[Patents] computer doing the inventing

[Gregory Aharonian srctran@world.std.com]

>It does seem strange to have computers inventing things though.

Why?  Suppose I have a general idea for a new circuit I want to
invent.  I hire 100 electrical engineers to randomly build
different arrangements of the proposed circuits, in the hopes
that one of them finds the arrangements of components and 
component values needed for the desired circuit effect.  One
of these people does find the best mode arrangement, for which
I apply for a patent.  They are under work-for-hire agreements
so all rights belong to me, at least as the assignee, if not
the inventor, since after all they are doing experiments under
my instruction.  In the end, I get the patent, and I suspect the
above arrangements occur often in industry.

I replace the 100 electrical engineers with a version SPICE driven
by one of Koza's genetic algorithms.  What's the difference?
Same starting point, same multiple testing of different arrangements,
same end solution (though maybe better with the computer - it can
do more experimental runs).  So why not the same patentability?

Greg Aharonian