D. 6

Gelfand – Grischuk
Baku FIDE GP Baku (3), 04.10.2014



Grischuk played 30...Ra8 which was criticized by Gefand. Gelfand analyses this endgame in Decision Making in Major Piece Endings (2020): "Grischuk chooses to go for the endgame where White is left with the c-pawn rather than the a-pawn. I am not a great expert in the most recent developments in the endgame with an a-pawn, which has been analysed a lot by such experts as Karsten Mueller, Mark Dvoretsky and Vardan Poghosyan after a nice discovery by FM Johannes Steckner". Gelfand argues that Black would have been better off going into an ending with a rook pawn rather than a bishop pawn - a view shared by Shereshevsky in the Russian edition of his book (p. 173). In fact, in this position the two options are equivalent, and the general supposition that a rook pawn offers a greater defensive advantage than a bishop pawn is unfounded.

Gelfand recommends here 30…Rc8. 

D. 6-1



White has here 2 plausible moves – 31.a4 (A) and 31.Kf1 (B). Gelfand considers 31.a4, while in Shereshevsky's view, 31.Kf1 is better. In my opinion, 31.a4 creates more problems for Black, than 31.Kf1.

A) 31.a4 c3 32.bxc3 Rxc3

Shereshevsky thinks that this position is an easy draw. In fact, the position is indeed drawn, but it is by far not so easy!

33.a5 Ra3! 34.a6

D. 6-2



According to Gelfand, "Black can play either 34...h5 with a classical position, or maybe 34...g5!? with the idea ...g4, when I don’t even see any plausible ideas for White".

In fact, 34...h5? loses, while Black holds only with Gelfand’s second suggestion, 34...g5!. It is the only move to draw and the path to a draw is far from simple. In position D. 6-2, White wins if it is his move: 1. Kf1 Ra2 (see D. 6-8, line 33.a4 c3 34.bxc3 Rxc3 35.a5 Ra3 36.a6!).

1) 34...h5?

D. 6-3
 


I believe that when Gelfand referred to the classical position after 34...h5, he was merely drawing a misleading parallel with the situation where the White rook stands in front of its pawn. In the present case, however, the rook supports the pawn from the side, and Black has no real counterplay on the kingside. If Black transfers his king to the queenside, White can advance the g-pawn after f3, either gaining space on the kingside with g4-g5 or creating a passed h-pawn.

35.h4 Kg7

35…Kf8 36.Kf1 Ra2 37.Ke1 Ke7 38.f3! Ra3 39.Ke2 Kd7 40.g4 Ke7 41.g5! Ra4 42.Rb7+! Ke6 43.a7! Ra3 44.Kd2 f6 45.f4! Kf5 46.Kc2 fxg5 47.fxg5! Kg4 48.Kb2 Ra6 49.Kb3 Kg3 50.Rc7 Kh3
51.Rc3+ Kxh4 52.Rc4+ Kxg5 53.Ra4 Rxa7 54.Rxa7+–.

36.Kf1! Ra2 37.Ke1 Kh6

37...f6 can be met by 38.f4 or 38.Rb7+ Kh6 39.a7 g5 40.Kd1! Kg6 41.Kc1! +–.

38.f4

38.f3 also wins for White: 38…Ra3 39.Ke2 Kg7 40.Rc6 Kh7 41.Kd2 Rxf3 42.a7 Ra3 43.Rc7! Kg7 44.Kc2 Kf6 45.Kb2 Ra6 46.Kb3 Kf5 47.Kb4 Kg4 48.Kb5!+–.

38...Ra3 39.Kd2 Rxg3 40.Kc2 Ra3 41.Kb2 Ra5 42.Kb3 Kg7 43.Kb4 Ra1 44.Kb5 f6 45.Rb7+ Kh6 46.Rf7 Rb1+ 47.Kc6 Rc1+ 48.Kb7 Rb1+ 49.Kc8 Rc1+ 50.Rc7 Ra1 51.Rc6 Kg7 52.Kb7+–.

2) 34...g5!

D. 6-4
 


Black must prevent h4, which would fix an unfavorable pawn structure for him.

35.Kf1 Ra2!

35…h5? releases the white king.
The impatient 35...g4? also loses: 36.Ke2 h5 37.Kd2 Kg7 38.Kc2 Rf3 39.a7 Ra3 40.Rb7 Kg6 41.Kb2 Ra6 42.Kb3 f5 43.Kb4 h4 44.Kb5 Ra2 45.Rb6+ Kg5 46.Ra6+–.

36.g4

36.Ke1 h5 37.h3 g4 38.h4 Kg7=

36…Kg7 37.Ke1 f6 38.h3 Kf7!

38...h6? 39.f3! Kf7 40.Kd1 Kg7 41.Kc1+–.

39.f3

D. 6-5



39...Ra3! 40.Kd2

40.Ke2 h6 41.Rb7+ Ke6! 42.a7 Ke5 43.Rh7 Kf4 44.Rxh6 Rxa7 45.Rxf6+ Kg3! 46.Rg6 Kf4 47.Kd3 Ra3+ 48.Kd4 Ra4+ 49.Kc5 Ra3 50.Rf6+ Kg3 51.Kd5 Kxh3 52.Ke6 Ra6+ 53.Kf5 Ra5+! 54.Kg6 Kh4! 55.Rf7 Ra6+ 56.Kh7 Ra3 57.Rf5 Ra7+ 58.Kg6 Ra6+=.

40...Rxf3! 41.Kc2 Ra3! 42.Kb2 Ra5! 43.Kb3 f5

43...Ra1 44.Kc4 f5! 45.gxf5 h5! 46.Kb5 g4! 47.hxg4 hxg4! 48.Rg6 Rb1+! 49.Kc6 Rc1+ 50.Kb7 Rb1+! 51.Ka8 Rb4 52.a7 Kf8! 53.f6 Kf7! 54.Rg7+ Kxf6! 55.Rb7 Ra4 56.Rb5 g3! 57.Rb3 g2! 58.Rg3 Ra2! 59.Kb7 Rb2+! 60.Kc6 Ra2=.

44.gxf5 h5! 45.Rb7+ Kf6! 46.a7 g4 47.Kb4 Ra1 48.hxg4 hxg4! 49.Kb5 g3 50.Rb6+ Kxf5 51.Ra6 Rxa6 52.Kxa6 g2=. 

B) 31.Kf1

Shereshevsky believes that this is a strong move which forces Black to retreat later to a8. However, his recommendation leads to a loss for Black.

D. 6-6



A) 31...c3

Shereshevsky gives this move a question mark, and states that Black's play 31...c3? would have been very risky due to 32.bxc3 Bxc3 Rc3 33.Rb3! (a) or 33.Rb2! (b), similar to the Karpov- Knaak game. According to Shereshevsky Black would most likely have to return to the idea of 31...Ra8! (B) 32.a3 Rc8!? with a loss of tempo. In fact, 31...c3 is a good move and the recommendation of Shereshevsky to retreat with 31...Ra8 32.a3 Rc8 is losing.

32.bxc3 Rxc3

D. 6-7



Contrary to Shereshevski’s opinion, this position is drawn. The analogy with the Karpov–Knaak ending is misplaced. The activity of the black rook in the two games differs radically. In the Karpov–Knaak game, the rook was confined to an ineffective position on the a-file in its own half of the board, whereas here it can harass the a-pawn from the side or from behind.

a) 33.Rb3 Rc1+ 34.Kg2 Rc2 35.a4 Ra2! 36.Rb4 h5 37.Kf3 Kg7 38.h4 f6 39.g4 Ra3+ 40.Kf4 hxg4 41.Kxg4 Ra2 42.Kf3 f5 43.Ke3 Kh6 44.f3 Kh5=.

b) 33.Rb2 Ra3

Black blocks the pawn and immediately starts active counterplay on the kingside.

34.Ke2

34.g4 g5 35.Ke2 h5 36.f3 f5 37.h3 fxg4 38.hxg4 hxg4 39.fxg4 Kg7=.

34...f5 35.Kd2 g5 36.Kc1 Kf7 37.Kb1 h5 38.Rb3 Ra4 39.Kb2 f4=.

B) 31...Ra8 32.a3

 
D. 6-8
 


32…Rc8?

This is a losing move. It wastes time and allows White to bring his king closer to the passed pawn.

33.a4

33.Ke2 (33.Ke1) c3 34.Kd1! cxb2 35.Rxb2! Ra8 36.Ra2 or 36.Rb3+–.

33...c3 34.bxc3 Rxc3 35.a5 Ra3 36.a6! Ra2

D. 6-9



37.h4 Kg7 38.Ke1 h5 39.Rc6 Kh7

39...Kf8 40.Kd1 Ke7 41.Kc1 Rxf2 42.Rc2 Rf1+ 43.Kb2! Rf6 44.a7 Ra6 45.Rc7+ Ke6 46.Kb3 f6 47.Kb4 g5 48.Kb5 Ra1 49.Rc6+ Kd5 50.Ra6 Rb1+ 51.Ka5 Kc5 52.Rc6+! Kxc6 53.a8Q+ Kd6 54.Qf8+ Ke6 55.Qe8+ Kf5 56.Qd7+ Ke5 57.Qe7++–.

40.Kd1 Rxf2 41.a7 Ra2 42.Rc7 Kg7 43.Kc1 Kf6 44.Kb1 Ra6 45.Kb2 Kf5 46.Kb3! Kg4 47.Rc4+! Kxg3 48.Ra4! Rxa7 49.Rxa7 Kxh4 50.Kc3+–.

Instead of 32…Rc8? Black makes a draw, entering an endgame with a passed c-pawn: 32…c3 (a), 32…Kf8 (b), or 32…Kg7 (c).

a) 32...c3 33.bxc3 Rxa3 34.Rc6 Ra2=.

b) 32...Kf8 33.Ke2 c3 34.bxc3 Rxa3 35.Kd3 Ke7 36.Rb2 Kd6 37.Kd4 Ra1 38.h4 Rf1 39.Kc4 Kc6 40.Ra2 Kd6=.

c) 32...Kg7?! 33.Ke2 c3! 34.bxc3 Rxa3 35.Rc6 g5=.

Shereshevsky’s remarks on the Gelfand–Grischuk game appear in the Russian edition of his book Associative Thinking. In the English edition, published this year, the Gelfand–Grischuk game is no longer mentioned, which suggests that during translation, editor Peter Boel discovered inaccuracies in Shereshevsky’s analysis and, in agreement with the author, decided to omit them. Before I became familiar with the English edition, I asked whether the analytical error in the Spassky–Torre game had been retained; Peter Boel responded promptly and kindly, explaining that, in agreement with the author, this section had also been excluded. In the next part of this article, I will point out several other inaccuracies found in the Russian edition, which have been omitted from the English edition.

I would like to briefly mention the famous Karpov–Knaak (1992) ending, which I will discuss in detail later in a separate thread.

D. 7



In his book Shereshevsky doesn't intend to delve into complex, specific variations, but rather to demonstrate the dangers for the defending side of placing the stronger side's rook on the same rank as its passed pawn. However, it would have been appropriate to indicate that after Karpov’s move 41.Rc3? the position is in fact drawn, as demonstrated in the 5th edition of Dvoretsky’s textbook by Karsten Müller (p. 239). The well-deserved praise for Karpov’s endgame technique naturally gives the impression that White’s win in this ending was straightforward. This ending will be examined in detail later, but for now it should be noted that 39...g5? was the decisive error, and that after 40.Kd2! Kg6, White could have won with 41.Ra4 or 41.Ra6 instead of 41.Rc3?.