Author: Rolf Tueschen
Date: 11:41:00 05/20/02
Go up one level in this thread
On May 20, 2002 at 13:56:10, Uri Blass wrote: > >I do not think that using book is cheating. >I believe that there are also GM's who remember some book moves without >understanding them. Before going to you promissing ideas below we must surely consider this aspect. My comment on it is, that if we are trying to find fair rules in a sport we should not justify something for one side with the exceptional existence of the same on the other side. This question here is the most important IMO. Now, we start directly on GM level, and that is right so. Now, as I repeated often enough, almost all GM of the top, say 50 players or such, have eidetical talents. Otherwise they have no chance to come into the top ranks. Now, this is something you won't read about in chess magazines, simply because the GM won't talk about it, it's their secret if you want, and the journalists don't even know what eidetics really is! Recently I read an interview with Karpov, where this topic was lightly touched after a lost game I think. He said of course he couldn't remember all the lines, and that he had simply forgotten. You see? Now, there is not much we can say. Where the public doesn't know of edetics at all, a GM can easily talk about his memory and its holes. But the experts know better what 'Karpov' means or 'Kasparov', make your choice. That is not having said that these giants will be able to activate their potential power at will and always. Some distraction and the level goes down a bit. Look at Kasparov in Prague. So we take such GM with eidetics and compare him/her? with the computer program. Would you still say, that these are the same or comparable? Of course not, because even the best book doctors I heard of are no GM and are unable to make comparable analyses to a human GM. The GM you mentioned might really play a book line without understanding but he knows when to start his own thinking. Or perhaps he had accepted a silent draw? ;) No, the books in the actual format have a single goal, to get Elo numbers of 2500-2600, after imbreding even 2700 and more. Without them they are in my personal view at 2350. (This number 2350 is personal but with a little background, that is never discussed by the critics. As I wrote in the first or second posting above the number is the result _after_ human chessplayers would create a computer related new chess, and that would be only happening after enough money would be reserved for that task. I am not gambling with you when I just choose 2350 to get you in some discussions. I discuss the topic having the standards of computerchess in mind and its destiny. The number is _not_ the actual reality! It's the potential reality.) > >It does not mean that they are going to play them in games without thinking but >knowing the book moves may be productive to give some direction which lines to >extend. > >I prefer to see chess programs play in 0 seconds book moves only when they >analyzed and learned the book moves are not clearly inferior relative to the >best choice(this learning may be based on a tree that the programmer gave them). > >Programs should use books also in other cases but in that case the book should >be used only for extensions. > >I believe that using this idea is going to give practically better results for >chess programs > >Uri If your idea would become reality it looks like we had a different computerchess. Although I would like that the machine created its own tree. Again, in games against human players it should not be a fight between the operating team and the player but between the player and the machine itself. Rolf Tueschen
This page took 0.01 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.