Re: computer-go: perfect play

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

> VD> Thus it is beyond anyone's power to prove anything about the strength of the
> VD> perfect player.

You don't  necessarily  need a proof.   Much  of  science is  based on
probability and statistics.  It's might be possible, for example, that
someone could present  overwhelming evidence that  a given lower bound
is clear.

In a few years, we might  be able to  prove (if it's true) that humans
are  relatively  weak as  Chess playing  entities.  We  can do this by
designing  computers that  crush the  best   human players.  This  was
suggested  by you in fact.  I  think this will take  some time, but is
inevitable.

A few years   ago, it  was  widely held   that computers would   never
approach  master level Chess.   This was a  stupid and naive assertion
and many of us knew it then.

But Go on the other hand  is pretty scary!   In principle everything I
said about Chess  applies.  But we are  starting with a much much more
difficult  problem.  I'll play it  safe  and not make any  predictions
about how good Go programs will get in the future.  However, I suspect
that humans are not  good Go players (in  the GoGod/Devil sense of the
word.)



Don