Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Perfect "God" Play ELO = 3800?
1 2 3 Previous Next  
Poll Perfect "God" Play ELO... (Closed)
< 3800. 5 8%
<= 3800. 0 0%
approx 3800. 5 8%
>= 3800. 4 7%
> 3800. 35 57%
No Idea!! 12 20%
Parent - - By Uly (Gold) Date 2010-09-19 06:54
Can you paste a PGN? I don't have Fritz installed...
Parent - - By mindbreaker (****) Date 2010-09-19 08:18
Actually, here are all the games...in PGN.
Attachment: TheDraw...Complete.pgn (461k)
Parent - - By Uly (Gold) Date 2010-09-19 09:39
Can you please paste a PGN of the drawn game? I'd just like to see it, I can produce some wins like those "on demand", it's the draw that would be interesting to see, and I wouldn't like to swim on PGN for that.
Parent - - By Mark (****) Date 2010-09-19 18:49
Here's the pgn.  (Just had to search for "1/2"...)

[Event "FB v R4L, 1'/60+1'/60+0'+1""]
[Site "CAVERN-3"]
[Date "2010.09.18"]
[Round "78"]
[White "Fritz 5.32"]
[Black "Rybka 4 x64 Exp 39 v1"]
[Result "1/2-1/2"]
[ECO "E15"]
[WhiteElo "2700"]
[BlackElo "3400"]
[Annotator "-0.06;0.03"]
[PlyCount "234"]
[TimeControl "60/60:60/60:0+1"]

{Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz 3465 MHz  W=11.9 plies; 5,
318kN/s; Perfect2010.ctg  B=17.1 plies; 275kN/s; 13,047 TBAs; Perfect2010.ctg}
1. d4 {[%eval 0,0] [%emt 0:00:00]} Nf6 {[%eval 0,0] [%emt 0:00:00]} 2. c4 {
[%eval 0,0] [%emt 0:00:00]} e6 {[%eval 0,0] [%emt 0:00:00]} 3. Nf3 {[%eval 0,0]
[%emt 0:00:00]} b6 {[%eval 0,0] [%emt 0:00:00]} 4. g3 {[%eval 0,0] [%emt 0:00:
00]} Ba6 {[%eval 0,0] [%emt 0:00:00]} 5. b3 {[%eval 0,0] [%emt 0:00:00]} Bb4+ {
[%eval 0,0] [%emt 0:00:00]} 6. Bd2 {[%eval 0,0] [%emt 0:00:00]} Be7 {[%eval 0,
0] [%emt 0:00:00]} 7. Bg2 {[%eval 0,0] [%emt 0:00:00]} c6 {[%eval 0,0] [%emt 0:
00:00]} 8. Bc3 {[%eval 0,0] [%emt 0:00:00]} d5 {[%eval 0,0] [%emt 0:00:00]
Both last book move} 9. Nbd2 {[%eval -6,11] [%emt 0:00:02]} O-O {[%eval 3,13]
[%emt 0:00:02]} 10. Bb2 {[%eval 6,10] [%emt 0:00:01] (0-0)} Nbd7 {[%eval -10,
13] [%emt 0:00:04]} 11. O-O {[%eval 13,10] [%emt 0:00:01]} Rc8 {[%eval -11,12]
[%emt 0:00:02]} 12. a3 {[%eval -9,10] [%emt 0:00:02] (Rc1)} c5 {[%eval -22,11]
[%emt 0:00:02]} 13. Rc1 {[%eval -6,10] [%emt 0:00:02]} Bb7 {[%eval -27,11]
[%emt 0:00:01] (dxc4)} 14. dxc5 {[%eval 6,9] [%emt 0:00:01]} bxc5 {[%eval -24,
12] [%emt 0:00:01]} 15. Nh4 {[%eval 13,9] [%emt 0:00:01] (Ne1)} g6 {[%eval -24,
13] [%emt 0:00:03] (Re8)} 16. Nhf3 {[%eval 19,10] [%emt 0:00:01] (Re1)} Qc7 {
[%eval -26,11] [%emt 0:00:02] (Rb8)} 17. cxd5 {[%eval 13,10] [%emt 0:00:02]
(e3)} exd5 {[%eval -22,13] [%emt 0:00:01]} 18. Qc2 {[%eval 3,11] [%emt 0:00:02]
(Ne1)} Rfe8 {[%eval -27,12] [%emt 0:00:02]} 19. Rfd1 {[%eval -3,10] [%emt 0:00:
02]} a5 {[%eval -26,11] [%emt 0:00:01] (Bf8)} 20. e3 {[%eval 9,9] [%emt 0:00:
01]} Qb6 {[%eval -25,10] [%emt 0:00:01] (Bd6)} 21. e4 {[%eval 3,9] [%emt 0:00:
01] (Bc3)} d4 {[%eval -26,12] [%emt 0:00:03] (dxe4)} 22. Nc4 {[%eval 9,10]
[%emt 0:00:02]} Qa6 {[%eval -18,11] [%emt 0:00:01] (Qc7)} 23. Nfd2 {[%eval -9,
9] [%emt 0:00:02]} Bf8 {[%eval -19,11] [%emt 0:00:01]} 24. Re1 {[%eval 0,9]
[%emt 0:00:01]} Bc6 {[%eval -18,9] [%emt 0:00:01]} 25. Bf1 {[%eval 0,9] [%emt
0:00:01] (a4)} Qa7 {[%eval -4,12] [%emt 0:00:02] (Qa8)} 26. e5 {[%eval 9,10]
[%emt 0:00:01]} Nd5 {[%eval -16,11] [%emt 0:00:01]} 27. Bh3 {[%eval 28,10]
[%emt 0:00:01] (Nd6)} a4 {[%eval -16,11] [%emt 0:00:01]} 28. Qd3 {[%eval 19,9]
[%emt 0:00:02] (e6)} axb3 {[%eval -2,10] [%emt 0:00:01] (Nc7)} 29. e6 {[%eval
19,8] [%emt 0:00:00]} fxe6 {[%eval -2,11] [%emt 0:00:01]} 30. Bxe6+ {[%eval 13,
9] [%emt 0:00:01]} Kh8 {[%eval 0,11] [%emt 0:00:00] (Kg7)} 31. Nxb3 {[%eval 28,
9] [%emt 0:00:01]} N5f6 {[%eval 0,10] [%emt 0:00:00]} 32. Nxd4 {[%eval 28,9]
[%emt 0:00:01]} cxd4 {[%eval 0,12] [%emt 0:00:00]} 33. Bxd4 {[%eval 44,9]
[%emt 0:00:01]} Bc5 {[%eval 0,11] [%emt 0:00:00] (Qa6)} 34. Bxd7 {[%eval 9,9]
[%emt 0:00:01]} Rxe1+ {[%eval 0,10] [%emt 0:00:00]} 35. Rxe1 {[%eval 13,11]
[%emt 0:00:00]} Bxd4 {[%eval 0,11] [%emt 0:00:00]} 36. Bxc8 {[%eval 31,10]
[%emt 0:00:01]} Bxf2+ {[%eval 0,11] [%emt 0:00:00]} 37. Kf1 {[%eval 19,4]
[%emt 0:00:00]} Bxe1 {[%eval -1,12] [%emt 0:00:00]} 38. Kxe1 {[%eval 38,10]
[%emt 0:00:01]} Qg1+ {[%eval -1,12] [%emt 0:00:01]} 39. Qf1 {[%eval 0,11]
[%emt 0:00:01]} Qd4 {[%eval -1,10] [%emt 0:00:00]} 40. Ba6 {[%eval 9,10] [%emt
0:00:02] (Qf4)} Kg7 {[%eval -38,11] [%emt 0:00:01]} 41. Qf2 {[%eval 6,11]
[%emt 0:00:02]} Qa1+ {[%eval -38,11] [%emt 0:00:02]} 42. Ke2 {[%eval 0,11]
[%emt 0:00:02]} Qa2+ {[%eval -38,12] [%emt 0:00:01]} 43. Nd2 {[%eval 3,11]
[%emt 0:00:02]} Qxa3 {[%eval -38,12] [%emt 0:00:01]} 44. Bc4 {[%eval -6,10]
[%emt 0:00:01]} Qd6 {[%eval -35,11] [%emt 0:00:01]} 45. Qa7+ {[%eval -6,9]
[%emt 0:00:01] (Nf3)} Bd7 {[%eval -25,13] [%emt 0:00:01] (Kh8)} 46. Ke1 {
[%eval -3,10] [%emt 0:00:01] (Nf3)} Qe5+ {[%eval -66,11] [%emt 0:00:01] (Ng4)}
47. Be2 {[%eval -3,10] [%emt 0:00:01] (Kf1)} Qc3 {[%eval -65,12] [%emt 0:00:01]
(Kg8)} 48. Bf3 {[%eval 6,9] [%emt 0:00:01] (Bg4)} Kh6 {[%eval -65,12] [%emt 0:
00:01]} 49. Ke2 {[%eval 0,10] [%emt 0:00:01] (Be2)} Bb5+ {[%eval -80,11] [%emt
0:00:00] (g5)} 50. Kd1 {[%eval -9,11] [%emt 0:00:01]} Nd7 {[%eval -72,13]
[%emt 0:00:01] (Bd7)} 51. Be4 {[%eval 0,9] [%emt 0:00:01] (Be2)} Ne5 {[%eval
-82,12] [%emt 0:00:01] (Nc5)} 52. Qb8 {[%eval 0,9] [%emt 0:00:01] (Qe7)} Qc5 {
[%eval -130,12] [%emt 0:00:02]} 53. Qd8 {[%eval -66,9] [%emt 0:00:02]} Kg7 {
[%eval -125,12] [%emt 0:00:02]} 54. Qd5 {[%eval -59,10] [%emt 0:00:01]} Qxd5 {
[%eval -131,13] [%emt 0:00:00]} 55. Bxd5 {[%eval -66,11] [%emt 0:00:00]} Ng4 {
[%eval -132,14] [%emt 0:00:00]} 56. Ke1 {[%eval -97,11] [%emt 0:00:01] (Bb7)}
Nxh2 {[%eval -129,15] [%emt 0:00:00]} 57. Ne4 {[%eval -122,11] [%emt 0:00:01]}
Bd3 {[%eval -130,15] [%emt 0:00:01] (h5)} 58. Nc5 {[%eval -134,11] [%emt 0:00:
01] (Bb7)} Bf5 {[%eval -134,18] [%emt 0:00:01]} 59. Nb7 {[%eval -144,11] [%emt
0:00:01] (Ne4)} Ng4 {[%eval -145,15] [%emt 0:00:01] (h5)} 60. Nd6 {[%eval -128,
10] [%emt 0:00:00] (Ke2)} Bd7 {[%eval -146,17] [%emt 0:00:01] (Kf6)} 61. Kd2 {
[%eval -144,11] [%emt 0:00:01] (Ke2)} Kf6 {[%eval -145,19] [%emt 0:00:02]} 62.
Bg2 {[%eval -150,10] [%emt 0:00:01] (Bb7)} h6 {[%eval -151,18] [%emt 0:00:02]}
63. Bf3 {[%eval -150,11] [%emt 0:00:01] (Kd3)} Ke6 {[%eval -151,19] [%emt 0:00:
02] (h5)} 64. Nb7 {[%eval -137,11] [%emt 0:00:01]} Ke5 {[%eval -152,19] [%emt
0:00:02]} 65. Nc5 {[%eval -131,11] [%emt 0:00:01]} Bf5 {[%eval -151,18] [%emt
0:00:01]} 66. Nb3 {[%eval -134,11] [%emt 0:00:01] (Ke2)} Nf6 {[%eval -161,19]
[%emt 0:00:02] (Nh2)} 67. Ke3 {[%eval -125,11] [%emt 0:00:01]} Be4 {[%eval
-160,19] [%emt 0:00:01]} 68. Bxe4 {[%eval -131,11] [%emt 0:00:01]} Nxe4 {
[%eval -156,19] [%emt 0:00:00]} 69. g4 {[%eval -116,14] [%emt 0:00:01]} Nd6 {
[%eval -147,21] [%emt 0:00:01] (Nf6)} 70. Nd2 {[%eval -116,12] [%emt 0:00:01]
(Kf3)} g5 {[%eval -146,22] [%emt 0:00:02] (Ke6)} 71. Kf3 {[%eval -122,15]
[%emt 0:00:01] (Nb1)} Kd4 {[%eval -146,23] [%emt 0:00:02]} 72. Nb3+ {[%eval
-119,14] [%emt 0:00:01]} Kd5 {[%eval -145,24] [%emt 0:00:01]} 73. Nd2 {[%eval
-119,14] [%emt 0:00:01] (Nc1)} Nf7 {[%eval -145,23] [%emt 0:00:01]} 74. Nb1 {
[%eval -128,15] [%emt 0:00:01]} Kd4 {[%eval -145,23] [%emt 0:00:01] (Ne5+)} 75.
Na3 {[%eval -116,15] [%emt 0:00:01]} Ne5+ {[%eval -145,23] [%emt 0:00:01]} 76.
Kg3 {[%eval -112,15] [%emt 0:00:01]} Nc6 {[%eval -130,23] [%emt 0:00:03] (Kc5)}
77. Nb5+ {[%eval -100,14] [%emt 0:00:01]} Kd5 {[%eval -125,22] [%emt 0:00:01]
(Ke5)} 78. Nc7+ {[%eval -109,13] [%emt 0:00:01] (Nc3+)} Ke5 {[%eval -125,24]
[%emt 0:00:02]} 79. Nb5 {[%eval -109,14] [%emt 0:00:01]} Nd4 {[%eval -125,23]
[%emt 0:00:02]} 80. Nc3 {[%eval -100,15] [%emt 0:00:01]} Kd6 {[%eval -125,23]
[%emt 0:00:01] (Nb3)} 81. Ne4+ {[%eval -109,14] [%emt 0:00:01]} Ke6 {[%eval
-125,24] [%emt 0:00:01]} 82. Nc5+ {[%eval -109,13] [%emt 0:00:01] (Nc3)} Ke7 {
[%eval -125,23] [%emt 0:00:01]} 83. Ne4 {[%eval -125,14] [%emt 0:00:01] (Nd3)}
Nc6 {[%eval -125,24] [%emt 0:00:01]} 84. Kg2 {[%eval -131,14] [%emt 0:00:01]
(Kf3)} Ne5 {[%eval -125,24] [%emt 0:00:02]} 85. Kh3 {[%eval -112,14] [%emt 0:
00:01] (Kg3)} Nc4 {[%eval -125,23] [%emt 0:00:02] (Ke6)} 86. Kg3 {[%eval -112,
14] [%emt 0:00:01]} Ke6 {[%eval -125,24] [%emt 0:00:01] (Kf7)} 87. Kf3 {[%eval
-109,14] [%emt 0:00:01] (Nc5+)} Ne5+ {[%eval -125,22] [%emt 0:00:01] (Kf7)} 88.
Kg3 {[%eval -119,15] [%emt 0:00:01]} Nd3 {[%eval -111,21] [%emt 0:00:01] (Nd7)}
89. Kg2 {[%eval -103,13] [%emt 0:00:01] (Nc3)} Ke5 {[%eval -110,21] [%emt 0:00:
01]} 90. Kf3 {[%eval -81,14] [%emt 0:00:01]} Ne1+ {[%eval -110,22] [%emt 0:00:
00]} 91. Ke3 {[%eval -112,15] [%emt 0:00:01]} Nc2+ {[%eval -110,22] [%emt 0:00:
01]} 92. Kf3 {[%eval -112,15] [%emt 0:00:01]} Nd4+ {[%eval -110,23] [%emt 0:00:
01] (Nb4)} 93. Ke3 {[%eval -100,15] [%emt 0:00:01]} Nc6 {[%eval -106,23] [%emt
0:00:01]} 94. Kf3 {[%eval -119,14] [%emt 0:00:01]} Ke6 {[%eval -106,22] [%emt
0:00:00] (Na5)} 95. Kf2 {[%eval -119,13] [%emt 0:00:01] (Nc5+)} Ne7 {[%eval
-106,22] [%emt 0:00:01] (Nd4)} 96. Ng3 {[%eval -109,14] [%emt 0:00:01] (Kf3)}
Kd5 {[%eval -106,22] [%emt 0:00:01] (Ke5)} 97. Ke3 {[%eval -91,15] [%emt 0:00:
01] (Kf3)} Ke5 {[%eval -105,22] [%emt 0:00:00]} 98. Kf3 {[%eval -100,15] [%emt
0:00:01] (Ne2)} Kd4 {[%eval -70,20] [%emt 0:00:01] (Ng6)} 99. Kf2 {[%eval -69,
16] [%emt 0:00:01]} Kc5 {[%eval -70,20] [%emt 0:00:00] (Ng8)} 100. Kf3 {[%eval
-62,14] [%emt 0:00:01]} Kd6 {[%eval -70,21] [%emt 0:00:00]} 101. Ke4 {[%eval
-75,15] [%emt 0:00:01]} Ke6 {[%eval -70,22] [%emt 0:00:00]} 102. Nh5 {[%eval
-84,15] [%emt 0:00:01]} Kf7 {[%eval -70,22] [%emt 0:00:01] (Ng8)} 103. Ke5 {
[%eval -84,14] [%emt 0:00:01]} Nc6+ {[%eval -66,22] [%emt 0:00:00]} 104. Kd5 {
[%eval -84,15] [%emt 0:00:01] (Kf5)} Nb8 {[%eval -66,23] [%emt 0:00:01]} 105.
Kd6 {[%eval -87,16] [%emt 0:00:01]} Na6 {[%eval -66,23] [%emt 0:00:00]} 106.
Ng3 {[%eval -91,15] [%emt 0:00:01] (Ke5)} Kf6 {[%eval -61,24] [%emt 0:00:01]}
107. Ne4+ {[%eval -91,14] [%emt 0:00:01]} Kg6 {[%eval -61,23] [%emt 0:00:01]}
108. Ke5 {[%eval -103,15] [%emt 0:00:01] (Ng3)} Nb4 {[%eval -5,20] [%emt 0:00:
02]} 109. Kd4 {[%eval -103,13] [%emt 0:00:01] (Nf6)} Nc2+ {[%eval -5,18] [%emt
0:00:01]} 110. Kd3 {[%eval -91,15] [%emt 0:00:01]} Ne1+ {[%eval 0,19] [%emt 0:
00:01]} 111. Ke3 {[%eval -84,14] [%emt 0:00:01]} Ng2+ {[%eval 0,20] [%emt 0:00:
00]} 112. Kd4 {[%eval -91,14] [%emt 0:00:01]} Nf4 {[%eval 0,21] [%emt 0:00:01]}
113. Ke5 {[%eval -72,13] [%emt 0:00:01] (Nd6)} Nd3+ {[%eval 0,23] [%emt 0:00:
01] (Ne2)} 114. Kd4 {[%eval -78,14] [%emt 0:00:01] (Ke6)} Nb2 {[%eval 0,24]
[%emt 0:00:01] (Nc1)} 115. Kc3 {[%eval -81,14] [%emt 0:00:01] (Nc3)} h5 {
[%eval 0,25] [%emt 0:00:01] (Na4+)} 116. gxh5+ {[%eval 0,16] [%emt 0:00:01]}
Kxh5 {[%eval 0,25] [%emt 0:00:00]} 117. Nxg5 {[%eval 0,15] [%emt 0:00:01]} Kxg5
{[%eval 0,59] [%emt 0:00:00] Draw accepted} 1/2-1/2
Parent - - By Uly (Gold) Date 2010-09-19 21:23
Thank you. Ahh, I blame the opening :)

It would be interesting to repeat the experiment with Rybka 3 Dynamic at Contempt 30, perhaps it has more luck at winning 100 games without a single draw (if it doesn't play that Queen's Indian :lol:).
Parent - By Mark (****) Date 2010-09-19 21:34
Too bad Fritz can't play FRC.  Then we wouldn't have to worry about openings.
Parent - - By Gaмßito (****) Date 2010-09-20 10:47

> Actually, here are all the games...in PGN.


Thanks for the PGN. I still remember with pleasantness those times when the great tactician Fritz 5.32 was the king. It was my favorite engine, an amazing strong engine in blitz games. I also remember when it beat Judith Polgar in a match of six games in 1999. Time control was not so fast, both had 25 minutes.

Regards,
Gaмßito.
Parent - - By mindbreaker (****) Date 2010-09-20 23:56
Do you remember was it Fritz 2 that drew Kasparov in a blitz game way back?
Parent - - By TheHug (Bronze) Date 2010-09-21 00:00
it was fritz 3 if i remember correctly
Parent - By mindbreaker (****) Date 2010-09-21 00:09
There is a version history but does not appear to be complete: http://en.wikipedia.org/wiki/Fritz_%28chess%29
Parent - By mindbreaker (****) Date 2010-09-21 00:02
Looks like there was a blitz match in 1992...that was probably what I remembered: http://www.chessgames.com/perl/ezsearch.pl?search=kasparov-fritz

Not sure which version was used in 1992.
Parent - - By mindbreaker (****) Date 2010-09-19 03:38
I still think Sonas' line (the first graph) is closer to reality than the Elo tables: http://www.chessbase.com/newsdetail.asp?newsid=562  Of course there must be some asymptote.

The performance final after 100 games (99.5-.5) is +920 Elo according to the Elo tables in Fritz.
Parent - By mindbreaker (****) Date 2010-09-19 03:43
So I guess he should have said: There will be draw :lol:
Parent - By Quapsel (****) Date 2010-09-19 16:19 Edited 2010-09-19 16:24
1core, 32bit Fritz5.32 to 1core, 32bit R4 means an ELO Difference of perhaps 550
http://www.3dkingdoms.com/chess/elo.htm
tells us, we can expect 96%:4% as a result of a match-result between those opponents

1core, 32bit Fritz5.32 to 6core,64bit R4 means an ELO difference of perhaps 700
We should expect 98,2%:1,8% between those opponents.

No, we have a good chance to see one win or one draw of poor Fritz5.32 within 100 games,
even if R4 runs on 64bit with 6 cores.

Weak engines are not damned to make bad moves, they only do not see, that other moves are stronger. They can prefer the good move even if the don't find the correct reason for it's correctness.
(And as far as I know Rybka4 has no logic to prefer moves that lay a snare! But, by the way, wouldn't this be a fine feature?)

Quap
Parent - By mindbreaker (****) Date 2010-09-19 00:52
Protector 1.3.5 is considerably stronger and I had to stop that run because it could not get anything.
Parent - By Uri Blass (*****) Date 2010-09-19 20:13
No

It does not mean it and elo is meaningless because it is clearly dependent on the set of games that you choose so I do not participate in this poll.

It is possible that at some level you simply do not lose with white so A can score 75% against B(by winning with white and drawing with black)
and B can score 75% against C(by winning with white and drawing with black) when A can score only 75% against C.

Rating is simply meaningless because if You play only A against B and B against C you get higher rating difference between A and C
relative to the case that you play only A against C.

Uri
Parent - By turbojuice1122 (Gold) Date 2010-09-19 00:41
I think that this is already disproven by the noticeable increase in draw percentage between computer programs, even much stronger ones, as the time control increases from blitz to tournament.  I haven't checked out current stuff on this, i.e. my estimate of 3800 was 2-3 years ago, but my feeling is that it's probably not a huge difference.  My feeling is also that if there is a difference, the Elo estimate would be higher, but not by more than a couple of hundred points.  We're talking tournament time controls, of course--obviously God's Elo would increase very substantially as you decrease the length of the game.
- - By Uly (Gold) Date 2010-09-18 19:53
Please note this is a recurring discussion, we had this version three months ago:

http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=264101

Assuming God just refers to the strongest opponent imaginable (that uses the best strategy and metastrategy against specific opponents conceivable) the Rybka Cluster would seem as a 1 core engine at 400Mhz, or something ridiculous, fighting against the cluster. People have problems conceiving this strength, but we keep rapidly advancing, and getting stronger all the time, and I haven't seem any limit yet, we're still very far from perfection.

Rybka keeps grossly mis-evaluating positions, and God would know it and would aim at them, with no reason for not succeeding every time, so at a CCRL scale (where a simple Rybka 4 at a Quad with a generic book at 40/40 TC already is at 3269 elo) I can believe God hitting 6000 elo.
Parent - - By stephanie (**) Date 2010-09-19 11:09 Edited 2010-09-19 11:11

>I can believe God hitting 6000 elo.


Cool, this means God can have some knight handicap vs me plus rybka on the fastest computer and God will still beat me most of the time :D

And probably beat or equalize in a queen odds vs a 2000 rated player if God is really 6000 elo. :P
Parent - - By turbojuice1122 (Gold) Date 2010-09-19 12:59
Yes, this is an obvious defeater to any argument about God having an Elo higher than about 4000...either that, or a proof that the Elo rating system breaks down in this region, making the discussion moot.
Parent - - By Mark (****) Date 2010-09-19 18:27
I don't think this defeats the argument about God having an Elo higher than about 4000 at all.  You would have to adjust how much a knight handicap is worth with a perfect player, though.  At levels near perfect, maybe a knight handicap is worth thousands of Elo??
Parent - - By turbojuice1122 (Gold) Date 2010-09-19 20:19
If that's the case, then things have blown up due to the winning nature of the position.  Similarly, the drawish nature of chess would cause huge potential strength differences (e.g. hardware differences among computers, or time differences among humans) to evaporate.
Parent - - By Mark (****) Date 2010-09-19 21:30
These are interesting discussions.  I wonder what the threads regarding God's elo will read like in 5 or 10 years?  My guess is 'the same as the 2010 discussions.'
Parent - By Uly (Gold) Date 2010-09-19 21:50

> My guess is 'the same as the 2010 discussions.'


No, I see some progress on these discussions, back on 2007 were were discussing how humans would have to use quantum charged dice to choose their moves to draw against God in some alternate reality.

Discussion no longer includes magic moves or theories about perfect opponents managing to win with 100% certainty against imperfect ones, we'll eventually refute some ideas that are flying around and discuss something else about the elo of a perfect entity.
Parent - By Uly (Gold) Date 2010-09-19 21:41

> either that, or a proof that the Elo rating system breaks down in this region


I think elo is already broken at the top, one just needs to see how Rybka's elo would be measured if you only take her results of the near engines and how her elo would look like if you only take into account her results against engines far apart. If it's not the same then ELO is flawed.

Anyway, I can still imagine some Stockfish 4 engine playing at 4000 elo and remaining so far away from perfection than we keep pushing perfect chess elo's up.
Parent - By mindbreaker (****) Date 2010-09-20 02:49
But as God knows the weaknesses we have and any engine and computer we may have, he can intentionally make less than perfect moves that we will trip over to guarantee a win. ;)
Parent - - By BankShots (***) Date 2010-09-20 18:03
Perfect "God" Play with respect to Odds = very interesting point! :lol:
Parent - By BankShots (***) Date 2010-09-20 22:55
:lol: i.e.:  What are God's Odds?!! :grin::smile::neutral::roll:
- - By sarciness (***) Date 2010-09-18 21:35
God would beat Rybka Cluster EVERY game.
Parent - - By ThudanBlunder (**) Date 2010-09-18 22:32 Edited 2010-09-18 22:34
Bet you also think God would win EVERY game of noughts-and-crosses (aka tic-tac-toe). :razz:
:
Parent - - By sarciness (***) Date 2010-09-18 22:46
Yeah, at tiddlywinks too!
Parent - - By Uly (Gold) Date 2010-09-18 23:45
Tic-tac-toe is different, as a perfect strategy is known to exist and it's trivial to follow. God can't beat me a tic-tac-toe, just like he can't beat me from this position:

3rk3/3p4/8/8/8/8/3P4/3RK3 w - - 0 1
Parent - - By Felix Kling (Gold) Date 2010-09-19 18:50 Edited 2010-09-19 18:59
what is your drawing idea in this position? depending on your answer I would actually give it a try to win it against you :D
Or do you mean correspondence chess with tablebase support?
Parent - By Uly (Gold) Date 2010-09-19 21:45

> Or do you mean correspondence chess with tablebase support?


Of course! Otherwise God may beat me from this position:

8/8/4K3/3R4/8/4kr2/8/8 w - -


By somehow attacking my rook with both his rook and king, or at least, I recall losing an OTB game like that once.

Anyway, the point is that since tablebases are accessible to anyone, God can't expect to beat anyone once the position reaches 6men and it's drawn.
Parent - - By mindbreaker (****) Date 2010-09-21 19:20
Ahh, but you are forgetting the lightning defense....KERZAPPE...Vytron looses on time ;)
Parent - - By Banned for Life (Gold) Date 2010-09-21 19:43
Not to mention, the pillar of salt defense!
Parent - - By Uly (Gold) Date 2010-09-22 00:55
Both illegal chess moves, I'm assuming God doesn't cheat.
Parent - - By mindbreaker (****) Date 2010-09-22 04:00
If things were set in motion before the game began, would it not just be preparation ;)  Might also be hard to find a TD who will side with you.  "You say, your opponent called down lighting on you?  Are you on crack?"  And there is the magical change to paragraph 3 of section II in the "laws of chess" which somehow only you did not notice ;)  That line permits a player, provided he has a cross on his king to request and receive divine assistance during a game ;)  Can't be cheating if it is in the rule book...heh...heh...heh.
Parent - - By Uly (Gold) Date 2010-09-22 08:19
What if I pray enough for him to let me win, would that allow me to beat God?
Parent - By mindbreaker (****) Date 2010-09-22 18:13
I think he would decide you needed some humility ;)
- - By jamerolle (**) Date 2010-09-19 00:22
Gods elo? to him the game is like tic tac tow every posibility has reached a conclusion.  However he did lay down his King (Christ)if you know what i mean.
Parent - By Uly (Gold) Date 2010-09-19 00:58

> to him the game is like tic tac tow every posibility has reached a conclusion.


Yes, but he has to develop a strategy that would beat someone that doesn't know the optimal strategy of tic tac toe, that's not trivial. Saying God would draw all his games is unsatisfactory as that would make him a mediocre player.
- - By Christoph Fieberg (*) Date 2010-09-19 12:39
It is easy to demonstrate that an increased search depth leads to increased ELO.

Ryka3 beats Rybka4 in a match 6:0 if search depth of Rybka3 is 7ply and that of Rybka4 only 5ply.

You can assume that a doubling in search depth means at least an increase of 600 ELO.

This would mean
Rybka at 15 ply ~ 3200 Elo
Rybka at 30 ply ~ 3800 Elo
Rybka at 60 ply ~ 4400 Elo
Rybka at 120 ply ~ 4800 Elo
Rybka at 240 ply ~ 5600 Elo

240 ply is not as much as it seems.

Remind that Marc Bourzutschky and Yakov Konoval found a position in the 7 men-endgame where it takes White 517 moves to win with best play by both sides.

I think we just scratched the surface of the deepness of chess...
Parent - By Not the real Kasparov (***) Date 2010-09-19 12:48
What about God playing itself? Would perfect play versus perfect play be a draw?

Is perfect play an objective thing that can be reached? Or will it always be subject to the person/entity giving their opinion on what perfect is?

At what point would you know that you have played perfect moves?
Parent - - By turbojuice1122 (Gold) Date 2010-09-19 13:03
This assumes that there is no draw barrier in chess.  To wit, this assumes the truth of the thing that you end up demonstrating.  More importantly, this assumption is disproven by the increasing number of draws at higher time controls.

I'm sure that from the starting position, a 60 ply engine versus a 120 ply engine will draw almost every game.  Even a 30 ply Rybka will probably draw just about every game if it plays relatively drawish openings, which is what tends to occur anyway when playing from the start position without book.
Parent - - By Christoph Fieberg (*) Date 2010-09-19 13:42
It was just a simple example. In fact I personally assume that there is a draw barrier in chess and that the increases in Elo are declining.

Maybe it will look as such (ply increases by factor of ~1.6 each step)
15 ply   ~3200 Elo
24 ply   ~3800 Elo (+600)
38 ply   ~4300 Elo (+500)
60 ply   ~4700 Elo (+400)
100 ply ~5050 Elo (+350)
160 ply ~5350 Elo (+300)
256 ply ~5600 Elo (+250)

We will be able to have a better evaluation if the speed of the computers will allow to reach higher plys.

I suggest to test Rykba 15 ply vs Rykba 24 ply.
Parent - - By Felix Kling (Gold) Date 2010-09-19 13:48
did you see my private messages? (you can find them in "Nachrichten" on the top.)
Parent - By Christoph Fieberg (*) Date 2010-09-19 14:21
Noted. Thank you.
Parent - - By turbojuice1122 (Gold) Date 2010-09-19 14:57
I don't think that ply is so meaningful as time of thinking, especially since different programs have different meanings with their ply values, and this might change depending on the phase of the game.  In any case, the real solution would be far lower in Elo than in your examples, as obviously any master would win if given queen odds against any entity at tournament time controls (or even faster).  This either (a) puts an upper limit of near 4000 on the Elo scale at tournament time controls or (b) shows that the Elo formula is meaningless above some certain point before that.
Parent - - By Christoph Fieberg (*) Date 2010-09-19 17:19
Let an engine (any engine!) calculate until it reaches ply (x).
Then let the same (!) engine calculate until it reaches ply (x*2) and compare.

The point is that ply (x*2) is stronger than ply (x) by a certain factor.
To analyse these differences and factors is meaningful.

Queen odds games have nothing to do with the above. This is not a criteria for normal playing strength. And they are not relevant for the Elo formula.
Of course an engine with Elo 5600 with queen odds could be not able to win a game against Elo 2000.
Nevertheless such an engine would score >95% against an Elo 5000 engine.
Up Topic Rybka Support & Discussion / Rybka Discussion / Perfect "God" Play ELO = 3800?
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill