Re: [computer-go] citation

[Don Dailey <drd@xxxxxxx>]

9x9 go  has been compared  to Chess in  complexity, but even on  a 7x7
board the branching  factor for go is initially 49,  and it's about 36
in  chess.  So  9x9 Go  has a  harder lookahead  in this  sense  in my
opinion because half of the game must be played out in order to get to
the same branching  factor.  By the time half the  game is played out,
the game is usually over.

The complexity of Go isn't  that much related with branching factor in
my opinion.  If Chess was played  on a much bigger board and with more
pieces, it would still play strong relative to humans.

In chess, a search will tend to  find the best move in a great deal of
cases,  even   when  the  program  doesn't   "evaluate"  the  position
correctly.  It's  a kind of  side effect of  a chess search.   That is
because pieces interact  in such a way that short  term issues tend to
dominate.   I have seen  my own  chess program,  and many  others play
flawlessly, even  while being completely  clueless wrong about  who is
winning.   Playing a good move in Go very often means having a really
go understanding of the whole board.

- Don
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/