Re: libdts patent issue?
On 7/19/05, Nathanael Nerode <email@example.com> wrote:
> Arnaud Engelfriet wrote:
> >Here's a claim that would _not_ be maths as such under European law:
> >"A method of encrypting a bitstream A using a key B that is the
> >same length as A, comprising computing A XOR B".
> That *is* math. If a judge has ruled that it isn't, he doesn't know what the
> hell he's talking about. The bitstream A, the key B, and the output are
> mathematical objects, and the "method" is a mathematical algorithm. Unless
> it involves measuring the local wind velocity or some other non-mathematical
I think you're missing Arnoud's point. It's not math, it's an
application of math to the problem domain of message encryption. That
makes it statutory subject matter for patenting, which math as such is
not. And it doesn't need to spell out some particular way of
converting back and forth between physical observation and bitstream;
the relevant courts (in the US and, AIUI, in Europe) have gotten over
their Platonic muddle when it comes to "abstract ideas" vs.
"processes" and "machines", and focus instead on the requirement that
the invention solve a practical problem.
> Everything I said remains true. If a judge really has ruled that that isn't a
> mathematical method, then there is indeed caselaw which is wrong as a matter
> of fact. Which would suck, but there you are.
Case law isn't dispositive in Arnoud's system, and where it is it is
only dispositive on points of law, not points of fact. Appeals judges
in the US have the discretion to review factual determinations only
under a "clear error" standard, and even that is constitutionally
denied them with respect to jury rulings -- but in any case that
"clear error" relates only to the facts in the court record of that
trial and no prior case or external knowledge is supposed to be
applied. A case could get the facts completely wrong and yet be an
excellent precedent on a point of law.