Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The death of computerchess.

Author: Dann Corbit

Date: 19:40:16 12/20/01

Go up one level in this thread


On December 20, 2001 at 22:26:47, Christophe Theron wrote:
[snip]
>"Copy, paste" is a metaphor.
>
>It's "copy, paste" of the METHODS not of the actual code.
>
>When I say "Crafty clone" I mean "Crafty methods clone".

Like what -- use of null move?  Use of negascout pvs search?  Is something else
really a lot better?  I have researched many search algorithms and I try them
all.  I also try to invent my own ideas.  Crafty does not use some other things
that I have tried (and will revisit again).  And a lot of techniques that you
see all over are the same ones as we see in crafty.  No wonder, since crafty has
been open source for ten years and most of the newcomers are far younger than
that.  Crafty and GnuChess were the only Winboard engines to look at for a
while.  I think that there are probably more 0x88 type engines than bitboard.

A lot of people like MTD(f), and I found another very simple interesting search
algorithm in a paper that I am going to try.  I also have some of my own ideas.
One example is to speculate by quickly tracing the pv (sort of go into auto-play
mode at [e.g.] one second intervals and go forward 10 plys or so like that.  The
idea is to help out move ordering and also to see some disasters that loom over
the horizon.  Obviously, this works better with more cpu's than with just one.
Also, if I have a CPU farm, I have an idea that I have discussed before, where
they cooperate and search the most promising lines.  With this type of search we
have much more information than alpha-beta will give, because I have a score for
every forward position root.


>For me it's the same. I do not see where the creativity is.

I don't think you have studied the source code of other engines.  All of them
are different.  All of them are creative.  All are interesting.  I have really
seen one or two engines which are just a quick clone with a tweak (I find this
out by binary snooping because those who will publish their code would not be so
brazen as to do this).  But almost all engines are interesting and highly
original.

In other words, I think you will find creativity if you look carefully.  Mostly,
I think you just are not very interested in what other people have done and
prefer to think things up by yourself.  That's certainly a noble attitude, but
one that does not work well for someone as lazy as me.

>I repeat that I consider writing a chess engine, even under these conditions, as
>a great personal achievement. It's something that counts in your life.
>
>What I do not understand is the reason why we should celebrate the number of
>such engines. They are extremely valuable for the persons who have been able to
>write one, but are they for other people?

Suppose that there are only two sculptors in the world.  Then suddenly there are
one hundred.  Mabye the first two are the best, but the new ones will get
better.

>Naturally there are exceptions. Amongst the Winboard engines there are very
>valuable ones.

All of them are valuable, if they are the object of enjoyment for someone.



This page took 0.01 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.