Author: Harald Faber
Date: 02:22:18 07/11/01
Go up one level in this thread
On July 10, 2001 at 13:32:17, G. R. Morton wrote: >>I think the best method to find good and bad lines is to take a look at the >>games. Start where the book line ends and look what happens in the next 5-10 >>moves. If the -1.2 arises to a justified +0.2 then the line cannot be bad just >>because the engine came out of book with -1.2. And so on and so on. >> >> >>>Uri > >This is a method I try to use, but I wonder if it would also be a method to >compare software, i.e., the better software would tend to be the one that can >can stay most consistent with its last book evlauations when you extend the >lines consitent with its own evelauations during the extention. I thought this was the adjusted opening book ever programmer dreams of? >It should also >be good to extend by known GM games and see what happpens. And you will be surprised how often programs mishandle such lines because the programs don't understand the strategy. Versus other programs this might not be a significant disadvantage because the other programs don't understand it either. Of course sometimes the programs exploit some holes in the opening lines of GMs. I have such a game running where the line ends after a black move and black won in the GM game later on. But the program with white plays a more logical attack and now has a won position. In another game the opening book didn't contain the last black move and the program played an alternative. It was an easy draw...remind me this evening (in 8-9h) to post the games, don't have them here now.
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.