Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A new chess program on the horizon?

Author: Bruce Moreland

Date: 19:09:01 11/16/01

Go up one level in this thread


On November 16, 2001 at 15:29:19, Tom Kerrigan wrote:

>On November 16, 2001 at 02:34:27, Bruce Moreland wrote:
>
>>On November 16, 2001 at 01:25:08, Tom Kerrigan wrote:
>>
>>>On November 15, 2001 at 23:10:27, Bruce Moreland wrote:
>>>
>>>>On November 15, 2001 at 19:56:45, Tom Kerrigan wrote:
>>>>
>>>>>Why do you need a great chess programmer to make a great chess program, and not
>>>>>several good chess programmers? I can understand how this might seem intuitive
>>>>>to you, but you don't seem to be going on anything more than intuition.
>>>>>
>>>>>-Tom
>>>>
>>>>Which would you rather have, one good heart surgeon or two not so good ones?
>>>
>>>Who would you rather have design your car? One great engineer or several good
>>>ones?
>>>
>>>-Tom
>>
>>One great one.  I believe there are a few wonderful examples.
>
>Hmm, maybe it was possible a century ago for someone to design an entire car
>from the ground up, but these days, if someone "designs" a car by himself, it's
>kind of like saying that Gateway "designs" computers. They buy a dozen or two
>parts from various sources, assemble them, and put a fancy exterior on the
>result. This is not what I meant, and is not analogous to the computer chess
>world. (You can not take a move generator from Program A and just drop it into
>Program B to "soup it up.")
>
>I believe that there is enough complexity in chess programs that if you get
>several good chess programmers to make a new one, they can specialize on
>different parts of the program and come up with something better than any one
>programmer could.
>
>Of course, this is just speculation, but so far it hasn't been disproven.
>
>-Tom

I don't want to get too crazy with this, since this is a discussion about which
beer is better, and that is always a matter of personal taste, but I think that
we're still at the stage in computer chess where the engine is a one-person job.

There are some logical places to break the work up, but they aren't in the
engine itself.  In particular, the opening book is a great example.

But doing something like breaking between eval and search is a bad idea.  And
eval and search are very connected to move generation and make/unmake.  So you
end up with something being one organic thing.

Perhaps someone could fiddle with eval terms, but a lot of search and
make/unmake stuff has significant influence on what you can do in eval and how
fast you can do it.

I believe that Dark Thought was broken between eval and search, and it's a
wonder that Peter, Ernst, and Marcus didn't kill each other.  As is, I'm not
sure that they all speak to each other.

bruce




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.