Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Opening book question
- - By mightyRYBKA (*) [ca] Date 2009-04-12 04:22
Hello, it seems lately that most of my opponents no matter where I play, use the same opening book( mybook - caro kann gives it way easily ) and the games usually end up in draw despite my opponent being on lesser hardware. Is there anyway to avoid this or better yet is there a book that normally does better against mybook?

Thanks, appreciate all the info.
Parent - By Sunny (***) [in] Date 2009-04-12 04:29
Better is hand tune your own book and then test it against mybook and remove the loosing or the draw lines from your own book.
Parent - - By M ANSARI (*****) [kw] Date 2009-04-12 05:57
Most opening books are based on Playchess games, and since most of those games were downloadable ... it is normal that everyone ends up with similar games.  This might change now that the Spaghetti site is not providing monthly book updates.  Also the fact that many book authors are graciously providing their books for download has also made it easy for anyone to even the playing field.  Sedat has really helped out a lot of people who were not able to get competitive books to get in the action, and CB should give him a medal for that.  At the moment I am developing a book based on the many thousands of games I have played on my internal network at home, and it is refreshing to see it play as it plays quite differently since many of those games are based on neutral books.  But as soon as I put that book on Playchess, again all the relevant lines get copied within a week.  I might release this book to increase diversity on Playchess once I clean it up a bit.  So far it has a plus score against all the top major books.

One thing to note about Playchess books is that they are optimized for fast time controls, and will not do well at longer time controls.
Parent - - By ppipper (*****) [es] Date 2009-04-12 10:15
Some months ago I decided to not to play 1...c5 and 1.e4 lines anymore.

Statistics about games and certain moves are really redundant because the lack of diversity in playchess's games. Of course the first few hundreds of games you tend to loose, but soon you start to appreciate the flavour of winning with some "strange" moves like 1.d4, 1...e6, 1...f5, and so on...

With all respect to all people who do it, I find no merit at all playing some of the mentioned lines, where you stay in book until move 38, specially when you are playing that line just because lot of people have done it before, not because you have worked your book so deep. Most of book makers around still do the same.

Regards
Parent - By ChessKnight (**) [ca] Date 2009-04-12 10:40
ppipper

>I find no merit at all playing some of the mentioned lines, where you stay in book until move 38, specially when you are playing that line just because lot of people have done it before, not because you have worked your book so deep.


-You're right.

After certain initial opening moves, chess tree starts growing up so hugh with major branches and sub-branches. So most of the people having wining strategy would like to play only proven lines from the past rather than exploring each branch and it's sub-branches to the deepest root level which is in fact very time consuming.  And only one or two game out of these invented lines doest not alter any opening book statistics much nor guarantee for sure play by an engine. Normally good books and good engine only plays those moves which has a strong history and statistics.

Still there are tons of major branches and sub-branches to be explored yet to the root-level.

>Most of book makers around still do the same.


Yeah, it's true unforunately.
And sad part of it is Book Blenders blending someones book completely in to theirs, so out come of books also contains good lines and draw-backs of previous ones as well.

CK

 
Parent - By ChessKnight (**) [ca] Date 2009-04-12 10:17
M ANSARI

>But as soon as I put that book on Playchess, again all the relevant lines get copied within a week.


- I agree 100%.

For CB books direct import seems to be major advantage for the book blenders and major disadvantage for genuine book authors who improve their lines-by-line by research and hand-tuning. It takes only few seconds for book blender freaks to import all novelty (newly researched) lines into their own books.

Any way it does not stop contributing goodies to the world as long as opening books are made public available for FREE.

CK  
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-04-12 11:20
If you really wanted to influence the lines you'd provide the games, not just the book, as a whole generation of new books would all incorporate them.

As for the My book question that prompted this, the simple answer is to find new lines.  The quickest way to do that is to play the Rybka-My book combination against other engines with short books, forcing My book to always play the line you want to beat.  The inferior engines will be slaughtered, of course, but every once in a while they may come up with something completely new that wins.  That's what you are looking for.  All books get worn out over time.  Remember RybkaII?  Perfect 15?  Both were cutting edge in their day.  Now they are easy marks.

 
Parent - - By M ANSARI (*****) [kw] Date 2009-04-12 11:44
Very true indeed!  Actually I have a way of promoting such games ... I use double, sometimes triple and even quadruple copies for such game to artificially enhance these games in a book.  Games that Rybka 3 loses against Dynamic or N4 or ZII or even DF 11 are very rare indeed (except for R3 Dynamic) and since most will be using R3 Default, books that are artificially blended in this manner work very well on Playchess ... again for a while that is.
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-04-12 12:14
If you really want to create mischief on Playchess it's very simple.  Find the very worst games you can find--human games between sub-2000 players only--and make that into a book with "Normal" settings.  If that idea doesn't make you smile you have no sense of humor.  It won't be a strong book, but everybody will be out of book so fast they won't know what hit them.
Parent - By Uly (Gold) [mx] Date 2009-04-12 12:17
I'd like to see Vas doing that with Rybka cluster, now, going out of book early AND losing the game would be a new dimension.
Parent - By Sunny (***) [in] Date 2009-04-12 15:25
I will try your idea Nelson. With black I have already stoped playing black and surprised many people who love b90 and b80 and this time Human games :)

Sunny
Parent - - By Banned for Life (Gold) Date 2009-04-12 17:51
Your protegee, Amando, has already demonstrated that atrocious opening moves with extensive preparation can achieve excellent results (does 1.h3 still have the highest win percentage?).  Some people have accused me of doing the same thing, but I deny it! (Or maybe I am in denial).
Parent - By Nelson Hernandez (Gold) [us] Date 2009-04-12 18:52
In my book 1.h3 does have the highest win percentage, but if you ELO-adjust the results the numbers come back down to earth, and if you look at the ELO-adjusted result over time (looking at games by month) you see that the days of easy pickings are long over.  1.h3 is basically a draw along the lines that have been created.  Not sure if it is forced but it gives that impression.  Of course you can deliberately transpose back into a main line and a playable game, but few do that.
Up Topic The Rybka Lounge / Computer Chess / Opening book question

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill