RE: [computer-go] Statistical significance (was: SlugGo vs Many Faces,newest data)

["Mark Boon" <tesuji@xxxxxxxxx>]

Not anymore about statistics really, but something Don wrote is about the
chess-endgame is important.

There's always more to a game than its moves and patterns. A decision to
play a certain move is often based on expert knowledge used in a variation
that never actually appeared in the game itself. Simple tesuji and
life-and-death situations rarely appear in expert Go games because the
situation is avoided by one side or the other. Still, to reach the expert
level you need this knowledge, even if it never gets put into real play at
that level anymore.

This has severe implications for anyone who tries (or already tried) to make
a computer learn from professional games, as is often proposed here. And an
important reason why they all failed, and will keep failing in the future.

    Mark Boon

-----Original Message-----
From: computer-go-bounces@xxxxxxxxxxxxxxxxx
[mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx]On Behalf Of Don Dailey
Sent: Thursday, September 09, 2004 21:27
To: s.yeates@xxxxxxxxxxxxxxxxx
Cc: computer-go@xxxxxxxxxxxxxxxxx
Subject: Re: [computer-go] Statistical significance (was: SlugGo vs Many
Faces,newest data)


So it seems to be the case  that a lot of the knowledge neccessary for
strong play is not explicitly represented in Grandmaster games (but is
of course hidden somewhere inside them.)


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