[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Patch: Value of edge josekis reduced
From: |
Daniel Bump |
Subject: |
Re: [gnugo-devel] Patch: Value of edge josekis reduced |
Date: |
Tue, 18 Dec 2001 08:18:22 -0800 |
Arend wrote:
> > - reduce patterns EJ2 to EJ4 from J to j
> Hmm, just looking at the patterns I'd say they are definitely worth a
> J classification, I guess they are more urgent than a lot of joseki
> moves. I think the problem with strategy:20 is that the alternative
> isn't valued high enough. (I can't see how this patch affect
>
> Maybe this is a general problem: GNU Go does not see recognize itself
> that J moves are worth 25pts or more. If then another move arises, that
> would deserver a J (or even U) classification, but just happens to be
> to special to have a joseki pattern, its comparison of move values is
> spoiled.
I agree with all these comments.
GNU Go makes many bad tenukis. This results in inconsistent play,
where a group will be developed, then abandoned because when a
threatening move is made, suddenly something else looks bigger.
strategy test 20 is a case in point.
I don't think these patterns should be changed from J to j.
Maybe strategy test 20 is a case worth thinking about. How
can GNU Go see that the connection at Q11 is absolutely
necessary?
Dan
Re: [gnugo-devel] Patch: Value of edge josekis reduced, Gunnar Farneback, 2001/12/19