Secret Hiarcs 11.2 Feature?

[Anabolic Karpov]

Well it certainly seemed like it looking at this game played on the Playchess server. Facing mate in 5, what could Hiarcs pull out of the bag? Is this some Super Swindle or Mesmerise the Opponent feature? Seems like someone handed Hiarcs a Get out of Jail card but in reality it isn't surprising Rybka failed to beat Zappa with engine errors like this. It's also a good reason why Hiarcs doesn't resign. Imagine if this happened in a major tournament.
AK

Black to play. Rybka couldn't win this position ...

[d]8/5K2/6p1/6P1/1k6/8/pp6/8 b - - 0 89


[Event "Rated game, 3m + 0s"]
[Site "Engine Room"]
[Date "2007.09.23"]
[Round "?"]
[White "Anabolic Karpov, HIARCS 11.2 MP."]
[Black "EL-SHADDAI, Rybka 2.3.2a mp"]
[Result "1/2-1/2"]
[ECO "B80"]
[WhiteElo "2497"]
[BlackElo "2723"]
[Annotator "-0.34;0.13"]
[PlyCount "195"]
[EventDate "2007.10.15"]
[TimeControl "180"]

{Rybka 2.3.2a mp: 16.3 ply; 538kN/s Intel(R) Xeon(R) CPU X5355 @ 2.
66GHz, (8 threads)} 1. e4 {[%emt 0:00:00]} c5 {[%eval 35534,0] [%emt 0:00:00]}
2. Nf3 {[%emt 0:00:00]} d6 {[%eval 35534,0] [%emt 0:00:00]} 3. d4 {
[%emt 0:00:00]} cxd4 {[%eval 35534,0] [%emt 0:00:00]} 4. Nxd4 {[%emt 0:00:00]}
Nf6 {[%eval 35534,0] [%emt 0:00:00]} 5. Nc3 {[%emt 0:00:00]} a6 {
[%eval 35534,0] [%emt 0:00:00]} 6. Be3 {[%emt 0:00:00]} e6 {
[%eval 35534,0] [%emt 0:00:00]} 7. f3 {[%emt 0:00:00]} b5 {
[%eval 35534,0] [%emt 0:00:00]} 8. Qd2 {[%emt 0:00:00]} Nbd7 {
[%eval 35534,0] [%emt 0:00:00]} 9. O-O-O {[%emt 0:00:00]} Bb7 {
[%eval 35534,0] [%emt 0:00:00]} 10. g4 {[%emt 0:00:00]} Nb6 {
[%eval 35534,0] [%emt 0:00:00]} 11. Qf2 {[%emt 0:00:00]} Nfd7 {
[%eval 35534,0] [%emt 0:00:00]} 12. Bd3 {[%emt 0:00:00]} Rc8 {
[%eval 35534,0] [%emt 0:00:00]} 13. Nce2 {[%emt 0:00:00]} d5 {
[%eval 35534,0] [%emt 0:00:00]} 14. Kb1 {[%emt 0:00:00]} Bc5 {
[%eval 35534,0] [%emt 0:00:00]} 15. e5 {[%emt 0:00:00]} O-O {
[%eval 35534,0] [%emt 0:00:00]} 16. f4 {[%emt 0:00:00]} Nc4 {
[%eval 13,13] [%emt 0:00:05]} 17. h4 {[%emt 0:00:00]} Qb6 {
[%eval -10,12] [%emt 0:00:05]} 18. Qg3 {(g5) [%emt 0:00:00]} f6 {
[%eval -7,13] [%emt 0:00:04]} 19. Bxc4 {(exf6) [%eval -34,12] [%emt 0:00:11]}
dxc4 {[%eval -14,13] [%emt 0:00:01]} 20. Rhf1 {
(Rh2) [%eval -31,12] [%emt 0:00:05]} fxe5 {[%eval -46,15] [%emt 0:00:04]} 21.
fxe5 {[%eval -42,11] [%emt 0:00:01]} Rxf1 {[%eval -46,15] [%emt 0:00:03]} 22.
Rxf1 {[%eval -75,13] [%emt 0:00:00]} b4 {[%eval -49,15] [%emt 0:00:07]} 23. Rd1
{[%eval -93,12] [%emt 0:00:00]} Be4 {(b3) [%eval -51,15] [%emt 0:00:04]} 24.
Qf4 {(Ka1) [%eval -107,13] [%emt 0:00:32]} Bd5 {[%eval -71,14] [%emt 0:00:07]}
25. Bg1 {(Qh2) [%eval -130,12] [%emt 0:00:00]} c3 {
[%eval -89,12] [%emt 0:00:03]} 26. b3 {[%eval -105,12] [%emt 0:00:02]} Qa5 {
[%eval -90,15] [%emt 0:00:03]} 27. Be3 {[%eval -102,12] [%emt 0:00:00]} Rf8 {
(Re8) [%eval -79,14] [%emt 0:00:08]} 28. Qg3 {[%eval -106,12] [%emt 0:00:05]}
Be4 {(Bb6) [%eval -71,15] [%emt 0:00:11]} 29. Bf4 {
[%eval -109,11] [%emt 0:00:04]} Bb6 {[%eval -70,15] [%emt 0:00:05]} 30. g5 {
(h5) [%eval -96,11] [%emt 0:00:00]} Bc7 {[%eval -84,13] [%emt 0:00:04]} 31. Qg4
{(Qh3) [%eval -132,12] [%emt 0:00:03]} Nc5 {[%eval -100,12] [%emt 0:00:02]} 32.
Qg3 {[%eval -126,11] [%emt 0:00:00]} Rd8 {[%eval -99,13] [%emt 0:00:02]} 33.
Rf1 {[%eval -114,12] [%emt 0:00:04]} Bxe5 {[%eval -91,14] [%emt 0:00:01]} 34.
Bxe5 {[%eval -114,13] [%emt 0:00:01]} Rxd4 {[%eval -81,16] [%emt 0:00:01]} 35.
Nxd4 {[%eval -105,12] [%emt 0:00:03]} Qa3 {[%eval -83,19] [%emt 0:00:02]} 36.
Rf8+ {[%eval -99,14] [%emt 0:00:00]} Kxf8 {[%eval -83,21] [%emt 0:00:05]} 37.
Qf4+ {[%eval -110,15] [%emt 0:00:00]} Ke7 {(Kg8) [%eval -85,21] [%emt 0:00:03]}
38. Qc1 {[%eval -94,15] [%emt 0:00:04]} Qxc1+ {[%eval -85,21] [%emt 0:00:08]}
39. Kxc1 {[%eval -93,19] [%emt 0:00:00]} g6 {[%eval -96,21] [%emt 0:00:06]} 40.
a4 {(Ne2) [%eval -92,16] [%emt 0:00:00]} Nb7 {[%eval -106,18] [%emt 0:00:01]}
41. Kd1 {[%eval -107,16] [%emt 0:00:04]} Kd7 {[%eval -98,21] [%emt 0:00:00]}
42. Ke1 {(Ke2) [%eval -113,16] [%emt 0:00:03]} Nd8 {
[%eval -115,18] [%emt 0:00:01]} 43. Ke2 {[%eval -121,16] [%emt 0:00:01]} Nf7 {
(Bf5) [%eval -108,21] [%emt 0:00:02]} 44. Ke3 {[%eval -104,17] [%emt 0:00:03]}
Nxe5 {(Bxc2) [%eval -136,24] [%emt 0:00:00]} 45. Kxe4 {
[%eval -128,18] [%emt 0:00:02]} Kd6 {[%eval -111,25] [%emt 0:00:02]} 46. Ne2 {
[%eval -128,19] [%emt 0:00:00]} a5 {(Nc6) [%eval -142,24] [%emt 0:00:03]} 47.
Nd4 {[%eval -132,19] [%emt 0:00:03]} Nf7 {(Nc6) [%eval -190,27] [%emt 0:00:03]}
48. Nf3 {[%eval -132,18] [%emt 0:00:03]} e5 {[%eval -180,28] [%emt 0:00:00]}
49. Nh2 {[%eval -139,18] [%emt 0:00:02]} Ke6 {
(h5) [%eval -197,28] [%emt 0:00:02]} 50. Ng4 {[%eval -124,18] [%emt 0:00:02]}
Nd6+ {[%eval -205,27] [%emt 0:00:00]} 51. Ke3 {[%eval -132,18] [%emt 0:00:02]}
Nf5+ {[%eval -218,27] [%emt 0:00:02]} 52. Ke4 {[%eval -125,19] [%emt 0:00:00]}
Ng3+ {(Nd4) [%eval -221,27] [%emt 0:00:04]} 53. Kf3 {
[%eval -160,16] [%emt 0:00:02]} Nf1 {[%eval -233,27] [%emt 0:00:03]} 54. Nf6 {
(Ke2) [%eval -179,17] [%emt 0:00:01]} Nd2+ {
(Kf5) [%eval -267,20] [%emt 0:00:01]} 55. Ke3 {[%eval -203,18] [%emt 0:00:04]}
h5 {[%eval -364,25] [%emt 0:00:00]} 56. Ke2 {[%eval -260,19] [%emt 0:00:06]} e4
{[%eval -460,26] [%emt 0:00:01]} 57. Ke3 {[%eval -276,19] [%emt 0:00:00]} Nf3 {
(Ke5) [%eval -470,26] [%emt 0:00:02]} 58. Kxe4 {[%eval -274,18] [%emt 0:00:03]}
Nxh4 {(Ne1) [%eval -470,26] [%emt 0:00:00]} 59. Nd5 {
(Ke3) [%eval -381,18] [%emt 0:00:06]} Ng2 {[%eval -506,21] [%emt 0:00:07]} 60.
Nc7+ {[%eval -448,18] [%emt 0:00:05]} Kd7 {(Kd6) [%eval -514,21] [%emt 0:00:00]
} 61. Nb5 {[%eval -448,16] [%emt 0:00:02]} h4 {
(Ne1) [%eval -562,19] [%emt 0:00:01]} 62. Nd4 {[%eval -600,19] [%emt 0:00:14]}
Kd6 {[%eval -580,20] [%emt 0:00:00]} 63. Ne2 {[%eval -630,16] [%emt 0:00:02]}
Ne1 {[%eval -580,18] [%emt 0:00:02]} 64. Kf4 {
(Nf4) [%eval -657,17] [%emt 0:00:00]} Nxc2 {[%eval -462,16] [%emt 0:00:00]} 65.
Nc1 {(Kf3) [%eval -1100,18] [%emt 0:00:07]} Nd4 {[%eval -590,14] [%emt 0:00:01]
} 66. Kg4 {[%eval -1100,17] [%emt 0:00:00]} Nf5 {
(c2) [%eval -593,11] [%emt 0:00:00]} 67. Kf3 {[%eval -1125,16] [%emt 0:00:01]}
c2 {(h3) [%eval -684,14] [%emt 0:00:00]} 68. Ke2 {
(Kf2) [%eval -1185,17] [%emt 0:00:02]} h3 {[%eval -648,11] [%emt 0:00:01]} 69.
Kf1 {(Kf2) [%eval -1341,19] [%emt 0:00:02]} Nh4 {
(Nd4) [%eval -749,14] [%emt 0:00:01]} 70. Kg1 {
(Ke1) [%eval -1274,17] [%emt 0:00:00]} Nf3+ {[%eval -746,11] [%emt 0:00:00]}
71. Kf2 {[%eval -1435,20] [%emt 0:00:02]} h2 {
(Nd4) [%eval -816,18] [%emt 0:00:00]} 72. Kg2 {[%eval -1436,19] [%emt 0:00:01]}
Ke5 {(Kd5) [%eval -816,16] [%emt 0:00:03]} 73. Nd3+ {
[%eval -1514,17] [%emt 0:00:02]} Kd4 {[%eval -1511,13] [%emt 0:00:02]} 74. Nc1
{(Nxb4) [%eval -2578,17] [%emt 0:00:00]} Ke3 {
(Kc3) [%eval -1821,10] [%emt 0:00:00]} 75. Na2 {[%eval -2650,18] [%emt 0:00:02]
} Kd2 {[%eval -1952,12] [%emt 0:00:00]} 76. Kh1 {
[%eval -3340,11] [%emt 0:00:00]} c1=Q+ {(c1R+) [%eval -1962,10] [%emt 0:00:01]}
77. Nxc1 {[%eval -3346,12] [%emt 0:00:00]} Kxc1 {
[%eval -1962,11] [%emt 0:00:00]} 78. Kg2 {[%eval -3346,14] [%emt 0:00:00]} Kc2
{(Kb2) [%eval -1962,8] [%emt 0:00:00]} 79. Kh1 {
(Kxf3) [%eval -3353,13] [%emt 0:00:00]} Kxb3 {[%eval -2230,8] [%emt 0:00:00]}
80. Kg2 {[%eval -32759,28] [%emt 0:00:00]} Kxa4 {[%eval -2664,8] [%emt 0:00:00]
} 81. Kxf3 {(Kf2) [%eval -3415,11] [%emt 0:00:00]} h1=Q+ {
[%eval -32752,7] [%emt 0:00:00]} 82. Kf2 {(Ke2) [%eval -2650,6] [%emt 0:00:00]}
Qf3+ {(b3) [%eval -32754,4] [%emt 0:00:00]} 83. Kxf3 {
[%eval -32755,14] [%emt 0:00:01]} b3 {[%eval -32756,1] [%emt 0:00:00]} 84. Kf4
{[%eval -32756,62] [%emt 0:00:00]} Kb4 {(b2) [%eval -32757,1] [%emt 0:00:00]}
85. Ke5 {[%eval -32756,11] [%emt 0:00:00]} b2 {[%eval -32758,1] [%emt 0:00:00]}
86. Kf6 {[%eval -32757,62] [%emt 0:00:00]} a4 {
(b1Q) [%eval -32760,3] [%emt 0:00:00]} 87. Kf7 {
(Kxg6) [%eval -32757,10] [%emt 0:00:00]} a3 {
(b1Q) [%eval -32761,3] [%emt 0:00:00]} 88. Kg7 {
(Kxg6) [%eval -32757,9] [%emt 0:00:00]} a2 {
(b1Q) [%eval -32762,3] [%emt 0:00:00]} 89. Kf7 {
(Kxg6) [%eval -32760,8] [%emt 0:00:00]} Kc4 {[%eval -32762,3] [%emt 0:00:00]}
90. Kf8 {(Kxg6) [%eval -2520,7] [%emt 0:00:00]} Kb3 {
(b1Q) [%eval -32762,3] [%emt 0:00:00]} 91. Kg8 {
(Kf7) [%eval -3223,7] [%emt 0:00:00]} Kc4 {[%eval -32762,3] [%emt 0:00:00]} 92.
Kf8 {[%eval -32762,9] [%emt 0:00:00]} Kb3 {[%eval -32762,3] [%emt 0:00:00]} 93.
Ke7 {(Kf7) [%eval -32760,9] [%emt 0:00:00]} Ka3 {
(b1Q) [%eval -32762,3] [%emt 0:00:00]} 94. Kf7 {[%eval -2014,4] [%emt 0:00:00]}
Ka4 {(b1Q) [%eval -32762,3] [%emt 0:00:00]} 95. Ke6 {
(Kxg6) [%eval -1985,1] [%emt 0:00:00]} Ka3 {
(a1Q) [%eval -32762,3] [%emt 0:00:00]} 96. Kf7 {[%eval -1710,1] [%emt 0:00:00]}
Ka4 {[%eval -32762,3] [%emt 0:00:00]} 97. Ke6 {
(Kxg6) [%eval -2000,4] [%emt 0:00:00]} Ka3 {[%eval -32762,3] [%emt 0:00:00]}
98. Kf7 {(Lag: Av=0.52s, max=3.4s) [%eval 0,7] [%emt 0:00:00]} 1/2-1/2

[Mike S.]

This looks like the incomplete tablebases problem... Rybka "keeps" a #5 but doesn't promote.

On my system, Rybka finds 89...b1Q #5 immediatly (a1Q is equal). I have 4-piece tbs. and R-R 5 piece tbs., but they are not accessed up to depth 11, with default engine settings:

Analysis by Rybka 2.3.2a 32-bit :

89...b1Q 90.Ke7 Qf5 91.Kd8 a1Q 92.Kc7 Qg7+ 93.Kd8 Qfd7#
-+ (-#8) Depth: 5 00:00:00
-+ (-#5) Depth: 11 00:00:10 2023kN

If I change tbs. usage from 'rarely' to 'frequently', no problem either:

Analysis by Rybka 2.3.2a 32-bit :

89...b1Q 90.Ke7 Qf5 91.Kd8 a1Q 92.Kc7 Qg7+ 93.Kd8 Qfd7#
-+ (-#8) Depth: 5 00:00:00
-+ (-#5) Depth: 11 00:00:10 2068kN, tb=39

Could the problem have been caused if your opponent had KPPP-KP but not the KQPP-KP tablebase? (I don't know how Rybka handles incomplete 6-piece sets.)

[Anabolic Karpov]

Hi Mike,

This looks like the incomplete tablebases problem... Rybka "keeps" a #5 but doesn't promote.

You could be right here. Unfortunately unable to see what was going on on opponents machine or what TB's were used. I don't know if another route other than a CBase GUI can be used to play on server but if not then it is clear GUI had not taken over so seems to confirm opponent did not have full 6 piece TB set.

It seemed as if Rybka could not see 3 fold repetition. Wonder if it gets bypassed when accessing TB's? Maybe on 8 thread machine there is a bug in that area. Is it possible that Rybka flags to other cores that mate is found and stops calculating in these short mate positions so there is potential for them not to get played out?

I also thought Hiarcs' part was interesting that it didn't take Black's g6 pawn as it still gave same depth to mate.

I have part 6 piece and full 3-4-5 piece TB's but like you I could not reproduce this end result with the 64 bit Rybka MP engine when playing over the game. So it could be state of hash. We'll probably never know but thought it interesting that this could happen and probably gives strength to other posted argument about Zappa playing on in what seemed to be a drawn position. If etiquette is applied to computer chess games, we may never see these bugs arise.

regards,
AK