Author: Thorsten Czub
Date: 02:05:07 02/19/02
Go up one level in this thread
On February 18, 2002 at 23:01:16, Antonio Dieguez wrote: >hello. > >thanks for posting this, I didn't read it before. It is very passional. > >about the best move, there is always a best move, for the program, otherwise why >it plays a worse move? (i know a best move generally doesn't exist, but the >program always pick one that it chose the best.) > >what do you call a plan here? how you think ccs play a best move for a plan >without preprocessing, that it is different from the way others good programs >play a best move? > >if the program thinks there are better chances over this or this variation then >they have a higher score. > >the game you post is pretty, but can be acomplished in a normal way, "only" :) >scoring well king attack for example. Your comments are so romantic. > >a. ok. the question is: do you see a game of chess as a number of test-positions, like an EPD-file, where your engine only has to find the accurate move, the BEST move, or that it avoids the wrong move ? is chess only a number of positions you have to find the BEST move. or is it possible to play moves, that are in the position the 2nd or 3rd best, they do not instantly lose, they let the score of the opponent increase because it saw: ough - not the "best" move here, that makes me happy. but your engine plays those NON best moves because they are a subtarget in a bigger plan, lets say a normal program would have to push a pawn in position move 30. pawn push 31. defend bishop 31. recapture piece 32. short-castle those moves are the best moves your engine sees in the POSITIONS. now lets say there is a different program, that has different mechanism. it wants to bring a knight from the queen side to the king side of the board. Why ? because the "target" is to use the knight in a king attack. instead of playing the always best moves it tries to do a manouever. this is not seen by the opponents because THEY expect your engine to play BEST moves. so they are blind ! but - of course it weakes your position. so you need extra knowledge to judge if the knight, when brought to the king side, makes sense. i think, when replaying games of lasker, tal and fischer, that i can see those behaviour of them in their games. that makes it for computers so difficult to reproduce lasker,tal,fischer games. because they do often not play the BEST move. they sac the best move for the plan-move. static against dynamic. for me chess is more than finding key moves in a position. i believe that beside the best moves, there exist interesting ideas, that can be realized if someone is interested in. this is what differenciates computerchess and human-chess. computerchess is often not creative. the programs do not follow an idea. they only compute the best move. they are not interested in ideas. they play chess like it is an EPD file they solve.
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.