Author: Mike S.
Date: 12:42:27 03/13/00
Go up one level in this thread
On March 13, 2000 at 15:10:53, Steve Coladonato wrote: >(...) how long it would take >a chess engine to develop a reasonable opening book from scratch? I guess that >would be like taking away the opening book and turning learning on. I think human experience, although it's results may not be 100% exact always, cannot be matched by engine generated books so far. For example: It's not likely that a program will ever decline a queens gambit by it's own means (assuming that the generation process is done by autoplay, or against other engines without book). Btw. the common method to take millions of master games and generate a book from these, does not make much sense I think. Not every move a master plays in the beginning, is theory - there's always an experts analysis needed, or even a hundred games for the variation to judge if it's good. Furthermore, it has been said that too large books (too long opening variatios) could lead to drawish, or already "shaped" positions, leaving not enough gameplay for the engines to show their strength i.e. earlier directly out of the development phase. I remember that Ch.Donninger (Nimzo) once wrote something similar as: he'd like better a small book of about 50.000 positions, which would be very well edited and tuned for the engine. Which programs opening book is considered to be the best? Regards, M.Scheidl
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.