Author: Uri Blass
Date: 12:51:07 02/20/05
Go up one level in this thread
On February 20, 2005 at 07:58:24, Sandro Necchi wrote: >On February 20, 2005 at 07:37:21, Uri Blass wrote: > >>On February 20, 2005 at 07:08:45, Sandro Necchi wrote: >> >>>Uri, >>> >>>do you think Kasparov can win without home preparation and without reading book >>>and check what's new and novelties as well? >>> >>>Well, believe it or not it is the same thing...I mean the chances to win without >>>this is like without a good opening book for a chess program. >>> >>>Sandro >> >>I did not claim that your book does not increase significantly the chances of >>shredder to win tournamnent. >> >>I only argued about the numbers. >>I agree that shredder has only small chances to win a tournament without book. >>I do not agree that the chances today are 0.000000001%(note that 1% is also a >>small chance). >> >>I also claimed that if the engine is strong enough relative to the opponents >>then it has good chances to win a tournament even without book(it does not say >>that book is not important for it to win tournament because increasing chances >>from 50% to 95% is important for winning tournaments). > >According to Vincent one engine should be at least 700 points stronger than the >competitors to do what you claim.... > >Of course if one engine is much much stronger than all the others it can do it >even without a strong book, but even if 700 points may be a lot it should be at >least 300 points if not more and this seems quite impossible today... > >> >>Uri > >Sandro 1)I will not argue about things that I do not know but an interesting experiment that is possible to do(unfortunately I do not expect it to happen) is if some strong program with no book and with positional leanring participates when the participants do not know about it's identity(it can be for example some personality of shredder9). It will be interesting to see what is it's performance relative to the top programs. 2)Another idea to check the value of book is to have 2 copies of Crafty in the next CCT's tournaments I chose Crafty because they it is a free source code so it will be easy to check no cheating later. one copy will have no book and positional learning when the second copy will have book by somebody who believe that he can give Crafty 300 elo in tournaments. In that case we will see if the copy of Crafty with book can perform 300 elo better than the copy of Crafty without book. Unfortunately I do not expect it to happen(we need 2 agreement when one of them is agreement to have 2 versions of Crafty in the same tournament and another agreement is to have some book maker who want to work for Crafty). Uri
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.