Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Rybka Discussion / Rybka vs Benjamin II - information center
1 2 3 Previous Next  
Parent - - By Victor Zakharov (*****) [ru] Date 2008-01-05 00:24
No progress in the game

[White "Benjamin"]
[Black "Rybka"]
[WhiteElo "2575"]
[BlackElo "3100"]
[Result "*"]
[GameID "240"]
[UniqID "673095"]
[WhiteClock "0:16:08"]
[BlackClock "0:33:36"]
[Stamp "5643"]
[LastMoves "51...Rb7 52.Kg2 -0.34"]

1.e4 c5 2.Nf3 e6 3.c3 Nf6 4.Bd3 Nc6 5.O-O d5 6.e5 Nd7 7.Bb5 Qb6 8.Qa4 Be7 9.d3 O-O 10.Bxc6 Qxc6 11.Qxc6 bxc6 12.c4 a5 13.a4 Rb8 14.Nbd2 g5 15.h3 h5 16.Re1 Kh7 17.g4 Ba6 18.Rb1 Rh8 19.b3 Kg6 20.Bb2 Bd8 21.Bc3 Rb7 22.Bb2 Kg7 23.Ba3 Bc7 24.Kg2 Rb6 25.Kg1 Rbb8 26.Nf1 Rb7 27.N1d2 Kg6 28.Kg2 Bd8 29.Bb2 Kg7 30.Ba3 Rb6 31.Rbc1 Rb8 32.Rb1 Kg6 33.Bb2 Bc7 34.Kg1 Kg7 35.Ba3 Rhe8 36.Bb2 Bd8 37.Ba3 Kg6 38.Kg2 h4 39.Kg1 Bb6 40.Rbc1 Rb7 41.Kg2 Rd8 42.Kg1 Rg8 43.Kg2 Rh8 44.Kg1 Rhb8 45.Kg2 Re8 46.Kg1 Bc7 47.Rb1 Rc8 48.Kg2 Rd8 49.Kg1 Ra7 50.Kg2 Rg8 51.Kg1 Rb7 52.Kg2 *
Parent - - By JohnL (***) Date 2008-01-05 00:36
Still a quite a lot to play for. Black can place knight on f4 and play f6 for example. Hopefully the contempt allows...
Parent - By Fulcrum2000 (****) [nl] Date 2008-01-05 00:39 Edited 2008-01-05 00:43
Yeah, we are waiting for the 50-move rule to kick in and force Rybka to show some action...

61. Kg2 Nf8 maybe something will happen now!
Parent - - By SR (****) [dk] Date 2008-01-05 01:40
[#] 1. e4 c5 2. Nf3 e6 3. c3 Nf6 4. Bd3 Nc6 5. O-O d5 6. e5 Nd7 7. Bb5 Qb6 8. Qa4 Be7 9. d3 O-O 10. Bxc6 Qxc6 11. Qxc6 bxc6 12. c4 a5 13. a4 Rb8 14. Nbd2 g5 15. h3 h5 16. Re1 Kh7 17. g4 Ba6 18. Rb1 Rh8 19. b3 Kg6 20. Bb2 Bd8 21. Bc3 Rb7 22. Bb2 Kg7 23. Ba3 Bc7 24. Kg2 Rb6 25. Kg1 Rbb8 26. Nf1 Rb7 27. N1d2 Kg6 28. Kg2 Bd8 29. Bb2 Kg7 30. Ba3 Rb6 31. Rbc1 Rb8 32. Rb1 Kg6 33. Bb2 Bc7 34. Kg1 Kg7 35. Ba3 Rhe8 36. Bb2 Bd8 37. Ba3 Kg6 38. Kg2 h4 39. Kg1 Bb6 40. Rbc1 Rb7 41. Kg2 Rd8 42. Kg1 Rg8 43. Kg2 Rh8 44. Kg1 Rhb8 45. Kg2 Re8 46. Kg1 Bc7 47. Rb1 Rc8 48. Kg2 Rd8 49. Kg1 Ra7 50. Kg2 Rg8 51. Kg1 Rb7 52. Kg2 Rh8 53. Kg1 Rbb8 54. Kg2 Rhe8 55. Kg1 Rb7 56. Kg2 Reb8 57. Kg1 Ra8 58. Kg2 Bb6 59. Rbc1 Rc8 60. Kg1 Rbb8 61. Kg2 Nf8 62. Rb1 Kh6 63. Bc1 Bd8 64. Nf1 Ng6 65. Be3 Be7 66. N1d2 Rg8 67. Kh2 d4 68. Bxg5+ Bxg5 69. Nxg5 Kxg5 70. Ne4+ Kh6 71. Nxc5 Bc8 72. Re4 Rd8 73. f4 Nf8 74. Rf1 Nh7 75. Rf2 Kg7 76. f5 Ng5 77. Ref4 Rb4 78. Rf1 Kh6 79. R1f2 Rb8 80. Rf1 Rb6 81. R1f2 Rb4 82. Rf1 Rb8 83. R1f2 Kg7 84. Rf1 Rb4 85. R1f2 Rb6

Benjamin clearly have seen deeper than Rybka in this position.  The calculating monster presumably does seem to realize that Bc8 is useless, while Nc5 hold the white position together as glue.
Parent - - By SR (****) [dk] Date 2008-01-05 01:57
Maybe white get into zugswang 86.f6?! Black plays his knight to d7 and after the knights are swapped  black doubles the rooks in b-file. The king has prevent the black king coming to f4. If white place the rooks on b1 and b2 he runs eventually out of moves.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-05 02:13
Hmmm...interesting idea...though the knight coming to d7 also leaves the d4 pawn unguarded, and when the rook takes that, I think it's curtains for black.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-05 00:49 Edited 2008-01-05 00:56
Looks like Joel's 63.Bc1? was a blunder and the game will go to Rybka...

Of course, I can't really see "myself" what would have been better, so I guess that even these Father-like games have to be played carefully.  Spike 1.2 Turin suggests Nf1 with -0.45, but I don't really see how that's better than Bc1, though I see that lots of lines open up in Bc1.

EDIT:  Oooh, looks like 64.Nf1?! really does it, as Ba3 seemed necessary...
Parent - - By Victor Zakharov (*****) [ru] Date 2008-01-05 00:56
Finally Rybka started actions

[White "Benjamin"]
[Black "Rybka"]
[WhiteElo "2575"]
[BlackElo "3100"]
[Result "*"]
[GameID "240"]
[UniqID "673095"]
[WhiteClock "0:07:25"]
[BlackClock "0:31:22"]
[Stamp "6174"]
[LastMoves "64.Nf1 Ng6 -0.82"]

1.e4 c5 2.Nf3 e6 3.c3 Nf6 4.Bd3 Nc6 5.O-O d5 6.e5 Nd7 7.Bb5 Qb6 8.Qa4 Be7 9.d3 O-O 10.Bxc6 Qxc6 11.Qxc6 bxc6 12.c4 a5 13.a4 Rb8 14.Nbd2 g5 15.h3 h5 16.Re1 Kh7 17.g4 Ba6 18.Rb1 Rh8 19.b3 Kg6 20.Bb2 Bd8 21.Bc3 Rb7 22.Bb2 Kg7 23.Ba3 Bc7 24.Kg2 Rb6 25.Kg1 Rbb8 26.Nf1 Rb7 27.N1d2 Kg6 28.Kg2 Bd8 29.Bb2 Kg7 30.Ba3 Rb6 31.Rbc1 Rb8 32.Rb1 Kg6 33.Bb2 Bc7 34.Kg1 Kg7 35.Ba3 Rhe8 36.Bb2 Bd8 37.Ba3 Kg6 38.Kg2 h4 39.Kg1 Bb6 40.Rbc1 Rb7 41.Kg2 Rd8 42.Kg1 Rg8 43.Kg2 Rh8 44.Kg1 Rhb8 45.Kg2 Re8 46.Kg1 Bc7 47.Rb1 Rc8 48.Kg2 Rd8 49.Kg1 Ra7 50.Kg2 Rg8 51.Kg1 Rb7 52.Kg2 Rh8 53.Kg1 Rbb8 54.Kg2 Rhe8 55.Kg1 Rb7 56.Kg2 Reb8 57.Kg1 Ra8 58.Kg2 Bb6 59.Rbc1 Rc8 60.Kg1 Rbb8 61.Kg2 Nf8 62.Rb1 Kh6 63.Bc1 Bd8 64.Nf1 Ng6 *
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-05 01:00
Some DPA now after move 64 is giving -3 in most of the lines...
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-05 01:35 Edited 2008-01-05 01:38
Well, after move 85...

I had seen Rybka winning material, but hadn't really thought about what Rybka would do after this happened.  It seems that bishop on c8 is, as it has been all game, still worthless and is, for all practical purposes, not actually on the board.  Neither Rybka nor Spike seem to recognize this fact.  For all practical purposes, Joel is currently up a pawn, I think...

EDIT:  In fact, if Joel plays 86.f6!!, a move that the engines don't seem to link, I think that this forces an immediate draw.
Parent - By Victor Zakharov (*****) [ru] Date 2008-01-05 01:45
Sorry, server will be unavailble for an hour.
Parent - By Krempov [ca] Date 2008-02-10 16:28
Rybka the king of kings. Spike seems to be in second place.
Anyways the only person I found on the internet to even come close to matching rybka is M, Hayward. His games are cool. [url=]www.chess-analysis.com[/url]
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-04 23:28
Rybka is stupid on this, and Spike 1.2 Turin has been laughing at Rybka for most of the game.  Rybka had a winning advantage, I think, on move 10.  It's strange that Qxc6?! was played, especially after wanting to keep queens on the board; Spike was giving bxc6 about -1.  There were a number of other instances, too, but I don't recall them all.

Anyway, Joel is probably laughing and I think he has an extremely easy game now--it's a Father-like game.
Parent - - By Werewolf (*****) [gb] Date 2008-01-04 23:32
only chance is to crack open the k - side
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-04 23:36
I'm really not sure that ...f6 would do much of anything in the long-term.

I am thinking that this may be exactly what Larry feared about playing the Sicilian.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-04 23:32
Actually, this seems exactly the type of game we all feared going into this match...(move 29 now)
Parent - - By Uri Blass (*****) [il] Date 2008-01-04 23:58
I wonder if rybka's anti-human is against trading queens.

Rybka's score with 2 options shows only small difference for Qxc6

depth 15:
r1b2rk1/pp1nbppp/1qB1p3/2ppP3/Q7/2PP1N2/PP3PPP/RNB2RK1 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. ³  (-0.28): 10...Qb6xc6 11.Qa4xc6 b7xc6 12.c3-c4 g7-g5 13.h2-h3 Bc8-a6 14.Rf1-e1 h7-h6 15.Nb1-d2 Rf8-e8 16.g2-g4 Ra8-b8 17.b2-b3
2. =  (-0.23): 10...b7xc6 11.Rf1-e1 Ra8-b8 12.Nb1-a3 Qb6-a6 13.Qa4-c2 h7-h6 14.Bc1-f4 Qa6-b6 15.Ra1-b1 Bc8-a6 16.h2-h3

(,  05.01.2008)

depth 16:

r1b2rk1/pp1nbppp/1qB1p3/2ppP3/Q7/2PP1N2/PP3PPP/RNB2RK1 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. ³  (-0.27): 10...Qb6xc6 11.Qa4xc6 b7xc6 12.c3-c4 g7-g5 13.h2-h3 Bc8-a6 14.Rf1-e1 h7-h6 15.Nb1-a3 Ra8-b8 16.b2-b3 Kg8-g7 17.Ra1-b1
2. =  (-0.21): 10...b7xc6 11.Rf1-e1 Ra8-b8 12.Qa4-c2 f7-f5 13.e5xf6 Rf8xf6 14.c3-c4 Be7-d6 15.Nb1-c3 h7-h6 16.Nc3-a4 Qb6-c7 17.h2-h3

(,  05.01.2008)

depth 17:

r1b2rk1/pp1nbppp/1qB1p3/2ppP3/Q7/2PP1N2/PP3PPP/RNB2RK1 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

1. =  (-0.22): 10...Qb6xc6 11.Qa4xc6 b7xc6 12.c3-c4 Bc8-a6 13.b2-b3 f7-f6 14.Bc1-b2 f6-f5 15.h2-h4 Ra8-b8 16.Nb1-d2 h7-h6 17.Bb2-c3
2. =  (-0.20): 10...b7xc6 11.Rf1-e1 Ra8-b8 12.Qa4-c2 f7-f5 13.e5xf6 Rf8xf6 14.c3-c4 Be7-d6 15.Nb1-c3 Qb6-c7 16.Qc2-e2 h7-h6 17.b2-b3

(,  05.01.2008)

Uri
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-05 00:14
I seem to recall that both of the anti-human machines were in favor of the trade on that move, though they only showed one variation, not two.
Parent - - By Victor Zakharov (*****) [ru] Date 2008-01-05 00:21
Analysis of two versions
supports Multy-pv

[(-0.33)]  d=15  1...Qxc6 2.Qxc6 bxc6 3.c4 a5 4.Nc3 a4 5.Bf4 h6 6.h3 Ba6 7.b3 axb3 8.axb3 (0:00.32)
[(-0.28)]  d=15  1...bxc6 2.Qg4 f5 3.Qf4 a5 4.c4 Ba6 5.Na3 h6 6.h3 Rab8 7.Nc2 (0:00.24)

[(-0.33)]  d=16  1...Qxc6 2.Qxc6 bxc6 3.c4 a5 4.Nc3 a4 5.Bf4 h6 6.h3 Ba6 7.b3 axb3 8.axb3 (0:00.41)
[(-0.24)]  d=16  1...bxc6 2.Qg4 f5 3.Qf4 Ba6 4.c4 Rfd8 5.h4 dxc4 6.dxc4 Nf8 7.h5 Rd3 8.Nbd2 (0:01.08)

[(-0.34)]  d=17  1...Qxc6 2.Qxc6 bxc6 3.c4 a5 4.Nc3 a4 5.Bf4 h6 6.h3 Ba6 7.b3 axb3 8.axb3 (0:01.30)
[(-0.25)]  d=17  1...bxc6 2.Qg4 f5 3.Qf4 Ba6 4.c4 Rfd8 5.h4 dxc4 6.dxc4 Nf8 7.h5 Rd3 8.Nbd2 (0:01.54)

[(-0.34)]  d=18  1...Qxc6 2.Qxc6 bxc6 3.c4 a5 4.Nc3 a4 5.Bf4 h6 6.h3 Ba6 7.b3 axb3 8.axb3 (0:02.22)
[(-0.21)]  d=18  1...bxc6 2.Qg4 f5 3.Qf4 a5 4.c4 Ba6 5.Na3 Rab8 6.h3 Qc7 7.Nc2 a4 8.Rd1 (0:05.06)

doesn't support Multy-pv
[(-0.30)]  d=20  1...Qxc6 2.Qxc6 bxc6 3.c4 a5 4.Nc3 a4 5.Bd2 h6 6.b3 axb3 7.axb3 Ba6 8.Ra4 Bb7 9.Rfa1 f6 10.exf6 Bxf6 11.Ra7 Kf7 12.g3 Ke7 13.Kg2 g5 14.h3 Bg7 15.Rxa8 (0:04.59)
[(-0.26)]  d=18  1...bxc6 2.Re1 f6 3.exf6 Rxf6 4.c4 Bd6 5.Nc3 h6 6.Kh1 Rb8 7.Qc2 Qb4 8.Re2 Ba6 9.b3 Bc7 10.h3 Rg6 11.d4 Rf6 12.dxc5 Nxc5 (0:04.00)
[(-0.32)]  d=13  1...bxc6 2.Re1 Rb8 3.c4 Qc7 4.Qc2 h6 5.h3 Bb7 6.b3 Rfe8 7.Nc3 d4 8.Na4 Nb6 9.Nxb6 (0:00.03)
Parent - By Vasik Rajlich (Silver) [hu] Date 2008-01-05 14:30
The contempt we used so far gave an effective penalty of roughly 10 centipawns for trading queens.

Raising this further has its drawbacks and starts to really cost Elo in engine vs engine play, which is the one area where we can collect hard data.

Vas
Parent - - By Roland Rösler (****) [de] Date 2008-01-05 00:47
You see 24. ... Rb6!?? When I see this, I´m also sure, it would be a long game. But what we see now, is this really chess? :(
PS: I hope, your 7-1 vote will become ash tonight :-).
Parent - - By Fulcrum2000 (****) [nl] Date 2008-01-05 00:54 Edited 2008-01-05 01:03

>PS: I hope, your 7-1 vote will become ash tonight


Rybka will win this one, so no... ;)
Parent - - By Roland Rösler (****) [de] Date 2008-01-05 02:05
Ok, now I see nothing (after move 85)!  But if Rybka will win this, it´s a shame! No clue about the game but winning. Is this the new (Rybka?) approach? And if you would be in the chess-forum like me, you will hate the chess fellows. Now I think, football is a sport for intellectuals and chess for the crazy ones! You can´t hold it out. 0.01 point increasing in one (of the three) of the Rybka engines eval and the crowd is sure, Rybka will win! I´m rather frustrated.
Okay, celebrate you win.
Parent - - By Fulcrum2000 (****) [nl] Date 2008-01-05 02:07
Evaluation jumped from -0.30 to -1.30 which is normally enough for a win, or at least some good winning changes. In this game Joel clearly knew what he was doing so like you say it will probably be a draw.
Parent - - By Roland Rösler (****) [de] Date 2008-01-05 02:25
I´m not sure it´s a draw. It´s like Verdun. Pressing blood!
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-05 02:27
It's a drawn position :-)

To avoid the draw, black must get his knight to d7.  To do this, he must move Rh8 to protect the pawn on h4.  To do this, he must abandon the protection of d4, which will lose the game.  I am assuming that Joel has played f6!! and that we have simply not seen the transmission yet.
Parent - - By Roland Rösler (****) [de] Date 2008-01-05 02:34
I´m not sure! There where so many chances for Joel to play f6, but he didn´t. So why yet?
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-05 02:36
Because that would end the game more quickly.  He is probably having too much fun chatting with Larry right now and occasionally making moves, and he doesn't want the game to end. :-)

I think that even as an unaided human, I could draw this against any centaur, and probably with massive time odds in favor of the centaur (though I'd still need an increment of at least 10 seconds!).  My only fear would be that the centaur, knowing that I'm "only a human", would allow d4 to fall, opening things up, and then I might make a mistake...
Parent - By Mark (****) [us] Date 2008-01-05 02:50
Still no news??

I suspect Joel is holding off on playing f6 to see if Rybka will somehow give the game away while trying to avoid a draw.
Parent - By Roland Rösler (****) [de] Date 2008-01-05 03:23 Edited 2008-01-05 03:25
I like this kind of humor very much!
Okay, I will try my very best :). When I heard of the rules of the match, I think: Fair, but rather hard for Joel with his second game of the day. But what we see now: Joel always "win" the second game of the day!
Now we will have a lesson about theory of conspiracy:
1. In the morning (before the first game) Joel gets some coffee (or tea) with tranquillizers from Larry.
2. For lunch (before the second game) Joel will have some (cooked?) fish :)
3. Place for others (real fanatic (and good :)) consparicy theoraticans).
Parent - - By BB (****) [gb] Date 2008-01-05 21:57
Game 6 so far: 1. d4 Nf6 2. Nf3 e6 3. Bg5 h6 4. Bxf6 Qxf6 5. Nbd2 d6 6. c3 Nd7 7. e4 g6 8. Bd3 Bg7 9. Nc4 O-O 10. O-O e5 11. dxe5 dxe5 12. b4
I figured he would play 12. b4 (though the engines don't like it that much). After seeing 11. dxe5, my hunch is that Joel will get nickeled and dimed (lose bits of equity here and there) and eventually lose.
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-05 22:37
11.dxe5 had been played four times before in human games, though black's response was always 11...Nxe5 with two wins and two draws; one example of a win is Schmidt (2340) - Bischoff (2500) Bundesliga 1994 (Spike 1.2 Turin thinks that 11...dxe5 was a mistake, and really likes white's position here, but says that Joel threw it to a slight black advantage with 12.b4, which you mentioned).  I think that 16.a4 was required on that move, as the engines seem to think that 16.Bc2? threw the game away (I really don't see the motivation at all for this move).  We now have Spike 1.2 Turin and others giving a score of -1 with a fairly open position--this should be a fairly quick loss.
Parent - By BB (****) [gb] Date 2008-01-05 23:04

> Spike 1.2 Turin and others giving a score of -1 with a fairly open position--this should be a fairly quick loss.


Not sure how quick it will be: 18. a4 a6 19. bxa6 Rxa6 20. Nc4 Nxa4 21. Ne3 Bc6 22. Bxa4 Rxa4 23. Rxa4 Bxa4 24. Nd5 could be a slow death.
Parent - - By NATIONAL12 (Gold) [gb] Date 2008-01-05 23:27
congrats to larry & vas on game 6, you followed our advice and gave her a couple of hours rest before this game.
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-06 00:29
Yes, a great day for the Rybka team--congrats!
Parent - By lkaufman (*****) Date 2008-01-06 00:51
Interestingly, today's version without contempt (and older versions) would have taken with the knight, but the contempt setting tries to avoid trades and hence the pawn recapture.
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-07 00:01
Congratulations to the Rybka team on the awesome performance against Joel in this match!  Joel played quite well, even though it might not have looked like it at times from the engine evals, and he made the games enjoyable to watch.
Parent - - By Zadig (**) [gb] Date 2008-01-07 00:18
6+ 2= 0- vs 2575 Elo is a 2,900 performance if I am not mistaken. Are the White odds supposed to cost that much Elo? Intuitively, I would have doubted it...
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-07 00:46
Having the white pieces is 65-70 elo better than having the black pieces historically.  I think that in the case where one side is searching simply for a draw, there is a scaling factor on this that is greater than one, possibly even 2 or more.
Parent - - By lkaufman (*****) Date 2008-01-07 01:16
Aside from the White advantage, we had to cripple Rybka somewhat by setting huge contempt factor to avoid draws. Also in general it is difficult to get a really high performance rating against much lower opposition. I think all in all Rybka scored pretty well. The final score matched my prediction, though it was a slightly optimistic prediction.
Parent - - By Mark (****) [us] Date 2008-01-07 01:37
I really enjoyed following this match!  Hope there are more in the future.

How many games do you think Joel would need to play before he won a game outright against Rybka?
Parent - - By lkaufman (*****) Date 2008-01-07 06:13
Maybe fifty to one hundred.
Parent - - By turbojuice1122 (Gold) [us] Date 2008-01-07 12:24
That's exactly the range that I was thinking when I first saw that question, but I wonder: given the history of the match, many of the games would be a Philidor Defense, right?  That's a double-edged sword.  In both of the games yesterday, Benjamin had a distinct advantage, especially in game 7.  I would think that he would definitely need a far lower number of those types of games to win a game outright, i.e. have one where he gets lucky and doesn't blunder, something that is difficult to do against a computer in those types of positions.  Indeed, the positions resulting from the Philidor seem to be "anti-anticomputer play", with Rybka giving an advantage to Joel (in a position that I get the impression it still doesn't "understand" very well) to avoid closing up the game effectively, but which has enough tactical nuances to make it difficult for the GM to play accurately enough.  However, it would seem that the probability for the GM getting "lucky" is fairly high...
Parent - - By Uri Blass (*****) [il] Date 2008-01-07 13:39
You forget that if the rule say that the target of rybka is to prevent Joel to win a single game out of 100 games then different opening may be used and Joel is not going to get the opportunity to play the same opening again and again.

Uri
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-07 14:52
True--then ironically, it could become Rybka playing for the draw :-).
Parent - By lkaufman (*****) Date 2008-01-07 15:56
Actually, Joel only got a good position in Game 7 because I didn't book his g4 move, thinking that he would not play so boldly in a draw odds game. In fact g4 is not a good move due to the reply b5! and if g5 b4!. Basically my book was not very deep as I had only a couple days to cover multiple defenses to both 1e4 and 1d4. As for the chance of Benjamin actually winning a game, if I kept the contempt setting as high as it was his chances would be more than 1-2%. My estimate assumes that only a small contempt setting would be used. Also of course the opening is critical. If we only want to avoid loss, probably I would choose the Petroff.
Parent - - By George Tsavdaris (****) Date 2008-01-07 18:12
Yes around 2913 but Rybka played all games with white so this should be considered too, although there is no sure way to do that.....
If Rybka just would win a game more instead of draw then it would have an 3045 ELO.

So you understand that these small number of games can't be such a sure way of measuring ELO of Rybka against GMs....
Parent - By turbojuice1122 (Gold) [us] Date 2008-01-07 18:23
Rybka played all games with white

I think you mean black here, judging from the (correct) point that you're trying to make...

If Rybka just would win a game more instead of draw then it would have an 3045 ELO.

Correct, and in fact, this is easily within the realm of chance--with two draws, we're talking about a standard deviation of 1.4 draws, so even given the same level of performance, there is on the order of a 10% chance that Joel gets no draws at all given our current data...
Parent - - By Roland Rösler (****) [de] Date 2008-01-07 23:08
Please, can we get the games with notation of depth, eval and expected move from the Rybka version that played the game and with time consumption by both sides like Alkele gave us that for the Mexican games against Zappa? So we can also see end of book.
Parent - - By lkaufman (*****) Date 2008-01-08 15:30
The evals showing on the playing version were totally unrealistic due to the way contempt factor worked here, so I didn't print or even save them. If the times and scores showing on the same version without contempt as displayed on the Chess Planet site were saved, they could be published. As for out of book move, I announced this each game (to Joel as well, though it was obvious anyway), though I don't know if anyone saved this information.
Up Topic Rybka Support & Discussion / Rybka Discussion / Rybka vs Benjamin II - information center
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill