Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Rybka Discussion / Are all opening books and theory obsolete after the release
- - By X3 (**) [se] Date 2008-07-13 21:25
Are all opening books and theory obsolete after the release of Rybka3?!

X3
Parent - By George Tsavdaris (****) Date 2008-07-13 21:30
No, why they should be?
Parent - - By turbojuice1122 (Gold) [us] Date 2008-07-13 21:35
No--opening theory is at least at the level of 3200-3300 elo, perhaps closer to 3500 or more for mainline stuff.  Rybka 3 will have a rating of perhaps 3150 elo on a Skulltrail (when equated to the FIDE Elo scale).
Parent - - By X3 (**) [se] Date 2008-07-13 21:47
Most Playchess opening books are made of games (and analyses) from games 15-60 m for a whole game.

Of course you can´t rewrite the theory for the beginning of the games (i.e. Ruy Lopez) but you have Playchess theory going as deep as 30 to 50 moves.

I guess that we could see some interesting things on the Playchess server in a few weeks!

X3
Parent - - By George Tsavdaris (****) Date 2008-07-13 22:38
Rybka 3 would have a chance to make books obsolete if it could find the 15.Bxf7!! move for example of the following position:
rnb1kb1r/1pq2p1p/p4P2/4pP2/2B5/2NQ4/PPP2p1P/2KR3R w kq - 0 15

This move draws at least and if black is not very very careful he would lose before he even realizes it.
Parent - - By lkaufman (*****) Date 2008-07-14 00:49
One of the games in the databases with this line is my game with GM Gormally, before he was a GM. I was on the Black side. Result was a draw.
Parent - By M ANSARI (*****) [kw] Date 2008-07-14 06:49
Books will not be obsolete ... but will be greatly challenged.  Most books made for computer chess are made from engine vs. engine games ... and by far the strongest engine will be Rybka ... so you can say that most books are made from Rybka vs. Rybka games ... and thus use the evaluation of Rybka 2.3.2a ... if now the evaluation changes, things will obviously still have to change.  I estimate that the holes in the "obsolete books" will take about 2 weeks to clean out.
Parent - By George Tsavdaris (****) Date 2008-07-14 10:43
Yes i see it now but you were lucky in the sense that your opponent didn't played the best moves to give you a hard time. You could also take the win from his suboptimal moves but you didn't. For example:
Move 18:
He played 18.Nd6?!
The at least drawing move that gives you a hard time since you have to find a difficult moves in order to survive, was 18.f6+!
rnb2b1r/1pq1k2p/p7/3QpP2/4N3/8/PPP2p1P/2KR3R w - - 0 18


After 18.f6+! Ke8 19.f7+ Ke7 20.Qd2! what move would you play? (hint: All moves lose except one that draws.)
rnb2b1r/1pq1kP1p/p7/4p3/4N3/8/PPPQ1p1P/2KR3R b - - 0 20


OK it's 20...Qb6!! that draws. But things are again not easy after this for either side as there are many "!" moves again....

So after the 18.Nd6?! move he played, you could even win.
Game correctly continued: 18...Bh6+ 19.Kb1 Kf6  And now he may play 20.Qf3 to give you a hard time about your 20th move.
What move would you play here? (hint: All moves lose, one move draws and one move wins.)
rnb4r/1pq4p/p2N1k1b/4pP2/8/5Q2/PPP2p1P/1K1R3R b - - 0 20


The best move you had to find was: 20...Bf4! that wins.  While the draw move was 20...Rd8.

Of course he didn't play 20.Qf3 but 20.Ne4+ and you accepted the draw with repetition playing 20...Ke7 and not risking anything.
rnb4r/1pq4p/p4k1b/3QpP2/4N3/8/PPP2p1P/1K1R3R b - - 0 20

If instead you have played 20...Kg7!  then after 21.f6+ Kf8!!(Kg6 draws) we have:
•22. Qd8+ Qxd8 23.Rd8+ Kf7 24.Rxh8 Nd7 25.Rxh7 Kg6 26.Rh8 Be3 and black has the advantage.
•22.Rhf1 Nd7 23.Rxf2 Rg8 24.Qe6 Nc5 25.Nxc5 Be3 26.f7 Bxe6 27.Nxe6+ Ke7 28.Nxc7 Bxf2 29.fxg8Q Rxg8 and black has the advantage.
Parent - - By Geomusic (*****) Date 2008-07-15 11:00
2.3.2a is already rated 3130 on CCRL rybka 3 is expected to be rated between 3220-3250 from the tests i've seen :)
Parent - By turbojuice1122 (Gold) [us] Date 2008-07-15 11:29
CCRL, however, does not equate to the FIDE list--in fact, I would say that it doesn't even remotely equate to the FIDE list for the top programs.  Even the CEGT list, which has lower numbers, is inflated for the top programs.  I would say that in terms of FIDE elo, Zappa Mexico would be rated just over 3000 on a Skulltrail machine (perhaps 3020), and Rybka 2.3.2a would be around 3050.
Parent - - By Quapsel (****) [de] Date 2008-07-15 13:14

> No--opening theory is at least at the level of 3200-3300 elo


1.) What arguments do you have for: the book authors created their Books for a lower ELO number!
They didn't create them for humans of their own strength. How do you want to detect, for which strength they are created?
That former engines had been weeker seems to be no good enough argument.

2.) Much has been taken out of GM-games.
And since Computers exists the style and the weak and the strong points of their play differs from that of humans.
So never such GM-based books had been fitting for engines.

But they worked rather good and so they were used.
And all the times bugs are found and have to be corrected.
so it was, and I think that so it will be.

Quap
Parent - - By turbojuice1122 (Gold) [us] Date 2008-07-15 13:18
When I say, "opening theory", I am referring to GM opening theory that has been checked by engines.  The elo of this has actually been roughly unchanged for the past 15 years.  Opening theory is almost entirely perfect play, and many of the moves are equally optimal, making it also perfect in the "strong" sense of the word.  3200-3300 seems likely to be an underestimate.  GMs are far stronger than computers in this phase of the game, and when they are given correspondence type time controls to improve, they aren't going to make very many mistakes, even small ones.
Parent - - By Banned for Life (Gold) Date 2008-07-15 16:53
Sure, its perfect play right up to the moment when its refuted and found to be totally wrong! :-)

Regards,
Alan
Parent - By Roland Rösler (****) [de] Date 2008-07-15 18:34
+1! :-)
Parent - By turbojuice1122 (Gold) [us] Date 2008-07-15 19:15
That tends to happen in the sharpest lines, and then we're dealing with two things.  First, these aren't main lines so much--it's hard to consider something that is 60 ply deep to be a main line, especially in GM play.  Second, even "refuted" variations are often drawn with correct play afterward--it just becomes more difficult to find the correct moves.
Parent - - By onursurme (***) [tr] Date 2008-07-14 08:44
It depends on how long you search a position with R3.
Parent - By onursurme (***) [tr] Date 2008-07-16 19:49
It is people not R3 who can make opening books obsolete.
R3 is a great help for who tries.
Parent - - By buffos (Silver) [gr] Date 2008-07-14 08:56
Opening theory is an evolutionary process. It can be enhanced but not made obsolete.
In some openings (e.g Kings Indian) Rybka (and other engines) are of no use at all.
Parent - - By Uri Blass (*****) [il] Date 2008-07-14 08:59
How do you know that rybka3 is of no use in the kings indian?
Do you have rybka3?

Uri
Parent - - By buffos (Silver) [gr] Date 2008-07-14 09:03
No, but i have not seen a singe engine that can understand KI.

There are certain keys to the position that no search can find. Just evaluation
Parent - By Uri Blass (*****) [il] Date 2008-07-14 09:07
Rybka3's evaluation is better and I doubt if no search can find.
I would like to see positions when you think all engines fail to see the right move
and also to know the right move(I do not play that opening so I do not know).

Uri
Up Topic Rybka Support & Discussion / Rybka Discussion / Are all opening books and theory obsolete after the release

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill