Re: EPO President: Patents unsuited for software

[PILCH Hartmut <phm@a2e.de>]

On Thu, Dec 05, 2002 at 11:56:55AM -0500, GregBot reiterated:

> >In an interview with the newspaper, EPO president Dr. iur. Ingo Kober
> >is quoted as saying that he is "sceptical about patenting of software"
> >and believes that copyright is more appropriate, and, if idea protection
> >is really needed, a tailor-made system would be more appropriate than
> >patents.
 
> Of course idea protection is what people want for their new software ideas,

Or for their political / musical / poetic ideas or whatever idea.
It is always nice to own a basic ressource.  The question, as Kober 
correctly put it, is: do we need such an "idea protection" ?

> since protecting the expression of software (the layout of the source code,
> such as indenting patterns, use of comments, variable names) is completely
> useless protection.  

It can become useless, if someone else with an "idea protection" tells me
that I may not use or distribute my copyrighted work.  Otherwise, with no
patents around, software copyright was a basis for the growth of the largest
and most successful software companies in the US, see statements of

        http://swpat.ffii.org/archive/quotes/


> This immediately make copyright pretty much totally inappropriate
> for protecting software ideas.

Copyright applies to whatever aspect of a work is individual, at
whatever structural level.  But it happens that when things get as
abstract as general ideas, there is little individual creation which
one could "protect".  And since "protecting" general ideas would mean
an undue infringement on property rights of others, copyright cannot
apply to structures at a very high level of abstraction.

> Indeed, the courts in Europe have tended to deny copyright
> protection to elements of software as have the US courts (especially
> if they understand software).  The infamous, and now totally
> ignored, Whelan decision here in America tried using copyright to
> protect the structure, sequence and organization of software.

It would be interesting to look at this decision.  If for some reason
some abstract-level structure in software could be deemed worth some
kind of appropriation, copyright would have the means for that, and
registration of property scope is also possible.  There are copyright
offices which register claims.  However if copyright is extended to
such a level, it needs to have a concept of half-life period: the
more high-level the structure, the shorter the term of appropriability.
The individuality in a high-level structure, if existent at all, tends
to become generality after some time (e.g. through de-facto standards).
Thus, as long as copyright is conceived to inflexibly last for 70 years,
it can only apply to implementation-level structures.  The only reason
for sticking to this inflexibility seems to be the difficulty of establishing
and revising an international standard such as the Berne Convention.

> The decision was so bad that sometime thereafter the US Copyright
> Office explicitly stated that it would ignore the decision and not
> allow such protection of software. Copyright is great for protecting
> the shipped executables of software (i.e. the products of software
> businesses), not the expression of the source code nor the ideas in
> the source code.

It is also useful for the source code.

Look e.g. at the business model of TrollTech, SAP and many others, who
publish source code with proprietary licenses.

> And why would anyone want to allow copyright to seriously protect
> software ideas, given copyright's excessively long term of
> protection?

Kober didn't propose to use copyright for ideas.

> You really want 100+ years of copyright control versus the 20 years
> of patent control?  You might respond that copyright doesn't give
> exclusive control, so 100+ years doesn't matter, which is a message
> to businesspeople to avoid copyright to protect their new software

Maybe a message to ignorant business people who will trust only
absolute "protection" with codified claims.  Not a message to real
software business people.

> ideas - ideas in the sense of the EC Copyright Directive and TRIPS
> prohibition against using copyright to protect ideas.


 
> So what then?  Some sort of tailor-made/sui-generis system for protecting
> software?  Nice try, but at least all of the sui generis systems for
> software proposed here in America (such as the Manifesto) pretty much
> reduce to some hybrid of patent and copyright, which is why they have

The reasons why they did not gain momentum are political ones.
Similarly the reasons why the patent movement has gained momentum are
rooted in the political economics of the patent community.

        http://swpat.ffii.org/analysis/inflation/

> never been seriously considered.  The Manifesto had its bit of playing
> games with definitions.  Instead of saying that software is a process
> (which immediately implies its lack of copyrightability and pretty much
> throws software into the patent world), the Manifesto said that software
> has "behavior", a word rarely if ever used by mathematicians and engineers
> to describe software.  Any tailored-made system for software runs into the
> problem of how to disclose and protect the processes and structures in
> software without using the language and techniques of the patent system.

Even if concepts such as codified claim scopes are used, that is not
specific to the patent system nor is it a problem.

The problem could be how to establish an integrated system with clear
division of roles and clear limits between copyright, sui-generis software
property and patents.  Which imho would not be so much a theoretical as
a political problem.  

Anyway, Kober did not say that a sui-generis system was needed but only
that "if property in ideas is needed at all", something new should be
designed instead of patents.

> Very hard to do so without introducing undefined terms, or terms that
> concede a lack of seriousness such as "as such".

???
 
> Since software is as much process, method, structure and system as
> hardware (especially when both software and hardware designs are expressed
> in languages such as VHDL/Verilog or SDL), if you are going to protect

VHDL/Verilog/SDL creations are software creations, which are, like all
software, ultimately embodied in some physical form.  The final chip is
a write-only data carrier.

> hardware with patents, you might as well protect software with patents.
> Dr. Kober might want to consider reading the reports from endless numbers
> of EC funded projects for system design tools that blur the boundaries
> (economic and engineering) between hardware and software.  Read any such
> paper and you will increasingly see the term "hardware/software" being
> used together in every aspect of product development from design to
> manufacturing - with no "as such" anywhere.
> 
> Just give the EPO (and PTO) examiners enough time, money and resources
> to do good searches and examinations, and require software patent
> applications to associate the software with a realworld application
> that is specific/concrete/tangible/useful, and most of the perceived
> problems of software patents go away.  

"Specific/concrete" seems to mean "no protection of abstract structures",
i.e. only coypright-level property.

If that is what is meant, the problems will indeed go away and you needn't
give examiners a lot of ressources (such as 5 billion eur / year for curing a
small part of the problem, see http://swpat.ffii.org/analysis/cost/).

But I am afraied GregBot is just playing with ill-defined words.

> Tests for novelty, inventive
> step and enablement can be extremely effective in rejecting the vast
> majority of software patent applications without having to invoke
> ill-defined phrases like "technical".  But tests require resources,
> resources currently being denied to patent examiners.
> 
> If all of the time and money spent in the last ten years on talking
> about software patents was instead directed to improving the quality of
> the issued patents, we would be having none of these discussions.  The
> problem of software patents (much like electronics patents in general
> which have serious quality problems as well) has always been logistics
> - the logistics of providing examiners with the time, money and resources
> to do their jobs.  Yet this logistics problem is pretty much ignored by
> most on both sides of the Atlantic, which is why I am skeptical that
> anyone really cares about software patent quality.  The problem is too
> easy to measure and solve to have to suffer these endless ill-defined
> philosophical arguments.
> 
> Here's a quick question - what is the average amount of non-patent prior
> art cited in EPO patents by IPC subclass?  No one knows.  How very sad.

Probably Greg knows.  It is one of these questions which you can
answer by fairly straightforward statistical methods.  But the answer
doesn't help us much, and the question seems to me like another
attempt at distraction and derailing of more basic discussions.

Apparently the patent expert metaphysics which has occupied too much
of my attention is not as powerful in clouding the minds as one might
think.  Politicians and PTO administrators are largely untouched by it
and still able to grasp the problems in more basic terms and to talk
common sense.

-- 
Hartmut Pilch, FFII & Eurolinux Alliance              tel. +49-89-12789608   
Protecting Innovation against Patent Inflation       http://swpat.ffii.org/
130,000 signatures against software patents      http://www.noepatents.org/