Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / what is your opinion about to import books?
- - By NOS_KOS (*) [gr] Date 2009-06-24 21:36
for example if i import games with high value in my current book that i use..
or to import another good book to my current one..
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-06-24 22:34
Importing games is always a good thing because, as I always say, more information is better than less information.
Parent - - By Jonas (****) [at] Date 2009-06-24 22:39
no, it also belongs on the information that is added.
importing games is a good thing if you import the right games. if you do it wrong you can completly destroy your book.
id say just make a backup copy and try it.
so... please be carefull
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-06-24 23:36 Edited 2009-06-24 23:40
"Right" games and "wrong" games--can you please define or quantify these terms?  For example, how could I find only those games that could "completely destroy" my book?

Isn't it true that in all but a tiny percentage of games that 1/2 point-losing or full point-losing moves are made by both sides?  (Think in terms of all positions being won, lost or drawn, as in an endgame tablebase.)

Isn't it likewise true the best way to arrive at truth in science is trial-and-error?  Is not every move sequence a kind of hypothesis?  And does not trial-and-error over time produce a body of knowledge, or more accurately a body of empirical results, that lead to theories that lead to something short of proof that lead eventually to proof?

I concede readily that game results are very misleading in probably 95%-99% of cases because both players probably made game outcome-changing errors.  But many of these errors are so subtle we may never discover them, while others are comparatively obvious and soon exposed.  But these truths can only emerge after a lot of trial-and-error.  My point is how can trial-and-error be "doing it wrong" when it is, in fact, part of the process of discovery?
Parent - - By Jonas (****) [at] Date 2009-06-24 23:49 Edited 2009-06-25 00:09
ok, heres my definition:
my guess is that only 1 of 4 games are decided in the opening.
half of them are draw and therefor suboptimum
so only 1 of 8 games are usefull. (or about 20% or whatever it doesnt mather at all how many those are, but its definatly already the minority)
then, they have to have the correct result, else the book plays weird deciding a move to be good just because the game was won.
so the won opening also must have a positive result. and by far not even 0,01% of all games have that.
those are then the 'right' games and those games will increase the books strengh.

u cant have 100% of those games (you can maybe make a book with 100 games in a week if you work that carful) but you should try to get as much as possible. so you can only guess what games are accurate (why am i always using this word when talking to you? lol). i prefer using rybka-rybka games and centaur games on high time control
Parent - - By Jonas (****) [at] Date 2009-06-24 23:57
yes, i must admit that in fact it is even a bit more complicated. but thats a short overview about the 'right' games
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-06-25 01:11

> in fact it is even a bit more complicated


This remark may be the greatest understatement in the history of the Rybkaforum.  Congratulations for this distinction!
Parent - By Jonas (****) [at] Date 2009-06-25 06:01
:) hey guys u got that? owned you all!! can't touch this naa nanana ;P
Up Topic The Rybka Lounge / Computer Chess / what is your opinion about to import books?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill