Re: [gnugo-devel] more questionable moves?

[Gunnar Farneback]

Evan wrote:
> OK, I have another question.  This time I actually did some checking with
> gnugo, making the move I thought better and seeing how gnugo played
> against itself.  Gnugo gives up on the group of three on the left at move
> 98, when if played against itself it can't find an attack to kill the
> group in the lower left anyway.  Basically, the OWL stuff shouldn't be
> counting the lower left group as critical, right?

Yes. The owl code incorrectly finds an attack at A2. This is worth
adding to the regressions.

/Gunnar