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!news.sprintlink.net!EU.net!sun4nl!aie.nl!news
From: geert@sparc.aie.nl (Geert-Jan van Opdorp)
Subject: Re: Chess - exhaustive searching
Sender: news@aie.nl (News Account)
Organization: AI Engineering BV
Lines: 42
Message-ID: <GEERT.95Apr18105447@sparc.aie.nl>
References: <scottecD6FAH9.2pp@netcom.com> <3ln0ol$ame@mycroft.rand.org>
	<D6GDIx.DM@cs.vu.nl> <3mjel3$q4e@nic.lth.se> <D6zECo.C24@mv.mv.com>
	<GEERT.95Apr14194948@sparc.aie.nl> <D71LAz.GzF@mv.mv.com>
In-Reply-To: sje@mv.mv.com's message of Fri, 14 Apr 1995 20:37:47 GMT
Date: Tue, 18 Apr 1995 08:54:47 GMT


>  geert@sparc.aie.nl (Geert-Jan van Opdorp) writes:
>   >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'? 

>   I think that state space enumeration is a shaky foundation to compare
>   the complexity of go and chess. 

That may well be so, it just makes wonder what makes go so much
more complex. 

>                  I think that go, while being
>   difficult to program, has not had the attention that has been seen by
>   chess and checkers (draughts).  That is the main reason why there are
>   not world class go programs. 

Your answer: Go is not much more complex. I don't think anyone who has been 
involved in writing go programs will agree with this. Btw, there are not
only no `world class go programs', the best programs are of interest
only to beginners!

>                                  I suspect that there is a possibility of
>   someone finding a fast pattern recognition algorithm for go that could
>   be easily programmed if only one knew what it was.  I do not think
>   there is a similar algorithm for chess or checkers.


You do not play Go I take it? I do not even understand
what use you would put your pattern recognition algorithm to. Maybe
finding the 'hot spot' in tactical situations (tesuji), but that does
not address the main difficulty in makeing a program play: the strategical
layer. Besides, even talking about tesuji, to call the problem of finding
a tesuji a matter of pattern recognition does not capture the
difficulty involved. Yes, if i have a database with shapes (as do the 
strongest programs) then a fast algorithm to match the actual
situation is usefull and exists. The problem is with the database I think.


I hope you did not seriously think of a nim-like solution!? :-)


