RE: [computer-go] Modern brute force search

["Mark Boon" <tesuji@xxxxxxxxx>]

> Go has the advantage that mirrorring is possible at the start of the game.
>
> I bet the average go pro can mirror his own signature from left to right
> and from top to bottom.
>

If you're implying a Go pro can revert to mirror-go let me tell you then
there are ways to defeat it. Not always obvious for an amateur, but very
obvious to a pro.

If you mean to say that the simmetry reduces the number of winning 1st
moves, that seems rather obvious. There are only 55 distinct starting moves
to begin with, but after that the number increases.

I don't know much about Chess, so I'm willing to take your word for it that
there's one starting move that wins for White, e4, but as Don has pointed
out, that doesn't mean there's a single winning line of play. Even if
there's only one winning move for White at each White ply, there are some 30
moves to refute at each Black ply. This was also already pointed out before
by Don I believe. So there's never such thing as a single winning line, it's
always a tree. When the average number of moves in these games is 20 moves,
the tree still gets a 30^20 size. Even if a pro has to remember only a tiny
fraction of this tree, the number remains huge.

Think about it, if the winning line was small and easy to remember, wouldn't
they have already discovered it?


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