Re: computer-go: NN GA was RE: Perl Module for next move.

[Daniel Bump <bump@xxxxxxxxxxxxxxxxxx>]

Matthew Corey Brown wrote:

> Okay thanx for all the help and suggestions. I've gone with making a
> perl wrapper that executes "gnugo --mode gtp --quiet" And I now have a
> computer trying to teach itself to play go. maybe in 8 weeks it would have
> learned enough to beat me =}. (wouldn't be a big effort as i only know the
> basics) 

Partly stimulated by the discussion in this list the scoring in
GNU Go has been revamped in 2.7.239. With --score last
actually two scores are reported, so for example we have:

$ gnugo -l score1.sgf --quiet --score last
Black moyo:  69
White moyo: 210
Black territory:  74
White territory: 214
Dead white stones:   5
Dead black stones:   4
WHITE seems to win by 145.5 points (old)
WHITE seems to win by 141.5 points (new)

In this example, the second score using the revised algorithm
is correct and the first score is wrong. Probably the old
scoring algorithm will be taken down in 2.7.240.

In the gtp mode, the command equivalent to --score last
is "newscore".

Scores are still unreliable when there is a seki on the
board.

The new scoring algorithm also reports a score before the
end of the game, based on the Bouzy 5/21 algorithm. This
algorithm was previously implemented in GNU Go (in moyo.c) 
but it has been reimplemented (and tested against the old
algorithm) in the new score.c.

Dan