[FOM] An Explanation of Computation Theory for Lawyers

[Francis Davey]

2009/11/14 Paul Elliott <pelliott at blackpatchpanel.com>:
>
> Are there any errors or over simplifications that should be pointed out
> to the author? Any other comments?

It seems pretty shallow to me I'm afraid (I am a practising lawyer
with some interest in intellectual property, but I used to teach
computation theory).

I agree that there does seem to be a view amongst some lawyers that
there is a clear distinction between mathematical algorithms and
computer programs that would not stand up to close inspection. I am
afraid the difficulties associated with a sensible discussion of
software patents are not so easily resolved.

Two specific points:

(1) The Re Alappat idea that a computer+program could be a new machine
 (in the patent sense) distinct from a general purpose computer  is an
entirely reasonable one. Its hard to make a principled distinction
between (say) a machine with levers, pulleys, cogs and so on carrying
out an algorithmic task, and a computer programmed to do it. If the
one is patentable, why should the other not be?

There might *be* reasons not to permit software patents, even of this
kind, but there's no necessary reason why the fact that something is
implemented in software means it can't function as a newly invented
machine.

The author's counter-arguments are fallacious since they assume that a
relatively obvious thing (like a car with steering wheel turned to the
right) could not be patentable subject matter, when of course they
could. The wheel-turned car would be rejected as unpatentable not at
the subject matter stage but later as being non-novel and obvious.

(2) While it is true that Microsoft Word is a proof in some carefully
chosen logic (via Curry-Howard) and running it on my computer is a
proof simplification, in practice we would never think of it in that
way. Since patent law contains a lot of fuzzy concepts, many of them
being matters of degree rather than possessing sharp dividing lines,
there's no reason why it should not reject Word as an example of a
mathematical algorithm even if in principle it really is.

I hope that makes sense. I have simplified rather so as not to bore
FOMers with law. Two notes of caution: (i) US patent law is based on a
different set of principles to European law: the latter specifically
excludes computer programs as part of the EPO treaty, whereas the
constitutional principles on which US patent law are based do not;
(ii) the terminology is also different between the two and I'm afraid
I tend to use them interchangeably.

If anyone wants further information, please get in touch.

-- 
Francis Davey