Author: Tim Mirabile
Date: 12:04:21 06/19/98
Go up one level in this thread
I guess the best method for selecting book moves depends upon what type of event your program will be playing in. If you are playing a match against Kasparov, you want a narrow opening book with the moves hand selected by a team of grandmasters. If you are playing automated blitz games on ICC, you probably want your program to play anything and everything, even slightly dubious stuff (especially with high "cheapo-potential"), so long as it doesn't play the same things over and over. Playing in a round robin event would require something in between. On June 19, 1998 at 11:14:27, Steven J. Edwards wrote: >Currently, I use a very simple method. I only store moves played by the winning >side; each move has a reference count (number of wins). At lookup time, all the >moves for a posiiton are scanned and the sum of the counts is calculated. Then, >a uniformly distributed psuedorandom number is picked so that the probability of >a move being selected is directly proportional to its count relative to the sum >of the counts. I've never been a big fan of this method. Even high level games tend to have results which do not necessarily indicate who played best in the opening. But it's hard to suggest another method that does not involve a lot of hand tuning. Perhaps you could just suck in all of ECO and Informant, with evaluations, which if not totally trustworthy would be more so than game results. Your objection to trusting games of the past is well founded. In those days the famous players would often play dubious openings (not necessarily intentionally, but because opening theory was not yet developed enough for them to know this) and still outplay their weaker opponents and win.
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.