[net.games.go] Programming Go

granvold@tymix.UUCP (Tom Granvold) (12/16/83)

-
     How about a dicussion of implementing a Go playing program in this
interest group. I have been toying with the idea for years now but have
not done anything. Some discussion and interest might get me going.

     Also, I would like to hear of any published material on Go playing
programs and related subjects. I will post a summary on this net.

      Tom Granvold
      Tymshare
      Cupertino, California
      decvax!ucbvax!oliveb!tymix!granvold

robison@eosp1.UUCP (12/20/83)

Allow me to throw down the gauntlet on programming go -- I believe
that because of the large number of available moves per turn, go will
never be played as well as chess by computers.

There are programs in Japan to play selected endgames at least.
Can anyone provide a survey of what programs exist?

				- Keremath,  care of:
				  Robison
			          decvax!ittvax!eosp1
				  or:   allegra!eosp1

mark@umcp-cs.UUCP (12/22/83)

The most successful Go playing program I know was done at the University
of Michigan several years.  It used some kind of funny data structure
called a 'webnet' to automatically propogate interesting information
around the board whenever a stone is placed.  But I don't think
it ever got to the 10 Kyu level, and I never saw anything actually 
published about it.  I did play it once though, so it was real.
Any of the U Mich folks on the net want to elaborate?
-- 
Mark Weiser 		
UUCP:	{seismo,allegra,brl-bmd}!umcp-cs!mark
CSNet:	mark@umcp-cs 	ARPA:	mark@maryland

jkm@linus.UUCP (Jonathan K. Millen) (12/29/83)

---------
There is a paper by Walter Reitman and Bruce Wilcox on the LISP
program developed at the U. of Michigan: "The  Structure and
Performance of the Interim.2 Go Program", Int. Joint Conf. on
A. I., 1979, pp. 711-719.  Reitman is now at NYU and Wilcox
at Intermetrics in Cambridge, MA.  According to Wilcox, the program
plays at about 5 kyu until the endgame, when it degenerates to 13
or so, because no special endgame heuristics had been developed.

Less ambitious Go programmers may be interested in an article of
mine on a beginner's level Go program in the April '81 issue of
BYTE.

				     Jon Millen
				     Bedford, MA