Author: Albert Silver
Date: 11:57:20 12/10/05
In a sense, I am almost apologetic to the members for posting yet another Rybka
post, but right in the second game of the match between Rybka "SP" and Fruit
2.2, the difference appeared as clearly as ever.
Please don't get me wrong. I think Fruit 2.2 is nothing short of a *super*
engine, and to suggest otherwise would be ridiculous. One thing that has
repeatedly struck me by Rybka, other than what has been mentioned already, is
the large difference in evaluations of positions I see *all* the time. I don't
mean one engine thinks it is +0.40 and Rybka thinks it is actually +0.20, I mean
they simply disagree on who is better period, and sometimes by very large
differences. i.e. well over 1.00 difference.
This position occured in the second game of the Nunn2 match against Fruit 2.2.
The game was only 10 minutes per side, with a 2 second increment. Fruit was
White and, as should be fairly evident, has some problems.
[D]4r1k1/pp4p1/6p1/3q1pP1/2rPp2P/1QPnP3/PP4K1/1R2BR2 b - - 0 29
Rybka played the next sequence right out of a textbook: f4!-f3 followed by
a5-b5-a4-b4. It had no doubts whatsoever as to how it stood:
Analysis by Rybka 1.0 "SP":
29...f4
µ (-0.91) Depth: 3 00:00:00
29...f4
³ (-0.55) Depth: 4 00:00:00
29...f4 30.exf4
µ (-0.77) Depth: 5 00:00:00
29...f4 30.exf4 e3+ 31.Kh2
³ (-0.63) Depth: 6 00:00:00 4kN
29...f4 30.exf4 e3+ 31.Rf3 a5
µ (-0.92) Depth: 7 00:00:00 12kN
29...f4 30.exf4 b5 31.Kg3 Rf8
³ (-0.45) Depth: 8 00:00:00 45kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Rf3
³ (-0.43) Depth: 9 00:00:01 82kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Rf3 Qf5 33.a4
³ (-0.36) Depth: 10 00:00:02 150kN
29...f4 30.exf4 b5 31.Kg3 Qf5 32.a4 a6 33.Ra1
³ (-0.36) Depth: 11 00:00:04 345kN
29...f4 30.exf4 b5 31.Kg3 Qf5 32.a4 a6 33.Ra1 e3
³ (-0.38) Depth: 12 00:00:07 614kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Rf3 Rf8 33.Qc2 Qe4 34.b3
µ (-0.85) Depth: 13 00:00:18 1491kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Rf3 Rf8 33.a4 a6 34.axb5
³ (-0.64) Depth: 14 00:00:28 2281kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Rf3 Rf8 33.a4 a6 34.axb5
³ (-0.70) Depth: 15 00:00:49 4072kN
29...f4 30.exf4 b5 31.Kg3 e3 32.Qc2 Qxd4 33.cxd4 Rxc2 34.Rf3
µ (-0.76) Depth: 16 00:01:32 7682kN
In contrast, here are a few other evaluations:
Analysis by Deep Fritz 8:
29...b5!
³ (-0.53) Depth: 9/23 00:00:00 672kN
29...b5 30.Bg3 a5 31.Qc2
³ (-0.54) Depth: 9/23 00:00:00 782kN
29...Re6!
³ (-0.55) Depth: 9/23 00:00:00 795kN
29...Re6 30.Bg3 Rb6 31.Qc2 Rcc6 32.b3 Nb4 33.Qd2 Nd3
³ (-0.60) Depth: 9/23 00:00:00 823kN
29...Re6 30.Bg3 Rb6 31.Qc2 Rbc6 32.Be5 Kh7 33.h5 Nxe5
³ (-0.46) Depth: 10/26 00:00:01 1630kN
29...b5!
³ (-0.47) Depth: 10/28 00:00:01 1910kN
29...b5 30.Bg3 a5 31.Qd1 Kf7 32.h5 Rh8 33.Be5 Rcc8 34.h6 Nxe5 35.dxe5 Qxd1
³ (-0.54) Depth: 10/28 00:00:02 2146kN
29...b5 30.Bg3 a5 31.Qd1 Kf7 32.h5 Rh8 33.Be5 Qd8 34.hxg6+ Kxg6
³ (-0.56) Depth: 11/30 00:00:05 5975kN
29...b5 30.Bg3 a5 31.Qd1 Kf7 32.h5 Rh8 33.Be5 Qd8 34.hxg6+ Kxg6
³ (-0.56) Depth: 12/32 00:00:10 11629kN
29...b5 30.Bg3 a5 31.Qa3 b4 32.Qb3 Rec8 33.h5 a4 34.Qd1 Kh7 35.hxg6+ Kxg6 36.Rh1
bxc3
³ (-0.49) Depth: 13/36 00:00:21 24261kN
29...b5 30.Bg3 a5 31.Qd1 Kf7 32.h5 Rh8 33.Rh1 Rcc8 34.Qb3 Qxb3 35.axb3 Rxh5
36.Rxh5 gxh5
³ (-0.60) Depth: 14/36 00:01:29 106991kN
Deep Fritz sees black is better, but doesn't see 29...f4!
Analysis by Fruit 2.2.1:
29...f4 30.exf4 e3+ 31.Kg3 Re6 32.Rf3 Rb6 33.Qc2 Qe4
² (0.43) Depth: 8/23 00:00:00 177kN
29...f4 30.Bf2 f3+ 31.Kh2 b6 32.Bg3 Rd8 33.Qa3 Qd7 34.Rbd1
² (0.27) Depth: 9/25 00:00:01 486kN
29...f4 30.Bf2 Rec8 31.Kh2 Qd7 32.Qc2 f3 33.Bg3 b6 34.Rbd1
= (0.25) Depth: 10/25 00:00:01 1166kN
29...f4 30.Bf2 Rec8 31.Kh2 Qd7 32.Qc2 f3 33.a3 Qd6+ 34.Bg3 Qd7 35.Rbd1
² (0.27) Depth: 11/27 00:00:03 1641kN
29...f4 30.Bf2 Rec8 31.Kh2 Qd7 32.Qc2 f3 33.a3 Rd8 34.Bg3 Qf5 35.Rbd1
= (0.18) Depth: 12/32 00:00:05 3445kN
29...f4 30.Bf2 b5 31.Qc2 f3+ 32.Kh2 Rc6 33.a3 Qa2 34.Kh1 Rec8 35.Be1 Rd6 36.Bg3
= (0.04) Depth: 13/40 00:00:20 14024kN
29...f4 30.Bf2 b5 31.Qc2 f3+ 32.Kh2 Rc6 33.a3 Qa2 34.Kh1 Rec8 35.Be1 a5 36.Kh2
Rd8
= (0.04) Depth: 14/40 00:00:28 19798kN
29...f4 30.exf4 e3+ 31.Kg3 Rf8 32.Rf3 b5 33.a4 a6 34.Qc2 Qe4 35.axb5 axb5 36.Ra1
Rxf4 37.Rxf4 Qxf4+ 38.Kg2
= (0.25) Depth: 15/40 00:00:51 36782kN
29...f4 30.exf4 e3+ 31.Kg3 Rf8 32.Rf3 b5 33.a4 a6 34.Qc2 Qe4 35.axb5 axb5 36.b3
Rxf4 37.Rxf4 Qxf4+ 38.Kh3 Qf3+ 39.Bg3 Nf4+ 40.Kh2
= (0.10) Depth: 16/44 00:01:24 61179kN
Fruit sees 29...f4! but not that it Black stands better. Same for Shredder 8,
which went from very enthusiastic about white, to about equal.
Analysis by Shredder 8:
29...Re6 30.Bg3 Rb6 31.Qc2 Qa5 32.a3 Rbc6 33.Qb3 b6 34.Bd6
± (0.83) Depth: 11/22 00:00:01 603kN
29...f4 30.exf4 e3+ 31.Kg3 Rec8 32.Bd2 exd2 33.Qxb7 Qxb7
± (0.82) Depth: 11/22 00:00:01 631kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Qa3 Rxf4 33.b3 Rxf1 34.Qf8+ Kxf8
² (0.47) Depth: 11/26 00:00:02 706kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Rd1 Nxf4 34.Qb5 Ne2+ 35.Kh2 Qxf3
36.Qxc4+ Kh8
² (0.35) Depth: 11/30 00:00:02 826kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Rd1 Nxf4 34.Qb5 Ne2+ 35.Kh2 Qxf3
36.Qxc4+ Kh8
² (0.35) Depth: 12/31 00:00:03 1143kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Kh2 Nxf4 34.Qa3 a6 35.Qd6 Rc8
36.Rd1
² (0.50) Depth: 13/31 00:00:04 1894kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Kh2 Rxf4 34.Rxe3 Rf2+ 35.Kg1 Rf1+
36.Kg2 Rf2+ 37.Kg3 Nxe1 38.Rbxe1 Qf4+ 39.Kh3 Rh2#
= (0.25) Depth: 14/36 00:00:07 2916kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Ra1 Nxf4 34.Kh2 Qd5 35.Bg3 Nd3
36.Rd1 b5
= (0.24) Depth: 14/36 00:00:08 3511kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Ra1 Kh8 34.Qa3 Rxf4 35.Rxe3 Rfxd4
36.cxd4 Qf4+ 37.Kg2 Qxe3
² (0.28) Depth: 15/35 00:00:12 5651kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Qb5 Nxf4 34.Kh2 a6 35.Qb3 b5 36.Bg3
e2 37.Kg1 Qf5 38.Qa3
= (0.16) Depth: 16/37 00:00:25 11336kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Ra1 Qf7 33.Kh2 Rxf4 34.Rxf4 Nxf4 35.Qb5 Rc6
36.Qb3 Re6 37.Bg3 e2
= (-0.07) Depth: 17/42 00:00:48 22568kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Kh2 Rxf4 34.Rxf4 Nxf4 35.Qb5 Qe6
36.Kg3 Ne2+ 37.Kh2 Nf4 38.Kg3 Ne2+ 39.Kh2 Nf4 40.Kg3
= (0.00) Depth: 18/42 00:01:17 36774kN
29...f4 30.exf4 e3+ 31.Kg3 Re4 32.Rf3 Qf7 33.Kh2 Rxf4 34.Rxf4 Nxf4 35.Qb5 Qe6
36.Kg3 Ne2+ 37.Kh2 Nf4 38.Kg3 Ne2+ 39.Kh2 Nf4 40.Kg3
= (0.00) Depth: 19/44 00:02:04 58678kN
This position arises a couple of moves later:
[D]4r1k1/pp4p1/6p1/3q2P1/2rPp2P/1QPnPp2/PP3B2/1R3R1K b - - 0 31
I can really find no fault in Rybka's approach with a5-b5 and a pawn attack on
the queenside, yet I found no other engine that played it, except Hiarcs 8 (I
still can't find my H9 CD ... Grrr...), and even it takes quite some time, and
more than that to start really giving Black a significant edge. Some engines
sometimes propose b5, but the line doesn't show the rest, and they stop with a6
or some other shuffling of the pieces.
Analysis by Hiarcs 8:
31...Rec8 32.Bg3 Qf5 33.Kh2 b5 34.Rbd1 a5 35.Kg1
= (-0.14) Depth: 8/20 00:00:00 155kN
31...Rec8 32.Bg3 Qf5 33.Kh2 b5 34.Rbd1 a5 35.Qc2 Qg4
= (-0.19) Depth: 9/23 00:00:01 422kN
31...Rec8 32.Bg3 b5 33.Kg1 Qf5 34.Kh2 a5 35.a3 a4 36.Qc2
³ (-0.26) Depth: 10/28 00:00:06 1794kN
31...Rec8 32.Bg3 b5 33.Rbd1 a5 34.Rd2 Qf5 35.Kh2 a4 36.Qa3 Qg4
³ (-0.33) Depth: 11/32 00:00:12 3673kN
31...Rec8 32.Kh2 b5 33.Rbd1 a5 34.Rd2 Qf5 35.Bg3 Kh7
³ (-0.31) Depth: 12/34 00:00:52 14760kN
31...a5 32.Bg3 b5 33.Rbd1 Rec8 34.Rd2
³ (-0.35) Depth: 12/35 00:01:37 27899kN
31...a5 32.Bg3
³ (-0.60) Depth: 13/37 00:06:28 99018kN
And Rybka's analysis gives
Analysis by Rybka 1.0 "SP":
31...Re6 32.Bg3 Rb6 33.Qc2 Qf5 34.Kh2 Ra6
µ (-1.17) Depth: 10 00:00:01 144kN
31...b5 32.Bg3 Re7 33.Kh2 Rb7 34.Qc2 Qe6
µ (-1.18) Depth: 10 00:00:03 243kN
31...a5 32.Bg3 a4 33.Qc2 Qe6 34.Kh2 Rec8
µ (-1.31) Depth: 10 00:00:04 311kN
31...a5 32.Bg3 a4 33.Qa3 Rd8 34.Kh2 Rd7 35.Rbd1
µ (-1.25) Depth: 11 00:00:05 427kN
31...a5 32.Bg3 a4 33.Qa3 Qe6 34.Kh2 Rd8 35.Rbd1 Rd7
µ (-1.34) Depth: 12 00:00:07 675kN
31...a5 32.Bg3 b5 33.Kh2 Rec8 34.a3 b4 35.axb4 axb4 36.Ra1
µ (-1.37) Depth: 13 00:00:16 1502kN
31...a5 32.Bg3 b5 33.Kh2 Rec8 34.a3 Rxd4 35.Qxd5+ Rxd5 36.Rfd1
-+ (-1.48) Depth: 14 00:00:30 2938kN
31...a5 32.Qc2 b5 33.Be1 Rec8 34.a3 Qd7 35.Qh2 R4c6 36.Bd2
-+ (-1.41) Depth: 15 00:01:28 8624kN
31...b5 32.Bg3 Rec8 33.Kh2 a5 34.a3 Rxd4 35.Qxd5+ Rxd5 36.Bf2
-+ (-1.60) Depth: 15 00:02:16 13048kN
Finally, below is the entire game. I'll post the full match and result as soon
as it is done.
Albert
[Event "N2-RybkaSP-F22-10+2"]
[Site "Private"]
[Date "2005.12.10"]
[Round "2.1"]
[White "Fruit 2.2"]
[Black "Rybka 1.0 Beta 32-bit"]
[Result "0-1"]
[ECO "A07"]
[WhiteElo "2830"]
[BlackElo "2830"]
[PlyCount "86"]
[EventDate "2005.12.10"]
[EventType "tourn"]
[Source "Silver"]
1. g3 d5 2. Bg2 Nf6 3. Nf3 c6 4. O-O Bg4 5. h3 Bh5 6. d3 Nbd7 7. Nbd2 e5 8. e4
Bd6 {[%eval -6,12] [%emt 0:00:15]} 9. g4 {[%eval -2,14] [%emt 0:00:17]} Bg6 {
[%eval -7,12] [%emt 0:00:10]} 10. Nh4 {(exd5) [%eval -2,14] [%emt 0:00:25]} O-O
{(Qe7) [%eval -5,12] [%emt 0:00:22]} 11. Nxg6 {[%eval 4,13] [%emt 0:00:20]}
hxg6 {[%eval -4,13] [%emt 0:00:41]} 12. g5 {(Re1) [%eval 29,14] [%emt 0:00:19]}
Nh5 {(Nh7) [%eval -2,13] [%emt 0:00:12]} 13. exd5 {[%eval 18,15] [%emt 0:00:25]
} cxd5 {[%eval 0,14] [%emt 0:00:16]} 14. Bxd5 {[%eval 32,14] [%emt 0:00:18]}
Nf4 {[%eval 9,13] [%emt 0:00:20]} 15. Ne4 {[%eval 26,15] [%emt 0:00:31]} Nxd5 {
[%eval 8,12] [%emt 0:00:04]} 16. Nxd6 {[%eval 23,15] [%emt 0:00:19]} Nc5 {
[%eval 16,12] [%emt 0:00:12]} 17. Nc4 {[%eval 37,14] [%emt 0:00:13]} e4 {
[%eval -4,13] [%emt 0:00:25]} 18. d4 {(dxe4) [%eval 28,14] [%emt 0:00:16]} Ne6
{[%eval 16,11] [%emt 0:00:08]} 19. h4 {[%eval 19,13] [%emt 0:00:13]} f6 {
[%eval -18,11] [%emt 0:00:29]} 20. Qg4 {[%eval 17,13] [%emt 0:00:13]} Re8 {
[%eval -28,12] [%emt 0:00:15]} 21. c3 {[%eval 18,13] [%emt 0:00:12]} Rc8 {
[%eval -7,13] [%emt 0:00:21]} 22. Ne3 {[%eval 47,13] [%emt 0:00:14]} f5 {
[%eval -20,13] [%emt 0:00:16]} 23. Qd1 {[%eval 40,13] [%emt 0:00:17]} Nef4 {
(Rc6) [%eval -7,12] [%emt 0:00:23]} 24. Bd2 {
(Nxd5) [%eval 41,12] [%emt 0:00:11]} Nxe3 {(Qd7) [%eval -12,11] [%emt 0:00:10]}
25. fxe3 {(Bxe3) [%eval 48,12] [%emt 0:00:11]} Nd3 {
[%eval -30,12] [%emt 0:00:09]} 26. Rb1 {(Qb3+) [%eval 46,13] [%emt 0:00:20]}
Qd6 {(Qd7) [%eval -49,11] [%emt 0:00:11]} 27. Kg2 {[%eval 46,12] [%emt 0:00:15]
} Qd5 {(Kh7) [%eval -51,11] [%emt 0:00:14]} 28. Qb3 {
[%eval 25,12] [%emt 0:00:12]} Rc4 {[%eval -61,11] [%emt 0:00:05]} 29. Be1 {
(Kg3) [%eval 28,12] [%emt 0:00:09]} f4 {[%eval -85,12] [%emt 0:00:10]} 30. Bf2
{(exf4) [%eval 2,12] [%emt 0:00:14]} f3+ {(b5) [%eval -119,12] [%emt 0:00:12]}
31. Kh1 {(Kh2) [%eval 11,13] [%emt 0:00:17]} a5 {
(Qf7) [%eval -134,12] [%emt 0:00:14]} 32. Bg3 {[%eval 24,12] [%emt 0:00:08]} b5
{(Rec8) [%eval -128,12] [%emt 0:00:14]} 33. Rbd1 {
(Kg1) [%eval -17,12] [%emt 0:00:10]} a4 {(Rec8) [%eval -204,13] [%emt 0:00:09]}
34. Qc2 {[%eval -58,14] [%emt 0:00:17]} b4 {[%eval -214,12] [%emt 0:00:08]} 35.
Be1 {[%eval -63,13] [%emt 0:00:06]} Rc6 {(a3) [%eval -230,12] [%emt 0:00:11]}
36. Ra1 {[%eval -88,13] [%emt 0:00:08]} Qe6 {
(a3) [%eval -269,11] [%emt 0:00:12]} 37. Kh2 {[%eval -184,13] [%emt 0:00:31]}
Rec8 {[%eval -315,11] [%emt 0:00:06]} 38. Kg3 {[%eval -197,13] [%emt 0:00:13]}
Rb6 {(a3) [%eval -389,11] [%emt 0:00:10]} 39. Qxa4 {
(c4) [%eval -316,13] [%emt 0:00:52]} bxc3 {[%eval -422,11] [%emt 0:00:07]} 40.
b3 {[%eval -311,12] [%emt 0:00:08]} c2 {[%eval -549,11] [%emt 0:00:18]} 41. Bd2
{[%eval -792,15] [%emt 0:01:39]} Qd6+ {(Ne5) [%eval -716,11] [%emt 0:00:14]}
42. Kh3 {[%eval -709,13] [%emt 0:00:18]} Ne5 {[%eval -716,11] [%emt 0:00:11]}
43. Rg1 {[%eval -709,12] [%emt 0:00:04]} Qe6+ {
(f2) [%eval -1342,10] [%emt 0:00:14]} 0-1
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.