Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: can GM's learn from programs

Author: Robert Henry Durrett

Date: 14:18:20 05/27/02

Go up one level in this thread


On May 21, 2002 at 18:14:44, Vincent Diepeveen wrote:

>On May 21, 2002 at 17:28:45, Peter McKenzie wrote:
>
>>On May 21, 2002 at 11:07:46, Vincent Diepeveen wrote:
>>
>>>On May 21, 2002 at 10:13:29, K. Burcham wrote:
>>>
>>>The answer is as simple as clear.
>>>
>>>They can use programs to analyze games with from other GMs
>>>and themselves. They can use it to analyze variations from their
>>>openings preparation with.
>>>
>>>But they can't learn from it.
>>>
>>>Whatever commercial propaganda says about chessproducts, they
>>>are weaker in all respects than a GM, except one: they always
>>>play a move at the same level.
>>>
>>>GM is everywhere better than a program, but doesn't have a
>>>level that is always above the constant line the program has.
>>
>>You overlook *at least* one area that the program is better in: any endgame that
>>is in tablebases!  Even GMs can have trouble with simple rook endgames, and they
>>are quite terrible at Queen endgames.
>
>No i don't overlook it. The question is whether they can learn from
>it. My answer is: NO the thing doesn't explain why a move is good.

Merely pointing out a move or line which the GM did not consider should be
sufficient.  GMs surely are not lazy.  Get them started on a new path and they
will fill in the details themselves by their own analyses.

>
>To explain it in children language:
>
>If EGTB says in position X: best move is Rh4 mate in 40, it doesn't
>explain why in the general casus this move is good.
>
>>
>>>
>>>So a GM can't learn anything from the program, except use it to
>>>analyze with.
>>>
>>>>
>>>>not discussing opening book moves.
>>>>not discussing egtb's.
>>>>not discussing a pawn promotion.
>>>>
>>>>
>>>>
>>>>lets start out saying GM eval is 0, and program eval is 0.
>>>>lots of time left on both clocks. time is not an issue.
>>>>
>>>>the GM makes a move and plays what he sees as a safe move.
>>>>GM took three minutes to move, but he cannot see the material winning
>>>>combination for the program.
>>>>the program takes three more minutes to see material winning move.
>>>>finally program plays winning move, and GM can see combination.
>>>>
>>>>program wins.
>>>>
>>>>If GM's cannot learn strategy from programs, if GM's can only learn tactics,
>>>>then the question is:  in the above described game , it would seem that most of
>>>>the time the GM cannot learn tactics either. if the combination is too deep for
>>>>the GM to see he loses material, then he cannot see this in another game.
>>>>
>>>>Just wondering how a GM can learn from home practice with programs.
>>>>it would seem that in a postion were the GM could play a losing move and he
>>>>cannot see the deep combination, but the program can, he could not have won
>>>>anyway.
>>>>
>>>>maybe some will answer that this is not possible. maybe some will say there
>>>>is no posiiton that the program can see winning material but GM cannot.
>>>>
>>>>of course i am not refering to an simple oversight by the GM. i am not refering
>>>>to a simple human blunder. i am refering to a situation that the GM says he
>>>>could not see the combination until it was played out.
>>>>
>>>>kburcham



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.