Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Go programming

Author: Roy Eassa

Date: 09:30:43 02/13/04

Go up one level in this thread


On February 13, 2004 at 11:11:24, Janosch Zwerensky wrote:

>
>>First, I think it's quite a bit easier to create an engine that plays Go legally
>>than it is to create one that plays chess legally.
>
>Do you include doing the scoring phase right with playing Go legally?

All your posts have been 100% spot-on and you are of course right again.
Scoring is a HUGE issue and I'd take the cop-out that some on-line programs
(like, ugh, Yahoo) do: ask each player to mark dead stones and don't end a game
until they agree.  Perhaps an alternative would be to use a standardized,
off=the-shelf scoring tool -- sort of like chess programs mostly all use the
same endgame tablebases?  Thus you'd free the many programmers from each having
to develop a scoring tool, freeing them up to work on the main playing engine
(which would THEN be quite simple to create to play legally, getting one's foot
in the door quickly and thus encouraging people to try).

BTW, I too find it surprising that people repeatedly bring up the relatively
tiny differences in rules (e.g., Korean vs. Japanese), which only pertain to
relatively rare board situations, as if that were the killer issue in designing
a good Go program.  And I also agree with you that the creation of a program
with great features or a strong engine need NOT necessarily be done by an Asian
or in Asia, even if it WOULD need to be internationalized for, and marketed in,
Asia to make real money.


>If you
>did, then I'd tend to disagree. Guarding against potential cheating by the
>opponent without boring them to death by requesting that completely obvious
>things be played out over the board is a nontrivial task as far as I know.
>I think that therefore people who are interested in Go programming in principle
>but who would like to have something that can play a complete game fairly
>quickly should take a look at the related (but different) game of Gonnect first
>(for details, see http://www.di.fc.ul.pt/~jpn/gv/gonnect.htm for example).
>Although I think that it would be easier to write a strong Gonnect program than
>to write a strong Go program, I guess that if someone were to write such a
>program they'd probably have learnt a lot of things useful for go programming on
>the way.
>
>Regards,
>Janosch



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.