Re: [gentoo-dev] European Patentability rules

[Stuart Herbert]

On Thursday 28 August 2003 9:59 am, Sven Vermeulen wrote:
> You have IP in the form of copyright and licensing.

Licensing is a subset of contract law, involving the lending of rights or 
assets from one business party to another.  IANAL, but as I understand it, 
licenses don't provide legal recognition of IPR.

> If they happen to find a similar algorithm doing the same thing due to
> their own research, they should be able to use it anyway.

And this is the crux of the real problems (imho) with software patents.  
Because software is typically made up of hundreds (if not thousands) of 
processes at a time, it seems impossible for programmers to avoid 
independently inventing the same processes time and time again.

This alone would seem to fall foul of the UK requirement that:

    "To be patentable your invention must ... involve an inventive step.
     An invention involves an inventive step if, when compared to what
     is already known, it would not be obvious to someone with a good
     knowledge and experience of the subject."

> IP protection is to _protect_ the _investements_ you've made so that other
> companies/people don't use your _inventions_ without having to go through
> the same timeline (idea, research, development, simulation, testing,
> production).

IP protection is to create a legal recognition of your IP.  It creates an 
asset, and a "territorial right" (quoting from the UK Patent Office) to legal 
protection of that asset.

> With normal patents, this IP protection is settled. However, software
> patents are much worse; they allow anyone to protect an idea, a
> general/generic look and feel of any algorithm. You prohibit others to even
> think about implementing something like your invention, even though they
> will have to go through the same cycle (irdstp). This is not IP protection,
> this is creating a monopoly.

Now we're back to the evils of the *system*.  I don't support the current 
system, and have already said that.  Let's move on.

IP protection *does* allow you to create a monopoly of sorts by its very 
nature.  That seems unavoidable, unless you wish to do away with IP 
protection.

> No, since remote access is a concept. Their implementation can and should
> be protected (dunno how it's called with HP), but they shouldn't stop
> others to develope a remote access implementation different of their own.

I agree that any protected IPR should be specific in nature.

> If I would have invented a way to predict register values before an
> algorithm has completed, I can implement this in a compiler to produce
> faster code. To protect my IP, I should protect my implementation. However,
> if I patented the concept of value-prediction, I make sure that no-one can
> ever develop something similar (or even better).

Similar, yes.  Better - no.  The vast majority of UK patents (and, I would 
wager, this is true for other systems too) are for incremental improvements 
to existing inventions.

> This is completely against the spirit of patents. Patents should promote
> research and development by giving the inventor a means to protect its
> investements.

You've confused me.  How is this statement in conflict with the idea of you 
protecting a concept of value-prediction?

> Even more, if _you_ develop software, do you check all existing patents to
> be sure that your software doesn't cross one of them? I'm sure that you
> don't.

You're right, I don't.  But there again, I live in a country where most 
programmers believe that software patents aren't part of our legal system.

> John Gage, the Chief Researcher and Director of the Science Office of Sun
> Microsystems, had a nice talk on the ONEday conference in Genval, Belgium
> on the 30th of July in which he described his PoV on software patents. One
> of they key items in his speach was that most _real_ software inventions
> (protocols, internet, ...) were inventions of students or people that have
> just graduated, people that don't think about patents. These inventions
> have changed the ICT-world (TCP/IP, WWW, ...).

I always thought TCP/IP was invented for the US DoD?  Vint Cerf does seem to 
be talking about the ARPA network in his 1973 paper where TCP is first 
mentioned.

I don't know what the situation was back then, or what it is now, but I do 
know that, in the 80's and early 90's, if NASA sold a piece of software, they 
weren't allowed to license it - you bought the whole thing, (non-exclusive) 
rights and all.  That's how NQS became the de-facto standard for UNIX batch 
processing.  NQS was written for NASA, who then sold implementations of 
Cosmic NQS to companies such as Cray, Sterling Software (both of whom 
released their own commercial versions) and Monsanto (who released the whole 
thing under the GPL).  When I succeeded Monsanto's John Roman as maintainer 
of the GPL'd NQS, I was led to understand that this ceeding of rights wasn't 
unique to NASA's charter, but more widely applied to work done by the US 
Government.  If true, wouldn't that mean that TCP/IP was never going to be 
patentable in the first place?

> Patents are for real, physic inventions,

Sorry, but that's not true, at least not here in the UK.  The oldest surviving 
patent issued in the UK was in 1449, for a process of manufacturing stained 
glass windows for Eton College.  Today, the UK Patent office very clearly 
states:

    "Patents are generally intended to cover products or processes that
     possess or contain new functional aspects; patents are therefore
     concerned with, for example, how things work, what they do, how they
     do it, what they are made of or how they are made."

Like copyrights, a patent is a business asset that can be licensed if 
required.  No asset - no license.

Best regards,
Stu
-- 
Stuart Herbert                                              stuart@g.o
Gentoo Developer                                       http://www.gentoo.org/
Beta packages for download            http://dev.gentoo.org/~stuart/packages/
Come and meet me in March 2004                 http://www.phparch.com/cruise/

GnuGP key id# F9AFC57C available from http://pgp.mit.edu
Key fingerprint = 31FB 50D4 1F88 E227 F319  C549 0C2F 80BA F9AF C57C
--