Author: Sandro Necchi
Date: 22:49:07 02/18/05
Go up one level in this thread
On February 18, 2005 at 14:17:11, Dann Corbit wrote: >On February 18, 2005 at 12:52:51, Volker Böhm wrote: > >>Hi, >> >>this shouldn´t be a problem would it? :-) >> >>Not joking: >>I think a good book could improve a chess engine, and a bad book can hinder a >>chess engine. >>My theory is that the stronger a chess engine is, the better must a book be to >>not weaken the strength of the engine. Thus a strong engine needs at least an >>acceptable book. The amount of elo a strong engine can gain from a "very strong >>book" (i.e. a book that is well tuned for the engine) is much less than the >>amount of elo a weak engine can gain. No, if a strong engine is playing agaist a very strong opponent, and gets a bad position from the book it may will not be able to recover. What you say is true only if facing weaker opponents. > >Originally, that was my theory also. > >But a good book seems to provide as good or perhaps even more benefit for Fruit >as it does for Glaurung. Now, Fruit is stronger than Glaurung, and so I would >have expected a bigger benfit for Glaurung. But I did not see that: >http://www.talkchess.com/forums/1/message.html?412441 In fact normally it is the opposite, if we are talking about computer games (computer against computer) and not computer against human. Why? Because the stronger the program is the best he will use the position out of book. If the program is very weak it may play the opposite plan and so on. Intermediate values will give limited success. > >Probably, at least with extreme examples, we must see something along those >lines, I would think. > >We cannot (for instance) benefit a 2700 Elo program by 400 Elo points, but >probably with a 1500 engine and an excellent, deep book we can do that. Only on specific cases and on limited amounts of games. Sandro
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.