Author: Matt Frank
Date: 23:51:53 02/27/99
Go up one level in this thread
On February 27, 1999 at 08:59:40, Ricardo Sant Ana wrote: >Hello > >A few weeks ago, I put an email asking how would Junior 5 increase it's rating >in ssdf if it uses Powerbook (like Fritz5.32!!). So, the answer I received tell >me that powerbook is not important - the reason was that powerbook was too >general. >Ok, I am not fully convincd of this, but it's hard to imagine that book is not >important at all. For instance, everybody know that Fritz5.32 is much more >"tactical"then the posicional Hiarsc7 : so, why not develop a book for H7 that >make it meets it's own style and run away from tactical positions ? >every human do this, so why not computers ? It can only increase it's own >strengh. > But for markeing it would be a good idea to develop two books: one that make it >stronger, and one mopre general and complete for people who is trying to learn >opening and variations - people interested in training for tournament need to >play every opening. > >Ok, the idea is easy, but how to implement it ? First of all, we would need to >know how to engine works and plays. Then , you would have to collect lot of >selected games from GMs who plays engine's style. > >Now, you would have to insert the moves into the book , and verif if the engine >would play the position in agreement with what the positions ask for (this is >for tactical chess engines and positional chess engines too). > >Ok, this is just an overview, but what really bothers me is: the afirmatives >above really convinced me: than I ask ? Who could develop a book for an engine ? >My only answer is: >Somebody who knows opening and can develop ideas and compare engine moves with >GM moves se what is right and wrong ! >So - a very good chess player can do this (GM or IM or even a FM) >Now I ASK: > >IS THERE ANY CHESS PROGRAM WHICH BOOK COMES FROM A GM WORK ??? > >Thanks >Ricardo Sant'Ana Setting aside the issue of tactical or positional comparative strength (i.e., Hiarcs 7 vs Fritz 5.32) I would like to agree with you that programs can be strengthened if the opening book is appropriate to the style of the program. In fact it is expected that good programers have run many test positions (with the program) based on opening theory that incorporates the "discoveries". I know that the Hiarcs team has reported that some book openings were developed based on Hiarcs analyses. Furthermore, if a program "learns" from its games then even playing a series of openings may improve the "book" based on the particulars of the soiftware itself and the parameters used in the analyses. Best regards, Matt Frank
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.