Author: Uri Blass
Date: 06:59:30 12/14/03
Go up one level in this thread
On December 14, 2003 at 09:26:24, Omid David Tabibi wrote: >On December 14, 2003 at 09:18:21, Uri Blass wrote: > >>On December 14, 2003 at 09:01:26, Omid David Tabibi wrote: >> >>>On December 14, 2003 at 08:34:31, Uri Blass wrote: >>> >>>>On December 14, 2003 at 08:10:43, Omid David Tabibi wrote: >>>> >>>>>On December 14, 2003 at 07:58:41, Uri Blass wrote: >>>>> >>>>>>On December 14, 2003 at 06:00:10, Tord Romstad wrote: >>>>>> >>>>>>>On December 13, 2003 at 19:18:58, Omid David Tabibi wrote: >>>>>>> >>>>>>>>On December 13, 2003 at 04:26:16, Tord Romstad wrote: >>>>>>>> >>>>>>>>>The easiest solution to this is to spend another afternoon on adding opening >>>>>>>>>book >>>>>>>>>code to the engine. >>>>>>>> >>>>>>>>How exactly, if you don't know anything about the format (ctg) you are going to >>>>>>>>code for?! >>>>>>> >>>>>>>Why is it so important to use exactly the same book? Just do like almost >>>>>>>everybody >>>>>>>else, and build a book from a PGN database. >>>>>>> >>>>>>>Tord >>>>>> >>>>>>I do not know how almost everybody does it. >>>>>>Did you look first in free source code like book.c of crafty to understand how >>>>>>other do it? >>>>> >>>>>Most programmers don't get into the hassle of writing their own book format. >>>>>Instead, the time is better spent improving the automatically generated book in >>>>>one of the common formats. >>>> >>>>I believe that most of the opponent of movei in Leo's tournament use their own >>>>book format but I may be wrong. >>> >>>I was referring to most programmers in WCCC. At least half the participants were >>>using CTG or BOK formats. >>> >>>Leo's tournament is in WinBoard, which does not have any book support. >> >>I know but the number of programs that participate in Leo's tournament is >>clearly bigger. >> >>I simply get testing time for my program thanks to the fact that I support >>winboard so I have more motivation to care about having book under winboard and >>not to care about having book under Fritz >> >> >>> >>> >>> >>>> >>>>I believe that using an existing format is wrong because it means that I have no >>>>freedom to change the book dependent on the time control. >>>> >>>>I believe that it is better to have a bigger book for blitz when the logic is >>>>that at longer time control I expect the engine to find more often by itself >>>>better moves when in blitz it may be better to trust some known alternative that >>>>was not played a lot. >>> >>>You can create several books, or set the interface to play a larger subset of >>>opening moves. >> >>I understand but I do not like to be dependent on interface and I have less >>freedom by doing it. >> >>I also think about not playing automatically book moves in 0 seconds(except >>moves that were checked manually) in order to avoid stupid blunders that I have >>bad luck to have in the pgn file and to use 1/10 of the normal time to calculate >>to see if I can trust the book. > >If you want to avoid book mistakes do the following: > >Take a pgn database of 2500+ rated players, and set the play threshold to 4 >games, and 2 wins at least. It is very unlikely that a move played by 4 strong >players is a "stupid blunder", especially if it results in a victory in 2 of >them. > >But I think that using such a high quality database, 1 win out of 2 (or 3) is a >better practical value than 2 out of 4. I understand but I have only 2 options not to trust a book move and to use normal search or immediately play it and I prefer to have a third option based on statistics and it is to play it only if some calculation suggest that it is not a blunder. Uri
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.