In the present position, White made an effective winning move.
21 years old young Paul Keres (07. January 1916 — 05. June 1975) won Alexander Alekhine (1892–1946), the World Chess Champion of 1927–1935 and 1937–1946, on the 23. move. Alekhine was an ex-champion at the time of the game, but he regained his title in the same year a few months later. At the end of the tournament Margate 1937, Keres shared 1st–2nd place with Reuben Fine, having 7½/9 points
(+6 −0 =3), and 1½ points ahead of Alekhine who earned the 3-rd place.
It was Keres's first and last win against Alekhine, who unexpectedly died in 1946.
Keres vs Alekhine
credit: Chessgames.com
This game, however, is highly controversial.
A Lot of Mistakes
Both the game itself and also Keres's comments to it contain surprisingly many mistakes. In the 1996 edition of Keres's book The Road to the Top, there are some comments by grandmaster John Nunn, but he has mentioned only some of the previously unnoticed mistakes.
Around 2010, the chess engines were already so developed that some previously unknown mistakes were discovered and discussed in the Chessgames "Kibitzer's Corner":
Keres has included this game into the collections of his best games, for example, into his book Valitud partiid, 1961 [= "Selected Games"].
My analysis with the chess engine Komodo (Chessbase, "tactical analysis", 30 min/move) gives an astonishing statistical result:
Precision: White: 42%
Black: 13%
Actually, Alekhine's last three moves 20...f6?; 21...Rxf6?; and 22...Qb4?? were all mistakes.
Some commentators guess that perhaps Alekhine was drunken — as it reportedly happened during the World Championship Match against Euwe in 1935 when Alekhine lost hist title. However, a half a year after the present game, Alekhine won his title back convincingly.
According to comments by Keres, Keres was in a zeitnot at the end of the game. So, perhaps Alekhine was in a zeitnot as well.
I do not know what happened.
Methodology
I have used the following sources:
- Keres, P. (1961) Valitud partiid. [= "Selected Games"]
- Keres, P. (1996) The Road to the Top. (Contains some comments by grandmaster John Nunn.)
In my analyzes, I have used the chess engine Komodo.
I have analyzed this game seriously like the correspondence chess players are doing this. It is a timeconsuming process. However, I have not been so serious as the correspondence chess players are about their own games.
It is not the final judgement. For example, I have not used some powerful supercomputers.
My analyzes have been presented in my Lichess study:
I have also composed a PDF file with my analyzes. Together with the diagrams, it is a book of 40 pages. You can download it from here:
As
there have been so many mistakes both in the game and in the
commentaries to it (even in 1996, only a few mistakes have been
noticed), it is obvious that human beings are unable to calculate the
variations in such complicated positions. At least, not during the game.
Therefore, we should rather
study the thought economics of the players: not how they did calculate the variations but how they avoided
calculating
the variations. For example, in several positions, the intuitive
evaluations of Keres were correct although he provided the variations incorrectly.
The Game
If you do not see the board above correctly, switch off your advertisement blocker or use the link below:
The Mystery of the d–Line Roentgen
Here, I shall only partly analyze the last phase of the game.
16...Qd8-e7!
Finally, Black is ready to castle 0-0-0.
Here, White can grab the pawn with the move 17. Nxg5??, but 17...0-0-0 follows.
Keres:
"...suddenly creating several unpleasant threats."
Now, White's knight N/g5 is attacked, while White's queen Q/d5 is threatened by Black's rook R/d8 after any move with Black's bishop B/d7 ("roentgen").
It is important to note that Keres's remark above makes it evident that he was aware of such tactical motive. Therefore, it is mysterious that he has not provided any comments about that motive when commenting on Black's move 20...f6 (see below).
Now, for example, 18. Nf3?? Bf5 and White's queen has been captured.
Thus, Keres did not fall into Alekhine's clever trap and played
17. Ba4xc6!
But his comments are misleading.
Keres provides this move with the exclamation mark "!" and writes:
"White's main attacking idea is to keep the enemy king fixed in the centre."
In the Estonian version, he even adds that White aims not to gain some little material advantage.
In actuality, after a few moves, White grabs the pawn P/g5 and still allows Black to castle.
The real idea of the move 17. Bxc6, as it turned out, was to win some time and improve the position of White's queen before taking the pawn Nxg5. White eliminates the knight N/c6 that attacks the central pawn P/e5 and does not allow White's queen to the squares d4, a5 and a7.
17...Bd7xc6
18. Qd4-d3
From here, White's queen defends the knight N/f3 and controls the square f5. It also attacks the pawn P/h7. However, note that the pawn structure of White's king's position has been weakened because of the move made c2–c4. The attack on the pawn P/h7 may be illusory because it would lead White's queen far away from White's king. The queen on h7 would be temporarily out of the play.
18...Bc6-d7?!
(0.72/29) Komodo
Better was 18...g4! 19. Nd4 0-0-0! 20. Qf5+ Bd7 21. Qxf4 +/= (0.36/30).
In the present position, Black would get sufficient initiative as compensation for the pawn P/h7:
19. Qxh7? 0-0-0 = (0.00/29).
The computer suggests 19. Qd4! as the only move keeping White's advantage.
As Black doesn't threaten Bxf3 anymore, White's queen can move on the better position. The idea is to play — after Black's 19...0-0-0 castling — 20. Qa7!
Surely, it is difficult to see such manoeuvre like Qd5-d3-d4.
However, while the computer initially assigns 0.72 centipawns to that move, later in the analysis, it is reduced to approximately 0.30.
Note that from Keres's comments to Black's 21. move it is obvious that during the game Keres was aware of the manoeuvre Qd3-d4-a7. And it would be consistent with his claimed aim "to keep the enemy king fixed in the centre". But he has not commented on the possibility 19. Qd4.
In the position shown above, after having exchanged Black's knight N/c6 and improved the position of White's queen (Black cannot play Bf5 anymore), Keres is finally going to take the pawn P/g5 with the move 19. Nxg5. He does not comment on this move. But it was a mistake, and Black can equalize after that.
19. Nf3xg5?!
Not only winning the pawn but also destroying Black's pawn structure. But White wastes time.
Still, perhaps it was a good decision if Keres was in a zeitnot already. Now, White is a pawn up, and it is difficult to lose any possible endgames.
19...0-0-0
There are several threats. It is the roentgen motive.
Here, the complications after 20. Nxh7 Bg4! 21. Qa3! = finally give Black full equality. Keres was right to avoid these complications. Moreover, he was in a zeitnot. Even Keres's and Nunn's analyzes contain some mistakes here.
But what about 20. Ne4 with the idea Ne4-f6-d5? Then 20...Bg4 -/+.
As Keres did not play 20. Ne4? was he aware of the roentgen motive? Granted, theoretically, it is possible that he did not see the move 20...Bg4, but regarded the equality after 20...Qxe5 21. Qd4 Qxd4 22. Rxd4 Bc6! = (-0.15/30) as unsatisfactory for White.
20. Ng5-f3
The d-line mystery position
As we saw above, Keres was well aware of Black's motive of moving his bishop B/d7 so that Black's rook R/d8 will attack White's queen Q/d3. However, it is strange that in his comments concerning the present position, Keres has not mentioned neither the move 20...Be6 nor 20...Bh3.
20...Be6 21. Qc2 Qc5 22. b3 Bg4 = (0.00/29) with full equality.
20...Bh3 21. Nd4 Bxg2 22. Rhg1 f3 23. Rge1 = (0.20/30) with full equality.
Suddenly, Alekhine made 3 mistakes consecutively:
20...f7-f6?
(0.75/33) Komodo
Keres provides this move with the exclamation mark "!"
According to Keres, Black gets rid of his doubled pawns and opens up a position to get even a little counterplay.
However, note that White's pawn P/e5 was a weakness as well, while Black's pawn P/h7 perhaps did not need a defence, as the move Qxh7 would lead White's queen out of the play.
21. e5xf6 Rf8xf6?
Better was 21...Qxf6. According to Keres, Alekhine was afraid of 22. Qxh7. After sharp tactical complications beginning with 20...Bf5!, the result is an endgame. Keres evaluates that White has some winning chances. But the computer shows 0.00 — full equality.
However, Keres was prepared to avoid such complications as he was in a zeitnot. His plan was to force the exchange of the queens with 22. Qd4. The resulting endgame is better for White.
It also demonstrates that during the game, Keres was aware of the motive Qd3-d4-a7. But neither did he play 19. Qd4 nor did he comment on this possibility.
In this game, Black has had difficulties with castling. Finally, he managed to castle: 19...0-0-0. In such a situation, it often happens that the player forgets that one's king may be unsafe even after castling. With his move 21...Rxf6? Black has weakened his back rank.
22. Rh1-e1 Qe7-b4??
23. Qd3xd7+
1-0
