|
< Earlier Kibitzing · PAGE 4 OF 6 ·
Later Kibitzing> |
| Feb-21-12 | | Penguincw: Stunning! Fischer won an endgame that looked drawn.
ENDGAME STATISTICS
    vs. 
The superior side (in this case white) collects the full point 44.8% of the time. A draw happens 51.8% of the time.
The inferior side (in this case black) collects the full point 3.4% of the time. White joined that 44.8% of the time that won. Of course, this game would've shifted the percentages around. |
 |
Feb-21-12
 | | Honza Cervenka: <tamar> <Fischer did not make progress with 52 Ra2, but the win was still there. Houdini (at 30 ply) shows that the Black's Kside pawns can still be vulnerable.> That is very long line and there are many moves, which I don't understand. But the first problem I can see here is a question whether the final position is really won for white. If you remove both Rooks from your diagram, then with black to move it's tablebase draw. Maybe white can win it in some way but in comparison to Balashov's clear cut way to win after 50.Ra6! Kc7 51.c5! bxc5 52.bxc5 Ne8 (52...Nf5 53.Bxf5 Rxf5 54.Rg6 is hopeless for black too) 53.Rg6! it is quite a long shot. |
|
Feb-21-12
| | Honza Cervenka: <Penguincw: Stunning! Fischer won an endgame that looked drawn.> It didn't only looked drawn. It was in fact drawn for some 30 moves before Taimanov's absolutely incomprehensible 81...Ke4?? I think that it would have never happened, if Taimanov would not have lost the third game of the match in a way he lost it before finishing (after the second adjournment) of this one. |
|
Feb-21-12
 | | tamar: <Honza Cervenka> The mere +1.61 advantage two pawns up does send out some warning signals. But I would say the diagram is a probable victory for White, and makes Balashov's statement that 50 c5 is "a surprising error for a player of such high class" over the top. |
|
Feb-22-12
| | Honza Cervenka: <tamar: <Honza Cervenka> The mere +1.61 advantage two pawns up does send out some warning signals. But I would say the diagram is a probable victory for White, and makes Balashov's statement that 50 c5 is "a surprising error for a player of such high class" over the top.> I think that Balashov has a point here and that Bobby, who a bit later dropped his extra c-Pawn in a somewhat desperate attempt to squeeze the win from a game, which he rightfully considered to be objectively won from opening or early middle-game, would not have disagree with a comment pointing 50.c5 as a positional howler. Even if the game would have been still winnable in some way, what I am frankly doubting about, the move 50.c5 runs directly into a position with quite simple, easily foreseeable and (in the best case) hardly surmountable defensive setup of black with blockade on c7, of course, instead of going for straightforward forced win after 50.Ra6 Kc7 51.c5 bxc5+ 52.bxc5 Ne8 (or 52...Nf5 53.Bxf5 Rxf5 54.Rg6  ) 53.Rg6 . And of course, concerning computer evaluation of final position, it is necessary to take them always "cum grano salis" especially in relatively simple endings, where they don't see any clear cut final. In the final position of your line, which is on the diagram, white has two extra connected Pawns g+h, but he has a bad Bishop (K+B+RP vs K ending would be draw here, if black King would not be too far and cut from h8) and misplaced King. Two connected Pawns and Rook against Rook usually win, even if one of those Pawns is a Rook-file Pawn, but two connected Pawns and Bishop against Knight are not quite often enough for win, if the weaker side can block the Pawns from squares of opposite colour to Bishop's colour like on the following diagram, which is dead draw (btw, evaluated by engines without tablebases as win for white). 
click for larger viewIn our position there are also some other possible drawing outcomes like above mentioned K+B+h-P vs K ending or some K+R+B vs K+R ending, if black manages to sac his Knight for both white Pawns. These are the reasons behind my deep doubts concerning white's winning chances here. |
|
Feb-22-12
| | tamar: <Honza Cervenka> I take it your larger point is about computer evaluations, and not limited to this particular position, which may or may not be won. It is a pity we don't have Bobby's notes to this game, as he could say whether his 50 c5 was wrong, or if he erred later. I did some more computer checking on the two pawns up ending, and find that
Houdini does not always know that it can't exchange rooks. After 12 hours, I still don't know if it is won or drawn. So it is still necessary to tweak analysis, and point the computer away from dead ends, which it does not see. |
|
Feb-22-12
| | tamar: Final evaluation of this position for example is difficult. For example take this diagram, two pawns up after exchange of rooks. 
click for larger viewDraw or win depends on whose move it is
With move, Black can draw with 1...Kg7
However, with White to move, even with rooks off
Mate in 47 moves 1. h6 Kg8 2. Kd6 Kh7 3. Ke6 Ng8 4. Be4+ Kh8 5. h7 Nh6 6. Kd5 Ng4 7. Kd4 Nh6 8. Bg6 Ng4 9. Bf5 Nh6 10. Ke4 Kg7 11. Bg6 Ng4 12. Kf4 Nf6 13. Bf5 Nd5+ 14. Ke4 Nc3+ 15. Kd4 Ne2+ 16. Ke3 Nc3 17. Be4 Nb5 18. g4 Nc3 19. Kf3 Kh8 20. Kf4 Ne2+ 21. Ke3 Nc3 22. Kf3 Kg7 23. Bf5 Nb5 24. g5 Nd6 25. Bg6 Nb7 26. Kg4 Na5 27. Be4 Nc4 28. Kf4 Nd6 29. Bg6 Nb7 30. Ke5 Na5 31. h8Q+ Kxh8 32. Be4 Nc4+ 33. Kf6 Nb6 34. Bc6 Nc4 35. g6 Nd6 36. g7+ Kh7 37. Bd7 Ne4+ 38. Kf7 Ng5+ 39. Kf8 Ne6+ 40. Bxe6 Kg6 41. g8Q+ Kf6 42. Qg7+ Kxe6 43. Qd4 Kf5 44. Kf7 Kg5 45. Qe4 Kh5 46. Kf6 Kh6 47. Qg6# [*] Probably beyond human capability to see if that one tempo is winnable, so Bobby made a practical decision to avoid the whole line. |
|
Feb-22-12
| | Honza Cervenka: <tamar> From human point of view and without help of tablebase I would guess that 1...Kg7 should lead to draw as black seems to be able simply to block both white Pawns from squares h6 and g5. On the other hand, to recognize forced win in your diagram with white to move would be much harder task, certainly not possible without proper analysis because despite of impossibility to set up a blockade on dark squares black has still some practical chances, which white must avoid with precise play. |
|
| Feb-22-12 | | RookFile: Regarding the previous diagram, a key question is whether black can give up the knight for the g2 pawn. If he can, it doesn't matter who's turn it is to move, it's a draw. |
|
Feb-22-12
| | Honza Cervenka: <RookFile: Regarding the previous diagram, a key question is whether black can give up the knight for the g2 pawn. If he can, it doesn't matter who's turn it is to move, it's a draw.> Yes, but it is not even necessary, if black is able to keep h6 and g5 under control as white Bishop cannot support his Pawns in crossing this line. On the other hand, with white to move the position on diagram is tablebase win for white and so he has some way to avoid not only blockade on dark squares but also the sac of Knight for white g-Pawn. |
|
| Feb-22-12 | | RookFile: Yes, just thinking of this as a player. We would not have the tablebase in front of us when playing. Make a long story short, black can draw either by setting up a dark square blockade or by sacrificing the knight for the g pawn. If black can't do either one of those things, he loses. I believe that's the "human reasoning" summary of all the long lines the computer gives us. |
|
Feb-22-12
| | tamar: A quagmire, I agree.
Suppose we point Houdini in the direction of a sure win TB endgame, even with Black to move. 
click for larger viewafter 52 Be4 Nc7 53 Kc4 Re5 54 Bf3 Ne6 55 c6 Rc5+ 56 Kb4 Rc1 57 Rd7+ Kc8 58 h4 Nd8 59 Kb5 Ne6 Here Houdini still at 30 ply chooses the naive 60 Be2, which allows Black to take the c pawn without an exchange of knight for bishop. So here we tweak Houdini, and choose another +1.60 range move that does not allow this. 60 Rf7
The idea is once you know that two connected pawns and rook is a win, you ignore other ideas in the position, and try to force that. Here 60...Nd4+ 61 Kb6 Rb1+ 62 Kc5 Nxf3 63 Rxf3 
click for larger viewBlack can win the c pawn, but has no way to prevent the tablebase position above. Alternatively, 60...Re1 61 Kc4 and the King begins wandering to the Kingside. |
|
Feb-22-12
| | Honza Cervenka: Btw, one very interesting point concerning tamar's diagram is that if the position is moved one file to the left, then it is draw even if it is white's turn to move. In case of analogical attempt to win the position by advance of g-Pawn by 1.g6, black saves the day by only 1...Ke7! 2.Kc6 Kf6! 3.Kd6 Nf8! with winning of g-Pawn and drawn K+B+P vs K+N endgame. In tamar's diagram this simple defense doesn't work as after 1.h6! Kf7 2.Kd6! Kg6 3.Ke6! Ng8 4.Be4+! Kxh6 5.Kf7! white wins the Knight, which has no escape from g8. Of course, in the same position moved one file to the left the Knight can from f8 retreat to h7. |
|
Apr-10-12
 | | Peligroso Patzer: <Petrosianic: *** I still can't quite wrap my head around the fact that Taimanov managed to lose this. How does a Top GM (or even a not-so-top GM... or even a Class A player, for that matter) forget that the King goes in the corner in a Bishop + Wrong Coloured Rook Pawn ending, and start running the other direction? The word "blunder" isn't quite grandiose enough. "Temporary Insanity", might cover it.>
[from Jun-16-09]
Here is what Dvoretsky has to say (commenting on the position after <81. Kxf6>):
click for larger view“The most accurate way to draw was <81. … Nd3! 82. h4 Nf4 83. Kf5 Kd6! 84. Kxf4 Ke7=> Also possible was <81. … Kd6 82. Be2 Nd7+ 83. Kf7 Ke5 84. h4 Nf6=>. “Amazingly, the highly experienced grandmaster sent his king off in the opposite direction. ***”
<Dvoretsky’s Endgame Manual, 3rd Edition>, by DVORETSKY, Mark, Russell Enterprises, Inc. ©2011, at page 87 [with move numbers in analysis conformed to move count from actual game]. Perhaps the temporary insanity that <Petrosianic> hypothesizes specifically involved confusing this ending (which, as <Petrosianic> says, is really a Bishop + Wrong Rook Pawn ending, despite the presence of a Black Knight) with the theoretically drawn B + P vs. [same color] B ending, which, as it happens, was the ending that arose when the teenage Fischer faced Taimanov for the first time, and in which he [Fischer] demonstrated a ready knowledge of the drawing technique: Taimanov vs Fischer, 1960. |
|
Jun-16-12
 | | Sneaky: When I was a kid, we didn't have no stinkin' "tablebase" to rely on. We had Bobby Fischer! |
|
Jun-16-12
 | | andrewjsacks: <Sneaky> Fischer was indeed great--but had nothing on Capablanca, Rubinstein, Lasker, or Korchnoi in the endgame, my friend. |
|
Jun-16-12
 | | Tiggler: I cannot understand why this puzzle is set at this point in the game. The only reason would be if 82. Bc8 is the only move that wins. Does not Be6 do just as well? It makes no difference in the line actually played, so what other defense can B find to Be6? |
|
Jun-16-12
 | | sevenseaman: Its tough for me to find the moves accurately w/o any help from Nalimov tablebases. The P must queen. First few moves look to be; 82. Bc8 Kf4(best) 83. h4 Nf3(best) 84. h6 Ng5 85. Kg6 N~ 86. h6 and the P seems to be on its way. White knows he can afford to lose the B but not the lone P. If necessary he will not mind dumping the B if he can lay his hands on the pesky N if it comes anywhere near. One scenario could be;
85...Nh6 86. Kxh6 Kxf5 87. Kg7 1-0 |
|
Jun-16-12
| | Tiggler: OK, I see: 82. Bc8 prevents 82... Nc3, which draws against 82. Be6 . <Eyal> already explained this on Jan-06-08. |
|
| Jun-16-12 | | bachbeet: Sometimes the most brilliant moves come when there are few pieces on the board and there is the danger of a draw. This is a fine example. |
|
Jun-16-12
 | | agb2002: Not a puzzle, but chess culture. |
|
| Jun-16-12 | | SamAtoms1980: Lucky I recently added this game to my collection of games from the book "Chess Endgame Quiz." 82.Bc8 Kf4 83.h4 Nf3 84.h5 Ng5 85.Bf5 Nf3 86.h6 Ng5 87.Kg6 and Black was in zugzwang. |
|
Jun-16-12
 | | SuperPatzer77: <Peligroso Patzer "...The most accurate way to draw was <81. … d3! 82. h4 f4 83. f5 d6! 84. xf4 e7=> Also possible was <81. … d6 82. e2 d7+ 83. f7 e5 84. h4 f6="...>> <Peligroso Patzer> Your analysis is absolutely perfect! Mark Taimanov missed the only move for a draw - 81...d3!. He blundered with 81...e4?? Black to move and draw - see diagram below:
click for larger view81...d3!, 82. h4 f4, 83. f5 d6!, 84. xf4 e7 (Black King finally gets back to the wrong corner) = <Peligroso Patzer> Cool, I wanna buy that book by Dvoretsky. SuperPatzer77 |
|
| Jun-16-12 | | The Big Lebowski: <andrewjsacks: <Sneaky> Fischer was indeed great--but had nothing on Capablanca, Rubinstein, Lasker, or Korchnoi in the endgame, my friend.> No Bobby was just better my friend. |
|
Jun-16-12
| | sevenseaman: A typo in my sol. <85. Bf5 N~ 86. h6> in lieu of <85. Kg6 N~ 86. h6> |
|
|
|
< Earlier Kibitzing · PAGE 4 OF 6 ·
Later Kibitzing> |
