|
< Earlier Kibitzing · PAGE 31 OF 79 ·
Later Kibitzing> |
| Jun-27-06 | | THE pawn: <In that game against Tony Kosten, Rybka just controlled the center and had the more active pieces, making chess look simple:> making chess look simple????
to me it looked like a maelstrom. Couldn't understand half the moves! |
|
| Jun-27-06 | | RandomVisitor: Here is a quick annotation of Kosten-Rybka:
[White "Tony Kosten"]
[Black "Goldbar"]
[Result "0-1"]
B90: Sicilian Najdorf: Unusual White 6th moves, 6 Be3 Ng4 and 6 Be3 e5 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be3 e5 7. Nb3 Be7 8. f3 Be6 9. Qd2 O-O 10. O-O-O Nbd7 11. g4 b5 12. g5 b4 13. Ne2 Ne8 14. f4 a5 15. f5 (last book move) Bc4 16. Kb1 a4 17. Nbc1 d5 18. exd5 Nd6 19. f6 gxf6 20. gxf6 Nxf6 21. Bh6 Kh8 22. Qg5 White has a mate attack Rg8 23. Qxe5 Ra5 24. Nf4 a3 25. b3 (25. Nb3 Bxb3 26. cxb3 Nde4 =) 25... Bxf1 26. Rhxf1 Nde4 27. Nce2 Bd6 [Black threatens to win material: Bd6xe5] 28. Qd4 Qb8 29. h4 White has a mate attack Bc5 [Black threatens to win material: Bc5xd4] 30. Qa1 White has a mate attack Bd6 Black has a mate attack 31. Qd4 Bc5 32. Qa1 Twofold repetition Bd6 33. Nd3 White has a mate attack Rxd5 34. Bg5 [White threatens to win material: Bg5xf6] Rg6 35. Kc1 Qe8 36. Rfe1? (36. Rde1 is the best option White has Rxd3 [Annihilates a defender: d3] 37. cxd3 Nxg5 38. hxg5 Qe3+ 39. Kc2 =) 36... Rxd3  37. Rxd3 Be5 38. c3 Nf2 39. Rf3 N6e4 40. Bf4 Bg7 41. Kc2 Qc8 42. h5 Rg2 43. h6 Bf6 44. Be3 Qd7 45. Bd4 Bxd4 46. Nxd4 bxc3 47. Re2 Qxd4 48. Qf1 f5 49. Qe1 Qc5 50. Rfe3 Ng4 51. Rd3 Rxe2+ 52. Qxe2 Nxh6 53. Qe3 (53. Rd8+ the last chance for counterplay Kg7 54. Rd7+ Nf7 55. Qh2 ) 53... Qxe3 54. Rxe3 Nf7 55. b4 (55. Re1 doesn't change the outcome of the game Kg7 ) 55... Nfd6 56. Re1 Kg7 57. Rh1 Nb5 58. Rc1 (58. Rg1+ hoping against hope Kf6 59. Rg8 ) 58... Kf6 59. Kd3 Ke5 60. Kc4 Nd4 61. b5 Nd6+ (61... Nd6+ 62. Kb4 N6xb5 ) 0-1 |
|
| Jun-27-06 | | psmith: <Randomvisitor> That is not exactly what I would call an annotation. It really doesn't help one to understand the complicated middle game. Note: "White has a mate attack" does not mean "White threatens mate." It means that white is winning and will mate soon. |
|
| Jun-27-06 | | RandomVisitor: <psmith>The comments were actually generated by a computer - I used Rybka in full analysis mode in the chessbase GUI. Some of white's problems come from the unusually weak move 36.Rfe1? that gives Black a nearly one pawn advantage in the evaluation. 30.Qa1 (and again on move 32) buries the Queen in the corner and when it comes out again (on move 48) the game is virtually lost. I usually generate analysis that has the evaluation score printed for each move, and this was an experiment. Rybka's strategy appears to be restriction of the mobility of its opponent's pieces, while maximizing the mobility of its own pieces. |
|
| Jun-27-06 | | psmith: <RandomVisitor> But after 30. Qa1 Rybka seemed happy to take a draw by repetition. The rules required white to win, so Kosten had to play on with 33. Nd3. But it isn't clear to me that Rybka had any advantage after 30. Qa1. There is also the question of how good Rybka's compensation for the exchange would be after 22. Bxf8 Bxf8 23. Bg2. This seems to be a critical variation. |
|
| Jun-27-06 | | RandomVisitor: Looking at the opening moves in Kosten-Rybka it appears that Kosten just selected the opening move that was the most popular (accoring to the chessgames opening book), rather than the highest winning percentage. 15.f5 has a 31% winning score, but is the most popular move. Perhaps better was 15.Kb1. Looks like his defeat started with the opening lines he chose. |
|
| Jun-27-06 | | RandomVisitor: <psmith>after 22. Bxf8 Bxf8 23. Bg2 Ng4 24.Ng3 Rb8 the game is roughly even. |
|
| Jun-27-06 | | psmith: <RandomVisitor> If you're right about the 22. Bxf8 line leading to rough equality (I assume that is just Rybka's evaluation, not base on further analysis), then how can it be that Kosten's defeat started with his opening move choice?? |
|
| Jun-27-06 | | notyetagm: Go checkout this incredibly well-played game by Rybka: IsiChess vs Rybka, 2006. |
|
| Jun-27-06 | | RandomVisitor: <psmith>The chessgames opening explorer gives the expected outcome of wins, losses and draws for the various openings. Since a draw is as good as a loss in an "armageddon" type game, Kosten has a 31% chance of a win at move 15 and a 69% chance of a draw/loss. He should have selected lines that maximized his winning chances and relied on the extra 15 minutes time on his clock to out-think his opponent. Failure to properly play the game as a "must win" (here I am criticizing a GM...) contributed to his loss. |
|
| Jun-27-06 | | RandomVisitor: <notyetagm>In IsiChess-Rybka 2006, Rybka knows that its King is safe and can therefore launch a Kingside pawn attack. Most chess engines penalize the King if its pawn shield has "defects" (Bob Hyatt's 'crafty' engine even uses this word in its source code). IsiChess seems to be unable to understand the principles of the Benko gambit, and therefore should never have even tried such an akward opening. |
|
| Jun-27-06 | | notyetagm: <RandomVisitor: <notyetagm>In IsiChess-Rybka 2006, Rybka knows that its King is safe and can therefore launch a Kingside pawn attack. Most chess engines penalize the King if its pawn shield has "defects" (Bob Hyatt's 'crafty' engine even uses this word in its source code).> Yes, this is -tremendously- impressive for a chess engine. I remember a famous Kasparov win over a computer (Deep Junior?) where the computer simply would not advance the pawns in front of its king when it was -clearly- the correct thing to do strategically and there were no tactical drawbacks. Rybka is much more advanced in its understanding of king safety then these other engines like Crafty. |
|
| Jun-28-06 | | rover: <RandomVisitor>It's dangerous to select moves based on winning statistics. A move with high winning percentage but few games played can mean one of two things: Either a dangerous new move for which no good defense has yet been found or an interesting idea which was successful for some time until someone basically refuted it. The problem is once a refutation is found people won't play the move again so the statistics can't go back to the defenders favour. |
|
| Jun-28-06 | | RandomVisitor: <rover>agreed. But if the White player plays an "armageddon" type game the same way he/she plays a standard game, the results will favor Black. White must change strategy to make the extra time work for him. Looking at won/loss statistics is one way to do this, but obviously there are other strategies. |
|
| Jun-28-06 | | Siddharta: <notyetagm> It was in the third game of the 4 games match between Kasparov and X3D Fritz (Deep Fritz 8 I believe) in november 2003. It's not in the chessgames.com database yet. Here's the chessbase report link:
http://www.chessbase.com/newsdetail...
(I believe the link to the game is broken)
<Kasparov made it look easy and really it seemed quite easy indeed. X3D Fritz never realized it had to attack with its kingside pawns.> And here's Mark Weeks' annotation of all four games:
http://chess.about.com/library/week...
In this game, a completely locked pawn structure was achieved. Fritz's only chances of counterplay was to advance the f and g pawns (just in front of its king). It never did that and saw no danger in Kasparov's infiltration through the a file for a while. A better conceived opening book wouldn't have let this kind of position arise. |
|
| Jun-28-06 | | Siddharta: Sorry, I didn't realize the games are not complete in the second link. Here you can find the games http://en.wikipedia.org/wiki/X3D_Fr... |
|
| Jun-28-06 | | Siddharta: This is the critical position of the game. I wonder what Rybka would play in this position. Kasparov vs X3D Fritz - New York 2003
Black to play (White has just played 14.Nd2-b3)
click for larger view |
|
| Jun-28-06 | | boo1: Analysis by Rybka 2.0 Beta 8 32-bit (at depth 19):
+0.14 14...Re8 15.Ne2 Nf8 16.Ng3 h5 17.Be2 h4 18.Nf1 Qd7 19.Nfd2 Qf5 20.O-O Qg5 21.Kh1 +0.20 14...h6 15.Ne2 Re8 16.Nxa5 Nh7 17.Rb1 Bh4 18.Nf4 Bg5 19.g3 Ndf6 20.Be2 Bxf4 +0.22 14...h5 15.Be2 Re8 16.O-O Nf8 17.Nxa5 Ng6 18.Bc1 Qd7 19.f3 exf3 20.Rxf3 Qe6 21.Rf2 +0.38 14...g6 15.Be2 Nh5 16.O-O Bh4 17.Nxa5 Qg5 18.Kh1 Ndf6 19.Bb2 Re8 20.Rad1 Be6 |
|
| Jun-28-06 | | Siddharta: Thanks for the analysis <boo>! Looks like even the mighty Rybka doesn't find the f5-f4 plan (at least at this point of the game). Which is understandable at "only" a 19/20 depth. Guess this is what they call "horizon effect". |
|
| Jun-28-06 | | Laocoon: A bit off the topic, but: does Rybka 1.2 or later find the stunning but quiet mate in three from this position? (white to play)
click for larger view
The reason I ask is because Rybka 1.1 found mate in twelve and stopped calculating, whereas other engines continue calculating for a shorter mate. |
|
| Jun-29-06 | | notyetagm: Rybka eval please for the following position from a recent chess club game of mine:
click for larger viewI have just played 25 ... ♗h6, threatening the crushing 26 ... ♖xf2! 27 ♖xf2 ♗e3 and the f2-square collapses around the White king. The culmination of an attack for which I had sacrificed the exchange to shatter the pawn shield around the enemy king (17 ... ♗xg2!?). The strongest OTB player I have ever beaten in a slow game (G/75), over 1800 USCF. 28 ... ♗e3!! is particularly pleasing. [Event "Cranston Chess Club June '06"]
[Site "?"]
[Date "2006.06.27"]
[Round "4"]
[White "Luis"]
[Black "Notyetagm"]
[Result "0-1"]
[Opening "Sicilian"]
[ECO "B50"]
[NIC "SI.01"]
1. e4 c5 2. Nf3 d6 3. c3 Nf6 4. Be2 Nc6 5. d4 cxd4 6. Nxd4 Nxe4 7. Nxc6
bxc6 8. Qa4 Nf6 9. Qxc6+ Bd7 10. Qf3 g6 11. O-O Bg7 12. Bc4 O-O 13. Re1 Qc7
14. Na3 Bc6 15. Qe2 e6 16. Bf4 Qb7 17. Bxd6 Bxg2 18. Bxf8 Bh1 19. Qf1 Bxf8
20. Nb5 Rc8 21. b3 Rc5 22. Re3 Rf5 23. Bd3 Rf4 24. b4 Ng4 25. Re2 Bh6 26.
h3 Nxf2 27. Rxf2 Rxf2 28. Qxf2 Be3 29. Rf1 Qg2# 0-1
|
|
| Jun-29-06 | | jhoro: <Laocoon> it doesn't answer your question (i only tried rybka 1.0 and it also stops early), but from the engines i tried mate-in-3 was found by: Spike 1.1 [<1 sec]
Ktulu 4.2 [<1 sec]
Toga II 1.2.X [31 sec] |
|
| Jun-29-06 | | notyetagm: <jhoro: <Laocoon> it doesn't answer your question (i only tried rybka 1.0 and it also stops early), but from the engines i tried mate-in-3 was found by:> I would be very surprised if Rybka 2.0 did not almost instantly find the mate in 3, since it is believed to be so concerned with piece activity/mobility. |
|
| Jun-29-06 | | boo1: <Laocoon>: Rybka 2.0 beta 8 finds the mate in 3 in <1 sec |
|
| Jun-29-06 | | jhoro: <notyetagm> the attack in your game was beautiful. 26.Qh3 was better for Luis but white is still around -1.90 down 26.Qh3 Rf3 27.Qxg4 Rxd3 28.Nd4 f5 29.Nxf5 exf5 30.Qc4+ Kg7 |
|
 |
|
|
< Earlier Kibitzing · PAGE 31 OF 79 ·
Later Kibitzing> |
Kibitzing
