Re: [gnugo-devel] Three stone games

[Daniel Bump]

> We know to beat 9 x 9 gnugo with 4 handicap stones, you must be 4 dan.  I
> expect the difference of one increased Handicap stone to be linear.

Ratings of Go programs are problematic. When a human plays a go
program many times, the human learns about the opponent's weaknesses but
the computer does not learn about the human's. So after some practice
the human can beat the computer much more easily.

GNU Go is hard to beat on 9x9 with 4 stones but probably with practice 
a middle kyu player can do it. See the Jean-Loup Gailly's page at:
http://gailly.net/go.html for examples of giving Go programs big 
handicaps. Or see the Lazarus/GNU Go games.

> what is a kadoban game ?

A kadoban is a game you have to win. Three game kadoban is
a system where when one player wins three games the handicap
changes. The third game is kadoban.

> I also find it very difficult to understand, why gnugo does not
> increase its strength with computing speed?  If gnugo plays
> against itself with different time limits, then how many minutes
> will it need to play 1 kyu stronger ?

We could ask for GNU Go's life and death algorithms to be more
accurate at higher levels. This doesn't seem to be true past
level 10 because we haven't really worked on making that happen.
But even if this were true, many of GNU Go's mistakes would
remain, because they have nothing to do with reading accuracy.
It is like saying a person can read better if the light is
turned on brighter. It helps up to a point but if there are
words in the book that the reader doesn't understand then it
doesn't help.

Dan