The hardware is necessary. Hence patentable.
The problem occurs
when the model being used to claim software is math doesn't include a random
number generator, and the hardware does. So all the patent attorneys are going
to add the phrase, "On a system with a potentially real random number
generator," and claim that makes everything patentable because the random number
generator isn't included in the model used to say that software is math. Based
on past experience, the Federal Circuit will likely say that sounds good to
them.
So we are back where we started, expect the phrase on a computer
has been replaced with on a computer with a random number generator.
If
anyone is really interested in ending software patents without assuming that
enough of the Supreme Court Justices have a clue, they should start working on
an extension of the Turing model that includes some random factors, networks,
and the like.
There is already a concept of turning machines with
oracles, but I don't know if it has been fleshed out enough.
My guess
is that this is quite doable, and would probably
make a good Ph.D. thesis for a
CS Grad student. But it something that needs to be done if you don't just want
the patent attorneys to yell, "It's coming straight for us." before shooting the
deer.
[ Reply to This | Parent | # ]
| 
