Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / This is just... buggy
- - By Ivang64 (*) Date 2009-10-28 01:03
[Event "60m + 15s, unrated"]
[Site "?"]
[Date "2009.10.27"]
[Round "?"]
[White "Rechenschieber, Rybka 3w91 clusterJ"]
[Black "Highendman"]
[Result "1-0"]
[ECO "E99"]
[Annotator "Trajkov,Ivan"]
[PlyCount "217"]
[EventDate "2009.10.27"]
[EventType "rapid"]
[TimeControl "3600+15"]

1. d4 {0.05/1 0} Nf6 {74} 2. c4 {0.05/1 0} g6 {23} 3. Nc3 {0.05/1 0} Bg7 {72}
4. e4 {0.05/1 0} d6 {74} 5. Nf3 {0.05/1 0} O-O {14} 6. Be2 {0.05/1 0} e5 {39}
7. O-O {0.05/1 0} Nc6 {50} 8. d5 {0.05/1 0} Ne7 {10} 9. Ne1 {0.05/1 0} Nd7 {45}
10. Be3 {0.05/1 0} f5 {11} 11. f3 {0.05/1 0} f4 {42} 12. Bf2 {0.05/1 0} g5 {29}
13. a4 {0.05/1 0} a5 {161} 14. Nd3 {0.05/1 0} b6 {21} 15. b4 {0.05/1 0} axb4 {
19} 16. Nxb4 {0.05/1 0} Nf6 {26} 17. Nd3 {0.05/1 0} h5 {21} 18. Nb5 {0.05/1 0}
Bd7 {282} 19. Qd2 {0.69/19 257} g4 {193 (Nc8)} 20. a5 {0.74/18 206} bxa5 {9
(gxf3)} 21. Rxa5 {0.78/18 189} Rb8 {3} 22. Rfa1 {0.85/18 215} Nh7 {229 (gxf3)}
23. Nb4 {0.85/18 218} Ng5 {9} 24. Na6 {0.91/18 232} Rc8 {154 (Rb7)
Rechenschieber reconnected. Rechenschieber reconnected. Rechenschieber
reconnected. Rechenschieber reconnected. Rechenschieber reconnected.
Rechenschieber reconnected. Rechenschieber reconnected. Black restart ok.} 25.
Kh1 {0.88/18 422} Ng6 {635} 26. Naxc7 {0.91/20 0} Rxc7 {17} 27. Nxc7 {0.94/21
60} Qxc7 {5} 28. Ra7 {0.94/20 187} Qd8 {5} 29. Qe1 {1.07/19 165} h4 {332 (Kh7)}
30. Bxh4 {1.21/19 40} Nxh4 {2} 31. Qxh4 {1.24/20 26} Nh7 {49} 32. Qxd8 {1.47/
22 38} Rxd8 {4} 33. Ra8 {1.52/25 144} Rxa8 {3} 34. Rxa8+ {1.48/24 11} Bf8 {2}
35. Ra6 {1.53/25 205} Bc8 {174} 36. Ra7 {1.51/27 0} Nf6 {14} 37. h4 {1.64/27
144} Bd7 {74} 38. Bd1 {1.63/26 70} gxf3 {158 (g3)} 39. gxf3 {1.76/22 21} Be8 {6
} 40. Bb3 {1.93/26 136} Kh8 {34 (Bh5)} 41. Rc7 {1.98/26 191} Bh5 {53 (Bg6)} 42.
Kg2 {1.93/25 16} Bg6 {3} 43. Rb7 {2.09/26 74} Bh5 {116} 44. Ra7 {2.05/26 0} Be8
{49} 45. Ba4 {2.05/27 31} Bh5 {10 (Bg6)} 46. c5 {2.59/24 166} dxc5 {10} 47. Ra6
{2.39/20 6} Kg7 {25} 48. d6 {2.72/24 0} Bf7 {88 (c4)} 49. Ra7 {2.72/23 29} Kg6
{39} 50. d7 {2.79/26 0} Be7 {22} 51. Ra8 {2.81/26 0} Nxd7 {10} 52. Bxd7 {2.93/
26 8} Bxh4 {23} 53. Be8 {3.31/27 0} Bxe8 {65} 54. Rxe8 {4.29/27 0} Bf6 {51} 55.
Kf1 {4.84/27 0} Kf7 {72} 56. Rc8 {5.12/26 0} Kg6 {35 (c4)} 57. Rc6 {5.09/27 46}
Kf7 {66 (c4)} 58. Ke1 {5.12/28 45} Ke7 {105 (Be7)} 59. Kf2 {5.12/28 44} Kf7 {
34 (Bh4+)} 60. Ra6 {5.12/28 43} Be7 {55 (Ke7)} 61. Ke1 {5.12/28 43} Bf8 {33
(Bf6)} 62. Kd2 {5.12/27 42} Be7 {8} 63. Rc6 {5.12/27 18} Bh4 {103} 64. Rc7+ {
5.12/27 0} Kf6 {65 (Ke6)} 65. Ke2 {5.12/28 41} Ke6 {19} 66. Rh7 {5.12/27 6} Bd8
{47 (Bg3)} 67. Rh8 {5.12/27 41} Bf6 {14} 68. Rh3 {5.12/27 11} Kf7 {42 (Bd8)}
69. Rh6 {5.12/28 40} Ke6 {26 (Kg7)} 70. Rh1 {5.12/26 39} Kf7 {13 (Kd7)} 71. Kd2
{5.12/27 38} Bg5 {11 (Kg6)} 72. Rh8 {5.12/28 38} Bf6 {11} 73. Rh6 {5.12/27 13}
Kg7 {29} 74. Rh3 {5.12/27 0} Kf7 {7} 75. Rh7+ {5.12/27 15} Kf8 {69 (Kg6)} 76.
Kc2 {5.12/26 37} Ke8 {13 (Kg8)} 77. Kb3 {5.12/24 37} Kf8 {22} 78. Rb7 {5.12/24
0} Be7 {13 (Bh4)} 79. Ra7 {5.12/26 37} Ke8 {18 (Bh4)} 80. Rc7 {5.12/26 36} Kd8
{11} 81. Rb7 {5.12/24 12} Bh4 {14} 82. Rh7 {5.12/24 8} Bf2 {27} 83. Kc4 {5.12/
23 0} Kc8 {10 (Ke8)} 84. Rh6 {5.12/24 36} Kd7 {23} 85. Rf6 {5.12/25 0} Ke7 {12}
86. Rc6 {5.12/24 11} Kd7 {22 (Bd4)} 87. Rg6 {5.12/24 36} Be3 {12} 88. Ra6 {5.
12/24 10} Bf2 {24 (Bg1)} 89. Ra7+ {5.12/26 35} Kd6 {15} 90. Ra2 {5.12/26 7} Be3
{13 (Bd4)} 91. Rc2 {5.12/27 35} Ke6 {19 (Ke7)} 92. Rg2 {5.09/27 34} Ke7 {9
(Kf6)} 93. Rh2 {5.09/26 34} Ke6 {20 (Bd4)} 94. Rh4 {5.12/25 33} Bf2 {9 (Bg1)}
95. Rg4 {5.09/27 33} Be3 {15 (Kf6)} 96. Rg7 {5.09/28 32} Kd6 {14 (Kf6)} 97. Rf7
{5.09/27 31} Bf2 {17} 98. Rh7 {5.09/28 3} Be3 {9 (Ke6)} 99. Rh6+ {5.12/24 31}
Kd7 {9} 100. Rg6 {5.12/22 11} Bf2 {18 (Bd4)} 101. Kd5 {5.12/23 31} Bd4 {58}
102. Rc6 {5.12/22 0} c4 {78 (Bb2)} 103. Rxc4 {5.12/22 31} Bb2 {22} 104. Ra4 {
5.12/20 0} Ke7 {13} 105. Ra6 {5.12/20 6} Kd7 {6 (Bd4)} 106. Re6 {5.24/15 31}
Bc3 {12 (Bd4)} 107. Rxe5 {5.94/15 13} Ba5 {24 (Bb4)} 108. Rh5 {6.42/15 31} Kd8
{19 (Ke7)} 109. e5 {-#1931/15 20 Highendman resigns  (Lag: Av=0.95s, max=2.2s)}
1-0

Vas, you should take a closer look at this :-D
Parent - - By Roland Rösler (****) Date 2009-10-28 01:12
Ask Highendman about it. How does he feel to be a mouse while the cat has some fun with the mouse before killing the mouse?

My conclusion: This "feature" was programmed only for HEM. Pain for pain! :-)
Parent - - By Highendman (****) Date 2009-10-28 09:05
LOL.

I think my h4 idea was a nice desparate attempt for a draw by creating a fortress. I was surprised the Cluster exchanged queens and allowed me to get to the position I wanted.
I believe the fortress I had was solid, but I got careless with that bishop shifting back and forth and allowed that c5 break and everything collapsed.

Why it took the Cluster another 50 moves to win after that - I guess either a bug or a feature. The feature resulted in a huge headache last night - so huge I woke up with it this morning.

Overall both games last night were a lot of fun for me. On the first, not yet posted, I got a nice winning position but again couldn't convert - it just defended so well! Maybe with more time I'd had spotted I'm reaching an end-game I can't convert and would have found Harvey's winning move that gets to a better form of that end-game.

And the second game - the one posted here: I read somewhere KID is good vs. engines. It was my first and last KID. I felt like resigning out of the book hehe.
However I quite enjoyed later trying to defend it with that fortress, up to the c5 break and subsequent torture.

While I really don't like loosing - and have now drawn 4 games and lost 2 vs. the new Cluster - I'm having a great time and really getting a feel as too how deep and far the Cluster feels. It's really an honour getting crushed by it. It's much stronger than the version I've won against. Although - it's not invincible. I believe I should be able to get a win or two every blue moon :)

Looking at the last six games vs. the new cluster - one I was lost out of the opening, one I was crushed in an amazing style - still can't figure out what I did wrong, one was drawn out of the opening, and in 3 I reached arguably better / winning positions, two of which I found a winning line post game that I couldn't convert on the board.

A lot of fun for me and thanks again to Lukas for playing them.
Parent - By Jiri Dufek (**) Date 2009-10-28 10:35
Thx for your very interesting comments!

There are little mine:

In all black cluster 3 times were played risky sicilian h5 line. In the first game cluster didnt play best dxe4 and position was verydangerous for him. In second you repeated same line and this time cluster found right dxe4 - position was equal for me, however you played very too much and paradoxicaly pawn on f4 made much weakness on white king than black pawn h5 (with pawn on f2  you had probably slight advantage). Last game was headache for me, because i sent newer version of book and forgot disable this line for this time. I thought, that cluster played very well, but still it was carelessly and you got strong attack on the black king, which was probably winning (i didnt checked Bxf7).

With white there is another story. In the first game (catalan with dxc4+e6/c6) you was under positional pressure and with sacrifice of pawn you got enough counterplay for getting draw. Second game you played very well established theory and you got slightly better engame but missed g5 and game ended in a draw. KID is another story - i dont show this is as book win. KID is quite playable for black, however you must have something prepared. You got little worse position which was tendency be worse :-). Fortress trick was nice but careless move Bh5 throw away all previous work. Latest 50 moves was torture for all who played or observed this game
Parent - By George Tsavdaris (****) Date 2009-10-28 11:12

>Overall both games last night were a lot of fun for me. On the first, not yet posted, I got a nice winning position but again couldn't >convert - it just defended so well!


Not so well. It allowed a most probably winning position(with Bxf7) so the reason this game was ½-½ and not a win for you is most probably not because of Rybka cluster's amazing defense but because you didn't find the move.

>And the second game - the one posted here: I read somewhere KID is good vs. engines. It was my first and last KID. I felt like resigning out of the book hehe.


King's Indian defense is indeed good against engines but i would say against not so powerful engines. You can create your attack in Kingside and engines not to worry too much until it's too late, but not with cluster i guess. It just sees too much. :-)

But i guess your 13....a5 wasn't and the most appropriate. I always play 13...Ng6 that it just has a much quicker attack. OK you are in big troubles then in Queenside but i prefer that kind of play.
Parent - - By M ANSARI (*****) Date 2009-10-28 04:55
As I went through this came, I couldn't understand what you guys were talking about.  The cluster played incredibly well ... that was up to move 56.  I am not sure what happened after that or why the pawn was not taken.  Sure seems like some sort of bug.  I hope this doesn't get a shoulder shrug because cluster won, and should be looked at carefully.  Something obviously went wrong.
Parent - - By Uly (Gold) Date 2009-10-28 07:05
It's the 5.12 bug, Rybka sees that several moves are >5.12, gives them the same score and picks pseudo randomly one of them, or something. IIRC it was corrected at some point and it's back.
Parent - - By InspectorGadget (*****) Date 2009-10-28 07:45

> IIRC


I think we need some sort of page to keep all these acronyms. It took me quite a while to figure this out.
Parent - By Uly (Gold) Date 2009-10-28 07:59
I like IIRC, it gives the impression that I'm saying some piece of factual knowledge, when it's the opposite :)

Anyway, Kullberg posted analysis where the bug was absent, the IIRC is just there because I could be wrong or my memory could be failing me.
Parent - By George Tsavdaris (****) Date 2009-10-28 11:14

>It's the 5.12 bug, Rybka sees that several moves are >5.12, gives them the same score and picks pseudo randomly one of them, or something.


????
Really?? That is horrid!!
Up Topic Rybka Support & Discussion / Rybka Discussion / This is just... buggy

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill