[gnugo-devel] test case 13x13.tst 52

[SP LEE]

In this regression test H10 is the correct answer and gnugo is doing right 
without my patch splee_3_20.1b. However, the owl analysis is wrong and gnugo 
is only doing right thanks of the shape factor. I'm just thinking if we can 
improve the owl analysis here.

The gnugo thinks N9 is a good defense. However, just when I want to see the 
interested sequence B:N9 W:H10, the owl node limit is reached. I think if 
gnugo wants to analyze the life/death of a group, the first thing to do is 
to consider the groups escape route. So for white,the first thing to 
consider is to minimize the escape route of black, thus H10 should be 
considered first.

I'm still not very clear of the software achitecture of gnugo, so I brought 
this case here for discussion. Your comments are welcome.

SP Lee

_________________________________________________________________
MSN 8 helps eliminate e-mail viruses. Get 2 months FREE*.  
http://join.msn.com/?page=features/virus