Author: Albert Silver
Date: 09:51:06 12/29/97
Go up one level in this thread
On December 29, 1997 at 06:08:38, Chris Whittington wrote: > >On December 29, 1997 at 04:32:17, Marty Hirsch wrote: > >>MCP7 is doing the "maximum" to produce strong, active and exciting >>chess. The Opening Library is designed to advance that goal by avoiding >>blocked positions and boring exchanges. >> >>In addition, the MCP7 Opening Library is designed for soundness, as well >>as variety. There are a large number of interesting variations >>included. These are "theoretical" lines in one sense or another. In >>fact, published "Chess Theory" is often not correct. M-Chess tries to >>improve on this with the lines it plays. >> >>M-Chess, like any chess-player needs to be able to continue what it >>starts. M-Chess has participated in public matches and tournaments, and >>has been available commercially for years. In competition it is >>commendable to be ready for anything your opponents, commercial or >>otherwise, might attempt. The strong showing of MCP7 at the Paris 1997 >>tournament seemed to validate that intent. >> >>The version 7 Opening Library is masterful and has excellent variety. >>Sandro has especially designed the MCP7 Opening Library to interest and >>challenge HUMAN chess students and opponents. I believe that it does >>so, well indeed. >> >>Starting with the 1995 World Champion M-Chess Pro 5.0, it has been my >>policy to include the tournament opening book with the latest program >>version, every year. There are book lines to move fifty, and some >>winning lines are included. Certainly, some of the lines are played by >>commercial opponents. But for every one that is, there are many >>interesting lines that are not played by any commercial opponents. >> >>Apart from computer/computer competition, is there any evidence that the >>M-Chess engine/book combination is effective? >> >>There is a record against human opposition. M-Chess was Best Computer >>at the Sixth AEGON Man/Machine Tournament, played at almost two hours >>per side. At the AEGON in 1995, M-Chess beat three IGM's: >>Christiansen, Cifuentes, and Sz. Polgar. In 1996 M-Chess won outright >>two tournaments pitting the top-rated programs against the Finnish >>National Chess Team, but lost a two-game match with IGM Nigel Short. >>In June 1997 M-Chess beat IGM Igor Efimov 3.5 to 2.5 in a match at >>ninety minutes per side. Recently, at 5 minute (plus 5 seconds per >>move), MCP7 beat IM Tom O'Donnell 9.5 to 0.5 at the office of the >>Canadian Chess Federation. Other consistently strong results have >>occurred also. >> >>At present, there is no definitive measure of the strength of chess >>programs. However M-Chess, like some other programs, has substantial >>value, for people who appreciate them and use them. >> >>-Marty Hirsch, author, M-Chess Professional > >Hmmmm. The problem with making a criticism of your post is that you go >away and don't come back, but ..... > >The questions (unanswered) are: > >1. does Mchess contain lines specifically put in the book to 'deal with' >other specific opponents ? > >2. if so, are these lines deliberately inserted, or have they just come >about as a sideline of 'improving' the book ? > >3. if so, do these lines have a material effect on the SSDF rating of >Mchess ? > The above 3 questions are of course the most relevant as many many people regard the SSDF ratings as a sort of Holy Grail, just as many players swear by the increase or decrease of 5-10 elo points in their own ratings. In a sense this is understandable as it is the best reference available as yet. >Other questions are: > >1. do you consider books in general to be just neutral data, or is there >some moral issue to consider ? I'll be honest Chris, I don't understand what you mean here (could be I need some coffee). > >2. do you have any comments on the issue of some kind of agreed basis >for what should and what should not be acceptable in book construction >(ie ECO only, or extra lines, or cptr-cptr game lines, or anything goes, >etc. etc.) > Frankly, the only thing I'm against are lines devised and included whose only purpose is to beat another machine. This can ONLY hurt the consumer. I had little in terms of opening reference material until recently, but did have 2 decent programs Genius 4 and Rebel 8. These programs were highly respected for their game and for their intelligent and solid opening books. Well, I began a few correspondance games in which one of them led to a line I had nothing on. So, I asked the programs what their theory on it was in order to make a choice. As it turned out, the line I chose after following the "theory" led to a won position for my side. I didn't expect my opponent to just follow this line all the way but I liked it and didn't bother to check it. As it turned out my opponent deviated immediately, but when I began to look at the move I had sent, I began to wonder what I would do if my opponent played a simple move that just left me lost. Turns out it was a killer-line worthless in itself, but only of use against a computer opponent. Ok, it didn't cost me the game, but I was infuriated. I realize theory changes as do evaluations, but I don't want to have to keep wondering if I can trust the computer's opening book (for purposes of study) because certain losing lines were deliberately included. If independant analysis brings about theoretical novelties that are subsequently included in the book, fine (I have seen this with Lang's programs), but deliberate, BAD computer killer lines?..... >Chris Whittington
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.