|
< Earlier Kibitzing · PAGE 38 OF 49 ·
Later Kibitzing> |
| May-13-11 | | Kinghunt: I would rephrase that - analysis is an exact science. Computers are only tools to be used in analysis and are only useful as far as they are used properly. Letting an engine sit on infinite analysis for a long time on a single position is not an efficient use of its power. Slide through the lines, move by move. If you do this, the engine will easily be able to eliminate most alternative moves and you can slide your way forward through the line. I gave you a 22 ply line, and even then, it ends at the imminent loss of material to avoid mate, and has not already happened. Trying to evaluate a line like that with infinite analysis from the very beginning is simply pointless. I wouldn't trust engine evals from the beginning of the line at all. The win is just too deep for the computer to find in a reasonable amount of time. But if you slide through, you will see that the main line wins for white, and all alternative black moves win even more quickly for white. In other words, what I'm trying to say is forget the simple engine evaluations and look at the <moves> they suggest. Follow through on the <moves> and evaluate the line for yourself. Does the computer find a miraculous save for black in the Ra7 line? Or does it find a hidden clear win for white in the Qc3 line? Because if neither, like I said before, the engine's evaluations of such a deep line from the very beginning are quite meaningless. |
|
| May-13-11 | | Jim Bartle: Thanks for that, AJ. This definitely is one of those games with all sort of surprising variations, and amazing resources for white to keep the attack going. Another game I remember which deserved such deep analysis was the first playoff game ivanchuk-Yusupov in the 1991 candidates quarterfinals. I'll take a look at World vs. Pogonina. |
|
May-13-11
 | | LIFE Master AJ: <<May-13-11 Jim Bartle: Thanks for that, AJ. This definitely is one of those games with all sort of surprising variations, and amazing resources for white to keep the attack going. Another game I remember which deserved such deep analysis was the first playoff game ivanchuk-Yusupov in the 1991 candidates quarterfinals.
I'll take a look at World vs. Pogonina.>> I analyzed the Ivanchuk-Yusupov game on my website, it won "best analysis" for that year from the CJA. |
|
May-13-11
| | LIFE Master AJ: Ivanchuk vs Yusupov, 1991 |
|
| May-13-11 | | Jim Bartle: Congrats for "best analysis." The variations are mind-boggling, even if Ivanchuk hadn't picked up the "wrong" knight. |
|
May-13-11
| | LIFE Master AJ: <Kinghunt> I think I know how to use an analysis engine ... been doing it for something like 15-20 years now. |
|
May-13-11
| | LIFE Master AJ: I even remember Fritz 2 and Fritz 3.
Fritz 5.32 was the first halfway decent chess program that I remember. |
|
| May-15-11 | | wetherulerz: doubt!!
30.Kb2 .... what can black do to save queen b3 check and then c takes b3 mate?? |
|
| May-15-11 | | Jim Bartle: 30. Kb2 in the actual game, you mean, not in the variations discussed? I think 31. Qd4 would end the threats. |
|
| May-15-11 | | wetherulerz: oh ya..how foolish of me! thanks :) |
|
| May-22-11 | | SMCB1997: What a game from Garry and Topalov. Each move was a great one from both players. The best game I've ever seen. :-) |
|
| May-30-11 | | Kinghunt: Anything new come up in analysis? I'm still not convinced there's any room in the present analysis for any improvement, but I'd be very happy to look at anything you came up with. |
|
| Jun-30-11 | | ksramesh: Can't understand why 29.... Rd6 is not played by Topalov. Could not find a satisfactory continuation for white. Infact white seems to not be able to hold ground after that. Please provide analysis. Thanks. |
|
| Jun-30-11 | | ksramesh: For instance 29.... Rd6 30.Bc8 Qc6!(... Qc4 would be a disaster in view of 31.Bxa6!! Qxc3 32.Bb7#) 31.Qxc6 Rxc6 32.Bxa6 Kxa3 33. Bxb7+ Kxb4 34.Bxa8 Rc3 and so on. I know Topalov is way smarter than I can ever be, but can't find why he chose 29.... Bb7. |
|
| Jul-02-11 | | dmac91: <ksramesh> 29...Rd6? 30.Kb2! (idea Qb3+ w/ mate to follow - 31.Qb3+ Qxb3 32. cb3#) 30... Qd4 31. Qxd4 Rxd4 32.Rxa6# So 29...Bb7 is necessary because it guards a6 and allows black to meet Kb2 with Qd4. |
|
| Jul-05-11 | | DrMAL: Count me in as one who considers this a favorite game. 24.Rxd4 is a fascinating positional sac, very much in the style of Tal. If it were performed in 1960 (particularly against Botvinnik) it would have been criticized for at least a decade with the win regarded as blind luck. OK a few points. The first I cannot resist for its entertainment value. Starting with the correct statement <Apr-15-11 Mozart72: Good morning my fellow detractors. Move 33 could have been a draw for bothe players.> the personal attack by AJ is certainly entertaining. Especially the part <Apr-20-11 LIFE Master AJ: I have always been of the opinion that Kasparov's move MIGHT be better ... yet I could never prove it.> and the ensuing discussion about engines. LMAO! On move 28 a quick look with Houdini 1.5 yields what AJ was first analyzing for 28.Ra7 (lines truncated): 20/61 01:17 176,941,639
+0.53 28.Re6 fxe6 29.Qe3
0.00 28.Qc3 Qxd5 29.Ra7
0.00 28.Ra7 Bb7 29.Qc3
At the next step five minutes into it, Houdini 1.5 gets the evaluation accurate with 28.Qc3 as third best: 21/70 05:07 724,925,827
+8.04 28.Ra7 Bb7 29.Rxb7
+0.96 28.Re6 fxe6 29.Qe3
+0.57 28.Qc3 Qxd5 29.Ra7
The only explanation is that AJ stuck the position on his engine for only a very short time, used it, and then tried to hide this fact afterwards, resulting in his spewing rubbish for awhile. My second point concerns the truth to <Mozart72>'s statement. Kasparov claimed back in 1999 that 28.Qc3 wins in all variations, buffaloing the public including the concensus among GMs. The position after 28.Qc3 is not very difficult to analyze, as the lines are quite forced for awhile if white wants anything and simplification has to occur. Leaving out tertiary lines, the best white can do is: 28...Qxd5 29.Ra7 Bb7 30.Rxb7 Qc4 31.Qxf6 Rd1+ 32.Kb2 Ra8 33.Qb6 (33.Rb6 a5 34.Rc6 Qd4+ 35.Qxd4 36.Rc7 (36.bxa5 Kxa5 37.Rc7 Rf8  ) 36...axb4 37.axb4 Rxb4+ 38.Kc1 Rd4 39.Rxf7 Kb4 ) 33...Qd4+ 34.Qxd4 Rxd4 35.Rxf7 a5 36.Be6 axb4 37.Bb3+ Ka5 38.axb4+ Kb6  After 38...Kb6 white's slight advantage is not enough to win. If in doubt, please check thoroughly with an engine to verify this for yourself before "pulling an AJ" on me LOL. My final point concerns the move 24.Rxd4 itself. As I stated it is a fascinating positional sac that, as shown above and in the game, can easily win if black blunders. 24...cxd4 was in fact a game losing blunder, it takes engines awhile to figure out why because the critical move 28.Ra7 comes seven plies later. This is what makes 24.Rxd4 particularly fascinating. Black's natural propensity is to recapture and, even though black's moves 25-27 are essentially forced, it is extremely difficult OTB to stall recapture, take the time to look ahead, see 28.Ra7 and then realize how it wins. If black plays correctly, 24.Rxd4 only draws. I will post the analysis in a few days after I have double checked it, this post is long enough and my second point above is enough keep the reader busy in the meantime. No matter what, after 24.Rxd4 the OTB fight in this game was fabulous. |
|
| Jul-05-11 | | DrMAL: Double checked, here is the line. If black plays correctly, white's best after 24.Rxd4 is an equal position: 24...Kb6 25.b4 (25.Nb3 Bxd5 26.Qxd6+ Rxd6 27.Rd2 (27.Rdd1 Rhd8 28.Re3 Nh5 ) 27...Rhd8 28.Rf2 Nh5 ) 25...Qxf4 26.Rxf4 Nxd5 27.Rxf7 cxb4 28.axb4 Nxb4 29.Nb3 Rd6 30.Re6 (30.f4 a5 31.Re6 (31.c3 Bd5 32.cxb4 Bxf7 33.bxa5+ Kc6 ) 31...Rxe6 32.Bxe6 a4 ) 30...Rxe6 31.Bxe6 Rd8 = To reiterate, the Tal-like sac in this game is still a fabulous one. White is not guaranteed any sort of win (even after the first blunder 24...cxd4 and then mistake 28.Qc3). In the game, Topalov blundered again with 31...Kxa3 (instead of 31...Rd1+) and Kasparov brilliantly finished him off. Like practically all of Tal's sacs that were criticized, 24.Rxd4 is much more than mere "shock value" its proper response is counter-intuitive and extremely difficult to find OTB. |
|
| Jul-05-11 | | Kinghunt: <DrMAL> Your analysis is excellent. White indeed had no more than a draw if black had declined the rook. Kasparov himself came to the same conclusion (see his analysis here: http://medlem.spray.se/tal0/replay/...) As for the Qc3 line, Kasparov analyzes the exact same line after 31...Rd1+ and concludes that it is enough to win. That statement is very tricky to evaluate. When I have my computer play this out against itself from the end of the line, white usually wins. However, I am convinced that despite this, black can hold with absolutely perfect play. The necessary precision is simply beyond the capacity of even the strongest engines. If you look back to page 27, <CharlesSullivan> gives some very convincing analysis that black can hold by forcing an extremely rare rook versus bishop and pawns position that looks dead lost, but is a tablebase draw. So I think in any kind of actual game, white would win from the end of that line. But objectively, yes, black can draw. |
|
| Jul-05-11 | | Kinghunt: To follow up on my last post, I'm checking the ending again. So far, I've found a promising looking improvement on the line <CharlesSullivan> gave in his last post with 43. Rg7!, saccing the b pawn in order to not be tied down to defending it. It's still too early to be able to say anything for sure though. I'll post again in a few days with my analysis of this line. |
|
| Jul-05-11 | | DrMAL: Thanx for the compliment and references <Kinghunt> I did not look at prior pages. In fact, I did not bother to look at Kasparov's analysis either LOL. And for some reason when I play out that link the text repeats on top of itself, becoming illegible. I did not thoroughly analyze 39...Rc8 because 39...Rf8 quickly worked out to be significantly better than that or other alternatives (39...Rd6 or 39...Re8 or 39...Rad8) I can't argue with Kasparov's analysis if he assumes a mistake for black. I will dump it on Houdini 1.5 overnight to make sure, cheers. |
|
| Jul-05-11 | | DrMAL: That link downloads into a ZIP, awesome, thanx again! |
|
| Jul-05-11 | | DrMAL: Seems already pretty clear that 39...Rc8 is a mistake, possibly giving white a chance to win (Houdini_15a_x64): 29/68 30:45 3,792,425,029
-0.35 39. ... Rf8 40.Rg7 Rf6
-0.66 39. ... Rd6 40.h4 Rad8
-0.67 39. ... Rc8 40.h4 Rd6
I will let it run overnight to see if it's line differs from what I came up with, but so far so good! |
|
| Jul-06-11 | | DrMAL: I am admittedly much weaker in endgame play finding it tedious and dull. I presented lines up to the end of move 38 because the number of viable possibilites for white grow so quickly from here. 39.Kc3 is a viable alternative to 39.Rxh7 but it seemed the latter is stronger and the former transposes with a weaker line for white anyway. Is there a reference for Kasparov's analysis from here? I assume his being quoted for writing 39.Rxh7 Rc8 verifies 39.Rxh7 as best. 39...Rf8 seemed more forceful to establish both rooks on the sixth rank. After learning of the discrepancy with Kasparov and running Houdini 1.5 last night to evaluate over 100 billion positions it strongly concurs with me so I will reiterate: Kasparov's 39...Rc8 is a mistake. 39.Rxh7 Rf8
On move 40. there are five viable moves for white. To mitigate going insane I had trusted Houdini 1.5's choice of Rh6: 32/74 2:08:14 18,917,253,193
+0.44 40.Rh6 Rf6
+0.35 40.Rg7 Rf6
+0.35 40.f4 Rf6
+0.32 40.Re7 Rxf3
+0.29 40.Kc3 Rd6
However, I liked 40...Rd6 and 41...Rf5 better as they seem more forceful to establish both rooks on the sixth rank: 40.Rh6 Rd6 41.f4 Rf5 42.h4 Rff6
This may not be black's best. Now white must try to exchange black's pawn at g6 and the straighforward way appeared as good as any other involving intermediate moves: 43.Bg8 Kc6 44.Bh7 Kc7 45.h5 gxh5 46.Rxh5 Rh6 47.Rxh6 Rxh6 48.Bf5 Rh1 49.Kc3 Rg1 50.g4 Kd6
click for larger viewAlthough white has bishop and three pawns for a rook, I could not find a way for anything other than a draw. Perhaps someone else better with endgames can help here, cheers. |
|
| Jul-06-11 | | Kinghunt: Beginning with the line <31...Rd1+ 32.Kb2 Ra8 33.Qb6 Qd4+ 34.Qxd4 Rxd4 35.Rxf7 a5 36.Be6 axb4 37.Bb3+ Ka5 38.axb4+ Kb6 39.Rxh7 Rc8 40. h4>, which is given by Kasparov as leading to a win, in light of <CharlesSullivan>'s analysis, most normal lines seem to draw. However, after playing around with some ideas, it seems that white may have an improvement from his analysis. After 40...Rd6 41. g4 Rf8 42. Bf7 Rd4, I propose 43. Rg7!, saccing the b pawn, as the most promising plan for white. The idea is that the increased king activity possible when that pawn no longer needs to be defended is worth far more than the pawn. By giving it up, white can decisively activate his king. I have found convincing wins in all lines I looked at. I will now briefly present my analysis.
click for larger viewThe above is the position after my improvement on <CharlesSullivan>'s analysis with 43. Rg7! My main line then continues 43... Rxb4+ 44. Kc3 Ka5 45. Kd3 Rf4 46. Bd5 R4f6 47. Ke3 Rh8 48. g5 Rf5 49. Be4 Rc5 50. Rxg6 Rxh4 51. f4 Rh3+ 52. Kf2 Kb4 53. Re6 Rcc3 54. g6 Rh2+ 55. Kg1 Rd2 56. Bd3, reaching the following position: 
click for larger viewNote how many problems the black rooks are having coordinating. The white bishop is just too powerful. The white king is where it needs to be, on the queenside, and the black king is cut off from where the decisive action will be. This position wins for white, and while I won't give a specific line here, as it's lengthy, I will post it with all the lines I looked at in their entirety in my next post. I also analyzed 45...Ra4 46. Ke3! to a win after both 46...Ra3+ and 46...Rh8. Diverting from the mainline with 46...Rxf3+ and with 47...Re8+ also led to white wins. I tried to be as thorough as possible and looked at most of the major alternatives for black, but I can't guarantee I didn't miss something. If you see a line that you think I missed, please point it out and I'll get to work taking a deeper look at it. My full analysis tree can be found in the following post. |
|
| Jul-06-11 | | Kinghunt: And here is my analysis tree, beginning from the position after <31...Rd1+ 32.Kb2 Ra8 33.Qb6 Qd4+ 34.Qxd4 Rxd4 35.Rxf7 a5 36.Be6 axb4 37.Bb3+ Ka5 38.axb4+ Kb6 39.Rxh7 Rc8 40. h4> It's a mess to look at, but just copy it into your GUI of choice and it should be clear and easy to use. Please let me know if you see any improvements for black in any of the lines I give. [Event "Endgame analysis"]
[Site "?"]
[Date "2011.07.06"]
[Round "?"]
[White "Kasparov"]
[Black "Topalov"]
[Result "1-0"]
[Annotator "Kinghunt"]
[SetUp "1"]
[FEN " 
click for larger view"]
[PlyCount "61"]
40... Rd6 41. g4 Rf8 42. Bf7 Rd4 43. Rg7 Rxb4+ 44. Kc3 Ka5 45. Kd3 Rf4 (45...
Ra4 46. Ke3 Rh8 (46... Ra3+ 47. Bb3 Rh8 48. Rxg6 Rxh4 49. g5 Ra1 (49... Rc4 50.
Kf2 Rc5 51. f4 Rf5 52. Kf3 Ra1 53. Ke4 Rc5 54. Rf6 Re1+ 55. Kf3 Rc3+ 56. Kg4
Kb4 57. g6 Rg1+ 58. Kf5 Kc5 59. Re6 Rcg3 60. Re5+ Kb4 61. Bf7 Rg4 62. Re3 Kc5
63. Rf3 R4g2 64. c3 Ra1 65. Re3 Rf1 66. Ke5 Rg4 67. f5 WI♘) 50. Rf6 Re1+ (
50... Rg1 51. g6 Kb4 52. Rc6 Rh5 53. f4 Ka3 54. Ke4 Re1+ 55. Kd4 Rd1+ 56. Ke3
Re1+ 57. Kf2 Rhh1 58. Kf3 Rhf1+ 59. Kg4 Rf2 60. Kf5 Rg1 61. Bd5 Kb4 62. Be4 Rh2
63. Ke5 Re2 64. Kd5 Rd1+ 65. Bd3 Rf2 66. Rc8 Rxf4 67. g7 Rg1 68. g8=Q Rxg8 69.
Rxg8 Rh4 70. Rb8 Kc3 71. Rxb5 WI♘) 51. Kf2 Rhh1 52. g6 Kb4 53. Rc6 Reg1 54.
Bf7 Ka3 55. c4 bxc4 56. Bxc4 Rg5 57. Bd3 Rh2+ 58. Ke3 Kb4 59. Kf4 Rg1 60. Ke5
Rh4 61. Bf5 Re1+ 62. Be4 Rg1 63. Rd6 Rg3 64. Kf6 Rh1 65. Re6 Rd1 66. f4 Rd8 67.
f5 WI♘) 47. Bxg6 Rxh4 48. g5 Kb6 49. Bd3 Rh8 (49... Rh2 50. g6 Rg2 51. Rg8
Kc5 52. f4 Kd6 (52... Rg3+ 53. Kf2 Rg4 54. f5 Kd6 55. Rb8 Rg5 56. Rxb5 Rh4 57.
Kf3 Rhg4 58. Ke3 Rh5 59. Be4 Rh3+ 60. Kd4 Rh2 61. Rb6+ Kd7 62. Kd5 WI♘) 53.
Kf3 Rg1 54. Bxb5 Ra5 55. Bd3 Rh5 56. f5 Rhg5 57. f6 WI♘) 50. g6 Ra3 51. Kd4
Rf8 52. Rf7 Rg8 53. Rf6+ Kc7 54. f4 Rd8+ 55. Ke5 Ra1 56. Re6 b4 57. Re7+ Kb6
58. Be4 Ra5+ 59. Kf6 Rf8+ 60. Rf7 Rb8 61. g7 Kc5 62. f5 Ra6+ 63. Kg5 Ra1 64.
Rf8 Rg1+ 65. Kf4 Rb7 66. g8=Q Rxg8 67. Rxg8 WI♘) 46. Bd5 R4f6 (46... Rxf3+
47. Bxf3 Rxf3+ 48. Ke4 Rh3 49. Rxg6 Rxh4 50. Rg8 Kb6 51. Kf5 Kb7 52. g5 Rh2 53.
c3 Rc2 54. Rg7+ Kc6 55. Rg6+ Kd5 56. Rg8 Rf2+ 57. Kg6 Rc2 58. Rb8 Rxc3 59.
Rxb5+ WI♘) 47. Ke3 Rh8 (47... Re8+ 48. Be4 Rd6 49. Kf4 Ree6 50. g5 Rd4 51.
Rh7 Rc6 52. h5 gxh5 53. Ke5 Rxe4+ 54. fxe4 Kb4 55. Rxh5 Rxc2 56. g6 Rc5+ 57.
Kf6 Rxh5 WI♘) 48. g5 Rf5 49. Be4 Rc5 50. Rxg6 Rxh4 51. f4 Rh3+ 52. Kf2 Kb4
53. Re6 Rcc3 54. g6 Rh2+ 55. Kg1 Rd2 56. Bd3 Ra3 57. Re1 Ra8 58. g7 Rg8 59. Re7
Kc5 60. f5 b4 61. Be4 Rd6 62. Kf2 Rh6 63. Rd7 Rh4 64. Ke3 Rh3+ 65. Bf3 Rg3 66.
f6 Rg6 67. Rc7+ Kd6 68. Rc6+ Ke5 69. f7 R6xg7 70. fxg8=Q Rxg8 WI♘ 1-0 You may have to manually change a couple things to undo chessgame's automatic revisions (ie, change the diagram back to a FEN, recognize that the Knights are just the Ns in the word WIN, etc.). |
|
 |
|
|
< Earlier Kibitzing · PAGE 38 OF 49 ·
Later Kibitzing> |
Kibitzing
+ sac: 24.Rxd4