software component is not patented per
se, then its use in another device (not covered by the patent) is not
subject to the patent.

My aversion to patenting algorithms would (largely) disappear if the
terminology would be adapted so that a programmer could read a patent
and understand it. You don't want to have each line of code you write
vetted by a lawyer to ensure you're not inadvertently infringing a
patent.

In addition, it should be almost impossible to be granted a patent on
algorithms that has been published (as text or as part of a program or
library) a year before the patent application is filed. There's
something profoundly amoral about being granted a monopoly on something
someone else has invented, and it should still be possible to publish
software as speech (articles, books) and as Free software without the
risk to see it appropriated by those who don't know software but know
how to patent it.

That being said, I am still doubtful that patenting is the best way to
promote advances in CS - it has done very well when there was no
protection at all, using the good old academic principles of sharing
knowledge and building on the work of others (which after all are what
the GPL tries to ensure for software). 

-- 
Stefaan A Eeckels
-- 
He who will not reason, is a bigot;
he who cannot is a fool;
and he who dares not is a slave.     (Sir William Drummond)