Author: Mark Young
Date: 21:12:20 04/17/99
Go up one level in this thread
On April 17, 1999 at 21:50:06, Steve wrote: >On April 17, 1999 at 01:25:24, Mark Young wrote: > >>On April 16, 1999 at 23:01:40, Steve wrote: >> >>>I'm a bit puzzled. Doesn't "positional understanding" require the ability to >>>conceptualize, meaning that no existing programs -- no matter how strong -- can >>>really have any "positional understanding?" Or has the art advanced beyond >>>that? >> >>Forget what you have heard from chess players and most book about "positional >>understanding" Then ask yourself what is "positional understanding"? >> >>You know your have the right answer when you figure out that a program does not >>have to conceptualize to have "positional understanding" > >I was thinking of comments that appear in annotated grandmaster games along the >lines of "White plays the pawn to h5 here, seeing that in all future rook and >bishop endings that may later arise, it will give him a favorable pawn >structure," and sure enough, when the ending is reached 20 moves later, it does. > Maybe it would have been more precise for me to say "long-range planning," or >are you saying that once a program can calculate far enough ahead, it can >accomplish the same things as "planning" humans do? Bingo! Search is everything in chess. The deeper the programs can search in the future the more "positional understanding" or "long-range planning" they will have without any added knowledge given to the program. Where the human player has to conceptualize, the chess programs will see tacticly, but the results will still be the same, "positional understanding".
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.