Re: [gnugo-devel] Surprisingly good, yet missing last chance

[Evan Berggren Daniel]

This is actually a very complicated position for Gnu Go, I believe.  Part
of the problem is that there are actually three groups involved -- J13,
E19, and H19.  After move 108 (b H19), Gnu Go believes that E19 is
unconditionally dead, and that H19 and J13 are unconditionally alive.  I
think the best result for both sides is actually for white to begin
filling the liberties on J13 (L19 at F10, for example); unless I'm
misreading, J13 dies, allowing E19 to escape, but H19 can get two eyes in
the process by catching the J17 stone.  Of course, I could be misreading
things.  Clearly, L15 and K17 are bad moves regardless.  If my reading of
the situation is correct, then I think the position may be too complex for
Gnu Go to understand without some serious revisions.

It's a very intersting position; thanks for sending it.

Evan Daniel

On Sat, 29 Mar 2003, SP LEE wrote:

> Hi all,
>
> I played a game again with GNUGO. At near the end, the GNUGO surprisingly
> let a group of its stones alive, while my group of stones die (of course due
> to my mistake). However, it seemed not having the knowledge of eliminate
> liberties of the opponent group and missed its last chance.
>
> I think this is rather a problem of calculation of liberties than
> complicated reasoning. Can you guys find out why this happened?
>
> SP Lee
>
> _________________________________________________________________
> Add photos to your e-mail with MSN 8. Get 2 months FREE*.
> http://join.msn.com/?page=features/featuredemail
>