Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question for chess programmers: Go

Author: Andrew Williams

Date: 04:37:25 02/15/02

Go up one level in this thread


On February 14, 2002 at 10:11:54, Gian-Carlo Pascutto wrote:

>On February 13, 2002 at 19:31:12, Angrim wrote:
>
>>On February 13, 2002 at 11:06:12, Edward Seid wrote:
>>
>>>This is a general question for chess programmers.
>>>
>>>Have you ever tried programming a Go engine?  Or do you even have an interest or
>>>understanding of the game of Go?
>>>
>>>Just curious.
>>
>>I became interested in the game of Go a while ago based purely
>>on the fact that people said it was extremely hard for a computer
>>to play.  After reading all that I could find about the rules of
>>GO, I decided that the real problem with it is that the sort
>>of people who write chess computers find the rules of GO too
>>vague and inconsistant and so most of them avoid it.
>>
>>Three things that I wanted from the rules that I was not
>>satified with:
>>1. I want to be able to find a unique set of moves in any position
>>such that the moves are legal.  The set of legal moves seems to change
>>from one country to the next, and even from one tournament to the next.
>>2. I want to be able to determine when the game is over.
>>As stated elsewhere in this thread, the game of GO is over when
>>both players agree that it is over, a condition that is hard to
>>detect in the search tree ;)
>>This one is actually not a huge problem, since the comp can just keep
>>playing and pass when it thinks that all legal moves would be
>>worse than no move, and at some point the opponent will also pass
>>which ends the game.
>>3. once the game is over, it would be nice to have an algorithm that
>>could determine who has won.  Most of the heuristics I have seen for
>>this either say "the experienced player will know" or involve the
>>players taking turns moveing stones around.
>
>I already talked with you on FICS about this, but I'll mention it
>here again for the other CCC'ers:
>
>There is a ruleset for Go called the Tromp-Taylor rules. It's both
>extremely short and still complete, and doesn't have any of the
>ambuigities the traditional human rules have.
>
>It's also close enough to the human game that a program that is strong
>by TT rules will also be strong with the more common human rules.
>
>It's been my impression that this ruleset would be the required
>'minimum supported rules' for the GTP (Winboard II equivalent for
>Go programs), so it looks like _the_ pick if you're going to develop
>a program.
>
>--
>GCP

This is very interesting. Thanks, GCP.

Andrew




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.