From: Eugene Leitl (eugene.leitl@lrz.uni-muenchen.de)
Date: Tue Sep 26 2000 - 00:33:56 MDT
Samantha Atkins writes:
> > You don't need a particularly fast computer to run digital evolution software.
>
> You sure as hell do if you are going to replay all of evolution from
> bacteria to human beings as was initially implied.
You don't have to start with bacteria, you could start with spiking
networks of automata.
But you need a really really large computer (say, a cubic foot of
computronium, or all nodes on the global Net in a 15-20 years) to
figure out an efficient way to even mutate the substrate. Afterwards,
the evaluation of fitness function will take progressively larger
resources, the smarter the critters become. Picking up a ball thrown
to it can be evaluated in under a second of subjective time,
evaluation of higher-order task performance could take hours of
subjective time.
Right now one would want to evaluate fitness using physical robots,
because, even considering time necessary to build them, simulating a
complex reality in realtime is not yet possible. But notice the
virtual league in robot soccer.
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:31:12 MST