Quoting Dave Dyer <[EMAIL PROTECTED]>:
I wonder if MC programs shouldn't prune game branches when
sufficiently large captures occur. The loss/win might not
be strictly allocated to the right player, but it certainly
means that the current game has entered sillyspace.
Are you talking about pruning a full board search tree or deep down a
simulation?
But the question might be: What move wins me the game even if something silly
happens? If the move examined at the root is good maybe I can still win if a
second silly thing favoring me happens?
Playing a random game does enter sillyspace immediately so I do not see an
argument in reducing sillyness. But the sillyness must be fair.
The MC method finds moves that are robust against random events. I see it
strongly linked to the ideas of playing honte and good shape. In short term
other mover are tempting but it is the moves that are efficient to the end of
the game that are truly good.
Also think of a position where two groups of opposing colors are dead. If the
smaller group has less liberties it will captured earlier and the
position will
often be misevaluated. So sufficently large as you wrote is probably
very close
to large enough to safely conclude the loss/win. But here there might be a
tradeoff between speed and accuracy which leaves it as an open question for
experimentation.
-Magnus
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
