Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: chess programmers: what info do you need to make your program better?

Author: Dan Newman

Date: 01:10:24 05/02/00

Go up one level in this thread


On May 01, 2000 at 18:08:21, stuart taylor wrote:

>On May 01, 2000 at 05:18:13, Jeroen van Dorp wrote:
>
>>I should know better by now, but I didn't pay enough attention and now I don't
>>know.
>>
>>I've seen Nunn tests, tournaments, all kind of kinky positions (supposed to be)
>>unsolvable for chess programs.
>>
>>Are all these things really important to programmers?
>>
>>I stopped playing engine vs engine tournaments like that because I see no reason
>>for them other than checking some of my own opening variations or chess ideas -
>>to be tested by a few programs, for my self, without any meaning for others here
>>around.
>>
>>But how about you programmers? If you would like feedback on your engines,
>>should the non-programming lot of us test, and if yes, what and how?
>>What information from actual chess playing (be it test suites or complete games,
>>or analysis) do you use to pick out the flaws in your engine?
>>
>>Do you have any help from the info a lot of people offer on this message board?
>>
>>I'm very curious, as some of these procedures may help even me to beat chess
>>programs easier or take advantage of weaknesses.
>>Not that this is possible of course, as mediocre players like me.... well, been
>>there, done that :))
>>
>>Jeroen ;-}
>
>good that someone asks this, as I don't think that any of the practicing
>programmers made even one comment about my question/thought on programming
>by getting the program to constantly shuffle round the peices to ideal-possible
>future positions and move according to that goal-if applicable according to much
>analysis and many criteria. OK, maybe it's a daft idea, but it must have been so
>daft that it isn't even worth trying to consider. Or perhaps it's so obvious
>that every programmer thought about it only as a baby. I, at any rate still
>don't know.
>S.Taylor

I think it's a good idea.  It's one I've thought about (a little) and has
been thought about at least as far back as 1972 (the Robin program I
mentioned in the other thread).  The problem I think is that it's a very
much more difficult sort of program to write than the usual alpha/beta
searcher.  (Partly because very few people have worked on chess programs
of this sort--so there is very little literature on it--and partly because
it's just inherently more difficult.)

I imagine that it would take a lot more work to become competitive with
such a program too, so that by the time you catch up to where the
alpha/beta searchers have already gotten, they will be even further ahead.
Still, I'd like to try it sometime...

-Dan.



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.