Author: blass uri
Date: 20:05:39 02/22/00
Go up one level in this thread
On February 22, 2000 at 19:04:08, Eelco de Groot wrote: >On February 21, 2000 at 12:31:36, Tom Kerrigan wrote: > >>On February 20, 2000 at 17:24:00, Eelco de Groot wrote: >> >>>They, Botvinnik and his programmers, obviously failed to generalize enough or >>>even at all the few rules they may have found for some positions. I don't know >> >>Evidently so. >> >>The program should have at least been able to play random moves. That's a >>starting point, however small. As far as I know, it couldn't even do that. >> >>-Tom > >Yes , sure, but if you want a program that somehow emulates human thought- >principles a move generator is not necessarily the first thing you want to work >on. I would guess that a human player first looks at the situation on the board, >the characteristics of the position. There probably follows an interplay between >"static" characteristics and possible moves that allow a good chess player to >quickly determine what parts of the board are important at the moment, which at >the same time helps him to narrow down the number of moves he should consider. I think that not considering all the possible moves in the first ply is a typical human mistake. There are cases when humans did not play a good move only because they did not think about them. I believe that humans can get better in chess if they use a small part of their time for generating the legal moves of them and their opponents(they cannot do it in blitz but they can do it at tournament time control). The problem of computers is not the fact that they consider the legal moves for the first ply but the fact that they do not know to prune illogical lines and the fact that they do not know to think about other things like planning where to put their pieces and generating moves later for this target. 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.