Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: So which programs beat which, only due to superior chess understanding?

Author: Uri Blass

Date: 10:01:22 05/06/02

Go up one level in this thread


On May 06, 2002 at 10:38:45, Victor Fernandez wrote:

>On May 06, 2002 at 10:16:16, Uri Blass wrote:
>
>>On May 06, 2002 at 09:59:13, Victor Fernandez wrote:
>>
>>>
>
>
>>
>>No
>>
>>Richard lang's program were better than the opponents in tactics and this is the
>>main reason that it won.
>>
>>Searching deeper also generates better positional moves so you can know nothing
>>based on watching the games.
>>
>>You need to give the opponent unequal hardware in order to get result of 50% and
>>only in this case there is a way to find the program that is better in tactics
>>based on a lot of games.
>>
>>
>>
>>If the winner was the first side to get a significant fail high then you can
>>count it as one tactical point for the winner.
>>
>>If the loser is the first side to get a significant fail low then you can count
>>it as a tactical point for the loser because it could see first the disaster.
>>
>>
>>It may be interesting to know information about the programs that have better
>>positional understanding but unfortunately today we have no information about
>>it.
>>
>>We need some objective test to know and the important thing in order to know is
>>to give programs unequal hardware.
>>
>>
>>Uri
>
>
>Sorry, Richard Lang says another different thing,
>
>"a chess program is an evaluation function"
>
>"There have been several changes in the evaluation function that have improved
>the results in the test"
>
>Victor

I think that these several changes were only after he had a strong program.

changes in the evaluation can help but most of the improvement comes from
search(I do not include simple chess knowledge that every top program has like
evaluating pawn strucuture).

It is also clear that richard lang believes in search because Genius3 is a root
processor.

Some people do not like these words and prefer to use do preprocesing but I
define for simplicty every program that does preprocessing that can change
significantly the score as  a root processor.

Explanation for people who do not understand what is a root processor:


When I say root processor I mean to say that the evaluation of Genius can be
changed significantly after the sides play good moves only
because of the fact that the root position is different and not because of deep
search in the new position.

If a program suggests 1.e4 e5 with a score of 0-0.1 for white at depthes 10-15
and if it also has after 1.e4 e5 a score of 1-1.1 for white at depthes 8-13 then
it is a root processor because the evaluation was changed not because of seeing
deeper.

quiet moves like 1.e4 e5 are usually not good candidates to prove that a program
is a root processor.

Genius sometimes change it's evaluation significantly after trading pieces or
castling so I consider it as a root processor.

Uri



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.