Re: [gnugo-devel] Patch: Value of edge josekis reduced

[Gunnar Farneback]

Inge wrote:
>  # incident 107
> +# L3 is correct in that it works as a combination attack.  However, it 
> +# breaks the principle that you should always attack the cheapest target 
> +# first.  I think L3 should be ok.  /iw
>  loadsgf games/incident107.sgf 216
>  8 gg_genmove black
> -#? [N1]*
> +#? [L3|N1]*

I think this one is within reach and that the correct attack point is
worth regressing for. Notice though that N2 may be a slightly better
attack than N1 and that K2, though no more correct than L3, also is
effective.

/Gunnar

Index: regression/atari_atari.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/atari_atari.tst,v
retrieving revision 1.8
diff -u -r1.8 atari_atari.tst
--- regression/atari_atari.tst  6 Dec 2001 16:57:28 -0000       1.8
+++ regression/atari_atari.tst  19 Dec 2001 16:42:03 -0000
@@ -1,9 +1,9 @@
 # N3 is ineffective. N2 and N1 differ in that the latter move gives
-# white the option to defend with a (disadvantageous) ko. L3 is also
-# effective but misses the point and wastes a good ko threat.
+# white the option to defend with a (disadvantageous) ko. L3 and K2
+# are also effective but miss the point and waste good ko threats.
 loadsgf games/incident107.sgf 216
 1 combination_attack black
-#? [N2|N1|L3]
+#? [N2|N1|L3|K2]
 
 loadsgf games/atari_atari01.sgf 235
 2 combination_attack black
Index: regression/strategy.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/strategy.tst,v
retrieving revision 1.21
diff -u -r1.21 strategy.tst
--- regression/strategy.tst     16 Dec 2001 15:57:36 -0000      1.21
+++ regression/strategy.tst     19 Dec 2001 16:42:03 -0000
@@ -38,7 +38,7 @@
 # incident 107
 loadsgf games/incident107.sgf 216
 8 gg_genmove black
-#? [N1]*
+#? [N2|N1]*
 
 # incident 109
 loadsgf games/incident108b.sgf 172