[Patents] Legal definition of "non-obvious".

[phm@a2e.de (PILCH Hartmut)]

> >For the concrete-physical solutions, the matter is different, because
> >they are based on new empirical knowledge that can't be obtained by
> >inference or any other computing operation.
>
> Once again, thanks for another reason why software and hardware should
> be equally patentable,

at best this can be a reason why it is difficult to distinguish, not why
both or any one of them should be patentable.

> for this assertion by Hartmut is false.  There
> are artificial intelligence programs that analyzing nothing more than
> raw data and basic operations in some field, can discover empirical
> knowledge.  For example, one AI program many years ago, given little
> more than sequences of integers and basic operations on integers (such
> as + - * /) was able to discover the existence of a set of integers
> with no divisors either than one, i.e. prime numbers.

A set of prime numbers =? physical knowledge

> That is an
> empirical discovery made by computer operation.

No, that is a discovery of mathematical knowledge.

> Another AI program discovered Ohm's law, voltage = current *
> resistance.

I assume that it deduced that law from other logically equivalent
knowledge which had already been fed into it.

> So Hartmut's proposed robot test applies equally as well to the
> "physical" as well as to the "informational" (as much as I hate using
> such mostly meaningless scientific terms).

These are not "scientific" terms but commonplace philosophical
terms that are at the basis of all discussions about science (=
epistemology).

> "Empirical knowledge" is
> not a criteria for distinguishing patentability - I can design
> computer programs in which a concrete-physical domain is specified,
> and then have humans and/or computer programs either discover new
> empirical knowledge or solve specific technical problems, be the
> domain be mechanical, chemical, electrical, software, hardware, genes,
> etc.

Exactly:  you can feed models of already-known physical knowledge into a
computer and let it solve problems on that basis.

> And for all such domains, such new inventions should be
> patentable.

Solutions that can be obtained by an inference machine based on
a formalisation of the problem and of prior knowledge should be
patentable?

Such solutions are, by current definitions, not novel, not inventive
and not technical.

> And based on his comments, it is obvious that the German courts are
> as unaware of developments in a variety of fields of AI, software
> and mathematics, as is Hartmut - all the more ironic because some
> of these development efforts have their leaders in Germany.  Your
> ignorance of your fellow German researchers is a shame, since they
> do some really good work.

Which development efforts?
How do they affect the current discussion?

Everything you said just proves my point.

I here get the impression that you are suffering from a memory deficiency.
You just remember conclusions you have once reached and decided to believe
in.  You no longer remember how you reached them nor how other people
reached their conclusions.  That's why you always need to reiterate and to
support your reiterations with assertions about some fancy "scientific"
authorities.

--
Hartmut Pilch
Federation for a Free Information Infrastructure        http://www.ffii.org/
Protecting Innovation against Patent Inflation        http://swpat.ffii.org/
For a software patent free Europe PLEASE SIGN http://petition.eurolinux.org/