Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!hookup!swrinde!cs.utexas.edu!utnut!utcsri!cdf.toronto.edu!g2jdr
From: g2jdr@cdf.toronto.edu (Rootham James Douglas)
Subject: Re: Chess - exhaustive searching
Message-ID: <D73C5H.Lz4@cdf.toronto.edu>
Sender: news@cdf.toronto.edu (Usenet News)
Nntp-Posting-Host: puck
Organization: University of Toronto, Computing Disciplines Facility
References: <scottecD6FAH9.2pp@netcom.com> <3mjel3$q4e@nic.lth.se> <D6zECo.C24@mv.mv.com> <GEERT.95Apr14194948@sparc.aie.nl>
Date: Sat, 15 Apr 1995 19:15:15 GMT
Lines: 29

In article <GEERT.95Apr14194948@sparc.aie.nl>,
Geert-Jan van Opdorp <geert@sparc.aie.nl> wrote:
>
>
>I've always thought it was to a large extend due to the branching factor
>that Go is so much harder then Chess. Am I wrong you think? Or do I 
>misinterpret `complexety'? 
>

There is debate on this subject. It recently surfaced in rec.games.go.

Certainly go has a larger branching structure than chess.  The other
critical question is about writing evaluation functions.  The
evaluation functions in go can be very unstable due to life and death
fights (it is frequently not clear from static analysis whether a
group is alive or not).  Uncertainty in this area can have a massive
effect on the value of the evaluation function.  One consequence of
this (or at least an attempt to alleviate the problem) would be to
search deeper in the tree.  But given that the branching factor is
already known to be larger this is unlikely to work well.

Jim Rootham
g2jdr@cdf.utoronto.ca



-- 
Jim Rootham   g2jdr@cdf.utoronto.ca

