gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] playing with handicap


From: Gunnar Farnebäck
Subject: Re: [gnugo-devel] playing with handicap
Date: Sat, 04 Jun 2005 15:38:27 +0200
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/21.3 (sparc-sun-solaris2.9) MULE/5.0 (SAKAKI)

Paul wrote:
> I'm not particularly familiar with those parts of the code, but I
> believe there is nothing like that conditioned specifically on
> handicap.

I'm pretty certain that there's nothing specifically dependent on the
handicap. There's some code that's in practice only applicable for
large handicaps (thrash around when everything looks dead) but that's
only an indirect effect.

> GNU Go is written to not depend on move/game history as much as
> possible, i.e. it tries to decide given the position at hand only.

Yes. The most notable exception is the code to suppress thrashing
opponent dragons.

> However, there are strategic decisions based on terms of score lead.
> If GNU Go believes it is ahead by a lot, it will value connections
> and certain other moves more.  Of course, it will consider itself
> far ahead in the beginning of handicap game when playing black.  That
> should be what you want, right?

On the other hand, when giving a large handicap and thus starting far
behind, we could need some adjustments to value cuts higher and more
aggressively try to break up moyos. This is often notable in KGS games
with 5-9 handicap stones.

/Gunnar




reply via email to

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