From Luuk.van.Dijk@xs4all.nl  Tue Aug 14 21:42:40 2001
From: Luuk.van.Dijk@xs4all.nl (Luuk van Dijk)
Date: Tue, 14 Aug 2001 22:42:40 +0200
Subject: [Patents] Re: EPO 1978: programs vs inventions (fwd)
In-Reply-To: <200108141839.OAA05410@world.std.com>; from srctran@world.std.com on Tue, Aug 14, 2001 at 20:39:26 +0200
References: <200108141428.KAA20155@world.std.com> <200108141839.OAA05410@world.std.com>
Message-ID: <20010814224240.A1336@wonder>

On Tue, 14 Aug 2001 20:39:26 Gregory Aharonian wrote:
> >Whether or not to allow software patents is a question of law and
> >politics.  NOT a question about physics.
> 
> I completely disagree.  The words used by lawyers and politicians to talk
> about software patenting, words such as 'process', 'method', 'machine',
> 'algorithm', are words whose meaning is defined and reasoned about by
> using mathematics and physics.  So one cannot have political and legal
> discussions about software patenting without discussions of the underlying
> mathematics and physics. 

It is interesting you should say so.  

At the workshop at the LinuxTag in Munich a couple
of weeks ago, someone spoke about the distinction between the physical and
the non-physical, to which a lawyer from the Max Planck Institute commented
that the scientific definition of 'physical' and 'natural forces' are of
no consequence to the legal definition of those concepts, and that lawyers
are free to pose their own definitions, without referring to the Real World(tm).

Of course, this was to support his opinion that software is 'naturally' patentable.

So, apparently, there are people who use exactly the opposite of Greg's arguments to
conclude that software should be patentable too.  This is to be expected
if this discussion is just considered a word game by some.

I think that means there is also the political choice of wether we want lawyers to invent
their own inconsistent and unreal world view and have it imposed on Innocent Law Abiding Citizens.

Now to return to Greg's mode of reasoning: 'process', 'method', 'machine' and 'algorithm' may
be defined in a strict mathematical sense, but none of the definitions I know of link
any of these four concepts to the known natural forces (being Gravity, Electroweak interaction
and the Strong interaction).

As a matter of fact, the theory of Turing and his universal
computing machine was intended to model /all humanly possible mental processes/.  It is the
widely accepted 'Church Thesis' that the class of algorithmically computable functions 
(in the intuitive sense) coincides with the class of partially recursive functions, i.e. the
class of functions that can be calculated by a Turing Machine. 

So we should not be surprised that when one tries to make programs (i.e. Turing Machines)
patentable, at once /all humanly possible mental processes/ turn out to be patentable, because
it is no longer possible to draw a line.  That line may already be missing in the US, but
in Europe some of us would like to keep it.

I happen to agree with you that legal concepts should ultimately be defined consistently
with the 'Real World', but that, imnsho, leaves only one possibility: to draw the line
between physical and pure mathematical, and software is on the non-patentable side of this line.
(for that matter: even if the software happens to be in VHDL, and implemented as a circuit
you can patent the circuit, but not the algorithm).

The political decisions that remain are
	a) do we want laws to be logically coherent
	b) do we want the non-physical patentable

But no matter how you turn it upside down: the distincion can be made, and can be made
logically coherent, and on sound physical principles.


Luuk van Dijk
___________________________________________________
Mind over Matter                     lvd@mndmttr.nl
The Netherlands                tel +31 6 224 97 227
___________________________________________________



