Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Importance of a book for an engine
- - By Indrajit (***) Date 2010-07-22 19:05
Here is something I am not able to understand. How is it that an opening book geared for an engine is necessary? Lets say, I start a tournament between two of my chess engines, say Rybka and Shredder. Both of them use the same book. After N number of moves, both the engines will move out of the book and start analyzing the positions. Since both the engines have used the book till Nth position, both are equally matched. So from the N+1 th position, whichever engine is able to analyze the board better will have a higher chance of winning. Am I right in this assumption?
Parent - By Erudito (**) Date 2010-07-22 20:03
Parent - - By theoak (**) Date 2010-07-22 20:04
the book that was created with an engine is an advantage in many ways:
a) the engine has already spent a big amount of time analyzing a position (in tournament this is not possible), therefore
the horizon is deeper and opening traps more rare.
b) a human brain is involved
c) the variations that make it in the book can be the ones that the engine feels "comfortable" to play or make the opponent
"non comfortable". Anti-human play e.t.c

many many others ...
Parent - - By Indrajit (***) Date 2010-07-23 01:47
I can agree with point (c) but not with (a) or (b), because if an engine and a human brain is involved to that extent in making the book, hence the moves would be the correct ones for both the engines. If there is an error in calculation while making the book, then either engine would be at a disadvantage in one out of every two games!
Parent - By Athinira (**) Date 2010-08-01 17:05 Edited 2010-08-01 17:07
Errors can (and should) be corrected, so i don't really see your point.

The engines them self are also made by a human, and can also contain errors putting them at a disadvantage.
Parent - By Banned for Life (Gold) Date 2010-07-22 20:25
Different engines play certain positions better than others. If a book is optimized for best results with Rybka 4, and you use this book to generate starting positions for a match between Rybka and Shredder, Shredder will be at a disadvantage. Likewise, if the book was generated for best results with Shredder, R4 will be at a disadvantage. It is easy enough to speculate why this might be the case. For example, if the opening results in a material advantage, this might be advantageous for Rybka. Maybe Shredder would be relatively better in closed positions, or in positions that are closer to the endgame.
Parent - - By titanium cranium (***) Date 2010-07-22 23:18
Rybka opening books or Aquarium books or MCO or any other books that combine human and engine games can only be used a general guidelines in engine v engine play. What we need are opening books based solely on R3 games or R4 games, etc. Otherwise, . . . loses
Parent - By Indrajit (***) Date 2010-07-23 01:50
That might not be correct - if your books are based solely on Rybka 3 / 4 games then there might a lot of situations which would not arise - where judgment of other types of engines / human mind would be necessary.
Parent - By Labyrinth (*****) Date 2010-07-23 10:02
Concerns like this are why I personally have a lot of respect for chess960 tests.
Parent - By Nelson Hernandez (Gold) Date 2010-07-31 23:07
Returning to your original question, your assumption is correct, but it omits one small thing that influences the outcome.

Both books may exit at the same position.  In this case they have used an equal amount of time (zero).  However all book-exits are not equal.  The starting position, for example, has an approximate +0.15 advantage for white.  So if your exit position is greater or lesser than that number it is apt to influence the end-result.

This is true in very broad statistical terms, and I only say it because we know from many millions of games that white has something like a 54.1% success rate, which translates to about +0.15 in Rybka.  Of course, all positions are objectively won, lost or drawn, and we know evaluations are merely a mathematical construct that represents the relative chance of success based on values in a chess program.  Thus many if not most evaluations are way off the mark, including the opening position, which most likely is correctly evaluated at 0.00.  However, speaking in generalities (and Rybka is a great program for generalities), there is a correlation and the goal of any bookmaker is to exit book with a strong evaluative advantage.
Up Topic Rybka Support & Discussion / Rybka Discussion / Importance of a book for an engine

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill