Computer Chess Club Archives




Subject: Re: Chessy

Author: Fabien Letouzey

Date: 13:40:08 09/11/02

Go up one level in this thread

On September 11, 2002 at 15:06:15, Dann Corbit wrote:

>Concerning this:
>"Chessy is a chess player. It is not very strong for a program, but has got an
>original way of playing (pawn oriented)."
>I have some questions.
>When you say that the program is pawn oriented, do you mean that you spend a
>great deal of energy calculating pawn structures or some other meaning?

I knew I should have removed that part of the page long ago :)

Although you can safely forget everything about Chessy, I owe you some
explanation (perhaps you collect information about all chess programs,
no matter how weak).  The statement on my web page referred to an old
program that I never really finished (it has at least one evaluation bug,
which makes my comment even more realistic ;) ).  It is true that I intended
to focus mainly on the pawn structure not because I thought this was the best
thing to do, but because I did not want to create yet another tactical program
(I believed that most programs were king-attack based).

That beeing said, I knew very little about the existing chess programs.
Because of the pawn hash tables any pawn-only computation is virtually free,
so my current bet is that any program nowadays (not only the top ones) knows
a lot more about pawns that I could possibly imagine at that time :)

>What are your future plans for Chessy?

Ok, you really seem to be interested in any chess program so I have to explain.
I mostly write Othello programs (I have just released one for Palm computers).
In order to learn new things about game programming (for instance see which
heuristics are game-dependent and which are not, study the impact of BF, ...),
I tried a bit of draughts and chess.  However, I will forever be unable to
write a strong chess program because I have a very strong aversion to eval
tuning (I get bored after 2 parameters).  This is not a problem in Othello
because machine learning is more efficient than manual tuning there, so I can
focus on algorithms and implementation; most of the tuning is statistic-based.

Now it sounds like my chess experiments were really negative, but this is not
the case; I did learn exactly what I was looking for.  A funny example is that
at the very beginning of Chessy development, I was easily able to find a PV bug
which I also had in all my other programs.  This gave me a better insight and
indirectly led to the top post of this thread :)

Again it looks like I am wandering aimlessly and am not answering your
question.  To me chess programming was just a "lab" for experiments about
search algorithms.  You can compare this with AliBaba if you want.  It is no
serious program at all, understanding is what I seek.  Maybe one day I will
find somebody interested in eval tuning, and then Chessy could have a future
again (but that would include a full re-write anyway).

>I notice that you have an Xboard version of Chessy.  Do you also have a Winboard
>version (Win32 port)?

No sorry, I do not use Windows.


This page took 0.02 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

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