Author: Laurence Chen
Date: 13:47:04 11/08/00
Go up one level in this thread
On November 08, 2000 at 14:55:24, Bob Durrett wrote: >Assume that a GM has developed and memorized a viable opening repertoire >for use against some specific chess program, such as Fritz on a PC. Assume >also that the GM has reached a position at the end of one of his/her repertoire >lines and is trying to prepare for an upcoming match with that computer program. > Finally, assume also that the GM's home computer is exactly half as fast as the >computer he/she will face in the upcoming match. Then, is there any reason why >that GM cannot experiment with alternatives on the >home computer until a forced win is found [from that position]? It might be >necessary to let the home computer think twice as long as the computer >will in the match, but that should not be a significant restriction when doing >home preparation. This question boils down to the technical issue of whether or >not the same moves would be found on a slower computer if the amount of time >allowed was increased to compensate for the slower speed of the computer. Also, >this question depends on the issue of whether or not chess computer programs, >like Fritz, are deterministic during play [after exiting the program's opening >book] versus the deliberate introduction of some form of randomness by the >programmers. Finally, it is possible that the GM might prepare for a computer >move which is better than the move the computer might actually play, due to >allowing too much time for the move during home analysis. Can anybody shed some >light on these technical issues? The answers might help everybody to understand >the extent to which GMs [as well as the rest of us] can prepare for matches >against computers. You're assuming that the same book opening is being used in the match and the one which is comercially sold with the product. It's possible that the programmer can change the book opening such as to exploit the weakness of the GM who the engine is going to play. Then, if the GM follows blindly with the assumption that the machine is going to play such move as he practiced in his home computer he can easily fall into a trap. I believe that the GM should study the games played by the chess engine and find out its weaknesses and strengths. Not to try to force the machine to fall for a specific book preparation. My 2 cents worth. Laurence
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.