Author: Christoph Fieberg
Date: 14:49:29 04/16/01
Go up one level in this thread
Mr. Frohlick, is it only pattern recognition? Or lack of search depth? I wonder what rank 1...c5 will have at search depth 20 or deeper. I think my comp will need some weeks to check it out. Or does such a result already exist?? Thanks, Christoph On April 16, 2001 at 16:20:45, Timothy J. Frohlick wrote: >Mr. Fieberg, > >The reason that chess programs do not understand the Sicilian is the lack of >pattern recognition. That is something the Russians are working on right now. >Computers don't "get" humor so how can they understand patterns. > >It will get better. > > > >Enjoy, > >Tim Frohlick > >On April 16, 2001 at 13:41:13, Christoph Fieberg wrote: > >>The best chess programs are able to play best opening books just by calculation. >>However, as far as I have seen, they have a problem to play Sicilian. >> >>The answer 1...c5 to 1.e4 is the most popular among the top chess players and >>also among all the others on earth. About 20% of all games are Sicilians. 1...c5 >>is the move with the best success rate. But why? And why chess computers do not >>like it? >> >>What is the secret of the Sicilian opening? Why is it better than 1...e5 which >>prevailed for decades? >> >>1...e5 is the favourite opening move of cess programs calculating. Not so the >>Sicilian opening. For Fritz6 this move is only the 8th choice (search deep 16!) >>behind the moves 1...e5, 1...e6, 1...d5, 1...d6, 1...Nc6, 1...c6 and 1...Nf6. >> >>Wouldn't it be a success to modify the code (evaluation, search functions) so >>that the programs would evaluate 1...c5 as best (or at least second best) move? >>Of course, the already achieved playing strength should not suffer from the >>modifications. >> >>Perhaps an even better strengths could be gained if programmers will find a >>method to teach the computers the Sicilian secret. >> >>Regards, >>Christoph
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.