Re: [computer-go] SlugGo v.s. Many Faces update

[Chris Fant <chrisfant@xxxxxxxxx>]

After I sent that, I realized that my description was needlessly
complex.  Sorry.  A higher level way to think about it is the error
between the overall score estimate and the actual final game score, at
each point in the game.  This just adds all the points/intersections
together.

The expected value of an open point at the begining of a game would be
close to 50% chance of black.  If you consider two such points, on
average, one of them would become black territory and one of them
would become white territory.  I suppose a perfect player would not
have a 0.0 metric, but but somewhere very close to that at least.

I suppose you could design an engine to thwart this metric entirely by
playing for a near tie, no matter if you are winning or losing.  Such
as by each player building life on one side of the board and then
spreading to the middle.
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/