gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] tsumego Evan Berggren Daniel


From: Evan Berggren Daniel
Subject: Re: [gnugo-devel] tsumego Evan Berggren Daniel
Date: Thu, 2 Oct 2003 14:13:41 -0400 (EDT)

On Thu, 2 Oct 2003, SP LEE wrote:

> I hope my following comment is not irritating. If black plays at '2'
> instead of '1', it is still one eye for white.

Not at all.  I suppose I didn't include enough context.  The point of the
problem is that the position is actually zero eyes for white, not one.
However, if black plays first, it is ko, not zero.  Here is the original
problem:

|.XXX...
|.......
|.XO.X..
|XXOOOX.
|OO..OX.
|.XOOOX.
|..OXXX.
|.OXX...
+-------

Black to play and kill.  Black throws in, white takes, black ataris the
two stones, white connects, black plays the 1-2 point, white catches the
three outside stones, and then black plays the 2-2 point.  Now white has
one eye from the capture, and none in the corner.  However, if white
claims life (which is what happens in the owl code), then black can only
disprove the claim by capturing from the outside or allowing the ko.  The
result is that gnugo (after a little pattern tuning to find the first
moves) believes that the group is ko, because it fails to understand that
there is no eye in the corner if black simply does nothing.

Evan Daniel




reply via email to

[Prev in Thread] Current Thread [Next in Thread]