gnugo-devel
[Top][All Lists]
Advanced

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

[gnugo-devel] ED66 considered harmful


From: Gunnar Farneback
Subject: [gnugo-devel] ED66 considered harmful
Date: Sun, 12 May 2002 15:19:16 +0200
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

The pattern ED66 has a J classification, which is harmful considering
how general it is. I'm not sure this pattern is needed any more but if
it is it should be made more specific. It really would have helped to
have some history telling where this pattern came from and what
problem it wanted to solve.

As examples of the problems it causes, I offer two test cases. The
game records have been added to CVS, but I'm not sure which test suite
to put the tests in. (nngs1.tst, strategy5.tst or a new nngs2.tst are
the most likely alternatives.)

# R9 is clearly worse than Q9.
loadsgf games/nngs/kumu-gnugo-3.2-200205070903.sgf 116
??? gg_genmove black
#? [Q9]

# G17 is clearly worse than G16.
loadsgf games/nngs/gnugo-3.3.2-mr23-200205120953.sgf 132
??? gg_genmove white
#? [G16]

I suggest we just remove ED66 and see whether it does cause problems
to do so. The identical pattern ED67 must go too. Although it's only
j-classified, that's also too high.

/Gunnar



reply via email to

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