Re: [Fsfe-uk] Fwd: UK Leading Patents?

[PILCH Hartmut]

> With electronics emulation, there is essentially a one-to-one mapping
> between the physical and the software: that mapping would allow you to
> implement any patented device without doing any work, and thus it
> wouldn't be worth patenting the device in the first place, even if we
> agreed that the patent was a good, novel, one.

A patent doesn't become good by just being novel. Apparently your
electronics idea is a non-invention: a new logical idea which is
translated into circuitry by known means.  I don't want logical ideas to
be patentable.  If the translation into circuitry (materialisation) is
novel, then that could be patentable.

> > bad aas such ;op. The problem is when you don't define "technical" as in
> > DispositionnenProgram (sorry, one day I'll learn  to spell it), but you
> > leave it open
>
> I disagree this is the problem. Attempting to fit UK law into German
> won't work; the two are incomparable. I've seen no example so far of how
> using 'technical effect' and case law rather than dispositionnprogram
> has led to greater bad patents.

What about the Nymeyer example from 1980 that is cited on

        http://swpat.ffii.org/vreji/gasnu/uk/

and that I cited here yesterday?

> Remember the following:

> * bad patents occur for a number of reasons - if a software patent is
> granted, it's not necessarily because it is deemed to have a technical
> effect;
> * the granting of a patent is a mere registration - there is no
> guarantee that the patent would hold up in court (that is the
> responsibility of the registrant), nor will the UKPTO protect
> patents/take sides in patent disputes.
>
> Thus, not only would it be possible to find bad UK patents, you'd
> probably expect it.
>
> You may not like the fact that bad patents can become registered only to
> need defeat in court later, but again, that is somewhat similar to our
> case law method. Bad patents will always be registered, as will software
> patents, no matter how obvious Article 52 is. The problem is when they
> are enforceable. I don't believe a software patent would be enforceable
> in the UK today.

With the doctrine outlined by Xavier and by the Dispositionsprogramm
decision

        http://swpat.ffii.org/vreji/papri/bgh-dispo76/

software patents wouldn't issue in the first place, because the doctrine,
unlike the novelty or non-obviousness test, is usually quite easy to
apply.

Why can't a proven good concept be applied to the UK, particularly when we
are talking about a european directive ?

> >  What do you  find contradictory exactly?. Confusing?. Unsupportable?.
>
> The fact that you can have two black boxes, and one of them is
> patentable and one of them isn't, even though they do precisely the same
> thing. The difference is one is built out of FETs and the other DACs and
> processors.

One of the fundamental problems of recent patent inflation is precisely
the patenting of black boxes.  You shouldn't get a patent for what you
achieve but for how you do it.  And this how should include forces of
nature as a constituent.  Software patents are in general functionality
patents, black box patents.  That is wrong in the first place.

> But, even though I find your argument contradictory, don't mistake that
> for me supporting the CEC proposals; I don't. I just don't support them
> for different reason: the fact that software is not deserving of any
> exploitation right due to small R&D costs. So, my black boxes would be
> in the same state as yours: one patentable, the other not. But I find my
> argument consistent.

If I understand correctly, you want the black box to be patented and thuse
the software solutions, which usually seem to exist in the cases that you
have in mind, to also fall under the patents.

Maybe I don't understand the cases that you have in mind.  I have never
seen any patents for which they would fit, and I have browsed through
hundereds of them in order to find difficult borderline cases where the
Dispositionsprogramm doctrine could be inapplicable.  Without success.
Could you perhaps name a few example patents?

> > But I don't think you can tell the patent office to  make an economic
> > assessment of each particular case and see whether it is worth it,
> > without any guidelines.
>
> Sorry, I didn't mean to suggest that was how I thought it should work.
> The basis for me being against software patents is that the
> justification for patenting in general doesn't hold for software. The
> patent office shouldn't be making an economic assessment, it should just
> throw the application out :)

If neither by economic assessment nor by a clear rule (the only candidate
so far mentioned being that of technical invention as in
"Dispositionsprogramm"), by what then?  guts?  feeling?  measurement of
time it took to come up with the idea?

> They don't decide policy. But, they do give advice on it, and ultimately
> are part of the same system. If the feedback they get is pro-patenting,
> then I would expect them to lobby for increased patentability. It is
> probably the case that the UKPTO hasn't been lobbied sufficiently by the
> anti-patent lobby.

The feedback they get is always pro-patenting.
They have before them a poor applicant begging for a patent and only
rarely someone disputing the patent in an opposition procedure.
More patents means more power for them.
The behavior of the UKPTO clearly that talking to them, even in a public
consultation, doesn't change their mind even by a millimeter.

> > So we better find a good criteria to decide what to patent. I think the
> > harnessing nature idea works well to make that choice.
>
> I would leave it at 'software is not included'. Technical effect or
> dispositionnprogram or whatever, it shouldn't matter.

which according to your criteria as exposed before may mean nothing.
You would at least have to come up with a list of examples of what should
be refused and what granted.  If you prefere precedent and analogy over
rules and theories, then that's the way to go.  A combination of both
would be even better.

> > Or more genuiney: what do we do with VHDL ? For me it is software, but
> > you may not call it so. It is logic, after all. So  we better exclude
> > logic and that will save us trouble
>
> VHDL I would leave patentable, I _think_ (I don't know it at all). To
> me, it appears to be a description of a system, not the actual system
> itself.

Patents are granted not for systems, but for inventions.
The form of materialisation (circuit or diskette) does not matter.

> VHDL compilers are supposedly very sophisticated, and I would suspect
> the actual device created would not actually be based on the VHDL
> aside from the fact that the devices are functionally identical.  At
> the end of the day, everything in this world is mathematical.
> Algorithms are what the universe is made of, everything we do is
> algorithmic. Defining what is mathematical and what is not is
> ultimately useless.

Not true.

We may be able to model a large part of the material phenomena. But the
ability to model them is already the result of a completed series of
discoveries and inventions.  We don't want to continue handing out patents
when everything is already modelled and there is nothing left to invent.
We have to a certain extent arrived at that situation already, and that is
one of the reason why patent offices are letting all the criteria erode.
If they went back to strict criteria of patenting only inventions that
deserve it, we might soon reduce the number of patents issued by 99% (the
technicity criterion alone, in its lightest version, does not necessarily
cut away more than 3% though).  You would however always have an area
where nature is not exhaustively modelled.  A quite large one by the way.
The universe is not created by the human mind and does not consist of
algorithms.

See also the Einstein quotation and the following text at

        http://swpat.ffii.org/stidi/korcu/index.en.html

> I would fall back on the economic argument: electronics R&D is
> expensive, as is manufacture. Electronics researchers seem deserving of
> patent protection; software researchers do not.

Maybe the electronics researchers you have in mind would be better served
with utility certificate protection or something electronics-specific,
more flexible than the one-size-fits-all patent system.  The patent system
imho is best reduced to its core area of harnessing nature.  As said
above, the reduction can take place with various bells and whistles,
cutting somewhere between 3% and 99%.

> > I want that in case you patent,say, an engine in which a sensor feeds
> > pressure data to an embedded computer that outputs control signals for
> > fuel valves,because it wasn't previously known that the pressure
> > sensed there would be an useful input to calculate the optimum input
> > of fuel, then I want  to be prohibited to sell the engine, but not
> > the embedded computer software

> This doesn't seem like an example of a combined physical/software
> patent. You did say that you had no problem with software forming part
> of a patentable system: if that is the case, though, taking the
> software away must make the patent invalid, because otherwise it's not
> part of the patent.

I don't understand that.
You mean "taking away the software must make the patented process
non-functional" ? or "cause the patent to be revoked" ?

Neither would make sense.
The patent is granted for a teaching about physical forces.
A computer could play a role in an embodiment of that teaching, perhaps
the preferable and only reallistically implementable embodiment, but still
the teaching would be independent of software.  It would say something
like "you mix chemicals A B and C together at environment conditions D E
and F according to formula G and thereby obtain H".  You get the patent
because the cause-effect relation of that teaching was not obtainable by
calculation based on known mathematical models.  However now you have a
model and a controlling algorithm.  To implement controlling algorithms on
a computer is normal here. Your inventon does not consist in having found
that the logic of the algorithm is correct but in having found that the
physical forces can be described by it.  The claimed objects derived from
that invention could be the chemical process and its product H, probably
a chemical.  The same kind of thing can also exist in electronics.

> Again, I find this argument confusing. I personally don't
> believe there is any such thing without admitting the possibility of
> partial software patenting in a similar manner to the UKPTO.

There is no such thing as partial software patenting, and the UKPTO is not
practising anything like it. You either uphold the limit between matter
and mind or you don't.  "Technical" algorithms are just as mathematical
and abstract as any other algorithms.  Attaching meanings to the entities
in algorithms is only a way to ensure triviality.  If you allow algorithms
to be seen as patentable inventions, then you'd better allow them
completely.  That could at least assure that we get only patents on real
advances in mathematics rather than on trivial applications of maths to
existing models of reality that any engineer would come across.

The inability of the UKPTO (and yourself) to come up with any rule for
differentiating between "technical" and other algorithms demonstrates this
point.  Read also Donald Knuths comments on this question in

        http://swpat.ffii.org/vreji/cusku/

-- 
Hartmut Pilch, FFII & Eurolinux Alliance              tel. +49-89-12789608
Protecting Innovation against Patent Inflation       http://swpat.ffii.org/
100,000 signatures against software patents      http://www.noepatents.org/