Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Quick loss to Shredder 9 at long time control
- - By Svilponis (***) Date 2008-08-09 16:13
[Event "40 Moves in 120 min; 20 Moves in 60 min; Rest in 60 min"]
[Site "R3 test2"]
[Date "2008.08.09"]
[Round "2.1"]
[White "Shredder 9 UCI"]
[Black "Rybka 3 1-cpu 32-bit"]
[Result "1-0"]

1. d4 {book 0s} Nf6 {book 0s} 2. c4 {book 0s} g6 {book 0s}
3. Nc3 {book 0s} Bg7 {book 0s} 4. e4 {book 0s} d6 {book 0s}
5. Nf3 {book 0s} O-O {book 0s} 6. Be2 {book 0s} e5 {book
0s} 7. O-O {book 0s} Nc6 {book 0s} 8. d5 {book 0s} Ne7
{book 0s} 9. Ne1 {book 0s} Nd7 {book 0s} 10. f3 {book 0s}
f5 {book 0s} 11. g4 {book 0s} Kh8 {book 0s} 12. Bg5
{+0.55/17 8:54m} f4 {-0.06/16 11:21m} 13. Bh4 {+0.29/17
3:25m} h6 {-0.07/16 5:33m} 14. Nd3 {+0.48/18 2:52m} b6
{-0.07/15 6:24m} 15. b4 {+0.40/19 6:21m} a5 {-0.10/16
7:27m} 16. c5 {+0.35/18 6:17m} Bf6 {0.00/15 5:41m}
17. Bxf6+ {+0.35/17 7:05m} Nxf6 {0.00/16 6:59m} 18. Qb3
{+0.14/17 3:38m} h5 {-0.06/13 3:27m} 19. cxd6 {+0.13/17
5:12m} cxd6 {0.00/12 0s} 20. g5 {+0.49/18 5:30m} a4
{+0.01/15 3:50m} 21. Qb2 {+0.73/19 6:44m} Nh7 {0.00/17
4:04m} 22. h4 {+0.82/19 3:10m} Nxg5 {0.00/17 2:19m}
23. hxg5 {+1.10/19 3:08m} Nf5 {0.00/17 2:23m} 24. Rf2
{+1.22/19 3:38m} Qxg5+ {0.00/16 4:07m} 25. Rg2 {+1.40/19
3:25m} Ng3 {0.00/15 0s} 26. Nf2 {+1.40/20 5:57m} h4
{-0.25/14 6:21m} 27. Kh2 {+1.48/18 2:59m} Qh6 {-0.32/14
6:06m} 28. Rc1 {+1.97/18 2:48m} g5 {-0.52/13 5:00m} 29. Qd2
{+1.93/18 3:01m} Rb8 {-0.70/14 5:15m} 30. Rgg1 {+2.22/17
2:08m} b5 {-0.77/15 6:54m} 31. Bxb5 {+2.72/18 1:24m} a3
{-0.75/15 2:52m} 32. Rb1 {+3.07/18 2:42m} Qg6 {-0.77/14
6:03m} 33. Qd1 {+3.16/18 4:39m} g4 {-1.05/12 1:43m}
34. fxg4 {+4.08/16 54s} f3 {-1.09/15 3:09m} 35. Qc1
{+3.72/18 3:20m} Bxg4 {-0.92/15 4:54m} 36. Qxa3 {+4.75/19
5:58m} Bc8 {-1.07/12 1:53m} 37. Qc1 {+4.96/18 3:09m} Kh7
{-1.26/12 1:31m} 38. a4 {+5.24/17 2:35m} Qf6 {-1.50/12
1:53m} 39. Qe3 {+5.36/17 2:23m} Qg7 {-1.57/12 1:04m} 40. a5
{+5.35/18 4:21m} Kh8 {-1.82/13 1:09m} 41. Bd3 {+6.23/17
5:15m} Rg8 {-2.46/12 4:32m} 42. b5 {+6.88/18 4:55m} Kh7
{-2.98/11 2:29m} 43. b6 {+7.37/17 3:23m} Kh8 {-3.85/12
3:03m} 44. Nb5 {+8.45/16 2:11m} Qg6 {-5.82/11 16:50m}
45. Na7 {+9.72/16 2:22m} Qg7 {-5.82/10 2:20m} 46. Nxc8
{+10.87/18 3:39m} 1-0
Parent - - By turbojuice1122 (Gold) Date 2008-08-09 16:20
Looks like a book loss to me...
Parent - - By George Tsavdaris (****) Date 2008-08-09 16:36
Definitely not a book loss but for computers it's almost mandatory that the 3...Bg7 should be avoided and 3...d5 to be played.....
Parent - - By Uri Blass (*****) Date 2008-08-09 17:40
I wonder if you think that humans can beat rybka in a match when rybka starts all games with black after 3...Bg7(rybka should be free to use an opening book)

Uri
Parent - By George Tsavdaris (****) Date 2008-08-09 18:09
My point is that Bg7 allows white if he wants to, to create a close position, where black has to play the right freeing moves the right time, and this is not easy for computers.

I also believe that Rybka 3 can win in a match against top humans after 3...Bg7. And actually the important question if Rybka 3 would be the favorite in such a match. I think it will be....
Parent - - By Venator (Silver) Date 2008-08-09 17:42
Yep. For humans 3... Bg7 is a '!?', but for computer programs definitely a '?!'.
Parent - By Svilponis (***) Date 2008-08-09 21:40
Ok, I see! But I still think, that draw should be possible even after this opening.
Parent - - By Bouddha (****) Date 2008-08-09 22:45
??

Grünfeld is good for black & computers !
Parent - - By Venator (Silver) Date 2008-08-10 07:43
Grunfeld: yes! But this was King's Indian.
Parent - By Bouddha (****) Date 2008-08-10 10:41
While I addmit that Kings Indian is an opening that computers have difficulty to play correctly because of its closed position.

I havent find a version of Kings Indian where white computer wins in most of the cases against black computer.

In computer vs computer games, the back side of the kings indian often wins. When looking at the game there are many places where a human intervention could improve
Parent - - By Priyanvada (***) Date 2008-08-10 08:42
One Should observe...

Upto 40 th move Rybka's evaluation doen't rise above -2.00 on negative side!!!,
...and just 6 moves latter "Game is Finished"!

...while shredder's evaluation almost perfectly reflects the situation of the game, and Game also progresses according to shredders mind!

what a mystery for a  3200 ELO engine!!!
Parent - - By Svilponis (***) Date 2008-08-10 08:49 Edited 2008-08-10 08:52
That was the reason why I posted the game. I saw that Rybka 3 misevaluates the game. But she reaches smaller depths also, and that might be the reason.
Parent - - By Arkansaw (***) Date 2008-08-10 09:02
one side reads +10 another reads -5
are you sure they are playing the same game?!
Parent - By Svilponis (***) Date 2008-08-10 09:47 Edited 2008-08-10 09:51
This is the way Shredder 9 GUI shows the evaluation. It gives evaluation for move, not for side (+ does not mean that white has dominant, but that moving side is dominating). + is positive evaluation, - is negative, and it is pretty normal, that if the position favours one side, then moves for one side are evaluated positive and at the same time moves for opposite side are negative.
Parent - - By George Tsavdaris (****) Date 2008-08-10 10:39
No mystery if you know how to read the evaluations and not only see -2.00 of Rybka and +5.00 of Shredder as "Oh Shredder sees it is a clear win while Rybka it doesn't."
Rybka's evaluations of more than 1.50 mean an expectancy of losing of about 90-95%.
So don't tell me that Rybka thought it would not lose.

If this i gave is correct then Rybka's -1.82 means she believes that white will win with a probability of more than 90% and that white's expected score anyway will be more than 95%.

Even a 1.00 score of Rybka 3 must mean that white's expected score is about 85% and that white's probability of winning is 70% more than that of black's.

So you see that these +1.00 score o fRybka that in other engine are considered small, for Rybka are huge.
Other engines evaluation of +1.00 can rapidly change and don't mean much, even +2.00 doesn't mean much for other engines. But for Rybka this is a different matter(excluding big evaluations in endgames though where Rybka's evaluations sometimes are nonsense).
Parent - By Priyanvada (***) Date 2008-08-10 13:03
I seeeeee
Up Topic Rybka Support & Discussion / Rybka Discussion / Quick loss to Shredder 9 at long time control

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill