Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: can a tree based on a chess program beat the chess program?

Author: Thorsten Czub

Date: 10:26:04 04/04/98

Go up one level in this thread


On April 04, 1998 at 01:10:41, blass uri wrote:

>when I give a computer hours to analyze a position
>often it does not analyze
>the position after 4 half moves in the main line more than a minute.
>this  is not logical so I think that a good tree based on
>the same program can beat the program.
>did anyone do experiment about this?

Any email-chess player I know uses his chess-programs to build those
trees.
They show exactly that today's algorithms in computerchess are nonsense
because although the programs search and search and search and come
really deeper, they do often NOT see the important and right things.
BY NOT following their main lines and instead planning OWN moves and let
them defend against those moves , you can create OWN trees. I have made
the experience that
a) those own trees are much more often relevant
b) later, when the chess programs are in the end position of those own
ideas and you give them really time, they will agree with your concept.

Unless the begin position and the end position are at least arround 13
or more plies apart from each other, the chance that a program gets an
idea about YOUR plan from the beginning is almost impossible. If you
want an example what I am talking about I can show you a position where
I - with the help of my programs, of course - made MY plan real and the
compus were unable to see these easy sac.




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.