That anti-patent pamphlet I mentioned

[Arnoud Galactus Engelfriet]

Niall Douglas wrote:
> On 10 Dec 2002 at 11:49, Arnoud Galactus Engelfriet wrote:
> > I would say: "despite the fact that the European Patent Convention
> > explicitly excludes computer software as such since 1973 and the 2000
> > EPC Diplomatic Conference made it clear the EPC Contracting States did
> > not want to change this. Unfortunately, the proposed Directive will
> > effectively render this provision null and void."
> 
> A bit of a mouthful for an opening paragraph. I'll see what I can do 
> with it whilst keeping it factually correct.

Maybe you can make it multiple sentences? The fact that the EPC has
had this exclusion since 1973 is a pretty strong argument that the
situation shouldn't be changed. Then as an additional argument you
provide the example that the Diplomatic Conference did not want to
change this, confirming that apparently people still think this way.

> Here is precisely the knub of where the world went wrong with 
> software and patents. You use "implementation" in the way they do but 
> logically it is not equivalent to "implementation" with respect to 
> other patentable devices.

Ok. I see an implementation as computer software itself (ie, just the
ones and zeros), but it has to be loaded into a programmable system
to actually do anything. A blueprint for a chip design doesn't do
anything either, but if you feed it into a chip manufacturing
machine, it spits out something that can exhibit patentable behavior.

> I was using "implementation" in its software context ie; the software 
> code itself. This would equally apply to the design of the circuits 
> of the hardware because as you previously pointed out, there is 
> considerable scope for the dividing point between hardware and 
> software in an overall system.

Can you make a definition that makes the distinction clear? I don't
get any further than "the software is the bits you can't kick if it
doesn't work". :-)

> Now if we used this correct form of "implementation", I would 
> infringe on MPEG2 patents during my implementation only if my 
> software or hardware bore a striking resemblence to another existing 
> implementation. This is *separate* from the algorithms.

Well, I think for infringement it would be sufficient if your program
carried out the same steps as the patented technique. But yes,
usually this means there is some degree of similarity.

> Obviously, with some care, there can be near infinite implementations 
> of identical algorithms just as there can be near infinite 
> implementations of a pump. Thus, the MPEG2 patents are precisely the 
> worst kind of patent much in the same way as patenting all possible 
> pumps is a stupid idea.

This is actually exactly what happens when somebody comes up with a
new hardware widget. In the early 1900s, the German company Singer
had the patent on the sewing machine. Before that, there were several
patents supposedly covering the steam engine, and development of
steam-based engines didn't take off until the courts determined which
one was the steam engine patent. In 1907 an American invented the
supermarket and patented it. The transistor, in all its shapes and
forms was patented, and nobody could use a transistor without paying
royalties to Bell. There is still a big fight going on about who owns
the patent to the semiconductor.

> If you don't get what I'm on about and how I'm making this 
> differentiation, please see 
> http://mailman.fsfeurope.org/pipermail/discussion/2002-
> December/003514.html. It's effectively about the difference between 
> an ideal thing and the thing itself.

I'm sorry, that message is a bit difficult to follow for me.

> > I don't understand this. Are you saying that patents on hardware
> > implementations of MPEG should also not be permitted?
> 
> On the specific design of circuits and wiring yes they should be 
> permitted. But the current situation where it's impossible to 
> implement MPEG2 in hardware without infringing a patent indicates the 
> definition of "implementation" is wrong - this is more patenting the 
> "generalised solution" than specific implementation.

This is normal in the patent world. If you're the first to come
up with a widget, you patent the widget itself and no one can use
it no matter how they improve it or what particulars their
implementation involves. The steam engine. The sewing machine.
The supermarket. Heck, even the light bulb was patented in its
most general form.

> > The counter argument would be that without the prospect of using
> > patents to get revenue from MPEG 4 implementers, the MPEG 4 drafters
> > would not have made the investments to develop the standard in the
> > first place. My company certainly wouldn't have.
> 
> No, you could still get patents for your specific implementation 
> which if it's the best and cheapest, others will licence it from you 
> and you get your money. No one else could sell a facsimile for 
> example, but they could sell a similar solution.
> 
> This would encourage competition and improved quality.

Encouraging competition and stimulate rip-offs, yes. If a patent
cannot protect you against differing implementations of your
invention, it is of little use. Other laws are already available to
protect against people making copies of your device. The whole idea
behind patent law is to give you protection for your technical idea.

Kind regards,

Arnoud Engelfriet

-- 
Arnoud Engelfriet, Dutch patent attorney - Speaking only for myself
Patents, copyright and IPR explained for techies: http://www.iusmentis.com/