1

Contents

Foreword
Introduction
Chapter 1 - ‘Masonry’ and the essence of associative thinking
Chapter 2 - Cooperation between knight and pawn
Chapter 3 - To hurry or not to hurry?
Chapter 4 - Benchmarks and errors
Chapter 5 - Associative thinking in the defence of inferior
positions; active or passive defence
Chapter 6 - Fundamental endgame principles: methods of play
Chapter 7 - Associative thinking in rook endings
Chapter 8 - The problem of exchanges in inferior positions
Chapter 9 - Associative thinking in superior positions
Chapter 10 - The difference between associative thinking and play
in typical positions
Acknowledgements

2
Foreword
What does Mikhail Shereshevsky’s new book teach? The same as
all his previous works: thinking! Do not mechanically memorize
information (languishing in the realization that it is impossible to
really learn even one opening), but study the decision-making
methods of the strongest chess players in the world. Notice their
favourite manoeuvres and techniques. Find those inconspicuous
‘landmarks’ that they are guided by when choosing one path or
another.

In one of the broadcasts of the popular YouTube channel Levitov

Chess, grandmaster Daniil Dubov highly appreciated
Shereshevsky’s previous book Endgame Strategy: ‘An amazing
book has just been published, I recommend it to everyone, and I
praise it every day. I think it’s suitable for any level; I found a lot of
interesting things for myself. I really like the way it’s constructed.
First, there is a thesis, then its proof, and then its refutation. The
book is structured in such a way that you understand that things are
not easy. Let’s say the first chapter is ‘Centralization of the King.’
First, it explains to you what the idea is, that it is very reasonable to
centralize the king. You think: okay! Then you are given a bunch of
examples where the author says: look, White centralized the king,
everything is great, it was correct. So he found it or didn’t find it,
but it was necessary to centralize the king. It would seem that you
have already mastered the topic, but on the last five pages there are
examples where the king was centralized, but it was not necessary.
You are told: but here it would have been better to play concretely.’
Analyse, do not take anything on trust. Compare, criticize and draw
your own conclusions. These are the principles to which Mikhail
Shereshevsky has stuck throughout his entire life and which he
3
taught to his numerous pupils. Think for yourself! This is an
essential habit, not only in chess, but in life generally.
Shereshevsky’s new work is devoted to a very important and at the
same time, surprisingly, little-researched topic – associative
thinking. During a chess game, we constantly have certain
associations, and sometimes even a feeling of déjà vu. We are not
talking about specific pawn structures, where you need to act
according to a once and for all verified plan, but about more abstract
things. How do knight and pawn interact best? Where should the
rook be placed – behind the passed pawn or to the side? Is it
necessary to seize space on the flank if the queens are about to
disappear from the board? What squares should you place your
pawns on if you have a bishop versus a knight? Generations of chess
players before us answered these and similar questions (of which
there are countless) and will seek answers after us. Sometimes, in
notes to their games, grandmasters tell when and what associations
they had, but these examples have to be collected bit by bit.
The author clearly shows that studying the works of classics and
contemporaries is a both exciting and very useful activity, especially
when the games are accompanied by their own paras or paras from
experts. Shereshevsky pays special attention to outstanding
endgame masters such as Ulf Andersson, Anatoly Karpov, Vladimir
Kramnik and Magnus Carlsen. The latter stands apart: he is a real
wizard who knows how to squeeze blood from a stone!
Of course, Shereshevsky devotes many pages to his favourite stage
of the game (especially rook endgames, and this is understandable –
they occur most often in practice), but this book is not about the art
of playing endings. More precisely, not only about that. Among the
topics discussed are such important and, again, little-studied ones as
‘The problem of exchanges when defending inferior positions’,
‘Associative thinking in better positions’, ‘Varying the rhythm of
the game’, ‘Playing to widen or narrow the game’. And everywhere

4
he offers practical advice on how to increase the ‘strength of the
average move’, avoid failures, and achieve that very ‘smooth
masonry’ that distinguishes outstanding masters. Let us emphasize
once again: the book does not offer ready-made recipes (although
they are also available), but mainly advice from a wise person on
how to expand your knowledge base and improve your playing
technique.
You may ask: what about flights of fancy, combinational fireworks,
deep and precise calculation of variations? All this, of course, is
very important, but hundreds of books have already been written on
this topic. However, there are not enough works on chess technique.
You must clearly understand that this is not some boring routine, but
an equally exciting part of the game – just different. To get a taste
for such play means to raise one or even several levels of skill at
once. After all, we have before our eyes the example of Magnus
Carlsen: the strongest chess player of our time is equally
enthusiastic about attacking and about looking for the slightest
chance of winning in a position in which many of his colleagues
would resign themselves to a draw. That’s why he’s the best.
I am sure that the attentive reader will obtain great pleasure (and
benefit, of course – practical benefit with bonuses in the form of
rating points) from studying this book. Enjoy reading!
Vladimir Barsky, International Master

5
Introduction
Today’s chess has become a much younger man’s game. This is due
to many factors, but the main thing, in my opinion, is the availability
of work with chess programs that are superior to any person in terms
of playing power. Still, in order to gain practical benefit from
working with these programs, a chess player requires high
qualifications. After all, the main difficulty in such work is that the
tree of variations suggested grows to exaggerated sizes. It is simply
unthinkable for a person to remember them all, and it is not
necessary.
The ability to cut out unnecessary things without compromising the
quality of analysis is highly valued in the chess world. This skill is
possessed by a small number of grandmasters, from whom, as a
rule, coaching teams are formed to prepare for world championship
matches. But most often such teams work only on the opening
preparation of chess players.

How does one achieve a high level of play in all stages of a chess
game? It seems that everything is simple: learn your openings, work
on typical middlegame positions, practice calculating variations and
improve your endgame technique. But as soon as you start working
on any stage of a chess game, you will encounter the same problem
– a huge load on the memory. As a result, you will most likely
significantly reduce the amount of information necessary for further
progress. Or you will learn this volume poorly, and you will have to
return to this material, losing some important points. In any case,
you will try to learn most of the information mechanically, which, of
course, is boring and ineffective.
What should you do? Stop rote learning, turn on your curiosity,
come up with a vivid image for the technique being studied and
6
remember it as an association. And most likely, at the right moment
you will recognize it and understand how you need to act.
Let me give you an example. Let’s remember two methods of
realizing an extra pawn on the flank in a rook endgame:

1. Move the white king away from his kingside pawns and free the
rook on the 7th rank.
2. Put the rook on the 8th rank and the pawn on the 7th and then
use zugzwang to win one of Black’s pawns.
During my classes at the junior chess club Sirius, I noticed that in
the first method, it was most difficult for young chess players to
remember the first word. So I asked them a question: ‘What do you
do at the disco?’ The answer was: ‘Move!’ After this, there were no
longer any problems with remembering the wording: everyone
remembered that in such positions the king must ‘break away’ from
the pawns and free the rook along the seventh rank.
And even though the associations are not always so emotional, and
sometimes they are even purely chess-related, they are very helpful
in reducing rote learning. Associative thinking, which is the focus of

7
this book, develops a chess player’s ability to recognize similar
techniques or methods of play in different positions, and also
improves the quality of play in all stages of the game.
In order to explain how this works, I offer a few examples for you to
examine.
Quite recently, I looked at the following game.
Queen’s Gambit Declined
Maxim Matlakov (2662)
Volodar Murzin (2547)
Dubai 2023
1.d4 d5 2.c4 c6 3. f3 f6 4. c3 e6 5. g5 bd7 6.e3 a5 7.cxd5
xd5 8. c1 xc3 9.bxc3 a3 10. c2 b6

In this position, grandmaster Matlakov played a novelty with the

surprising move 11.h4!? .
This move is not among the first three lines of my computer. Let us
consider its point by using associative thinking. It is clear that White
does not intend a kingside pawn storm in the near future, as he has

8
no grounds at all for that. So why was the move with the rook’s
pawn made? To seize space and take play into a favourable
endgame! I immediately recalled another game of Matlakov’s.
Queen’s Gambit Declined
Maxim Matlakov (2702)
David Howell (2667)
European Championship, Minsk 2017
1.c4 e6 2. c3 d5 3.d4 e7 4. f3 f6 5. g5 h6 6. h4 0-0 7.e3
e4 8. xe7 xe7 9. c1 c6 10.g4!? d7 11.h4 e5 12.cxd5 xc3
13. xc3 cxd5 14.g5 e4 15. d2 hxg5 16. h5 g6 17. xg5 xg5
18.hxg5

All of White’s activity on the kingside is directed at transposing into

a favourable endgame.
Here is an earlier example on the same theme.
English Opening
Ulf Andersson (2600)
Andrey Sokolov (2635)
9
Bilbao 1987
1. f3 f6 2.c4 b6 3.g3 c5 4. g2 b7 5.0-0 e6 6. c3 d6 7.b3
bd7 8. b2 e7 9.d4 cxd4 10. xd4 xg2 11. xg2 c7 12.e3 a6
13.f4 0-0 14. f3 ac8 15.g4 c5 16.g5 fd7 17. ad1 fe8 18.h4
f8 19. d2 b8 20. fd1 c6 21. xc6 xc6 22. xc6 xc6

Once again, White’s seemingly dangerous initiative on the kingside

has served as the basis for a transition into a superior ending.
Let us return to Matlakov-Murzin. After 11.h4!? events developed
as follows:
11... a6 12. xa6 xa6 13. f4 0-0 14.h5 ad8 15.h6 g6 16. e2
xe2+ 17. xe2

10
White’s position can be preferred. Volodar Murzin could not cope
with the defensive problems and lost. The games Matlakov-Howell
and Andersson-Sokolov also ended in victories for White. They are
examined in detail in the second edition of Endgame Strategy .
What do these three games have in common? After all, the positions
we are interested in do not resemble each other in pawn structure
and arose from completely different openings. But associative
thinking is directed at exposing the logical essence of the ideas of
Matlakov and Andersson.
In Chapter 9, ‘Associative Thinking in Superior Positions’, we will
discuss spatial advantage. In positions with an advantage in space,
the side with this advantage should, as a rule, strive to exchange
heavy pieces. First of all, the queens. If you are constrained, then
you need to strive to exchange minor pieces.
What unites these games is the method White used in the opening .
He advanced the kingside pawns in order to gain space in a future
queenless middlegame or endgame, and not at all in order to
checkmate the opponent’s king with a direct attack. By the threat of
such an attack, he persuaded the opponent to exchange queens to
11
White’s advantage. The Andersson-Sokolov game was to some
extent ahead of its time. And Matlakov’s fights with Howell and
Murzin demonstrate a modern approach to playing openings, using a
computer. By the way, Maxim Matlakov was part of Ian
Nepomniachtchi’s coaching team at the world title match with Ding
Liren.
It is important to note that associative thinking is different from
playing in standard positions. There are many books on standard
structures and the methods of play in them. Such books are useful.
They teach a healthy pattern of action in typical positions. But this is
not associative thinking, which contains the image of some
technique or method of action in a wide variety of situations. This
book ends with a short chapter: ‘The difference between associative
thinking and playing in standard positions’.
I hope this book will help you accelerate your chess improvement.

12
Chapter 1
‘Masonry’ and the
essence of associative
thinking
In the slang of Russian-speaking chess players, ‘masonry’ is the
strength of a chess player’s average move. Why this name? Perhaps
this term moved into a chess context by analogy with bricklaying: a
good mason skilfully and quickly lays brick to brick, and the result
is a beautiful structure. For a good wall, it is essential that all the
bricks are lined up accurately and equally spaced. An amateur will
perhaps achieve this most of the time, but if even one brick in ten is
misplaced, the effect will be to make the wall look a shambles.
Chess is similar – the player who plays every move at a constant
level will do better than the player who plays outstanding moves
nine times out of ten and then a really poor move.
But it seems to me that chess masonry is inherently closer to
snooker masonry: a strong player must not only beautifully knock
the balls into the pocket, but also carefully think through a whole
series of shots and place the cue-ball in the right position each time.
After all, if a snooker player in an advantageous position does not
think through the subsequent placement of the balls, then even a
single beautiful pot will take him away from victory.

13
Likewise, in chess, one often observes how a chess player plays an
excellent game, makes good, logical moves according to a correctly
constructed scheme, and achieves a tangible advantage. But then
suddenly there follows one move ‘in the wrong direction’. And
although it is not a serious mistake, the position is instantly levelled,
and all the fruits of the previous play disappear.
On the contrary, if a chess player manages to maintain a high quality
of move throughout the entire game, then he will naturally achieve
success. This reminds me of television advertising, which
emphasizes the high quality of the products: ‘Cucumber to
cucumber! Mushroom to mushroom!’ In relation to a chess game, I
would like to add: ‘Move to move!’
In modern high-level chess, the opening preparation of most
grandmasters is systematic and takes into account possible surprises
from the opponent. Nowadays, it is very difficult to come up with a
conceptual novelty in the opening, and it will only be a surprise for
a very short time. Therefore, it is rarely possible to truly surprise
your opponent in the opening, and more often than not, chess
players of the level 2650-2750 emerge from the opening with
approximately equal chances, having a wide variety of playing
opportunities. And although the position remains approximately
equal for some time, the advantage gradually comes to the player
whose average move is stronger. He seizes the initiative and begins
to outplay his opponent.
To better remember this process figuratively, it will be useful to
imagine a scale. If you mentally weigh your opponent’s moves on
the scales, then the chess player who has the best ‘masonry’ will
have each move several grams heavier than his opponent’s move.
After about a dozen such moves, the evaluation of the position
moves from the symbol ‘=’ to the symbol ‘ ’ or to ‘ ’.
Of course, in modern hard and sporting chess, much depends on the
time spent thinking about a move. The right move must be made at

14
the optimal time. An exception may be critical moments in which
the outcome of the game depends on the move made.

As an example, let’s look at the first game of the 2021 World

Championship Match between Carlsen and Nepomniachtchi, which
clearly demonstrates the mighty masonry of the sixteenth World
Champion.
In doing so, we will rely on the deep and sometimes ironic paras
made by grandmaster Alexander Khalifman in the magazine 64 No.
2 for 2022.
Ruy Lopez
Ian Nepomniachtchi (2782)
Magnus Carlsen (2855)
Dubai Wch 2021
1.e4 e5 2. f3 c6 3. b5 a6 4. a4 f6 5.0-0 e7 6. e1 b5 7. b3
0-0 8.h3

15
‘... Magnus was very, very pragmatic in his choice of opening this
time. Unlike the two previous matches, he did not seek to prove to
his opponent that he was stronger in general in all elements of the
chess game, but looked for exactly those types of positions in which
he was comfortable (my italics – MS)... He was well aware that it
was in closed positions of the Spanish type that Ian would at some
point become bored.’ (Khalifman)
8... a5
‘Carlsen "takes liberties". This version of the pawn sacrifice has not
previously been seen in grandmaster practice (admittedly, Alexey
Bezgodov played it a couple of times last century, but before he
became a GM) and, frankly, I doubt that it will be seen much in the
future.’ (Khalifman)
9. xe5 xb3 10.axb3 b7 11.d3 d5 12.exd5 xd5 13. f3 d6
14. f1! fb8 15. xd5 xd5 16. d2 c5

‘Here the opponents began to think for themselves, and it makes

sense to sum up preliminary results. Yes, two bishops and a slightly
better pawn structure give Black almost (an important word) full
16
compensation for the sacrificed pawn, and yet... as should be said in
such cases, a pawn is a pawn, and Black does not have direct active
play. Yes, of course, with best play by both sides, the game should
end in a draw, but it is Black who will have to go through some
unpleasant moments on the way to a draw. I am sure that if Magnus
had White, he would be extremely pleased with this position and
would press his opponent for a long time and sophisticatedly,
although, of course, without a guarantee of victory (my italics –
MS). Ian was also pleased at this moment, but... still, the game was
not quite in his spirit.’ (Khalifman)
17. f3
‘Over the next few moves, the advantage remains around , not
reaching either = nor . There is little point in discussing every
micro-nuance worth a tiny fraction of a pawn.’ (Khalifman)
I would like to draw your attention to the assessment of the position
at move 16, at the end of the home preparation. White has a small
but stable and pleasant plus ( ). It is at this moment that the
‘masonry’ starts. It is interesting to compare this position with that
which arises in a further 16 moves.
17... d8 18. c3 b4 19. ec1 ac8 20. e2 c6 21. e3

17
White is ready to change the pawn structure with the move 22.d4,
after which the position will be more stable. In any event, the scope
for play will be reduced.
21... e7!
High-class masonry! The sixteenth World Champion demonstrates a
deep understanding of the problem of exchanges. He allows the
exchange of the dark-squared bishops, which deprives him of one of
his main trumps, the bishop pair.
22. f4?!
On 22.d4 Black would reply 22... f5, but as Khalifman shows,
White should have maintained the position with 22. g3!? d5
23. d2, keeping the assessment around ‘ ’. If in reply to 22. g3!?
Black exchanges on f3, then after 22... xf3?! 23.gxf3 the
assessment of the position would immediately change to ‘ ’. The
exchange of the b7-bishop for the f3-knight is clearly in White’s
favour – but only with dark-squared bishops on the board .
22... xf3!

18
An important subtlety. In a knight ending with rooks, Black’s better
pawn structure fully compensates for the pawn minus.
23.gxf3 xf4 24. xf4 c6

After a series of exchanges, a position that was slightly better for

White has become equal. Up to now, as we have seen, the
assessment was ‘ ’.
25. e1 f5 26.c3 h4
‘Magnus understands that the situation has changed and he refrains
from simplification with an immediate draw: 26...g5 27. e5 gxf4
28. xf5 xd3.’ (Khalifman)
27. e3 f8 28. g2 f5!?
‘A fighting move. It was slightly simpler to achieve a draw by the
immediate exchange of knights: 28... xg2 29. xg2 cd6 30. d1
b4 31.c4 d4 32. f1 h4.’ (Khalifman)
29. e5 g6 30. e1

19
From Khalifman’s previous comments, we saw that Carlsen had
already begun to look for chances to fight, not being content with a
draw. Try, dear reader, to find the next move of the sixteenth World
Champion. I gave this position as an exercise to young chess
players, and not even all of the grandmasters coped with the task.
To understand how to proceed, you need to ask yourself the
prophylaxis question: ‘What does your opponent want, if it were his
move?’ Then the idea 31. e4 and 32.b4 appears, and our thought
comes to the ideal arrangement of the knight and pawn: the knight is
on e6, and the pawn is on f5. This arrangement of pieces allows us
to control the maximum number of important squares in the
opponent’s camp: d4, e4, f4, g4. At the same time, the move is not
at all striking. After all, on f5 the black knight stands beautifully on
a strong point, whilst on e6 it seems to be ‘hanging in the air’.
Even so, Magnus played:
30... g7! 31. e4 f5 32. e3 e6
It turns out that the knight is one square behind and to the side of the
pawn... in my opinion, this idea did not occur to Carlsen out of

20
nowhere. As far as I know, the sixteenth World Champion studied
the classics seriously and has an amazing memory. I would venture
to suggest that this image was inspired by the classical examples
Saidy-Fischer, Reti-Rubinstein and Fischer-Taimanov. This
interaction between knight and pawn was so firmly rooted in his
mind that this image at the right moment worked as a technique
almost automatically.
Maybe some people would call it intuition. But I think that intuition
is too broad a concept. For me, what Carlsen achieved in this game
is an example of brilliant associative thinking. A chess player with
well-developed associative thinking can quickly and accurately
grasp the essence of a position and find the necessary analogy from
the store of acquired knowledge. This analogy – whether it is
regrouping pieces, changing the rhythm of the game, using
prophylactic techniques, and much more – serves as a reliable
support for making the optimal decision in the face of a constant
lack of time to think.
How to acquire a wealth of knowledge that will evoke the necessary
associations? Truly, high-quality baggage consists of key points that
you yourself managed to find in instructive games and explain
verbally. Specific variations get forgotten, but a vivid landmark
image remains in the memory for a long time. In this work, you and
I will collect just such baggage.

21
Chapter 2
Cooperation between
knight and pawn
Now we will leave behind the game Nepomniachtchi-Carlsen and
look at some similar examples of cooperation between knight and
pawn. In the process, we will revisit some examples from my book
Endgame Strategy
Salo Flohr
Igor Bondarevsky
Leningrad 1939

22
A move earlier, Flohr exchanged his active f8-rook for Black’s
passive e8-rook. But here Capablanca’s aphorism is very
appropriate: what matters is not what comes off the board but what
stays on .
The next important question White should ask himself is: who is
Public Enemy no.1? Of course, it is the pawn on h7. This means that
White’s first aim must be to eliminate the enemy passed pawn.
63. f4 b6

Remember the device used by Flohr in this position:

64. a4!
He exchanges the weakness of b7 for the weakness of a6. Such
moves should be played automatically, without any worries or
calculations.
Mikhail Shereshevsky
Vladimir Malaniuk
Gomel 1981

23
Here White exchanged his bishops for the knights, transposing into
a winning ending:
41. xb4! axb4 42. f5+ e7 43. xd7 xd7 44. e3
The game was adjourned here, but Malaniuk resigned without
resuming. White needs to know only one important technical
device:
44... e6 45. f5 f7 46.d5 g6

24
 analysis diagram
47. h4!.
The analogous idea on the kingside: White exchanges the weakness
on g7 for one on h6 and wins.
Let us now continue our analysis of Flohr - Bondarevsky after
64. a4!

25
64...bxa5 65.bxa5
Now, instead of the weak pawn on b7, Black has a weak pawn on
a6.
65... f7 66. c5 c4 67. g5 e2 68. h6 d8 69. xh7 e7

26
So, White wins a pawn. To realize this advantage, he should employ
the device of ‘schematic thinking’. White’s plan can be formulated
briefly as follows: he needs to transfer the king to b4 and have his
king and knight exchange places. And he must not advance the e5-
pawn!
This plan is fully in accordance with the process of realizing an
advantage, formulated wisely by Boris Gelfand: minimize the
quantity of difficult decisions .
There followed:
70. g6 f1 71. f5 h3+ 72. e4 g2+ 73. e3 f1 74. d2 f7
75. c3 g6 76. b4

Remember this position.

76... f5 77. b7 1-0
Now let us examine another classic endgame.
Anthony Saidy (2336)
Bobby Fischer (2780)
US Championship, New York 1963

27
24... f8 25. e2 e6 26. d3 h5!
White has one easily defensible weakness – the d4-pawn. In order to
have a reason to play for a win, Black needs to try to create a second
weakness for his opponent. Where can he count on such
‘happiness’? Only on the kingside. This means that the black king
must be there. Fischer leads the king closer to the centre on the
square f5 by the shortest route, at the same time preventing the
movement of the white pawn to g4.
27. e3 h7 28.f3 g6 29.a4 f5 30. e2 g5

28
Note that Fischer has achieved the intended set-up by the shortest
possible route, without wasting a single tempo.
31. f2
White lacks any sort of counterplay, but he can quietly wait. Now it
is time for Black to decide on a winning plan.
Let us consider this position. White’s position contains a weakness
on d4, but how can Black get at it? To do so, he needs to remove the
pawn from f3 and obtain control of e4. He can try to do this by
pushing his pawn to g4. But first he needs to fix the pawn on f3 by
playing his own f-pawn to f4. In the process, he can seize more
space. Then he can transfer the knight to f5 and prepare the
breakthrough ...g5-g4. He would prefer in the process not to allow
the opponent to take space with h2-h3 and g2-g4.
But what does Fischer do in this situation? He intends the above
plan of action, but first he observes the principle ‘Do not hurry!’ to
give Saidy a false sense of security.
31... d8!? 32. d2

29
Saidy was probably thinking along the lines of ‘Why did Fischer
play this move? Probably he wants to bring the knight via c6 to a5.
That is probably not so dangerous, but why not prevent it anyway?
So I’ll play 32. d2 and on 32... c6 I can ‘dominate’ the knight
with 33. c3.’ And he did not guess that Fischer’s intention was
rather more prosaic – to weaken his opponent’s attention!
32... g6 33. e3 e6!
Again, we can imagine Saidy thinking, ‘He is going back. His
manoeuvre has failed, aren’t I a star!’
34. d3 f5!
‘Fischer returns the king to f5. He does not know what to do. Soon
I’m going to make a draw with Fischer!’
If only Saidy could have guessed how far from the truth he was!
Fischer simply fooled him using the principle of ‘Do not hurry’.
Saidy might never have heard of this, but the future eleventh World
Champion intuitively understood how to use this principle, having
learned it from Petrosian.
Commenting on a game with Petrosian, Fischer wrote in his book
My 60 Memorable Games : ‘Petrosian likes to play cat-and-mouse,
hoping that his partners will make a mistake when there is no direct
threat. The most paradoxical thing is that they usually do!’
A very vivid, but logically not the most correct description of the
‘Do not hurry’ principle. I would clarify that the principle is most
often used to make a better position out of an almost equal one.
‘Having caught the mouse’, the ‘cat’ immediately gains a decisive
advantage and wins the game.
To continue:
35. e3 f6 36. e2 g6 37. d3 f5 38. e2 f4 39. f2 g7 40.h3
f5 41. d3

30
Fischer could have obtained much the same position within five or
six moves of starting active operations. However, he preferred to
double the number of moves involved, in order to lure his opponent
into a false sense of security. And at this seemingly quiet moment,
the unexpected breakthrough followed:
41...g4! 42.hxg4 hxg4 43.fxg4 h6

31
‘So what happened?’ Saidy probably thought. ‘Fischer, out of
desperation, decided to exchange a couple of pawns. After all, one
way or another there is a dead draw on the board.’
In fact, the game has suddenly come to a critical moment. Now
everything depends on White’s move. There is only one way to
draw. And if Fischer had not applied the ‘Do not hurry’ principle,
then his opponent would have been attentive and careful. And
without White’s help, it will not be possible to win such a position.
Saidy might otherwise have been able to find 44. e2 xg4 (
44... g5 45.g3 ) 45. g1!. Only here! The bishop must control h2
after the white king comes to f3 in the line 45... f5 46. f3. But
White cannot change the move order: if he plays 44. g1? (instead
of 44. e2), then after 44... g5! 45.g3 f3! 46. e3+ xg4!! 47. xh6
xg3 48. e3 g2 Black wins the pawn ending. The same happens
in the variation 44.g5? xg5 45.g3 f3!. But all of these variations
can be found, if the white player senses the danger of losing, as in
that case he is concentrated and attentive. Saidy had been weakened
by Fischer’s previous moves and calmly retreated the bishop to e1.

32
44. e1 xg4 45. d2 f5 46. e1

What must Black play to win here?

Answer: swap the places of the king and knight!
And now, dear reader, try to remember the position from the
possible variation in the game Flohr-Bondarevsky

33
How does White win here? He does the same thing in each case –
swap the positions of king and knight: the knight moves to b7 and
the king goes to c5. This is another example of associative thinking.
We return to Saidy - Fischer
46... f6 47. h4 h5 48. e1 g4 49. e2

49... g3+!
It seems to me that it was only here that White understood that life
can sometimes be very hard, especially when you play Fischer.
After the planned 50. f2 Black wins with 50... f5 51. c3 e3.
50. d3 f5 51. f2 h4 52.a5 xg2 53. c3 f3 54. g1 e2
55. h2 f3 56. g3 e3! 0-1
Black’s next move will be 57... f5, and the f3-pawn promotes.
A few years ago, Boris Gelfand published an excellent book,
Technical Decision Making in Chess . In the first chapter, about
Akiba Rubinstein, he gives the game Réti-Rubinstein, played in
Gothenburg over a hundred years ago. The top grandmaster, who

34
has a highly respectful attitude to his chess hero, gives the following
commentary on the position arising after the move 27... e8:
Richard Réti
Akiba Rubinstein
Gothenburg 1920

28. xe8
‘If White refrains from the rook exchange with 28. d2, then Black’s
best reaction is 28... d7!, which of course is a conception that
Rubinstein did not understand ...’ (italics MS). The great endgame
master Akiba Rubinstein, Gelfand’s favourite player, did not
understand Black’s conception in this ending. This means that
something very complicated is involved, something beyond
Rubinstein’s powerful logic.
28... xe8 29. e1 e7 30. e3

35
30... e6?
‘Rubinstein’s only bad move in this game, which allows White to
offer resistance. Stronger was 30... d7!, fixing the white pawns on
dark squares. Here they will not be objects of attack for the black
bishop, but they will also not prevent the black king penetrating into
the enemy camp, which is White’s greatest problem .’ (Gelfand,
italics added).
When I analysed this ending with FIDE Master Kirill Shoshin, the
young player suggested 31.d4! in reply to 30... d7.

36
 analysis diagram
We could not prevent the transfer of the knight to e3, since the pawn
ending after 31... f5 32. d2 e4 33. d3 ( 33.dxc5!? ) 33... xd3
34. xd3 is probably tenable. Nor does White lose the pawn ending
after 31...c4 32. g2 h3 33. f2 xg2 34. xg2.

37
 analysis diagram
The essence of this ending is that White can’t allow a pawn
blockade on the kingside. The following sample position is lost for
White regardless of whose move it is:

38
 sample position
Black carries out the pawn break ...b5-b4 with his pawn on a5 and
the white king on a2. For this, there is a reserve tempo (with the
pawn on a7).
White, however, should react with counterplay on the kingside. For
example, 34... d6 35. h3! c6 36. g4 or 34... e6 35. f3 f5 (
35...h5 36.g4 ) 36.g4 f6 37.gxf5 xf5 38.h3 with a draw.
31.g4! d6 32.h3 g6 33. d2 d7

34. f3?!
This move does not lead to defeat, but misses the chance to build a
defence based on the scheme suggested by Gelfand, with the aid of
the knight transfer to e3 via g2. The point of Gelfand’s idea is as
follows: 34. g2! d4 ( 34...h5 35. h4! ) 35.cxd4 cxd4 36.c4! dxc3+
( 36... c5 37. e1 b5 38.cxb5 xb5 39.h4 ) 37. xc3 g5 38.f5

39
 analysis diagram
38... c6 ( 38... e5 39. e3 c6 40. c4+! f4 41. d6= ) 39. e3
f3 40.d4 c6 41.a3.

analysis diagram
40
White has managed to build a fortress. It’s a draw.
I recommend everyone to examine this instructive endgame in
Gelfand’s book. Studying his splendid work can only lead to the
conclusion that he is a perfectionist. In my time, when I studied the
classics, nobody dug so deep in search of the truth. At that time,
there were no computers to quickly point out mistakes in long
variations. In his books, Gelfand achieves a good balance between
the wise thoughts of the author and computer variations.
I would like to draw your attention to his note to the move 30... a4-
d7, ‘ which, of course, was a conception not understood by
Rubinstein ’.

But the move 30... e6? instead of 30... d7! is not discussed in
Razuvaev and Murakhveri’s book on Rubinstein, nor is the
counterplay after 34. g2!, instead of 34. f3, looked at there.
So, which concept did Rubinstein not understand? First, we will try
to explain what happened and then I will share with you the
associations that arise with me.

41
Rubinstein played 30... e6? instead of 30... d7!. Note that on the
queenside, White has virtually irreparable pawn weaknesses on the
light squares . The pawn on c2 is hardly likely ever to move. But as
we know, one weakness is rarely enough for the stronger side to
win. He needs a second weakness on the opposite flank.
Gelfand hints that the side with a bishop against a knight can be
given the following recommendation: strive to ensure that the
opponent’s pawns on one flank are fixed on squares of the colour of
the bishop, and on the other flank – on squares of the opposite
colour.
In our example, the bishop on d7 controls a path of light squares for
his king, which is White’s second weakness. It prevents the
opponent from moving the kingside pawns to light squares.
Of course, everything depends on the specific situation, but I want
to share the association that arose when I was analysing the famous
fourth game from the match Fischer-Taimanov.
Bobby Fischer (2760)
Mark Taimanov (2600)
Candidates Match, Vancouver 1971

42
24. e5! b6?
An inaccuracy which has serious consequences. Now all of the
black queenside pawns will be fixed on dark squares and lose their
mobility. Correct was 24... d6!, and White’s advantage is not great.
25. f1! a5 26. c4! f8 27. g2
Only four moves have been made, but Black’s position has gone
from slightly worse to seriously inferior. His rook has to defend the
pawn on f7, his queenside light squares are weak and the pawn on
b6 can be fixed at any moment by means of a2-a4. The only plus for
Black is the centralized position of his king. Much will depend on
whether he can find a good post for the knight.
It was important to understand that the ideal place for it was the
square d6, from where it prevents the white king reaching the
queenside, cooperates with the pawn c5, and defends the f7-pawn.

43
27... d6
The only move. The immediate attempt to transfer the knight to d6
is premature. After 27... e8 28. b5+ d8 29. d5+ c8 30. d2
c7 31. d7+ b8 32. g4 the black pieces are driven into passive
positions and it is impossible to ensure their cooperation. We can
only wonder whether Taimanov in general realized the necessity of
getting the knight to d6. Fischer, on the other hand, paid careful
attention to his opponent’s possibility of bringing the knight to d6
and tried not to allow it.
Instead of 27... d6, it was worth considering the far from obvious
manoeuvre which starts with the move 27... g8!?. Black
immediately protects the square e7 against the entry of the enemy
rook and prepares to bring the knight to e8.
There could follow: 28. b5+ d6 29. e2 d8 30. f3 e7 31.c3
c8.

44
 analysis diagram
The black pieces are more harmoniously placed than six moves ago.
White must reckon with possible counterplay, for example 32. e8
f5 33. e5 c7 34. e2 g5 35. g6 (stronger is 35.fxg5! with the idea
to create a passed pawn with h2-h4) 35...gxf4 36.gxf4 g8 and the
game becomes sharper. In this case, the advantage remains with
White, but Black is no longer a passive observer.
28. f3 d7 29. e3 b8 30. d3+ c7 31.c3 c6 32. e3 d6
33.a4 e7

45
The black knight moved from f6 to e7 in five moves instead of two
from the previous variation. During this time, White centralized his
king, fixed the black queenside pawns with a2-a4, and took control
of the central square d4. But according to the principle of two
weaknesses, in order to win, White needs to create problems for his
opponent on the kingside as well. This would seem easy. He needs
to advance the pawns by g3-g4, h2-h4-h5, then play g4-g5 and g4,
pin the opponent, and then act according to the situation. The plan is
good, but it would force the opponent to think about moving the
knight to c8. After the straightforward 34.g4 c8 35.h4 c7 36. b5
d8 37.h5 f6

46
 analysis diagram
it would not be easy for Black to defend. Please note that all black
pawns are on squares of the same colour.
My engine at depth 43 gives a score of +1.33 in White’s favour,
although, in fact, the endgame is very unpleasant for Black, and
such an assessment is very merciful for him. In this situation,
exchanging rooks will immediately lead to a draw, whereas the rook
or pawn endgame may turn out to be in White’s favour.
But Fischer sought a different position. He intended to place the
opponent’s kingside pawns on the light squares, since on the
queenside Taimanov’s pawns were fixed on the dark squares . This
is exactly the positioning of the pawns that Gelfand called for in the
previous example in the Rubinstein game when he proposed the
move 30... d7!. We will speak in more detail about this in the note
to White’s 35th move.
Now let us analyse the endgame in which all of Taimanov’s pawns
are fixed on dark squares (see the previous analysis diagram). I
analysed this position during lessons with the Azeri women’s team.
47
With the computer’s help, the ladies managed to pose me some
difficult problems. The analysis ran roughly as follows: 38. a6!?.
As we have already noted, the rook ending is lost for Black. Here I
played 38... d6, but after 39. e7+ d7 40. e6 d8 41. b5!, the
black position is very difficult. The rook ending is almost certainly
hopeless, whilst the threat of the breakthrough g4-g5 should decide
the game in White’s favour.
Black probably needs to seek salvation in another rook ending, after
38... d7!? 39. xc8+ xc8.

analysis diagram
Then the play might develop something like this: 40. d3+ (
40. e4? e8+ ) 40... e6 41. e4 b8! 42. d5 c8 43.f5+ e7
44. d3 c4+ 45. d4 c6 46. e3 c5 47. d2

48
 analysis diagram
47... e5+! ( 47... c7 is hopeless because of 48. d4! followed by
49. e4 and 50. d4) 48. d4 d6 49. xc4+ c6.

analysis diagram
49
It’s quite possible that this endgame is winning for White, but Black
has activated his rook at the cost of a pawn sacrifice and retains
definite hopes of saving himself.
For the stronger side, the correct method of playing for a win in
such positions is to exchange the bishop for the knight. And the
weaker side needs to know that the rook ending may be lost for him,
so he must keep a permanent watch on the exchange of minor
pieces.
Associative thinking reminded me of an instructive example.
Hikaru Nakamura (2793)
Anish Giri (2769)
Stavanger 2017

For a long time in this game, the white bishop was stronger than the
enemy knight. Anish Giri had conducted a heroic defence with the
knight, which is significantly weaker than the enemy bishop, with
rooks on. He only needed to meet the threat of 48. d8, going into a
pawn ending. The Dutch GM incautiously played:
47... e7?
50
Correct was 47... c5 or 47... f6. Nakamura immediately
exchanged bishop for knight:
48. xd7!
and, exploiting his space advantage, easily won the rook ending.
Returning to Fischer - Taimanov 34.h3

This move to some extent accords with the principle ‘Do not hurry’,
giving the opponent the illusion that he has done everything
correctly.
34... c6?!
It was better to play 34... c8!, but this thought only seems to have
entered Taimanov’s head when the position had already become
indefensible.
35.h4!
When I first studied this game, I had the impression that Fischer had
persuaded Taimanov: ‘Come on, play ...h6-h5, else I will put my
pawn there and you’ll be under a press.’
35...h5?
51
The threat is stronger than the execution! He should have played
35... a7!?. From this square, the knight aims for c8, as it does from
the square e7, but this time it also controls the important square b5.
The knight move to a7 was also stronger than 35... e7 two moves
earlier. But even after 35... e7 36.h5 c8 37.g4 f6 Black retains
the chance of resisting.
36. d3+ c7 37. d5 f5 38. d2 f6 39. e2 d7

Fischer has managed to lure the enemy kingside pawns onto light
squares, making them an object of attack for his bishop. And on the
queenside, the white king should hunt after the enemy pawns, which
are fixed on dark squares. The path via d3 and c4 is covered by the
white bishop. Fischer only needs to exchange rooks and bring his
king to b5, before the enemy knight reaches d6.
40. e3?!
Fischer slightly lets go of the reins. Move 40. In this match, the
American GM played fast and Taimanov was probably in time
trouble. Correct was to play 40. d2+ c7 41. b5! and gradually
bring the king to the queenside.
52
At this moment, Black could sharply reduce the opponent’s
positional advantage by means of 40... a7!, followed by bringing
the knight to d6 via c8. However, in time trouble Taimanov played:
40...g6? 41. b5! d6 42. e2 d8?!
In a bad position, Black allows the exchange of rooks and he is
immediately lost. It was better to return the rook to f6, although
even in this case, it was extremely difficult to save the game.

43. d3! c7 44. xd6 xd6 45. d3

Black just lacks one small thing – a knight on d6!!
45... e7 46. e8 d5 47. f7+ d6 48. c4 c6 49. e8+ b7
50. b5 c8
Too late. The white king already breaks into the enemy camp.
Admittedly, now he is threatened with mate on d6.
51. c6+ c7 52. d5!

53
52... e7
In the event of the counterattack 52... d6+ 53. a6 e4 54. f7
xg3 55. xg6 e2 ( 55... c6 56. e8+ c7 57. a7 d8 58. c6
c7 59. f3 ) 56. xh5 xf4 57. f3, the bishop easily deals with the
knight.
53. f7 b7

54
Black hangs on with his last breath, but Fischer had long since seen
a decisive zugzwang. Now, with computer-like accuracy, he leads
Black to his inevitable end.
54. b3! a7 55. d1 b7 56. f3+ c7 57. a6 g8 58. d5 e7
59. c4! c6 60. f7 e7 61. e8
I am convinced that Fischer foresaw this position when he offered
the exchange of rooks on move 43, and maybe even earlier. The aim
is achieved. An energetic finish follows:
61... d8 62. xg6!
62. xb6?? xe8 63. xa5 d7 64. b5 d6 65.a5 d5 66.a6
c7+ 67. b6 c4! 68. b7 d7 leads to a draw.
62... xg6 63. xb6 d7 64. xc5 e7 65.b4! axb4 66.cxb4 c8
67.a5 d6 68.b5 e4+ 69. b6 c8 70. c6 b8 71.b6
Black resigned.
Fabiano Caruana (2792)
Richard Rapport (2763)
Wijk aan Zee 2022

55
21. d4?
A tactical oversight in a complicated position. After 21. hg1 g8,
22. d4 would be fully possible, since the continuation 22... xg4
23. xg4 e5 does not bring Black any special benefit. By means of
24. d2 fxg4 25. d3 White could maintain the balance. It is not easy
for Black to defend the pawn on e5, whilst 25...g3?! 26.hxg3 h3 is
unfavourable for him because of 27.g4.
21... xg4 22. xg4 e5!
Probably Caruana missed this tactical resource of his opponent.
23. d2 fxg4 24. d3

56
24... h6?!
A human decision. The computer choice is much more decisive:
24... f5!! 25. xe5 e6! 26. xc6 (if 26. d3, Black simply crushes
the opponent with his pawn mass after 26... xd3! 27. xd3 af8
28. g1 g3 29.hxg3 h3 ) 26... e4 27. d4+ e5 28. g1 g3 29.hxg3
h3 30. h2.

57
 analysis diagram
Despite the opponent’s two extra pawns, the black pieces
completely dominate the board. The computer assessment is -3.22.
Let us continue with the game.
After Rapport’s 24... h6?! , there followed:
25. xe5 e6 26. xg4 e4 27. f2 xc4+ 28. c2 xc2+ 29. xc2

58
The position has simplified and the game has become rather quieter.
Black has the advantage. The main factor is the difference in
strength between the powerful black bishop and the white knight,
with play on both flanks in an open position.
29... e6 30. g1 f8
The attempt to take the game into what seems, at first sight, a
promising rook ending, with the aid of a pawn sacrifice, does not
bring Black success. After a line such as 30...g4 31. xg4 g8 32.h3
xg4 33.hxg4 f6 34. d3, a draw becomes inevitable.
31. d3 g4 32. f4 f5+ 33. d2 d8+

59
34. e2?
The American grandmaster played this after just two minutes.
Maybe he was short of time. But either way, he did not sense that
the final critical moment had arrived.
White needed to think out his defensive scheme thoroughly in this
position. And its basis should be the transfer of the knight to d3 and
the advance of the pawn to e4. After the moves 34. c3 f6 35.e4
c8 36. d3 we could say hello to the games Saidy-Fischer, Réti-
Rubinstein and Fischer-Taimanov. White would have retained good
drawing chances.
After the move 34. e2, played in the game, White’s position
deteriorates sharply. Maybe Caruana could have put up more serious
resistance at some point, but defending such a position against a
strong and ingenious opponent, when short of time, is an almost
impossible task.
34... f6
This is stronger than 34...g3 35.hxg3 g4+ 36. f2 h3.

60
35.b3
The move 35.e4 already fails to 35... e8. Maybe he should have
advanced the b2-pawn two squares, but then he would have had to
reckon with the attack 35... b8 36.a3 c5.
35...a5 36. e1 e5 37. g2
If 37. e2, then 37...g3! 38.hxg3 g4+ 39. f2 h3 is strong.
37... h8 38. f4

The Hungarian grandmaster ends the game in his favour with a

series of tactical blows. Technique is short-range tactics!
38...g3! 39. g2
If 39.hxg3, then 39...h3 40. h1 h2 41. g2 e4 42. f2 f5! with
zugzwang ( 43. h4+ xh4 ).
39... e4 40. d2 h3 41.a3 d8!! 42. b2 f3!
White resigned.
The placement of the knight behind and to the side of the pawn by
one square allows you to control the maximum number of central
squares – four – and greatly limits the opponent’s options.
61
I hope that after working through the games Saidy-Fischer, Réti-
Rubinstein, Fischer-Taimanov and Caruana-Rapport, the image of
this effective alliance will be firmly fixed in your memory and will
help you in the art of manoeuvring using associative thinking.

62
Chapter 3
To hurry or not to
hurry?
The ‘Do not hurry’ principle is one of the fundamental principles of
the endgame. Playing according to this principle requires a high
level of understanding of chess from a player, and it can only be
used when the opponent is deprived of counterplay. We saw it in its
classic form in the game Saidy - Fischer
It is important to keep in mind that the rhythm of play in a chess
game is a very delicate thing. And situations often arise when the
‘Do not hurry’ principle should be forgotten. You are required to
hurry, and hurry a lot!
So as to see how this works in practice, let us return to the first
game of the match Nepomniachtchi-Carlsen.

63
I offer you an exercise. Guess Carlsen’s next move.
Carlsen again demonstrates splendid knowledge of the classics. In
terms of the plan, this game is very reminiscent of the 10th game of
the match Lasker-Capablanca, Havana 1921, in which the Cuban
genius won with Black.
Emanuel Lasker
José Raúl Capablanca
World Championship Match, Havana 1921

64
Black’s position is superior. His rook and knight are more active
than the opponent’s. Try to explain in words the essence of the
position and indicate a scheme for further play.
Answer: White has two pawn weaknesses, but the distance between
them is minimal. Assuming that the white king gets to d2, then the
isolated white pawns will have a reliable defender. In this case,
Black cannot win the game. This means that Black’s task is, first of
all, to prevent the enemy king from centralizing, and secondly, to
attack the opponent’s weak pawns. In other words, Black must
hurry, and hurry a lot! There can be no question of centralizing your
king, improving the placement of the pieces, or any embellishment
of your own position. You need to play extremely specifically and
actively; otherwise, the advantage will gradually evaporate.
From this moment, dear reader, try to guess Black’s moves. It seems
as though with our next move, we should prevent 36. f1 with
35... a6. But Capablanca played:
35... b6!

65
Why? The fact is that after 35... a6?! White has the move 36. b2!.
There could follow 36... a1+ 37. h2 d1 38.b4 xd4 39. xd4
xd4 40.b5 c4 41.b6 c8 42.g3 g5 43. g2 with sufficient
counterplay for a draw. Instead of the rook check, however,
36... d6! would have been very strong.
36. d3
If 36. b2? b4! 37. f1,

analysis diagram
Black will not hurry to win the d4-pawn, because after 37... xd4
38. xd4 xd4 39.b4 c4 40.b5 c7 41. e2 White has every
chance of a draw. Black will play 37... f8!, and if 38. e1 e7
39. d2, then he takes the pawn only now: 39... xd4 40. xd4
xd4+ 41. c3 f4 with a winning position. In the event of the
white e-pawn advancing, the black king will blockade it.
This device of blockading the enemy passed pawn with the king is
worth remembering.
Boris Maryasin

66
Peter Korzubov
Minsk 1982

Here is an analogous case. In this position, Boris Maryasin correctly

placed his rook on the third rank:
1. c3!
Why didn’t White support his pawn from behind as advised by
Tarrasch with 1. a2 ...? After all, if 1... a5? 2.a4 xe5 3.a5 d5
4.a6 d8 5.a7 a8 6. e2 f8 7. d3 e7 8. c4 d7 9. b5 c7
10. a6 Black loses.
Korzubov would have replied 1... a4! 2. e2 f8 3. d3 e7
4. c3 d7 (premature is 4... e4 because of 5.a4 d7 6.a5 c7
7.a6 b8 8. b2+ and 9. b7(+)) 5. b3 e4. Now is the critical
moment.

67
 analysis diagram
The black king can blockade the a-pawn. The position is drawn.
Let us return to Lasker-Capablanca.

68
Now it seems to be time for the advance of the enemy king into the
centre, but Capablanca played:
36... a6!
The white pieces are tied down to defence. He needs to bring his
king into play.
37.g4 hxg3 38.fxg3 a2 39. c3 c2?! 40. d1?
40. b5! was more tenacious, which is why 39... a1+ 40. f2 d6
would have been the most precise.

The key moment of the game. Black has firmly seized the initiative,
but White defends stubbornly. Emanuel Lasker was a splendid
defender. Black has to find a way to strengthen his position without
allowing any counterplay.
Capablanca finds a superb manoeuvre, improving the position of his
knight. From the previous examples, you have probably already
realized that we are talking not about the lovely square e4 but about
the modest though key square c6, behind and to the side of the
central pawn d5.

69
40... e7!
This is not one of the moves recommended by Stockfish, but it fully
accords with the essence of the technique of realizing an advantage
recommended by Gelfand – minimize the number of difficult
decisions.
So, once the black knight reaches the ideal square c6, White will not
be able to prevent the attack on the b3-pawn by the black rook from
b1 and the knight from a5. After winning the pawn on b3, Black will
not have to calculate variations, and so risk making a mistake.
41. e3
The move 41.b4 leads to the loss of a pawn after 41... c1 and
42... b1.
41... c1+!
Capablanca understood every bit as well as us that it is undesirable
to let out the white king, which is cut off by the rook on the second
rank. But after 41... b2 White has the strong reply 42. c3, not
allowing the black knight to the square c6.
42. f2 c6 43. d1 b1!

70
The b3-pawn is doomed. The appearance of the black knight on a5
cannot be prevented.
44. e2?
Lasker was twenty years older than Capablanca and the match in
Havana turned out very badly for him. He was not only inferior in
playing strength to the Cuban, who was in his prime, but was also
worn out physically. This serious mistake in a difficult position was
the result of these factors. After the correct 44. e1 a5 45. d2
xb3 46. xb3 xb3+ 47. c3 a5, White retains some chances of
a draw, although I do not believe that he would have saved the
knight ending a pawn down.
44... xb3! 45. e3 b4

71
Black has won a pawn and retained the rooks. Now the rhythm of
the game sharply changes and the strength of the ‘Do not hurry’
principle grows. Let us see how calmly and quietly Capablanca
plays the next seven moves.
46. c3 e7 47. e2
47.g4 would have been met by 47...g5.
47... f5+ 48. f2 g5 49.g4 d6 50. g1 e4+ 51. f1 b1+
52. g2 b2+ 53. f1 f2+! 54. e1 a2
I would like to draw your attention to this point. It is the ‘Do not
hurry’ principle in action. For Capablanca, there is nothing that is
too small to bother about. If the rook is a little better on a2 than on
b2, then you should not be lazy and you should place it there. After
all, White has practically nothing to do. The move 55. f3 loses to
55... f2, and 55. e2 to 55... a1+.
Why is it better for the black rook to be on the a-file than on the b-
file? Let’s assume that Black wants to bring his king to d6. In this
case, with a black rook on b2, White could play a3, with slight

72
signs of counterplay. With Black’s rook on a2, White does not have
b3 due to the fork ... d2+.
55. f1 g7
It is interesting to note that for 20 moves in this ending, Capablanca
did not make a single king move. This is an extremely rare thing in
his games.
56. e3 g6 57. d3

White has no moves and can barely hold on. Most likely, at this
moment Capablanca saw the final scheme of action, involving the
advance of the f-pawn one square and the transfer of the king to d6,
followed by ...e6-e5. After such a development of events, Black will
have everything under control and no difficult decisions will be
required of him.
57...f6! 58. e3 f7 59. d3 e7 60. e3 d6 61. d3 f2+!
62. e1 g2 63. f1 a2 64. e3 e5!

73
The advance of the e-pawn is more effective if the white rook is on
e3. This is precisely what was forced by the manoeuvre 61... f2+!,
62... g2 and 63... a2.
Once more I repeat: when realizing an advantage, never
underestimate the importance of trifles .
65. d3 exd4 66. xd4
Nothing is changed by 66. e2 a1+ 67. g2 c5 68. xd4 a2 or
68. xd4 c4.
66... c5 67. d1 d4 68. c1+ d5 69. d1 g3+ 70. e1 g2 0-1
A wonderful example of a human approach to technical problems!
Let us examine a modern example in which the question ‘To hurry
or not to hurry?’ is especially relevant.
Aleksandra Goryachkina (2504)
Ding Yixin (2432)
World Women’s Team Championship, Astana 2019

The leaders of the Russian and Chinese women’s teams reached this
position soon after the opening. We will present here the comments

74
of Alexandra Goryachkina, published in the magazine 64 No. 2 for
2019:
‘Here I thought for a long time. It would seem that White has so
many advantages, but there is too little material left, Black still
holds all the pawns, and the knight needs only one tempo to enter
the game. For myself, I understood that as soon as the knight makes
one move, the position will be equalized if White does not have time
to win anything. Therefore, you need to attack with every move, so
that Black does not have time for this.’
The strongest Russian chess player very clearly expressed the idea
that White needs to play very specifically here and to hurry. She
managed to find a series of precise and beautiful manoeuvres that
ended in a disguised trap. If you, dear reader, want to practice
developing the initiative, then think for ten minutes and try to find
her moves.
Answer: 26.Re1!
26. c7+ f6 27.f3 d8 28. b3 d1+ 29. f2 a6 30. c6 b1
31. xb6 b2+ 32. g3 xa2 is drawish.
26... e8 27. f3 f6 28. d2! d8
After 28... c6?! 29. e4+ g6 30. b1 b8 31. b5 Black’s position
is unpleasant.
29. e4+ g6 30. c3
White does not stop attacking. The continuation 30.f3 leads to a
draw after 30... a6 31. b1 c5 32. xb6 xe4 33.fxe4 and now the
most forcing line is 33... d1+ 34. f2 d2+ 35. f3 xa2 36. xe6+
f7 37. b6 g4+ 38. g3 e2.
30... f7
A human move, although the engine prefers 30... f5 or even
30... f6.
31.f3 d3

75
Here the move 31... a6?! runs into the unpleasant 32. b1.
32. e4

The pawn on g5 is under attack. What is the instinctive move for

Black? Correct – 32... g6. The Chinese player played exactly that
and thereby fell into a hidden trap! Do not forget that she still had to
make nine moves to the time control and, very probably, was short
of time.
32... g6?
After the retreat of the rook to d5 followed by 33... d7, the game
would probably have ended in a draw, but even so, Goryachkina’s
last few moves deserve the highest praise. She has consciously done
everything that she could to squeeze blood from a stone.
33. f2! d6 34. e5!
It turns out that the threat of 35. xg5+ wins White a pawn.
34... c6
The natural reaction. The engine, of course, defends better than the
human and suggests the cold-blooded 34... f7. There could follow:
76
 A) 35. xg5 d2 36.a3 a2;
B) 35. e4 d1+ 36. f2 d7 37. xg5+ ( 37. xg5 a1= )
37... f6 38. b5 d2+ with a draw;
C) 35. b5 d7 36. e4 d1+ 37. f2 g4! 38.fxg4 d4 39. e3
a4 40. b2 c5, and Black should survive.
But such play requires calmness and thinking time, and the Chinese
player had neither one nor the other.
35. e4 d7?!
Only after this natural move do Black’s chances seriously
deteriorate. The computer suggests 35... c1+ 36. f2 c2+ 37. g3
d7! 38. xe6+ f6!, but in time-trouble it is very difficult to
appreciate that this position is defensible.
How do we place Black’s idea in question? After all,
36. xg5+ f7

gives the impression that White cannot avoid the exchange of the
queenside pawns with a probable draw. However, Alexandra
managed to find a way out of the situation with the help of a tactic:

77
37.a4! c1+ 38. f2 a1 39. g4!
As we know, technique is short-range tactics!
39... e5
The rook ending after 39... c5 40. xc5 bxc5 41. e3 is most likely
lost for Black.
40. d6+ e7 41. d4 a2+ 42. g3 c2

White already has a healthy extra pawn, but she still needs to watch
out for the exchange of the queenside pawns. How can she
strengthen her position and begin to slowly push back her opponent?
Let’s try to visually evaluate all the white pieces. The rook and
knight are positioned actively and interact well, but they lack pawn
support. They seem to ‘hang in the air’. White’s king is not too
active and is busy defending his pawns, but in the future she is ready
to support their movement forward since she has an extra pawn on
the kingside.
Now let’s talk about Black. All of her pieces are perfectly
positioned and interact well. But there are no clear guidelines for

78
further actions in the position. She will have to manoeuvre her
pieces based on the opponent’s actions.
It is not easy to verbally formulate the essence of the position, but
let’s try to do it. In my opinion, White needs to try to worsen the
interaction of the opponent’s pieces and not take drastic active
actions for the time being. Black’s task is easier to describe, but
more difficult to accomplish: she needs to catch the moment the
opponent’s active actions begin and react to them correctly.
One can sum up the position in the diagram as follows: White’s
winning chances are roughly as good as Black’s drawing chances.
Whoever plays better will achieve her aim. That is, it is all a
question of masonry – she who makes the stronger average move,
wins.
43.h4!? g6!? 44. h3!? c1!?
For the moment, the opponents are well matched. White tries to
seize space on the kingside and Black reacts accurately.
45.f4
Goryachkina immediately goes over to active operations.
The three great Ks – Karpov, Kramnik and Carlsen (Translator’s
note: in Russian Cyrillic, Carlsen’s name is written with a K, not a
C!) – most likely would not have rushed with this move, but would
have found a way to ‘hold’ the position and test the quality of the
opponent’s average move.
Here I would like to quote from Mark Dvoretsky’s notes to the
game Karpov-Pomar, Nice 1974: ‘Often experienced masters try to
"hold" a position without defining or dramatically changing its
design (and forcing events only when this promises obvious
benefits). In uncertain situations, the probability of an enemy error
is higher. This is where one of the most important principles of the
technique for realizing an advantage comes into play: "Don’t
hurry".’

79
But it is also possible to play 45.f4, as Alexandra Goryachkina did.
45... c3?!
The rook on c1 occupies an excellent position and should not have
been moved to a worse one. The engines correctly suggest that
Black immediately retreat the knight to f7 or d7, with good drawing
chances. After all, the advance of the pawn to f4 weakens the third
rank, and Black can try to exploit this to transfer the knight to f5
after, say 45... f7 46. e4 h6.
46.g3 d7

47. c4!
In my view, if White had tried to transfer his knight to e5, then
Black could have saved the rook ending after 47. c4 b3 48. e5
xe5 49.fxe5 e3!. The only move. White has established a pawn
on e5, squeezing the enemy position. If it is not removed at once,
then after 50. g4 or 50. e4 Black loses. For example, 49...b5?
50. g4 with a winning position. But after 49... e3 50. b4 xe5
51. xb6 e4! 52.a5 a4, Black maintains the balance.

80
47... d3 48. c8+ d8 49. a7

Up to now, Black has defended relatively comfortably, without

making any serious mistakes. All of her moves have been of a fairly
high level. But here she commits a serious positional mistake:
49... e3?
The quality of this move is significantly lower than the previous
ones.
It was necessary to quietly play 49... e7!, and White has not
achieved much.
Try, dear reader, to play like Goryachkina and exploit Black’s
mistake.
Answer: 50. c6+ e8 51. d4!
The end! The coordination of the black pieces is hopelessly
disrupted. Now 51...e5 52. xe5 xe5 53.fxe5 xe5 is no good
because of 54. d6, while White herself threatens to transfer the rook
to d6 and play e5.
51... f6 52. d8+ f7 53. d3! e4 54. d4 e3 55. e5+ g7
56. d3! e4 57. d6 d5 58. xe6 xa4
81
The pawn ending after 58... xf4 59. xg6+ h7 60.gxf4 xf4+
61. g4 is hopeless for Black.
59. xg6+ h7 60.h5
The result of the game has long been obvious.
60...b5 61. g4 a7 62. d6 e3+ 63. f3 f5 64. d3 b4 65. e4
Here is an interesting note by Goryachkina on the topic of
minimizing difficult decisions:
‘I saw the variation 65.g4 b3 66.gxf5 b2 67. d1 a1 68. d7+ g8
69.f6, but did not wish to test myself this way. After 69...b1
70.f7+ g7 71.h6+ White wins.’
65... g7 66.g4 e6 67. d6 c5+ 68. d4 c7 69.g5 a4 70.g6+
g7 71. g4
Black resigned.
It is time to return to the game Nepomniachtchi-Carlsen . After the
sixteenth move, when the players had just emerged from the
opening, the assessment was ‘ ’

position before 17. f3

82
 position after 33. g2
In the second diagram position, the knight on e6 hangs in the air
somewhat and the instinctive move is 33... f7. But then the white
king would move towards the b3-pawn and take it under guard. By
analogy with the game Lasker-Capablanca, Black needs to hurry:
33...b4!
And within a couple of moves, Carlsen had regained the b3-pawn.
Remarkable feeling for the rhythm of the game!
If in the first diagram the assessment is ‘ ’, then in the second it is
closer to ‘ ’, than ‘=’. Yet Ian had not made a single serious
mistake. The reason for this change probably lies in micro-
inaccuracies and the imperceptible error 22. f4. To my mind, this
endgame is a striking demonstration of the difference in average
move strength.
Let us look at the rest of this game.
34. e2 b8 35. d2!

83
Correct. Bad was 35. b1? a5 36. d2 bxc3+ 37.bxc3 a4 38. c2
axb3+ 39. xb3 d4+!, winning the exchange and the game.
35...bxc3+ 36.bxc3 xb3 37. c2 b7 38.h4

At this moment, an anxious mood reigned on Russian-language

websites, sometimes turning into panic. Black had a wonderful
opportunity to play 38...a5!?. Now 39. xa5 is losing: 39... d4+
40. d1 b1+ 41. d2 cb6! (perhaps Magnus simply did not notice
this move) 42. a2 b3+ 43. c2 c1+ 44. b2 f1. Instead of
39. xa5?, correct is 39. ee1 cb6 40. a2 a4 41. e3 a6 42.h5 f4
43.hxg6 hxg6 44. d1 d7 45.d4 cxd4 46. xd4 xd4 47.cxd4 with a
quick draw.
It seems that the main cause for alarm was the impression of the
apparent ease with which the World Champion outplayed the
challenger. In fact, White’s margin of safety turned out to be quite
large. After Carlsen’s quiet move
38... f7
the game was drawn without undue adventures:

84
39. ee1 f6 40. e3 d7 41. c4 e7 42. e5 d6 43. c4 c6
44. e5 d6 45. c4
Draw.
So, what unites the three seemingly dissimilar games in this
chapter? The rhythm of the game is an association much more subtle
than the pawn structures. It is primarily about prioritizing. In all of
these examples, the main priority was to thwart the opponent’s
plans. And for this it was necessary to hurry, and hurry a lot.
The games Nepomniachtchi-Carlsen and Lasker-Capablanca can be
called ‘twins’ in their ideological essence. In both games, Black
could not afford to lose a single tempo, otherwise the white king
would have time to defend the pawn weaknesses on the queenside.
Likewise, Goryachkina very clearly understood in her game that the
slightest delay could lead to a draw.

85
Chapter 4
Benchmarks and
errors
Attention to trifles leads to perfection and perfection is no longer a
trifle. – Michelangelo
There are different periods in a chess game. Often you have to
defend a slightly worse position, look for chances to develop the
initiative in an equal situation, test your opponent’s ability to hold
the position – and there is no poetry in this. All this resembles
menial work, and many young chess players are not interested in
doing it. But without high technique in these elements of the game,
you won’t be able to become a great chess player, because the
strength of the average move is determined by your mastery of all
the elements of chess.
Let’s take a look at how future World Champions Magnus Carlsen
and Vladimir Kramnik demonstrate top-class skills with calm,
seemingly unimpressive actions and thereby outplay decent
grandmasters. We start with the defence of inferior positions by
looking at two more games from the match Carlsen-
Nepomniachtchi:
Ian Nepomniachtchi (2782)
Magnus Carlsen (2855)
World Championship Match, Dubai 2021 (3)

86
To reach this position, in Game 3, Black had to solve some difficult
problems. Most of Nepomniachtchi’s opponents, even those in the
top 5 or top 10 by rating, manage to solve such problems only once
in a while. The 2022 Candidates Tournament in Spain is indicative,
where Nepomniachtchi had five wins and six draws after eleven
rounds.
It seems Black will have to suffer a lot to achieve a draw in this
game. But Magnus acts like a computer in these positions, if not
better.
28... ab8 29. b1 f6 30. xb8
As Khalifman pointed out, equality results from 30. e2 h5 31. xb8
xb8 32.h4 f5 or 30... f5 31. xf5 xf5 32. xb8 xb8 33. e3.
30... xb8 31. b1 xb1+!
‘The simplest. Black’s position can be held relatively easily with
rooks on too, but why bother?’ (Khalifman)
32. xb1 e5 33. e2

87
33...f5!
The opponents put a lot of strength into this game. Black has almost
achieved a draw and it seems he can relax. However, if White is
allowed to play 34. e3 and then 35.f4+, then Black may have
problems. Magnus always asks himself his favourite prophylactic
question: ‘What does the opponent want if it were his move?’ The
black pawn wants to move to f4 and take the e3-square under
control.
34. c2 f4
It was possible to agree the draw already, but the rules required
waiting until move 40.
35. b1 c5 36. c2 d7 37.f3 f6 38.h4 e5 39. f2 f6 40. e2
e5 41. f2 ½-½
In his next Black game, Carlsen In his next Black game, Carlsen
changed the move order compared with Game 3. However, Black
did not manage to avoid problems. On move 20, White played too

88
slowly and Black immediately exploited this inaccuracy and
managed to go into a slightly inferior endgame.
Ian Nepomniachtchi (2782)
Magnus Carlsen (2855)
World Championship Match, Dubai 2021 (5)

White’s pieces are more active than his opponents’ and Black needs
to play accurately to hold the position. If Carlsen had the white
pieces, then he would doubtlessly have been able to squeeze
something from the position. But here he is Black, and he decides to
establish an unbreachable fortress:
26... b8 27. a6 e8 28. f1
If 28. f5, then 28...f6 and 29... f7.
28... f8! 29. f5 e6 30. c4 d8!
I have run training sessions on the subject of ‘masonry’ with many
promising young players, including GMs. In these sessions, I
suggest that they play the black pieces here, starting from move 27.
They all find the moves 27... e8!, 28... f8! and 30... d8! very

89
difficult. Of course, at the time they studied this World
Championship Match, but did not pay enough attention to these
moves and had to search for them afresh.
As dsaid, Carlsen suffered some unpleasant moments and used a lot
of energy to go into an inferior endgame. But then he played
faultlessly, demonstrating the highest class in his assessment of the
positions. It is to such small and not very interesting moments that
one must pay attention, so as to improve one’s technique.
31.f3
‘Both here and subsequently, there are no significant improvements
to be suggested for White. 31. g3!? f6 32.h4 f7 33. e2 f8
34.h5 d6 35. xd6+ xd6 36. f5 f8 37.g3 b8=.
31...f6! 32.g4
32.h4 h5! ( 32... f7 33.h5 f8 34. e2 d6 35. cxd6+ xd6
36.g3= ) 33.g4 g6 34. h6+ g7 35.gxh5 gxh5=.’ (Khalifman)
32... f7 33.h4 f8 34. e2 d6 35. cxd6+ xd6 36.h5 f8

The construction of the fortress is completed.

90
37. a5 e8 38. d5 a8
Of course, not 38... xd5?? 39.exd5 with a decisive advantage to
White.
39. d1 a2+ 40. d2 a1 41. d1
A draw, achieved on the basis of ‘masonry’ of the highest level.
At the press conference after the fifth game, Magnus said: ‘I don’t
mind defending a fortress like this at the end of the game. Yes, it’s a
bit of a perverted taste, but I like to do it from time to time.’
When analysing games of outstanding grandmasters, young players
most often do not pay enough attention to such a low-key, ‘boring’
style of play. But in a confrontation between rivals who are close in
strength, rapid attacks on the king or combinational positions, where
imagination and the ability to calculate variations come to the fore,
do not always happen. Very often, the fate of the game begins to tilt
in someone’s favor during a manoeuvring struggle in a relatively
calm and approximately equal position. It is in such situations that
the strength of a chess player’s average move clearly manifests
itself.
In my opinion, defending a worse position against a skilled and
motivated opponent is the most difficult skill in chess, and one that
not all world champions have mastered. Only a few of the world’s
chess elite are able to identify from afar the opportunity to build an
impenetrable fortress and accurately foresee the active capabilities
of an opponent in the nervous atmosphere of a tournament game. At
the same time, as grandmaster Lev Psakhis objectively and aptly
noted, most often a chess player either defends well or succeeds in
exploiting a small advantage. Magnus Carlsen combines both of
these unique qualities in his playing.
Let’s take a look at Carlsen’s masonry in superior positions.
Magnus Carlsen (2765)
Pavel Eljanov (2692)
Wijk aan Zee 2008
91
The position is almost equal, with a typical pawn structure in which
each side has a pawn advantage on a flank. White’s pieces are
positioned a little more actively, he has a slight spatial advantage
and Black’s light squares are slightly weakened. But all this is too
insignificant to justify immediate action. For example, after the
specific attack 23. b5, Black has the replies 23... ed8 or 23... ad8,
which do not promise White any real benefits.
In such situations, it is important for the stronger side not to force
the play, but to hold the position in order to wait for the most
opportune moment to become active. This is powerful masonry, and
it is more difficult to learn than beautiful tactical strikes or long
calculations of variations.
23.g4!
A typical device, which we have already seen many times.
23... e4
Possibly not the best decision, but I will not criticize it. There is
definitely logic in it. White’s activity will sooner or later be
associated with a knight transfer from d4, but then the black rook
92
gets the wonderful square a4. If we keep the rooks on d8 and c8,
then White will put his king on f3 and the pawn on h3 or h4 and
then transfer the d4-knight via e2 to d5. In any case, Black will have
to contend with his powerful opponent in terms of the strength of
the average move.
24. f3 ae8

Try to guess Carlsen’s next move.

Answer: 25.h3
There is nothing special in this move. It is simple, but players in the
region of 2300-2400 Elo almost never play such moves.
Now what should Black do? The best thing is to wait and see what
the opponent will do. But many players find this difficult
psychologically. They don’t like to mark time and prefer to do
something, or exchange something. But their opponents sometimes
do not permit this, and that is when mistakes are likely to occur.
When there are no clear points for orientation, many players get
nervous and start playing below their normal level.

93
25...f6?!
Black wants to bring his king closer to the centre, but in doing so he
seriously weakens his kingside pawn structure. He should have
bided his time and not made such committal advances. It was more
sensible to play something like 25...a6 or 25... a4 26. a1 c5
27. c3 a4 28. e1 c5 29. c2 h6, adopting waiting tactics.
Magnus would probably have found some way to force his opponent
to make some important decisions, but Black had no need to
voluntarily weaken his kingside like this.
26. a3 f7 27. c2!
Carlsen does not hurry. He plays another quiet but subtle move,
covering the a2-pawn, after which the threat of 28. b5 is in the air.
27... a6?!
Eljanov wants clarity. With the pawn on f6, it was already difficult
to bide his time.
28. xf8 xf8 29. c6
At the first opportunity, White occupies the square c6 with the rook.
It was also possible to put the knight here. White does not wish to
allow a four-rook ending after 29... b4, but after 30. xb4 xb4
31. c6 White should be winning.
29... g7 30. b5 4e7 31. dd6 c5!?
A sharp response. Black defends with the help of tactics.
32. c7
Of course, not 32. xf6?? because of 32... d7.
32... f8

94
Again, an interesting point. White has a tactical opportunity at his
disposal that leads to winning a pawn. By 33. xc5 bxc5 34. e6+
xe6 35. xe6, he could get into a nice rook ending. Black will lose
the a7-pawn after both 35... b8 and 35... c8. But we know that in
rook endings the weaker side always has a chance to draw. For
example, in the case of 35... b8 36. c6 b6 37. c7+ g8 38. xa7
c4 39. c7 a6 40. xc4 xa2, Black’s defence would be difficult,
but the position is most likely drawn. However, Carlsen does not
give such chances to the opponent if he has an alternative option at
his disposal that allows him to maintain the tension in a better
position.
I remember how, during classes with the strongest young chess
players in Russia at Sirius, Vladimir Kramnik described the
difficulties of defending against Carlsen. He said something like
this: ‘He never lets go, doesn’t allow you to simplify and move into
a difficult but drawn endgame a pawn down. It seems like it’s a
draw! But when you start to delve deeper into the position, you see
that he has the opportunity to avoid drawing simplifications.

95
Defending against Carlsen takes a lot of energy and nerves! You get
tired and start making mistakes’.
In the ending against Eljanov, Magnus avoided the simplifications:
33.h4! ff7 34. d5 d7 35. xd7 xd7

Again, an interesting situation. There is a threat of the check

36... e5+. White needs to move his king from f3 but to which
square: f4 or g3? This is a difficult question.
36. g3!
Masonry! After Carlsen’s move, my engine assesses the position as
+1.52, whereas after 36. f4 the assessment is +1.03.
Of course, each of the two king moves can be explained logically.
For example, from g3, the white king does not prevent the
movement of the f-pawn. But on f4 it is closer to the centre, and
king centralization is a fundamental principle of the endgame. But it
is unlikely that great chess players are guided by too general
reasoning. They simply feel such moments, thanks to their well-
developed intuition.

96
When I analysed this episode, I had an association with a fragment
of a game Caruana-Gelfand, analysed in Boris Gelfand’s book
Technical Decision Making in Chess .
Fabiano Caruana (2697)
Boris Gelfand (2739)
Amsterdam 2010

In this position, Gelfand played 36... b5 instead of 36... c4!. This

is what he wrote about it: ‘The move in the game is not a mistake,
but I could have played the more precise 36... c4!, so as to put the
pawn on b3, not giving the opponent the square c3. I had to see this
half-move variation.’
Gelfand goes on to give a wonderful example of masonry: ‘The
issue is not one of what you see or don’t see, but of how often you
play the correct move. Carlsen plays the right move a bit more often
than everyone else and this is why he is number one. Nobody can
play the right move all of the time’.
Let us return to the game Carlsen - Eljanov . I would remind you
that Carlsen played the correct move 36. g3!

97
Eljanov returned the knight to the secure outpost:
36... c5 and Carlsen played 37.f3! , strengthening his position in
the centre and on the kingside, and at the same time offering his
opponent the chance to change the character of the struggle.
37...h6?!
I would speculate that Black was already short of time. He should
not have weakened the g6-pawn. This is easy to say, but to put
oneself in Eljanov’s shoes and suggest something better is not so
simple. As one of my students aptly noted, ‘in such positions, it
would be better to be able to pass’. In case of 37...f5, there is the
unpleasant move 38.g5. Black would do best just to wait and see
what his opponent does, after 37... b7, but psychologically this is
very hard to do. Much the same situation arose after 25.h3!, but then
the situation was rather more favourable for Black.
38. f4!
White immediately starts playing against the weakened pawn on g6.
38...g5

98
Little is changed by 38... h7 39.h5.
39. h5+ g6

Here the engine recommends the prophylactic move 40. h3, not
allowing the variation 40.f4 e4+ 41. f3 e7 42. e6! xe6 43.f5+
h7 44.fxe6 d6 45. xf6+, although outwardly this position looks
winning for White. I do not want to get tangled up in the thickets of
analysis, but from a human viewpoint the move 40. h3!?? instead
of
40.f4
,
as played, looks like going a bit too far in decentralizing the white
king, although it does enable the manoeuvre h5-g3-f5. But let us
not forget that we are talking about the 40th move.
40...gxf4+?
Now it is all over for Black. Good or bad, the move 40... e4+ was
his only saving chance.
41.exf4 h7
99
Here already, 41... e4+ 42. f3 e7 does not help Black because of
43.f5+ f7 44. f4, with the threat of 45. e6.
42.f5
Not falling for the beginner’s trap 42. xf6+?? xf6.
42... g8 43. f3 d7 44. e4
Now it is time to centralize the white king. Black’s position is
absolutely lost and Carlsen easily wraps things up.
44... f8 45. c8+ e7 46. d5 b5 47. h8 b6+ 48. c6 c4
49. a8 e5+ 50. c5 d7+ 51. xb5 d6 52. xa7 f8 53. b4
c5 54. c4
Black resigned.
The analogy for play in this pawn structure may come from a 1994
Bundesliga game by another World Champion, Vladimir Kramnik.
Vladimir Kramnik (2725)
Christopher Lutz (2580)
Germany Bundesliga 1993/94

29. d1 c5 30.g4
100
A familiar motif.
30...g6
A logical move. It is important for Black to drive away the
dominating knight. 30... c8?? is impossible because of the check on
e7. But even if the black king stood on f8 or even e8, the rook move
to c8 would be impossible; for example, 30... f8 31.h4 e8 32.h5,
and 32... c8 is not possible due to 33. xa7 a8 34. c6 xa2
35. d8#. Therefore Black preferred the move 30...g6, after which
the kingside pawns on the sixth rank become more vulnerable.
31. g2 g7
Now Black is ready to play 32... c8, since the check on e7 is not
possible, whilst after 33. xa7 there follows 33... a8.
It is important to note the following: Black, in principle, is ready to
exchange the pawns on the queenside (two for one!), but Kramnik is
not. Therefore he avoids the move 32.a3, which would allow the a3-
pawn to be defended from b5 after 32... c8 33. xa7 a8 34. b5.
In a lesson with the strongest young Russian players, he pointed out
that the pawn on a2 should only be advanced if absolutely
necessary.

101
32. d2!
How this reminds us of the move 27. c2!
Carlsen played against Eljanov

27. c2!

102
Back to the game Kramnik-Lutz.
32...a6 33. d6!
White skilfully maintains pressure on the opponent’s position. A
critical moment is approaching in the game. What should Black do?
Should we start active play with 33... e4, regardless of the loss of
the f7-pawn, or continue relatively passive play, making moves like
33... c8...? It is always very difficult to answer such a question. In
such situations, the probability of mistakes increases sharply even in
games of the strongest chess players in the world.
It is not surprising that the German GM did not sense the
importance of the situation and played:
33... c8?!
The engine recommends 33... e4! 34. d4 e8 35. d7 c8 36. e5
( 36. d4 c3 37. e6+ f6 38. d8 c2 39. xf7+ e5 40.a4 b5= )
36... f6 37. xf7 c2 38. xh6 xf2+ 39. g1 xa2 40.h4 b5
41.g5+ e5 42. e7+ d5 43.h5 a1+ 44. g2 a2+ with a draw.
Calculating such variations and assessing the sides’ chances is very
difficult, indeed practically unrealistic.
Kramnik replied:
34. d4!
It is important to be able to find and verbally formulate the purpose
of the prophylactic moves of great chess players. It is no
coincidence that we marked this, the previous and next move of
White with exclamation marks. The move 32. d2! prevents
32... c8; 33. d6!, protects the knight on c6, which the opponent
was just about to attack, and 34. d4! repels the attack 34... e4.
We are watching how Vladimir Kramnik constantly plays against
his opponent’s next move. This style of play gives excellent results
and is the most important element of powerful ‘masonry’.
34...b5?!
34... a4! .
103
35.h4!

Over the last two moves, the situation has changed dramatically.
Black stands badly. And if you consider that the German
grandmaster’s opponent was the rapidly progressing Vladimir
Kramnik, then he is simply lost. Black has missed the opportunity to
launch a counterattack, and White has many opportunities to
improve his position on the king’s side and weaken the opponent’s
pawn configuration on this part of the board. The Kramnik masonry
is far superior to the average Lutz move. The decisive mistake
followed immediately:
35...b4?!
and after
36. b6 a5 37. b5
White won a pawn. The realization was short and precise:
37... d3 38. xa5 e1+ 39. g3 c2 40. b3 a3 41. a4 c4
42. d4! c2 43. f3 c5 44. a7 g5 45.h5 g8 46. d2 a3
47. e4 c2 48. b7 xa2 49. xb4

104
Black has managed to exchange all of the queenside pawns, but on
the other wing, his structure is fatally compromised.
49... c2 50. b6 h7 51. b7 g8 52. d6 c6 53. xf7 c4
54. d7 f6 55. d4
Black resigned, since after 55... xe3 White keeps two extra pawns
with 56. xh6+.
Now let’s look at another game in which two good grandmasters of
comparable playing class met. If you remember, when we were just
getting acquainted with the concept of ‘masonry’, we said the
following: ‘...
Most often, chess players of the level 2650-2750 come out of the
opening with approximately equal chances, having a wide variety of
playing opportunities. And although the position remains
approximately equal for some time, the advantage gradually comes
to the player whose average move is stronger. He takes the initiative
and begins to outplay his opponent.’
This is exactly what happened in the next game.
Sicilian Defence
105
Maxim Chigaev (2615)
Maxim Matlakov (2685)
Russian superfinal, Moscow 2020
1.e4 c5 2. f3 d6 3.d4 cxd4 4. xd4 f6 5. c3 a6 6. e3 e5 7. f3
e7 8. c4 0-0 9.0-0 c6 10. b3 h6 11. h4 a5 12. g6 xb3
13. xe7+ xe7 14.axb3 b5

We leave the opening stage of the game without commentary. The

players have emerged with roughly equal chances.
Maxim Chigaev played
15. d5
and after
15... xd5
White faced a choice between recapturing on d5 with the pawn or
the queen.
16. xd5
From here we begin to closely monitor all the decisions of the
opponents in order to assess the strength of the average move of
106
Chigaev and Matlakov. Both recaptures appear to be roughly
equivalent, but they lead to different types of positions. After
16.exd5, each side would have a pawn advantage on the flank.
Evgeny Tomashevsky aptly described the structure that arises in this
case something like this: ‘The pawn structure itself leads the game
to a draw.’
His opinion completely coincides with the assessment of the modern
engine. Here is an approximate computer variation: 16... d7 (
16... f5 ) 17.c4 fb8 18. d2 bxc4 19.bxc4 b3 20.h3 f5 21. fc1
b7 22. a2 f6 ( 22... b8 23. a5 ) 23. h2 h7 24. ca1 g8. At
no point did my computer score deviate from 0.00. After the game
move 16. xd5, my engine, at a depth of 40, gives the assessment -
0.17. In other words, it considers the position to be roughly equal,
but one should understand that Black’s game is easier to play .
16... e6!?
This move looks less natural than 16... b7, but it makes it harder for
White to take control of d5, by not allowing him to put a pawn on
c4.
17. d3 c7 18. fc1 fc8!

107
Black has won the fight ‘for and against the move c2-c4’, a small
psychological victory. A chess game is a struggle of characters,
which in people, unlike engines, evokes a range of conflicting
feelings. Now Black is going to put his queen on c6 and play ...d6-
d5. It is difficult for White to prevent this advance. From a human
point of view, White’s position has begun to deteriorate, although
the engine is absolutely calm and gives a score of 0.1 at a depth of
55. That is, based on the engine’s assessment, for White the
situation is no worse, if not better, than three moves ago. But the
human perceives the current situation differently, and Maxim
Chigaev was already worried!
19. d2?!
Curiously, almost all the young talented Azerbaijani grandmasters
with whom I analysed this game in individual lessons experienced
similar feelings. White strives to tactically change the nature of the
fight, instead of calm and cold-blooded play to neutralize the
opponent’s initiative. It was logical to play 19.f3 c6 20.b4 d8
21. d1 d5 22.exd5 xd5 23. c5. The engine assessment is 0.25 at a

108
depth of 45, but Chigaev would not have been concerned about the
outcome of the game from this position.
19... c6 20.h3 f6!
If 20...d5, then 21. g3.
Now, a critical moment in the game has been reached.

21.c4?!
Chigaev, as we have already noted, did not like his position, and he
decided to sharply change the rhythm of the game. From calm
positional manoeuvring, he transformed the position into a
calculation battle by sacrificing a pawn. The main difficulty for the
defending side is that it needs to correctly grasp the moment for
such a transition.
The ability to feel that now is the time to move from passive and
patient defence to active play is a great art that few chess players
master. In this game, White did not need such radical changes. Calm
actions like 21.f3 d5 22.exd5 xd5 23. e2 f7 24.b4 would have
allowed him to maintain an approximate balance. My engine gives -
0.2 at depth 50.
109
But in this case, the opponent would seize the initiative by carrying
out the programmed advance ...d6-d5. This would mean that Black
had managed to win a psychological victory in the strategic conduct
of the fight. Since chess players have a highly developed sense of
battle, during the game they strive not to concede anything to their
opponent. This approach to the game is fraught with a loss of
objectivity in assessing the position, although outwardly it seems
courageous and creative. But a modern computer is devoid of all
human emotions and does not approve of such play. If White really
wanted to compete in a tactical battle, then he should have found the
hidden opportunity to sharpen the game without positional
concessions, associated with the move 21. a3!. After 21...d5
22.exd5 xd5 23.c4! e4 ( 23...bxc4 24. g3 ) 24. g3 h7 25.f3
f5 26.c5, we have approximate equality on the board. But finding
continuations such as 21. a3! that defy human logic is still the
prerogative of the computer.
21...bxc4 22.bxc4 xc4 23. g3 xe4 24. xh6 h7

110
It’s actually not that bad for White. But the tasks he now has to
solve have become disproportionately more difficult than ten moves
ago. Why did this happen? The answer is simple: Chigaev’s average
move in a relatively calm position was inferior in quality to
Matlakov ’s average move.
The position in the last diagram is a good training exercise for
finding defences in a difficult situation. I suggest you take ten
minutes to try to find the optimal continuation for White.
Answer: 25. g4!
White’s first move is correct.
25...f5
The strongest reply. On 25...d5 there would follow 26.b3.
26. d1?
But this is a mistake. It was necessary first to retreat the queen to f3,
so as to force a weakening in the black central pawn position. After
26. f3! e4 27. d1! White would be able to hold the balance.
But this is significantly more difficult than the simple lines 19.f3
c6 20.b4 d8 21. d1 d5 22.exd5 xd5 23. c5 or even 20. d2 d5
21.exd5 xd5 22.h3 with an almost equal position.
26...d5 27. g5
And this is a more serious mistake. White could have kept fighting
with 27. d2!.
27... g6 28.b3 xg5 29.bxc4 d4

111
Black’s position is winning. My engine assesses it as 2.2 at a depth
of 35. The rest of the game is outside our theme and we give it
without commentary.
30.c5 c6 31. c4 ac8 32. ac1 h7 33.g3 d8 34. a4 f6
35. d1 d5 36. b4 h6 37.h4 h5 38. d2 d3 39. a4 h6
40.g4 dxc5 41. xc5 xc5 42.g5 h5 43. xd3 c1+ 44. h2 e2
45. g3 f4 46. g2 e1 47. h3 h1+ 0-1
In none of the examples given in this chapter did we see any bright
combinations, bold sacrifices, or swift attacks on the king. The
outcome of the games emerged gradually, and as a result, the scales
tipped in favour of the grandmasters, whose average move was
stronger than their opponents’.
It is important to note that neither Nepomniachtchi, nor Eljanov, nor
Lutz, nor Chigaev made obvious mistakes that would have
immediately affected the results of the games. Rather, they
committed inaccuracies, the detection of which sometimes required
serious analysis. But problems accumulated with these inaccuracies,
and in the end, they did not achieve the desired result. Chigaev’s

112
inaccuracy 16. xd5 instead of 16.exd5 is in reality just a trifle, but
it is precisely such trifles which influence the strength of the average
move. And in this chapter, we wanted to help you develop an
awareness of such ‘trifles’.

113
Chapter 5
Associative thinking
in the defence of
inferior positions;
active or passive
defence
Here, we will finish looking at examples in which the connection
between active and passive defence was clearly visible. We will
continue our study of rook endings in a large separate Chapter, No.
7, ‘Associative Thinking in Rook Endings’. But before that, I would
like to help you not to get confused by the terminology. Let’s
analyse it in the next chapter.
In the second edition of my book Endgame Strategy First, let us
take another look at the game Kramnik - Lutz that we already saw
in the previous chapter.

114
Black’s salvation consisted in switching to a rapid counterattack
associated with the move 33... e4!. This is all far from obvious and
it was hardly realistic to expect Black to play this instead of
33... c8.
That game was played almost thirty years ago, but now we will
examine several relatively recent examples of defence in inferior
positions with the use of associative thinking.
Anish Giri (2772)
Samuel Shankland (2708)
Wijk aan Zee 2022

115
In this virtually equal position, perhaps just slightly more pleasant
for White, Shankland played
19... c7?! ,
bringing his king towards the centre. This was probably played by
instinct, without great thought. We know that centralization of the
king can hardly ever be wrong, but it can be untimely, and this is
one such case.
Black should have asked himself the prophylactic question of what
his opponent wants. In this case, a player as strong as Shankland
would fairly easily have found the move 19...h5. Now, if the bishop
retreats to h3, Black can play 20... h4, seizing space on the kingside,
whilst if 20. e2 he has 20... c7 21. f3 f8 22. c4 f6 / f5.
In chess, hard thought in difficult positions periodically gives way to
relaxation in simpler ones. A human cannot permanently maintain
the tension and concentration at the maximum level, this is only
natural. But sometimes relaxing can lead to a loss of concentration
and mistakes then occur.
Giri replied:

116
20. f3!
Probably only now did Black realize that after 20... f8 21. e6! f6 (
21... d8 22. d5 ) 22.g4! he has a bad position.
20... d8 21.h4!
An interesting situation. Black probably already appreciated the
difficulties of passive defence in such positions and he had to decide
how to continue.

Shankland decided it was ‘now or never’ – and wrongly.

21...d5?!
The idea of Black’s counterplay is logical and interesting, but the
timing was poorly chosen. His pieces are not yet active enough to
support the initiative or build a reliable blockade on the dark
squares. He should have been patient a little longer, but during the
game, it was not at all obvious.
Let’s think about how Black needed to maintain a passive defence.
To begin with, my favourite question is: ‘What does the opponent
want?’

117
White will most likely place the rook on f6, paralysing the
opponent’s pieces. Then he will begin to further constrain his
position with the moves h4-h5 and possibly f2-f4, followed by
centralization of the king. Eventually Black will develop additional
weaknesses and his position will collapse.
What is the chief weakness of the black position? The lack of space.
This means that we should begin by widening the pawn chain. We
start with 21...h5!?. The rook gets the square h7. 22. h3. Now
what? It would be interesting to put the pawn on b5, but after 22...b5
White can react concretely: 23. f1 a6 24.a3 f8 25.c4.

analysis diagram
Black loses a pawn on the queenside (after 25...bxc4 26. xc4 a5
27. c3 White is winning), whilst the possibility of transferring the
knight to d4 only partially compensates.
But instead of 22...b5, why not play the useful move 22...a6...?
Suppose that White then carries out his plan and puts the rook on f6
– 23. f6. Black needs to defend the f7-pawn with 23... h7.

118
 analysis diagram
Play might develop thus: 24. f1 c6 25. e2 e7 26. f3 d5!?. In
this situation, the attempt to break out of White’s press is more
effective than the way Shankland did it in the game. After 27. d3
d6 28.exd5 h8 29.c4 b5 30.b3 b8, Black has good chances of a
successful blockade, though White retains chances of winning after
31. g2 followed by c4-c5+ and d5-d6.
Let us return to the game after 21...d5?!
22.exd5 f5 23. h3 d6?!
It was better to bring the knight into the game at once. After the
strong move 23... f7! 24.c4 e4 25. a3 e5 26. f1 a8!, things
would be far from all over for Black. But such a precise series of
non-obvious moves can only be established with the help of the
computer or the necessary higher-class level of masonry. Moreover,
White has a strong alternative in 24. e3!, forcing 24... d6 anyway.
24.c4

119
24...b6?!
Trying to create a blockade, but missing Giri’s wonderful reply:
25.g4!
With the knight on d8 and rook on h8 there is no answer to this
powerful blow against the pawn structure. The same reply would
have been possible against the strongest continuation 24... f7
instead of 24...b6?!. Now, however, the black position starts to
collapse.
25... f7
Too late, but even after 25...e4 26. a3 a5 27. b3 c7 28.gxf5 there
is no real hope of salvation.
26.gxf5 g5 27. a3 gxh4 28. xa7 g5 29. g2 c8 30.b3 b5
31. a6+ d7 32.f6+ xh3 33. xh3 bxc4 34. c6!
Technique is the mastery of short-range tactics. Black resigned.
Richard Rapport (2763)
Alireza Firouzja (2759)
Stavanger 2021

120
White has the advantage of two bishops and a better pawn structure.
Black’s queenside is pretty battered, and the backward c7-pawn is
pinned down by the white rook. If White is allowed to make several
calm moves and consolidate the position (a2-a3, f1-c1, e2-d3) in
some sequence, then Black’s position will become strategically lost.
Therefore, Firouzja immediately begins active counterplay:
19...a5! 20. d3!
If 20.a3, there could follow 20...axb4 21.axb4 a2 22. xb5 ca8
23. c3 e5! 24.dxe5 b8 25. d3 xc3 26. xc3 xe5 27. b3.

121
 analysis diagram
White has an extra pawn, but all of Black’s pieces are active. He has
the opportunity to play for a simplification of the position associated
with the exchange of rooks, or to strive to capture the second rank
with the pair of rooks. Black’s defensive plan needs to be thought
through, but visually all of his pieces are active, the bishop has a
stronghold on d6, and his chances of a draw here are higher than
White’s chances of winning.
20... f6!
A concrete and strong move. Alireza refrains from the nasty
opposite-coloured bishops ending with four rooks, preferred by
Stockfish, after 20... d6 21. c3! axb4 22. xb4 xa2 23. xd6 exd6
24. xb5.

122
 analysis diagram
Here, compared to the previous diagram, we have material equality,
but all of the white pieces are much more active than the enemy
ones. Visually, White’s chances of success in a practical game are
higher than Black’s chances of a draw. If Black manages to
exchange one pair of rooks, then White will try to activate his king
and push the pawns on the kingside.
How strong White’s attack can become in such a scenario can be
seen in the next game.
Magnus Carlsen (2848)
Sergey Karjakin (2775)
Wijk aan Zee 2013

123
71. d5 h2+ 72. g3 h3+ 73. xg4 xd3 74.f5 e3 75. xg6+
h7 76. g8+ h8 77. f4 c3 78.f6 d3 79. e3 c4 80. e6 h7
81. f5 c2 82. g2+ h6 83. xc2 dxc2 84. xc2 g5 85. d4
a3 86. xc4 b2 87. d5 f4 88.f7 a3 89.e6 g5 90. c6 f6
91. d7 g7 92.e7
Black resigned.
Let us return to Rapport - Firouzja

124
21. xb5 axb4 22. c4
In the light of subsequent events, it made sense for White to
consider starting with 22. a6!, to remove the defence from the
pawn on c7, and then to choose the moment for the transfer of the
bishop to b3 via c4. It is quite possible that the Hungarian GM
underestimated or simply missed Black’s next move.
22... e4!
As if nothing had happened, the knight returns to the centre and
takes control of the important squares d2, d6 and c3.
23. c1
White does not wish to agree to a draw after 23. b1 d2 24. xb4
xc4 25. bxc4 xa2 26. c3 a7 27.g4 f8 whilst in the event of
23. d1 he would have to allow the manoeuvre ... e4-d6-b7-a5.
23... d2!
Black does not permit his opponent to play the stabilizing move
24. c2. With active play, he has achieved a great deal and has
almost equalized.
125
24. a6 cb8 25. 1c2 b6
Here, it was possible to retreat with the knight to its favourite square
– 25... e4, and White would not have had the bishop transfer to b3
via c4 due to 26...b3. The knight could then move to d6 with various
options for counterplay. Such a line does not look as creative as the
game move 25... b6. However, when defending inferior positions, a
balance must be maintained between active counterplay and calm
defensive measures. True, few manage to achieve such harmony.
Associative thinking immediately reminds us of a similar situation
from the end of the following game.
Magnus Carlsen (2834)
Maxim Matlakov (2718)
Wijk aan Zee 2018

Up to now, Black had played actively and pretty concretely. Here he

needed to calm down and change the rhythm, with the relatively
passive but solid move 34... b8!. Then, ...b7-b5 would have allowed
him to exchange a pair of pawns. Also good was 34... g8! 35. xa7
g2, or 34... d8- d2 with the same idea.

126
Instead of this, in time trouble the strong GM Matlakov, suffering
from a rush of adrenaline, played
34... c4? 35. xb7 xa4 36. a7!
(with the threat of 37. b4) and gradually lost.
Returning to Rapport - Firouzja
26. xb6 cxb6

27. b7!
Under the enemy’s pressure, White has not even managed to make
luft for his king. The French GM had prepared a cunning trap. On
the natural move 27. d3 there would have followed 27... b3!
28. b2 xa2 29. c4 a1! 30. xa2 xc2, and the game would
have ended in a draw. With his last move, White forces the enemy
rook to leave the eighth rank.
27... a7 28. c6
28. d5 e6.
28... b1

127
Here 28... b3? 29. b2 xa2?? does not work because of 30. d5
a1 31. c8+.
29. d5! c3 30. b3
30. xc3 c7.

The white bishop has still made its way to b3. Rapport has stabilized
the position, and all he has to do now is to deprive his opponent of
the only counterplay associated with activating the black bishop by
...e7-e5 with f2-f4.
Firouzja has already achieved a lot. He has avoided the squeeze, got
rid of his pawn weaknesses, activated his pieces and simplified the
position. All this was achieved through timely and very concrete
active operations. But White still has the advantage.
Black again faces a choice: to give up a pawn in order to reach an
opposite-coloured bishops ending by means of 30...e5 31. xc3 bxc3
32. xc3 exd4 33. c8+ f8 34.exd4, or maintain a passive defence
with the move:
30... c7?!

128
with material equality and the opponent having the advantage of two
bishops.
Alireza chose the second path and was wrong. First of all, from a
practical point of view. The difficult and concrete play over the last
ten moves took a lot of energy and thinking time. And the time
control was unique: two hours for the entire game with a 10-second
(!) increment after the fortieth move. Playing in a relatively simple
position with opposite-coloured bishops and rooks a pawn down in
time trouble is much easier than in a multi-piece endgame with a
positional advantage for the opponent.
Due to the lack of time to think, the further part of the game went
beyond the boundaries of classical chess.
31.f4 e6 32.g4 f6 33.g5 e7 34. g2 g7 35. f3 h6 36.h4 c8
37. b2 b5 38. c1 d8 39.a3 a5 40.axb4 xb4 41.e4 a5 42.d5
exd5 43.exd5 hxg5 44.hxg5 b4 45.d6 d8 46. d1 b6 47.d7 f8
48. d6 a7 49. c4 c5 50. c6 d4 51. c8

Up to this moment, both sides have played well. White has the
advantage, but after 51... b6 there would still have been a whole

129
battle ahead. But Firouzja was already playing on increment and
here he made a terrible oversight:
51... e7?? 52. xf7 xd7 53. c4 e7 54. xg6 d6 55. xb4
And the game ended in Black’s defeat.
Dmitry Andreikin (2724)
Vladimir Fedoseev (2696)
Russian Championship, Ufa 2021

White’s situation is difficult. Black has an exchange for a pawn, but

that is not the main thing. All the white pieces are pressed in by the
enemy forces on the second rank.
37.d5!?
Andreikin poses his opponent a difficult exchanging problem, whilst
in time-trouble.
37... f7?!
Fedoseev decides to go into what at first sight looks like a very
favourable rook ending and thereby makes a mistake.

130
When thinking time is insufficient, one must base one’s play on
general principles and intuition. General considerations suggest to
us that in positions with an extra exchange, the stronger side should
strive for an exchange of rooks, whereas rook endings are notorious
for their drawing tendencies. After 37... xf3+ 38.gxf3 exd5
39. xd5 c1+ 40. d1 xd1+ 41. xd1 f4! 42.exf4 g7 43. e1
a3! 44. e2 h6

analysis diagram
the black king is transferred to h5, and White, due to zugzwang, will
be forced to either give up the f3-pawn or play f4-f5. In the first
case, his position looks completely hopeless. In the second, even if
he moves the bishop to the h4-d8 diagonal, it will not stay there with
the black pawn on h6.
But let’s follow the events in the game.
38.dxe6+ xe6 39. e2 e4 40. d3 xg3
If 40... c8 41. xe4 fxe4 42. f1 the white position would be
transformed into a fortress.

131
41. xc2 xc2 42.fxg3 xg2

The time control has passed. White needs to find a defensible set-up
in the rook ending, a pawn down.
43. d8!
A difficult move, found by a strong GM with the help of logical
(human!) thinking. White tries to force the advance of the black
pawn to h5, so that the enemy king cannot go there after the loss of
the white pawn on g3.
The move 43. d8! is undoubtedly an active method of defence. The
computer suggests starting with the passive move 43. f1, and after
43... xg3 44. f2 g4 (a draw results from 44... h3 45. d4 f7
46. d7+ g8 47. d4 g7 48. f4 h6 49.e4! fxe4 50. g2! e3
51. f2 f3+ 52. xf3 exf3 53. xf3; this line was found by the
extremely talented young GM Aidin Suleymanli during a training
session with the Azeri youth team) White should go over to active
defence with 45. h1 f6 46.h5! (otherwise the black king will
come to this square) 46...g5 47. b1!.

132
White’s counterplay should be enough for a draw, but even now this
is far from obvious.

analysis diagram
Seeing from afar the variation 47... h4 48. b6+ g7 49. g3 xh5
50.e4!! fxe4 51. g4 h6 52. b7+ f8 53. xg5 e6 54. f5 e8
55. xh7 with the subsequent elimination of the e-pawn was
completely unrealistic after 40 moves and in conditions of serious
time-trouble.
Let us return to the game.

133
43...h5!?
It seems as though Fedoseev has tricked his opponent and is about
to set about the realization of his two extra pawns. For this, he
needed to reject the tempting move 43... xg3. After 44. f2 g4
45.h5! gxh5 46. h8 g7, it seems that White cannot save himself.

134
 analysis diagram
But this is not so: 47. f3! h4 48.e4! h3 49. a8! ( 49.exf5+? f7 )
49...h2 50. a6+ e5 51. h6 fxe4+ 52. e3.

analysis diagram
135
Black has three extra pawns, but in view of the variation 52... g3+
53. f2 a3 54. xh2 a2+ 55. e3! a3+ 56. f2 d4 57. xh7
a2+ 58. f1 d3 59. d7+ e3 60. b7, the position is theoretically
drawn.
We return to to the game after 43...h5!? . Now there follows the only
saving move:
44. d1!
Beautiful! Fedoseev could not have foreseen such cunning. It runs
contrary to any human’s natural thought. Now the white rook is
quietly directed to h1, not fearing the transfer of the enemy king to
h5. White goes over to passive defence and confidently holds his
unbreachable fortress. The rest of the game needs no commentary.
44... e5 45. f1 xg3 46. f2 g4 47. h1 e4 48. h2 d3
49. h1 d2 50. h3 d3 51. h1 e4 52. h2 f4 53.exf4 xf4+
54. g2 g4+ 55. h3 ½-½
Andreikin’s manoeuvre 43. d1-d8! and 44. d8-d1! is hard to find
during a game, but it is not so difficult to find analogies for the
development of associative thinking. If you analyse attentively the
ending of the game Shuvalova-Maurizzi , Wijk aan Zee 2022 (see
Chapter 7),

136
then (instead of the game move 46...b5 ) the manoeuvre 46... b2!
47. e3 b5!! 48.axb5 d2 is a more complicated version of
Andreikin’s. The black rook in essence controls the play on the
second rank and only then will it take the enemy b-pawn along the
b-file.
But it seems to me that the most striking association with ‘an active
rook manoeuvre and a sudden retreat backwards’ is seen in the
following ending.
Hans Böhm (2445)
Jan Timman (2590)
Amsterdam 1977

137
In this complicated endgame, it seems White is in trouble. He loses
after 53. a1 a8 54. e1 e3 or 54... f3. However, the manoeuvre
53. a6!! e5 54. a1
allowed him to save himself miraculously. The move 53. a6!!
forces the black pawn to move forward and so give the white king
the possibility of attacking it from the square d5. The rest is simple:
54... a8 55. e1 f3 56. d5! xa7
56...e4 57. xe2=.
57. xe5

138
The black rook can eliminate either of the white pawns but the white
king succeeds in supporting the other and saving the game.
57... b7 58.g4 xg4 59. e4
59. xe2.
59... e7+ 60. d5 ½-½

139
Chapter 6
Fundamental
endgame principles:
methods of play
Knowledge of terminology leads to knowledge of things – Plato
When analysing games, it is very important to remember and
classify typical playing techniques, thereby expanding your
knowledge base for using associative thinking. But remembering
these techniques in the form of variations is difficult, and they are
quickly forgotten. You need to be able to explain them to yourself
verbally. And here there is often confusion in terms.
It is important to remember: there can be a great variety of methods
(or plans) in a game. For example, in sharp rook endings, the correct
method of action is the following: the king should go with his own
passed pawns, and the rook should fight the enemy’s.
And there are only four fundamental principles of the endgame:
‘Thinking in schemes’, ‘Centralization (activation) of the king’,
‘The principle of two weaknesses’ and ‘Do not hurry’. This topic is
presented in detail in my work Endgame Strategy.
To effectively memorize these principles, I suggest chess players
analyse the game I played with the Belarussian master Vladimir

140
Veremeichik. This game is interesting primarily because it is
possible to trace the action of each of the four principles.
Trompowsky Opening
Mikhail Shereshevsky (2475)
Vladimir Veremeichik (2354)
Minsk 1978
1.d4 f6 2. g5
In 1978, I had little time for serious work on the openings. I had just
started work on my first book Endgame Strategy. In addition, I had
to work as a trainer to the republic junior team, as well as play for
the republic’s adult team myself and spend time with my family.
Therefore I decided to build my opening repertoire by refraining
from fighting for an advantage as White. Of course, once my book
was finished, I returned to the normal move 2.c4.
2...c5 3. c3
Nowadays it is hard to surprise the opponent on move three, but
then this was rarely played. White’s idea is to bring the game into
Sicilian-type waters after 3...cxd4 4. xd4 c6 5. h4 followed by
6.e4 and 7.0-0-0, if the opponent allows it.
3... a5 4. xf6 gxf6 5.e3 f5
After the g-pawn comes to the f-file, Black has weak squares on h6
and h5. It is hard for White to get at h6, but h5 can be a useful
outpost. Ideally, he would like to get a knight there, but the queen is
also not bad.
6. h5 cxd4 7.exd4 b6 8.0-0-0 h6+ 9. xh6 xh6+ 10. b1 d6

141
The sides have exchanged queens and a pair of minor pieces. We
have a queenless middlegame and it is time for White to settle on a
piece set-up; in other words, to employ schematic thinking.
11.g3!?
The course of my thoughts was something like this: White has the
superior pawn structure and some advantage in development.
Black’s main trump is the advantage of the two bishops. The enemy
dark-squared bishop has no opponent, and this means that it is
useful to build a pawn barrier against it on the dark squares. In
addition, my bishop becomes a sort of ‘Catalan bishop’ on the long
diagonal.
11... d7 12. g2 c6 13.d5! d7 14.f4 g7 15. ge2 h5 16.h4
Such a move should be played automatically. One should not give
the opponent the opportunity to push his pawn to h4 and obtain
leverage for pressure on the white position. The weakness of the g3-
pawn is illusory, because it’s hard for the black knight to get to e4.
16... a6 17. he1 c7 18. d4 f8

142
An important point is that if in this position you ask yourself the
prophylactic question: ‘What does the opponent want if it were his
move?’, then the answer I have both today and forty-five years ago
will be the same. Black wants to play 19... f6 and then 20... g7, in
order to establish interaction between the rooks, which can currently
only communicate with each other ‘by telephone’, since the black
king prevents their direct dialogue.
Can we prevent this in some way? Yes, there is sense in trying to do
this with the help of tactics.
19. e3! f6?! 20. h3!

143
20... xd4?!
In Endgame Strategy , I gave the following explanation of this
move: ‘Now play becomes all one-way. Black is doomed to passive
defence. In the event of 20...e6 there could follow the piece sacrifice
21.dxe6 fxe6 22. xf5!? exf5 23. xd6, although of course this is not
essential. Even so, Black would in this case obtain some counter-
chances in a tactical battle.’
I can still agree with this comment today, but the stronger engine
allows one to add some corrections. Firstly, it shows that the threat
to win the f5-pawn is not yet dangerous; for example, after
20... d8!?, it is bad to play 21. xf5 because of 21... e8!, with the
terrible threat of 22... g7. Secondly, the piece sacrifice after 20...e6
21.dxe6 fxe6 22. xf5 only leads to equality after 22...exf5 23. xd6
e6 24. xf5 d4 25. dxe6 xe6 26. xe6 xc3 27.bxc3 e8
28. g6 g8 29. h6 xg3 30. xh5 xc3.

144
 analysis diagram
Instead of 22. xf5, White retains the better chances by means of
22. f3 e7 23. g5 e8 24. e2.
We continue our analysis of the game.
21. xd4 e8

145
I thought about this position for quite some time. My thoughts ran
something like this: ‘Black’s position has two pawn weaknesses –
on e7 and f5. But since the distance between them is minimal and
the black king can easily defend both from the square f6, we will
consider them as one weakness. To play for a win, I will need to
create a second weakness in my opponent’s position. Where can this
be done? On the queenside. To create such a pawn weakness, I
would like to force the opponent to advance the a-pawn or the b-
pawn. This will be a hook we can grab onto. The opponent will not
voluntarily advance the queenside pawns.’
I repeat, I am conveying to you my feelings from forty-five years
ago. At that time, as they say, ‘the grass was greener than today, the
girls were more beautiful’, and my mind was not clouded by the
computer vision of the game.
The modern engine indicates that if Black manages to advance his
pawns to a5 and b5, then he will obtain counterplay sufficient to
equalize. It hints that it would be a good idea for White to start with
22. c4, calling for, say, 22... a6. It would then be beneficial to

146
advance the a- and b-pawns to the fifth rank. After this, we can try
to weaken the c6-square and place the knight there.
In those distant times, such a direction of the game, associated with
preventing the opponent’s counterplay, would not have occurred to
me. In addition, during the game close psychological contact is
established between the opponents. I felt Veremeichik was afraid of
me and it was unlikely that he would voluntarily show activity on
the queenside. Therefore, I was more interested in how to force one
of my opponent’s pawns to advance on this part of the board, in
order to create a second weakness there. For this purpose, I decided
to send both rooks there. But before that, it was necessary to
improve the positions of the remaining pieces as much as possible.
Once again I had to ask myself important questions: ‘Who? Where?
Why?’
Here are some sample answers:
1) King. Loitering (if one can use such a rude word in relation to
His Majesty) on square b1. With four rooks, it is not safe for him to
walk around the board. But let him at least cover the g3-pawn on the
kingside. This is not a royal job, of course, but everything has its
time.
2) Bishop. Keeps his eye on the black pawn on f5. But this work
can also be done from d3, where it will be more active.
3) Knight. Its place is behind and to the side of the f4-pawn. On
e3, it will strengthen the important d5-pawn, increase pressure on
the black f5-pawn and will not get in the way of the rooks, which we
will send to the queenside. In addition, with this position of the
knight, the path will be open for the advance of the c-pawn if the
need arises.
These were my reasonings, in which I was guided by the principles
of ‘schematic thinking’, ‘centralization (activation) of the king’ and
‘the principle of two weaknesses’.
Let’s start improving our position!

147
22. c1
The engine recommends 22. g2, 22. c4 or 22.a4.
22... g7 23. d2 f6 24. e1 c8
And here it positively insists on the moves 24...b5 or 24...a5.
25. f1 e8 26. d2 g7 27. d1 he8

As I suspected, Black prefers passive defence and does not advance

his queenside pawns to the fifth rank.
28. a3 a6?
Black’s previous series of moves was not best, but was at least
patient. Now, however, he should have stuck to passive defence
with the move 28... a8!, not weakening his pawn configuration on
the queenside.
29. b3!?
The computer prefers 29. e3, but I thought I should strike while the
iron was hot.
29... c7?!

148
Black should have settled for 29...b5 , but with limited thinking
time, it was hard to make such a move. After all, Black must reckon
with various possibilities, including 30.a4.
I would probably have gone for the manoeuvre 30. a3!? a8 31. a5
with play for two weaknesses.
30. b6!
The start. The b7-pawn is fixed and with a high probability can turn
into a second weakness in Black’s position.
30... d8 31. e2 g6 32. e3 e8 33. d4 f6 34. d3 g8
35. e2?! a8?! 36. db4 a7
In this section of the game, Black’s average move was not very
good.

White’s plan is a complete success, and his position is won. My

engine estimates that White’s advantage is greater than two pawns.
At this moment, I had about 15 minutes left for four moves until the
time control, and I started thinking about finding a way to win.
Black’s pieces are all motionless except for the bishop, which, in my
opinion, had a sad choice between the squares c8 and d7.
149
When your opponent has almost nothing to hope for, it is important
to ask yourself the right questions and find the right answers to
them. One of these questions should be: ‘Which piece is the
opponent’s defence based on, or who is my No. 1 enemy?’ A short
reflection will quickly lead us to the answer. Black’s only piece that
holds his defence together is the c7-rook. This means that it must be
exchanged with the move c4. Then the white king will move to the
d4-square, and the game will end with the march of the white c-
pawn to c5 – with or without the preliminary move b2-b4.
I answered correctly the strategic question of how to win the game.
It remains to be understood how exactly to exchange the black rook
on c7. Should I play c4 right away, or should the a-pawn first be
advanced to a5? Should I place my b-pawn on b3 to recapture the
rook on c4? These are all tactical questions. I was thinking, and the
time on the clock continued to slip away, and there were 7 minutes
left for four moves.
And here a great idea came to my mind: isn’t it time to apply the
‘Do not hurry’ principle? That is, to hold the position, make four
moves and adjourn the game? At home, I could easily find the
optimal order of moves, depriving my opponent of random
counterchances. It’s a really good idea, but to play ‘Don’t hurry’
you have to be absolutely sure that your opponent has no
counterplay. If I had played 37. d2!, then this would have been the
case. Here is a sample variation: 37... g7 38. d4 g6 39.a4 a5
40.b3 c8 41. c4 xc4 42. xc4! ( 42.bxc4 d7 43. c3 g7
44. xf5+ f8 45. b3 ) 42... xd5 43. b5 b4 44. xa5.
Instead, I immediately played...
37. d4?
... and was hit with a blow:
37... b5!
Here, the best thing to do was to go 38.c4, maintaining a decent
positional advantage. But this square was intended for the white
150
rook to exchange the opponent’s only active piece! I hated the idea
of playing after 38.c4 d7, because in my mind I had already won
this game. And then the a7 would break out of its captivity, and we
would have to start all over again! In the remaining 2-3 minutes, I
made an emotional, creative decision, but objectively not the best
one, and sacrificed the exchange.
38. xb5!? axb5 39. xf5 a4?

With his last move, Black committed the decisive tactical error. He
had to play 39... xa2, after which the engine assessment is +0.43 at
a depth of 50.
Now try to find the way to exploit my opponent’s mistake. I, of
course, played 40.b4? as the final move before the time control,
having no more than a minute left to think.
Answer: 40. xd6+!!.
‘Serious chess culture’ prevents one looking at such a move. The
exchange sacrifice eventually eliminates the e7-pawn, so as to leave
the opponent with five crippled pawns: 40... g7 41. f5+ h8

151
42.d6!! (lovely!) 42...exd6 43. xd6 a6 44. d8+ g8 45. d2 b6
46.a3.

analysis diagram
The engine assessment is +6.97 at a depth of 37.
Let us continue analysing the game.
40.b4? xa2

152
41. f3
White again plays according to the principle ‘Do not hurry’. Here
too, the computer recommends taking the pawn on d6, followed by
42. f5+ and 43.d6, but without the a-pawn, it considers White’s
advantage to be close to two pawns.
41... c3 42. xe7+ g7 43. f5+ f8 44. e2!

153
In the adjourned position, White has some initiative, but according
to the engine, the sides have equal chances. I clearly understood that
in White’s position, there are two guidelines for correct play. First,
White needs to avoid exchanging rooks. In positions where the
opponent has an extra exchange, this is an unwritten rule. Secondly,
he needs to keep the pawn on c2 and the bishop on d3 – this is its
key support.
The placement of the pawn on c2 and the bishop on d3, or the pawn
on c7 and the bishop on d6 on Black’s side, arises from many
openings and is often a key element of the position. Perhaps one of
the most striking examples on this topic that associative thinking
suggests to me is the following ending.
Wei Yi (2733)
Ian Nepomniachtchi (2771)
Moscow 2019

154
White’s position is winning. He only needs to find the bishop
transfer to the secure outpost on d3: 32. f2! e8 33. f1!, and Black
would have no chance of saving the game. But this was a rapid
game and the Chinese GM was deflected by a kingside attack.
32. g4+?! h6 33. c7 a3+ 34.c3?
34. h2.
34... e8
And Nepomniachtchi managed to scrape a draw.
The cooperation between a bishop on d3 and a pawn on c2 can also
be seen in the game Firouzja - Aronian in Chapter 8.
Let us return to the game Shereshevsky-Veremeichik. In the
position of the last diagram but one,

155
Black could have broken the Gordian Knot by means of
44... axc2+!? 45. xc2 xc2+ 46. d3 c1 47. d2 c4 and made a
draw.
But Veremeichik played:
44... c8 45. d2! d8 46. xb5 a1 47. c3! e1 48. d3

156
It is essential to protect the square e4 and stabilize the position.
You have probably noticed that I constantly mark the white king’s
movements with exclamation marks, although according to the
computer’s assessment the position remains equal. Black’s average
move was somewhat inferior in strength to White’s, but the
computer has not yet reacted to this.
Instead of 45... d8 , the following variation was possible: 45... e8
46. c1 a1+ 47. b2 d1!,

157
 analysis diagram
forcing White to exchange rooks after 48. xb5 xd4 49. xd4 with
a probable draw.
I would like to draw the engine’s attention to the white king moves,
to which we have awarded exclamation marks: 44. e2!, 45. d2!,
47. c3!.
Back on move 44, I asked myself the question: ‘With what active
operations can I create problems for my opponent on the queenside
and count on seizing the initiative?’ After all, the pawn on c2 cannot
be moved forward, so as not to deprive the bishop of support and
expose the king. The answer was something like this: ‘The only
piece that can be used to disrupt the opponent’s defensive formation
is the king! But for his safety, at least the knights need to be
exchanged. I hoped my opponent would help me with this. After all,
for him, Enemy No. 1 is my knight on f5, and there is hope that he
will want to exchange it for his knight.’
48... g8?

158
The decisive mistake. Veremeichik could not ‘dig up’ the difficult
move 48... b1!, in order after 49. c4 to reply 49... c8+ 50. b5
c5+. He did not expect the advance of the white king.

49. c4!!
As my engine shows at a depth of 35, after the move 49. c4!! the
assessment of the position immediately jumps from 0.00 to +4.26.
The rest is easy.
49... e7 50. b5 xf5
50... c8 51. c4 and 52. c7.
51. xf5 e7 52. b6 f6 53. c7! g8 54. d3 e7+ 55. xd6
c8 56. c4!
The threat was 56... ee8.
56... d8+ 57. c5 c7+ 58. b6 cd7 59. c7 xd5 60. xb7 xd3
61.cxd3 xd3
It was possible to resign at once.
62.b5 xg3 63.b6 b3 64. c5 b4 65. c7 1-0

159
In this game we saw all four basic endgame principles in action. I
noticed that in the book Decision Making in Major Piece Endings , a
very interesting work for skilled chess players, Boris Gelfand I did
not include this principle among the most important ones. It seems
to me that advancing pawns in the endgame is too natural. To me, it
sounds like ‘a person must breathe in order to live.’ After all, the
main goal of the endgame is almost always not a mating attack, but
to queen a pawn. But can achieving the main goal of the endgame
be the fundamental principle of playing the game? In my opinion,
this question is rather more of a philosophical one.

160
Chapter 7
Associative thinking
in rook endings
Schematic thinking
In rook endings, the connection between the same techniques or
methods of action can be made relatively easily, although the
positions will lack external similarity in pawn structure. Our task
will be to learn to identify this connection.
Polina Shuvalova (2498)
Gunay Mammadzada (2472)
Nicosia 2023

161
This position immediately calls to mind an association with the
following game.
Boris Spassky (2625)
Eugenio Torre (2535)
Hamburg 1982

48.f4!
In this position, Black loses because of the diabolical position of his
pawn on f6. White will exchange on e5 and win the isolated pawn
by bringing his king to e6. Black inevitably falls into zugzwang and
cannot keep his rook on a5.
A sample position:

162
So, Black has to exchange pawns on f4, but then the outcome of the
game is decided by the passed e-pawn, which does not even need
the support of the white king.

163
In this position, it doesn’t matter whose move it is. But if the black
pawns were on g6 and f7, then White could not get a passed pawn
on the e- or f-file and the game would end in a draw.
It is time to return to the game Shuvalova-Mammadzada. Let us
compare these two endgames.
Shuvalova played:
45. c8 c2 46.c5

The pawn moved to f6 on the twenty-ninth move, and Black can’t

change anything here. It is obvious that White is trying to carry out
Spassky’s plan.
What should Black do to avoid losing the game? She should draw
an analogy between positions with White’s extra passed pawn on
the a-file and on the c-file and try to grasp the difference on which
to base a defensive plan.
Try, dear reader, to cope with this task.
To begin with, you should understand that the manifestation of
activity on your part can only weaken the e-pawn, so you must first

164
mentally follow the actions of your opponent. Let’s say White
carries out her plan:

It is your move and you play 1... g6. White exchanges on e5 –

2.fxe5 fxe5. And now it transpires that she cannot get at the e5-
pawn with the king, since the key square c4 is controlled by the
black rook. This is the key to the position and the vital difference
from Spassky-Torre. Black has to stick to passive waiting tactics,
because the e5-pawn is invulnerable.
Now let us follow the events in the game:
46... h7 47.c6 g6 48.c7 g7 49. h2 c1 50.g3 h7 51. g2

165
51...g5??
Gunay prevents the move 52.f4. This would have been the best
practical decision, had the white pawn been on a7 and the rook on
a8, or the pawn on b7 and rook on b8.
Here, however, she should simply have played 51... g7 and kept
the rook on the c-file.
52. f8! xc7 53. xf6 gxh4 54.gxh4 c5
The black rook cannot remain on the fifth rank once the white king
gets to d3.

166
White played the rest of the game confidently and Black could not
put up real resistance in time-trouble:
55. f5 g6 56. f2 a5 57. e3 a3+ 58. e2 a5 59. d3 h6
60. c3 a1 61. d3 h1 62. xe5 xh4 63. e3 g6 64. a5 f6
65. f2 g6 66. g2 h6 67. a6+ g7 68. a1 f4 69. g3 f7
70.f4 e7 71. f3 b7 72.f5 b2 73. f4 g2 74. f3 g5 75. a2
f6 76. g2 h4 77. xg5 xg5 78. g2 1-0
Before proceeding to the analysis of the next instructive ending, I
will remind you of the methods (or plans) of playing in positions
with an extra pawn on the queenside.
The stronger side in positions such as in the diagram below...

167
has two winning plans:
1. Break away with the king from the kingside pawns and free the
rook along the seventh rank;
2. Place the rook on the eighth rank, and the pawn on the seventh,
and use zugzwang to win the opponent’s isolated pawn.
In the diagram position, Black does not have an isolated pawn. This
means that White has to follow Plan 1 .
Let us follow the possible sequence of events:
1. f3 f6 2. e3
The white king emerges from its initial position.
2... e6?!
Let us assume that White reacts to his opponent’s play as follows:
3. d4! xf2 4. c7 a2 5.a7 f5!

168
6. c4!
The capture 6. xf7+? loses an important tempo, whilst 6. c5?!
leads to a draw (please take me at my word). The move 6. c4! is a
purely prophylactic one. White asks himself what his opponent
would do if it were his move. A simple analysis shows that Black
has nothing better than:
6... g4
6...f6 is bad because of 7. b5 with the threat of 8. c5+ and a
bridge-building exercise via a rook check from c6 or c4.
7. b3!
This zigzag manoeuvre is the idea of the move 6. c4. Events can
now develop according to the following scheme:
7... a6 8. c4+ xg3 9. a4 xa7 10. xa7 xh4

169
11. c2! g3 12. d1 h4 13. e1 g2
13...h3 14. f1 h2 15. a3+.
14. a2+ g1 15. a4
and 16. g4+, winning for White.
Now let us look at another position.

170
In this position, Plan 1 makes no sense. If the white king tries to
break away from the f4-pawn, then after 1. d3? xf4 2. c7 a4
3.a7 the white rook is free on the seventh rank but Black is ready
with counterplay in the shape of the passed pawn on f5. At the same
time, the black king does not need to eliminate any of the white
kingside pawns. The position is a dead draw.
But here White needs to employ Plan 2, the target of which is the
isolated pawn on f5 . We put the rook on a8 – 1. a8!. Say, 1... g7
2.a7!. Then the white king heads towards the f5-pawn and wins it.
After this, the game is won without the help of the white king, by
simply advancing the f4-pawn to f6.
At a time when not two, but six chess sessions a year were held at
Sirius, the strongest young chess players in Russia often came there.
In addition to general chess lectures, active work was carried out
with small groups. In one of these groups were Andrei Esipenko and
Alexey Sarana.
We looked at the methods of playing with an extra passed white
pawn on a6 using the previous example, after White’s second move:

171
In fact, this position is drawn. The only mistake is the move
2... e6?. The correct way to defend was suggested by GM Rustam
Dautov. For Black, the path to a draw is quite narrow and so, to save
space, we will look at an extract from the analysis by Dautov and
Mark Dvoretsky:
2... a4! 3. d3 g5!

172
4.hxg5+ xg5 5. c3 g4 6. b3 a1 7. b4 a2 8. b5 xf2
9. a8
9. b7 xg3 10.a7 a2.
9... b2+!
Not 9... xg3? 10. g8+ f3 11.a7 a2 12.a8 + xa8 13. xa8 h4
14. h8! g3 15. c4 and White wins.
10. c4 a2 11. g8+ f3 12. b5 b2+ 13. c6 a2 14. b7
b2+ 15. a8 f5 16. g5 f4 17.gxf4 h4 18.f5 f4 19. h5 g4
20. h8 xf5 21. xh4 e6 ½-½
After analysing the difficult variations of this endgame, Alexey
Sarana shared an important note. It turned out that he believed that
the weaker side in the initial position should keep the rook not on
the second, but on the third rank. That is, do not let the enemy king
onto the e3-square – the starting square for breaking away from
White’s own kingside pawns.

173
I did not find a way to refute the young GM’s idea. So it was all the
more interesting for me to analyse the following Internet game,
where the Russian GM Grischuk used a similar method of defence
in a position with colours reversed.
Alexander Grischuk (2777)
Maxime Vachier-Lagrave (2784)
Chess.com 2020

174
This game had great sporting significance, incidentally. It was
played in the final round of the preliminary tournament. To qualify
for the eight-player play-offs, White needed a draw, Black a win.
30.g3
I think the strongest move in this position is 30.g4!?, but that is quite
a different story.
30... xa2 31.h4 h5 32. a7 a5 33. g2 a4 34. f3 g7

175
35. a6!
Grischuk does not allow the enemy king to come to e6.
35...a3 36. g2 a1 37. f3 f8 38. a7 e8 39. g2 f6 40. a6
f7 41. f3 a2 42. e3 g5

176
Up to here, White has defended correctly, but at this moment he
commits the decisive positional mistake, by exchanging pawns on
the square g5.
43.hxg5?
Why is this wrong? After all, the weaker side, it seems, should strive
to reduce the number of pawns.
Let’s figure it out. We have already mentioned Black’s main plan
for playing to win: he needs to break away from the kingside pawns
with his king and free the rook along the second rank. How will this
plan guarantee success? Only when the weaker side, having gained
an extra pawn on the kingside, fails to quickly create a passed pawn
there and gain counterplay. Otherwise, as we have already observed,
the struggle takes on an unclear character, and the fate of the game
can be decided by one extra tempo.
And when does the strongest side get this ‘lucky’ chance? When he
has a spatial advantage on the kingside.
For example, in the position in the diagram below, the white pawn
on e5 squeezes the opponent’s position on the kingside and provides
a decisive spatial advantage.

177
After 1. d4 xf2 2. c7 a2 3.a7 it’s time for Black to give up,
since his counterplay on the kingside is clearly too late. The white
pawn on e5 squeezes the opponent’s position and gives his forces a
large spatial advantage. Along with the passed white pawn on a7,
this can be characterized as the second weakness of Black’s
position. In fact, White has an ‘extra’ king, which Boris Gelfand
equates in strength to a rook in the endgame if its safety is
guaranteed.
When choosing Plan No. 1, we were guided by the principle of ‘
Schematic Thinking.’ The ‘Principle of Two Weaknesses ’ brings
victory to White thanks to the e5-pawn, which guarantees a spatial
advantage on the kingside.
Now let’s return to the game Grischuk - Vachier-Lagrave

178
 analysis diagram
Let’s assume that Black plays 44... a1!. With this move, he
expresses his readiness to move to either of the two winning plans at
his disposal.
How can White play now? He can make a wait-and-see move:
45. a6. 45. e3 also looks good, but he cannot just retreat with the
king to the g2-square.
White can put the king on this square and allow himself to be
pressed in with the move ...g5-g4 only when the black pawn is
already on a2, and the rook is on the a1-square. If the square a2 is
not occupied by a black pawn with a black rook on a1, then Plan 1
will almost certainly lead the stronger side to victory with the black
pawn on g4 and the white king on g2.
Let’s look at the position after White’s mistake 45. g2?.

179
 analysis diagram
Black should silently offer thanks to the Almighty and immediately
grab the fleeting chance by playing 45...g4!. Then events could
develop as follows: 46. a6 a2 47. a5 f5 48. g1 f6 49. g2 e6
50. g1 d6 51. g2 (if 51. xf5? c2 52. a5 a2 53. g2 c6
White’s position is hopeless) 51... c6 52. xf5.

180
 analysis diagram
White has regained the pawn, and on his next move he will take a
second. In such situations, when the stronger side sacrifices two
pawns to free his rook on the seventh rank, the rook stands best one
file away from his passed pawn. The reason why will become clear
shortly.
52... b2! 53. xh5 (otherwise he can resign at once) 53... b5! (the
first plus of the move 52... b2! ) 54. h6+ b7!. The second plus.
Now White cannot get his rook to the first rank: 55. e6? a2 56. e1
b1. We continue the analysis: 55. h7+ a6 56. h8 a5 57. a8+
b5 58. b8+ a4 59. b1.

181
 analysis diagram
Here Black needs to ask himself the prophylactic question: ‘What
would White do if it were his move?’, and not allow the freeing
move f2-f3. The simplest is to play 59... c5!, so as to meet 60.f3
with 60... c2+.
Now all White’s hopes rest on the passed pawn. But after 60. h1 a2
61.h5 xh5 62. xh5 a1 the following position is reached:

182
 analysis diagram
We will not go into details of the analysis of this ending. According
to Nalimov’s tables, Black wins.
The main winning idea is to sacrifice the queen on the e3-square in
approximately this position:

183
 analysis diagram
After 1... xe3 the computer promises mate in 38.
Now we return to the game Grischuk - Vachier-Lagrave
43...fxg5

184
44. a5 a1! 45. f3 g4+!
Black constantly asks his opponent difficult questions.
46. f4!
As we have already noted, the black rook is perfectly located on the
square a1. It allows the stronger side to use either of the two
winning plans that we have described. In the case of 46. g2? Black
would respond with 46... a2! 47. xh5 c2. Next, Black’s king
would go to the a-pawn, and Black’s rook would already be free
along the second rank.
If the white king retreats to e2, then Black would place on a2 not a
rook, but a pawn: 46. e2? a2!. We see Plan No. 2 in action. The
rook is placed on the first rank, the pawn on the second, and with
the help of zugzwang the opponent’s isolated pawn is won. White
does not have an isolated pawn here, but his king is isolated from
the safe squares g2 and h2!

By sacrificing two pawns in the next moves, Black opens the

position of the opponent’s king along the second rank. After this, the
rook is won with the move ... a1-h1.
185
The skilful play of the French grandmaster evoked in me an
association with a game from the World Rapid Championship
between Magnus Carlsen and Alexey Dreev
Magnus Carlsen (2855)
Alexey Dreev (2635)
Warsaw World Rapid 2021

35.b6
Carlsen demonstrates his readiness to place his kingside pawns on
e3 and h3, and then implement Plan 1. That is, to break away with
the king from the f2-pawn with the move f3-e4 with the black
rook on b2 and free the rook along the seventh rank by b7-d7.
Alexey Dreev thwarted his opponent’s plans with:
35...f5
The stockade of black pawns will now not allow the white king to
reach his passed pawn.

186
But Black, as they say, has gone from the frying pan into the fire.
Magnus instantly – let me remind you that the game was played in
the World Rapid Championship – noticed the main flaw in Black’s
position, which appeared after the advance of the black f-pawn. Its
neighbour, the e6-pawn, suddenly became isolated! This means that
it’s time for White to move on to Plan 2 in order to win this ending.
But first we need to protect the g4-pawn:
36.h3! h5 37.gxh5+ xh5 38. b8 g6 39.b7 g7

187
White’s plan was successfully implemented. All he has to do is
approach the enemy e-pawn with the king and take it. But before
that, he needs to fix his own pawns:
40.e3! e5 41.f3 b1
Tactical ideas such as 41... b2+ 42. f1 b1+ 43. e2 b2+ 44. d3
( 44. d1 ) 44...e4+ 45.fxe4 g4 46.hxg4 fxg4 do not help Black;
after 47. c3 b1 48. c2 b5 49. d2 g3 50. e2, the white king
succeeds in stopping the enemy passed pawn.
42.e4 fxe4 43.fxe4 b3 44. f2
Dreev did not wait for the white king to reach d5 and simply
resigned.
Again, let us return to Grischuk - Vachier-Lagrave 46...a2!

188
Now how should White defend? Black threatens 47...h4 48.gxh4 g3
to open up the position of the white king for checks along the first
rank.
47. g5?
Four moves ago, White made a strategic mistake, but then there was
a significant margin of safety in his position to save himself in a
concrete tactical struggle; the position was still objectively equal.
Now his mistake is not strategic, but tactical. However, not a single
chess player can understand in a matter of seconds that the white
king does not have to hide in the shadow of the enemy pawns from
the seemingly decisive check of the black rook. It is much more
important for the king to control the e4-square, protecting the rear
from the opponent’s king. Only a computer can find this solution so
quickly!
With the correct move 47. f5!! White could have saved the game:
A) If Black responds with 47... e7, it makes sense for White to
change the defensive scheme: 48. e5+ d7 49. e2!.

189
 analysis diagram
The white king stands rooted to the spot on the square f5. He is
protected from the opponent’s attempt to remove his pawn cover,
which means he is protected from a deadly vertical check with the
f1-rook via a pawn breakthrough. In response to the continuation
49...h4 50.gxh4 g3, White has the important defensive resource
51.f3 or 51.f4. Black can only try to break the opponent’s defensive
stance with the help of a zugzwang.
Let’s imagine this position:

190
Black has managed to force the white king to leave the f5-square,
and he has reluctantly moved to f4. But the essence of the position
will not change. It lies in the possibility or impossibility of White
controlling the key square e4, which opens the way for the black
king to the f2-pawn! But, I repeat once again, it was impossible to
understand this in seconds;
B) Let’s continue our analysis with the move 47...h4.

191
 analysis diagram
It seems as if Black is winning. He will manage to expose the white
king and win the rook. But all is not so simple. White plays 48.gxh4
g3 49. a7+! (the black king must be driven as far as possible away
from his passed pawns) 49... e8 50. a8+ d7 51. a7+ c6
52.fxg3 f1+ 53. g6 a1 54. xa1 xa1.

192
 analysis diagram
Let us engage associative thinking on the basis of the material we
have just looked at. Somewhere we have seen something similar. Do
you remember where?
Let me help you. Remember Plan No. 1. After the prophylactic
move
6. c4!
, we reached the following position:

193
Then play continued
6... g4 7. b3 a6 8. c4+ xg3 9. a4 xa7 10. xa7 xh4
11. c2 g3 12. d1 h4
and we reach a position which should be remembered:

194
The white king manages to reach e1:
13. e1
So as to meet 13...h3 by occupying f1 – 14. f1, stopping the black
passed pawn.
Let us now compare the two positions:

position A

195
 position B
In position A, White wins after 13. e1. In position B after 55.h5
Black lacks one tempo to get the king to e8 and so the game ends in
a draw.
For a long time, I thought Grischuk’s mistake 47. g5? led to defeat,
and that 47. f5!! saved the game. But it turns out that White has
another drawing move.
During an individual training session with the Azeri women’s team,
one of their leaders, Ulvia Fatalieva, who has great imagination,
suggested another idea in the diagram position:

196
47. f5+!!. Now if 47... e6 White plays 48. e5+! d7 49. e2! with
a drawn position, because the black king cannot reach e4.
But what happens after 47... g6 48. g5+ h6...?

analysis diagram

197
It appears that White is losing. After all, in reply to 49. a5, Black,
by means of 49...h4, opens up the position of the white king. But
after 50. xg4!! (calmness, only calmness!) 50...h3 51.f4! we have a
draw. White gives checks on the 5th rank and if the black king goes
to the d-file, again, e2! follows.
Quite recently, at the European team championship in 2023, the first
round match between England and Georgia featured an endgame
quite similar to Grischuk-Vachier-Lagrave.
Luke McShane (2631)
Nikoloz Kacharava (2538)
European Team Championship, Budva 2023

Play continued:
43. f1 a2 44.a6 g7 45. g2 a3 46. f1 a2 47. e1 f6 48.f3
a3 49. e2 g7 50. f2 f6 51.g4

198
51...hxg4?!
All as in Grischuk-Vachier-Lagrave. It is interesting that also here,
the computer does not criticize this move, because it does not throw
away the draw. Of course, after 51... a4! Black’s task would be
rather easier.
52.fxg4 a4 53.g5+ f5
53... g7 is a losing move because of 54. a7!.
54.a7! f4+
Again, Black’s problems would be more easily solved with
54... f4!. The Georgian GM chooses the approach suggested by
Ulvia Fatalieva.
55. g2 g4+ 56. h3 a4 57.h5

199
57... a3+??
The decisive mistake. The players were both in time trouble. Correct
was 57... xg5 58.h6 f5!!, and if 59. g3, then 59... g4+ 60. f3
f4+ 61. e3?! e4+ and 62... e7, like in the previous example,
with reversed colours.
58. h2 a2+ 59. h1??
Nerves! Why advance the white king in order to retreat it into the
corner?
59... a1+??
Time pressure. After 59... g4! 60. g1 g3 61. f1 gxh5, Black
could have made a draw from a position of strength, but now it is all
over for him.
60. g2 a2+ 61. g1 a1+ 62. f2 a2+ 63. f1 a1+ 64. e2
a2+ 65. e1 a1+ 66. d2 a2+ 67. d1 a1+ 68. c2 gxh5
69.g6 fxg6 70. f8+ g4 71.a8 xa8 72. xa8 f3 73. d2 f2
74. h8 g5 75. xh5 g4 76. f5+ g2 77. e2 g3 78. g5 h2
79. f3 1-0

200
It only remains for us to see how the game ended after Grischuk’s
mistake 47. g5? , one that any GM could make in time trouble. The
game continued:
47... e6 48. a8 e5!
The black king does not allow his opposite number to get to f5.
Another way to win was 48... d5, walking to b2 and then f2.

49. e8+ d4 50. e2 d3 51. b2

We have already established that the essence of this position is
control over e4. If it were White’s move now, the game would end
in a draw after 52. f4. But after...
51... e4
the black king breaks into the enemy camp at g2 and takes part in
the extermination of the white pawns.
52. e2+ f3 53. b2 g2 54. c2 f1 55. xa2 xf2 0-1
I hope you gained a lot of useful knowledge from understanding this
complex and exciting endgame.

201
Now let’s try to consolidate our success – let’s practice one more
ending. During our analysis, we briefly discussed this position

and came to the conclusion that it is hopeless for Black, because

White has the e5-pawn squeezing the opponent’s position. Plan 1
will lead White to an easy victory.
Now let’s move the central pawns of both sides one rank down.

202
In this position, White will still have to carry out Plan 1 to play for a
win, since there are no isolated pawns in Black’s position. However,
White’s central pawn no longer squeezes the opponent’s pawn
structure, so Black does not have his usual second weakness,
namely the lack of space on the king’s side and, as a result, the
black king’s non-participation in the game. General reasoning
suggests that Black should have a good chance of a draw in this
position, since he does not have the second weakness.
Let’s do a quick analysis of the position.
1. e3 f6 2. d3 xf2 3. c7 a2
Obviously, in this position Black cannot grab the second pawn:
3... f3+? 4. c4 xg3 5.a7 a3 6. b5 and he achieves nothing
because of the threat of a bridge from the square e6.
4.a7 e6 5. c4 f5!

203
Black has managed to organize counterplay. But his position hangs
on an important tactical resource. It seems that after 6. b5 , he
loses. However, by means of 6...fxe4 7. c6+ d5 8. a6 b2+
9. a5 c5! 10. c6+ d4! ( 10... d5 11. c2 )
Black makes a draw.
It remains to examine the move 6. c6+ ( 6.exf5+ xf5 ) in the
position of the last diagram. After 6... f7 7.exf5 xa7 ( 7...gxf5??
8. c7+ e6 9. b5 ) 8. xg6 d7! 9. h6 d4+ 10. c3 g4 11. d3
( 11. xh5 f6 12. g5 xg5 13.hxg5+ xf5= ) 11... xg3 + 12. e4
g4+ 13. xe5 xh4 14. h7+ g8 15. a7 b4 (or 15... g4 ) it is
again a draw.
Now I propose to make the task harder. In the position where we
started our analysis, let’s move the white pawn from h4 to g5 and
try to analyse that position.

204
The main factor in playing for a win for White is the g5-pawn
squeezing the opponent’s position. The black pawn on e5, although
it has formal characteristics of being isolated, is not completely
isolated, since the e6-square is under the control of the f7-pawn.
Without this square, the white king will not be able to get to f6 and
put the opponent in a zugzwang position when carrying out Plan No.
2.

205
Black is in zugzwang.
This means that we must choose Plan 1 – bring the king away from
the pawns and free the rook on the seventh rank. Proceeding
directly:
1. e3 a3+ 2. e2 a2+ 3. d3 xf2

206
We see that in order to carry out our plan we have to sacrifice not
one but two kingside pawns. Therefore we play not 4. c7, but:
4. b7! ,
by analogy with the game Grischuk-Vachier-Lagrave. Remember
this device!
4... f3+ 5. c4 xg3
If 5... a3 6.a7, it is time for Black to resign.
6. b3!

207
Again a familiar motif. Other continuations throw away the
advantage.
6... g1 7. a3 c1+ 8. d5 c8 9.a7 a8 10. c6! f6 11.gxf6+
xf6 12. b7 xa7+ 13. xa7 h4 14. c6 g5 15. d5 g4 16. h7 h3
17. h8!
And White wins.
The end of the game Grischuk - Vachier-Lagrave is very important
for understanding the methods of play in a position in which the
stronger side has an extra a-pawn on the queenside, and on the
kingside each side has three pawns in a standard position. If you
carefully analyse the techniques of play in this endgame, they will
become your guiding star in such positions. For this reason, we will
recall this fight more than once in our book, using associative
thinking.
In my opinion, this endgame is very instructive, first of all, in the
method of defence chosen by Grischuk. In such situations, a chess
player has to take one of two paths:

208
 1. Allow the king to the starting square e3 (e6) to move away from
his kingside pawns. In this case, you need to know well the
‘squeeze’ from the analyses of Dautov and Dvoretsky, not make a
single mistake in playing this ending, and make a draw with only
moves.
2. Choose the method proposed in a conversation with me by
Alexei Sarana and chosen in the game by Alexander Grischuk. But
at the same time, avoid making the mistake associated with the
move 43.hxg5? instead of the correct 43. a5!.
Personally, I think that the second method of defence is much more
rational. If the first method requires mechanical memorization of
variations, then in the second a logical explanation (the enemy king
must be forced to break away from the h5-pawn) replaces the load
on the memory.
Let’s look at the consequences of choosing an irrational method of
defence:
Alireza Firouzja (2759)
Evgeny Najer (2654)
Riga 2021

209
Here you can appreciate the position of the black pawn on g5. It has
captured space on the kingside and fixes the white pawns on f3 and
h3.
The only reasonable winning plan for White is to break through with
the king to the queenside and give up one or two pawns on the
kingside.
Why doesn’t Black use his advantage in space and deprive White of
this plan by playing 42...h5...? After all, if White loses a pawn on
the kingside, Black will have ready counterplay on this part of the
board thanks to the active g- and h-pawns. And the move 43.h4 is
not dangerous for Black because of the simple 43...f6. There could
follow 44.hxg5 fxg5 45. b7+ g6 46. a7 a2 47. a8 g7 48.a7
a4.

210
 analysis diagram
White’s only threat – the creation of a passed pawn on the f-file – is
easily repelled by Black. Exactly this ending with a change of
colours arose in the game Shafigullin-Pogonina, Izhevsk 2019. It is
analysed in detail in my book Endgame Strategy .
49. h2 h7 ( 49...h4 ) 50.g3 a2+ 51. h3 g4+!. Thus, Black does
not allow the white king to be separated from the kingside pawns
and easily achieves a draw.
But Najer played:
42... a2?!
As it turned out later, this move does not lose the game. But I don’t
like it. Now Black will have to calculate the consequences of pawn
races. After all, the white rook is not on a7 – that is, it has already
been ‘freed’; however, along the sixth, and not the seventh, rank.
Everything can be decided by one tempo.
43.f4 gxf4 44. xf4 xg2 45. e5 a2 46. d6 f5

211
All Black’s hopes are now connected only with this pawn. Neither
the king nor the other black pawn can help it.
If we are talking about playing in sharp rook endings, then this rule
applies: the king must go with his own pawn, and the rook fights the
enemy pawns. According to this rule, White’s position should be
won, unlike the position in the previous diagram after Black’s move
42...h5!. But sharp endings are sharp because there are many
exceptions to the main rules of playing them.
Visually, Black’s position looks very bad, but there are some details
that are invisible from afar that allow him to count on salvation.
47. c7 f4 48. b8 f3 49.a7 f2 50. b1

This development of events could have been predicted. The passed

pawns have simultaneously reached the penultimate rank, but
White’s is supported by his king, and the black king cannot reach
the f2-pawn. It seems that the only chance for salvation is the rook
sacrifice 50... b2+. Otherwise, it’s time to give up.

212
Evgeniy Najer was probably very tired, he had little time to think,
and he did not believe in the possibility of perpetual check. He
didn’t want to admit defeat, so he played:
50...f1 ?
In vain! Firstly, after 50... b2+ 51. xb2 f1 52.a8 f8+ or
52... f4+, the engine sees no way for White to avoid perpetual
check or losing the rook on b2. But a person is not a computer and,
under time pressure, will most likely lose. Secondly, and this is the
main thing, a draw could be achieved in another, much simpler and
more natural way: 50... g6 51.a8 xa8+ 52. xa8 g5 53. f1
h4 54. xf2 xh3.

analysis diagram
Paradoxically, there is a draw on the board, although it is hard to
believe:
51. xf1 b2+ 52. a8 b3 53. c1 g6 54. c7

213
It would have been more technical to play 54. c6+ g7 55.h4 b4
56.h5 b5 57.h6+ f7 58. c8 g6 59. b8 f5 60. b6+, but the
move in the game also leads to victory.
54...h5 55. b7 g3 56. b6+ g5 57. b7 a3 58. a6 b3+
59. c7 1-0
As we have just seen, preventing the enemy king from breaking
away from his pawns can be much easier than calculating the
consequences of this distance in sharp tempo endings. Now we will
again continue to observe complex rook endings using associative
thinking.
Let’s look at the ending of the game Andersson-Hübner, which is
most important for understanding the methods of playing in a rook
endgame.
Ulf Andersson (2640)
Robert Hübner (2580)
Ter Apel 1997

If the black f6-pawn had been on f7, and the king on the g7-square,
the game would undoubtedly have ended in a draw. However,
214
Black’s pawn structure is compromised and Ulf Andersson outlines
a deep plan that leads to success.
As we already know, in such positions, with the help of thinking in
schemes, one can look far into the future.
38.g4!
Creating the preconditions for isolating the pawn on e4 or f5. White
probably plans to advance the passed pawn to a7 with his rook on a8
and then use zugzwang to win the pawn e4/f5.
38...g5
It was scarcely any better to give the white king a path to the square
f4 after 38...f5 39.gxf5 gxf5.
39.a4 a1 40.a5 a2 41.a6 g6 42. a8 a1 43.a7 g7

An interesting position. If the black rook were on a2 and the white

king on f1, everything would be absolutely clear. The white king
would come to d5 and then, with the aid of zugzwang, White would
win another pawn.

215
Here is a sample variation of such a development of events: 44. h2
a2 45. g1! h6 46. f1 h7 47. e1 g7 48. d1 h7 49. c1
g7 50. b1 a5 51. b2 h7 52. b3 g7 53. b4 a1 54. c5
a2 ( 54... h7 55. d5; 54... a5+ or 54... a4 55. b6! ) 55. d4!
a4+ 56. d5 h7 57. e6 g7 58. f5.

analysis diagram
Zugzwang! White wins a second pawn and with it, the game.
Returning to the game:
44. g3
This is possible, but 44. h2!? leads to the intended aim one move
quicker.
44... g1+
It was better to play 44... a2!, in the hope of provoking White into
the active continuation 45.h4?. The further course of the game
shows that after the exchange of the white pawn on a7 for the black
one on f6, the white h-pawn is better placed on h3 than h5,

216
paradoxical though this is. After the move 44... a2 White should
retreat his king with 45. g2, with the threat of 46. f1.
45. h2 a1 46. g2
By triangulation, Andersson puts his opponent in the first zugzwang.
46...h6 47. h2 h7
47... a2 48. g1 a1+ 49. g2 is the second zugzwang.

This is the position for which the great Swedish endgame master
had been striving. With some beautiful manoeuvres, White has
created the possibility of exchanging the pawn on a7 for the pawn
on f6.
48. f8?! xa7 49. xf6

217
49... g7?
This natural move is the decisive mistake. After 49... e7! Black
retains every chance of a successful defence. From e7, the black
rook would not only defend the pawn on e4 but also cover the
seventh rank, not allowing White to use it freely to attack the h6-
pawn from h7.
50. e6 a4 51. g3 b4 52.h4

218
I think it was this precise position that Andersson was striving for.
White will give Black a weak pawn on either h6 or g5, temporarily
retaining the pawn tension, by not advancing the h4-pawn to h5.
Then the black position is gradually pulled apart on the principle of
two weaknesses.
52...gxh4+
Let us examine the attempt to maintain waiting tactics: 52... c4.
Now not 53.h5? b4 54. e7+ f8 55. d7 b6, and Black holds,
because the weakness on h6 would have been fixed prematurely.
After all, the point of White’s attack is that he has no need to hurry
in changing the position of the pawn on h4! That way, Black has to
be prepared for both h4-h5 and h4xg5, and this already clearly
strains him.
Therefore, correct is 53. e7+! (instead of 53.h5? ) 53... f6 54. h7
g6 55. d7! c6 56. e7 c4 57.h5+ f6 58. h7, and White wins.
53. xh4
Now the white king has access to f4.

219
53... a4 54. g3 b4

White just needs to find the decisive regrouping of his forces and
Andersson easily copes with the task.
55. e5! g6 56. f5 b8
Otherwise 57. f4 and 58.f3.
57. f4 e8 58. e5! xe5
58... f8+ 59. g3.
59. xe5 g5 60. xe4 xg4 61. d3 1-0
In the variation 61... f3 ( 61...h5 62. e2 h3 63. f1 ) Black gets
mated: 62.e4 h5 ( 62... f4 63.f3 ) 63.e5 h4 64.e6 h3 65.e7 h2
66.e8 xf2 67. e4 g1 68. g4+ f2 69. h3 g1 70. g3+
h1 71. f3+ g1 72. e2 h1 73. f2#.
After seeing 49... e7, and not being able to find a precise winning
plan for White, I decided to look for another way to win. Let us
return to the position after 47... h7.

220
White does not have a great choice. Other than the exchange of a7
for f6, we can try exchanging the f2-pawn for the black pawn on e4.
To start with, it is necessary to move the king from h2 to g2, playing
48. g2 g7 49.f3 exf3+ 50. xf3.

analysis diagram
221
Then play could develop according to the following scheme:
50... a4 51.e4 a3+ 52. e2 a2+ 53. d3 a3+ 54. c4 a4+
55. b5 a1 56. c6.

analysis diagram
Black cannot avoid the transition into the pawn ending, the point of
which requires explanation. 56... c1+ (otherwise 57. d8) 57. d6
a1 58. c8 xa7 59. c7+ xc7 60. xc7.

222
 analysis diagram
Black would not be lost in this ending if there were no pawns on h3
and h6 on the board, and it was White’s turn to move. Why?
In this case, the black king would have the additional free square h6.
The whole point is that when the white king attacks the f6-pawn,
Black must be able to defend it while maintaining the opposition.
The corresponding squares are e6 – g6, e7 – g7, e8 – g8, c7 – g7, d7
– h7. When the white king is positioned on d7, from which the three
most important squares e8, e7 and e6 can be attacked, the black king
must be on h7 with White to move. If there are pawns on the h-file,
the white king makes a move to the d6-square, but the black king
does not have access to the h6-square. He has to retreat to h8, but
now after the moves e6 – g7 and e7, Black ends up in
zugzwang and loses.
Black’s attempt to exchange the h-pawns in the position of the
diagram does not bring success. After 60...h5 61.gxh5 ( 61. d7??
hxg4 62.hxg4 h7!= ) 61... h6 62. d7 xh5 63. e7 g6 64. f8
White wins.
223
I was confident that I had found a clear path to the win. But
International Master Vasily Gagarin, who took part in editing the
book Endgame Strategy, made an important clarification, for which
I am sincerely grateful to him. In the position in the analysis
diagram after 50. xf3, the move 50... a4? is bad. Correct is
50...h5! 51.gxh5 (otherwise after the exchange of pawns on g4
Black need not fear the pawn endgame, as shown in the commentary
above) 51...f5!,

analysis diagram
and the best White can hope for is a difficult but completely drawn
endgame.
From a practical point of view, Ulf Andersson was right to exchange
the a7-pawn for the f6-pawn. Robert Hübner went wrong in playing
49... g7 instead of 49... e7. We can only speculate whether he
would have found the move 50...h5! in the position of the last
diagram. In both cases, the probability of a mistake by Black is very
high.

224
It seems that everything is clear. The position after 38.g4 is a draw,
but White has two approximately equal opportunities to win. One of
them was demonstrated in the game by Ulf Andersson. The second
possibility is to move to the pawn ending. In it, as Vasily Gagarin
pointed out, Black also achieves a draw by 50...h5!. Both of these
defensive possibilities were discovered retrospectively, by
reviewing and studying the course of events in the game.
During the next online session of the Kramnik school, I discussed
this ending with a group of players from the eastern regions of
Russia. The next day after the class, Alexey Grachev
I would like to present to you the main ideas of
Grachev
’s analysis. We start again after Black’s move 47... h7.

As we have already seen in the game, White uses zugzwang to

ensure that the black king leaves the g7-square for h7, after which
White can exchange his passed a7-pawn for the f6-pawn.
Here the Swedish grandmaster made the move 48. f8 , but the
correct decision was 48.h4!!.

225
Now White threatens to play 49.h5!, fixing the black pawn on h6. In
this way, he is going to secure a transition to a won pawn ending.
The subsequent undermining of the e4-pawn by f2-f3 cannot be
prevented. The idea of Black’s counterplay with ...h6-h5, proposed
by Vasily Gagarin, will become impossible if the white pawn moves
to the h5-square. Therefore Black is forced to take the pawn:
48...gxh4 .
Only now does White play 49. f8! and after 49... xa7 50. xf6 we
reach the following position:

analysis diagram
With a subtle move order, White has forced his opponent to move
the g5-pawn to h4. Now, as we know from the events in the game,
White needs only to move the rook to f5 and take the h4-pawn with
the king. Black is unable to prevent this, since 50... a5 is still met
by 51. f5! , and the pawn ending after 51... xf5 52.gxf5 is hopeless
for Black.
Conclusion: in the position from which we began our analysis of the
Andersson-Hübner ending, White has a clear path to victory. Of
226
course, it was discovered only after carefully studying what
happened in the game.
The next ending which we will consider, in its logical content, is in
some points similar to Andersson-Hübner, and in some other ways
to Grischuk-Vachier-Lagrave.
Anton Demchenko (2600)
Levon Aronian (2782)
Riga 2021

This position can be easily assessed at first glance. Black has a weak
pawn on e6, a bad king, and a pawn less than his opponent. By all
indications, his position is absolutely lost.
If we talk about White’s winning plan, then the isolated black pawn
on e6 serves as a reference point for carrying out Plan No. 2. Thus,
White needs to place the rook on a8, the pawn on a7 and, with the
help of zugzwang, win the black pawn along the e-file and then
create a passed pawn on the f-file.
We can also try to carry out Plan No. 1 – break away with the king
from the kingside pawns and free the rook along the seventh rank.

227
Of course, first White will have to advance his passed pawn to a6.
This plan should also lead to success for White. After all, his king
does not effectively break away from anything – the opponent will
not be able to capture any of the kingside pawns, since then White’s
rook will move to c7 or b7. Because of the terrible position of the
black king, the white passed pawn will queen.
But it’s not without reason that they say that the best is the enemy of
the good. The right to choose also means the right to make mistakes.
Instead of carrying out the standard Plan No. 2 for such situations,
Anton Demchenko decided to break away from the f2-pawn with his
king. Levon Aronian managed to find defensive resources. After
mutual mistakes, the game ended in a draw, although Black’s
position was objectively lost for a long time.
What was White’s fundamental mistake? He chose Plan No. 1, with
which the game soon took on a calculative character. But the
opponents had already passed the first time control and were playing
on increment, and they were of course very tired. In this state, it is
much easier to think in schemes than to calculate long variations.
And as we have already noted, Plan No. 2 would be very useful
here, with which the calculation of variations would have been
reduced to a minimum.
Let’s analyse this instructive ending.
42...h5!
Black seeks counterplay. He threatens to shut out the enemy king
with 43... a1+ 44. h2 h4. White reacted with an intuitive move:
43.h4?
A serious mistake. Why? He gives up the chance to carry out Plan
no. 2!
Let us try to explain this assertion. In order to do so, we must use
associative thinking and remember the following endgame.
Lev Polugaevsky (2585)

228
Evgeny Vasiukov (2509)
Tbilisi 1966

In this interesting position, with the help of schematic thinking, we

can look far into the future. White has the opportunity to bring his
passed pawn to the b7-square. The black rook will have to go to the
rear of the passed pawn and stand on b4. Then White will sneak his
king onto d5 and win the black pawn on e5. Then all he has to do is
create a passed pawn on the f-file and force Black’s king from the
safe g7-square with the f5-f6+ check:

229
 sample position
But in this apparently simple element of a lengthy piece of
schematic thinking, an unexpected problem arises. It is enough for
Black to advance his pawn on h6 one square to prevent his
opponent’s idea.
Imagine this position:

230
White will not be able to exchange his g3-pawn for the pawn on h5,
because checks by the black rook on rank or file will prevent this.
Even with inaccurate play by the opponent, the best White can reach
is something like this position:

1.g4 a3+ 2. g2 hxg4?! 3.h5 gxh5 4.f5.

231
It seems that White will just win by pushing his f-pawn. But Black
can save himself in an instructive way: 4... a2+ 5. g3 a3+!. Here
it is important to be able to explain to oneself in words what the
defensive method consists of! Black strives t o drive the enemy king
away from his passed pawn on g4 . 6. h4 (if 6. f4 g3 7.f6+ f7
the black passed pawn is in no way weaker than White’s passed
pawns) 6... h7! 7.f6 g6. Black threatens mate on h3; White has to
force a draw with 8. g8+.
We should note that 4...g3? loses the game after 5. h3.
42.b5?!
Now Vasiukov, instead of 42...h5! with a drawn position, replied:
42... b4 43.h5!
Exploiting his chance; Polugaevsky won after a less than faultless
struggle.
This device has a place in an endgame we have already examined,
Spassky-Torre. After White’s 47th move, this position arose:
Boris Spassky (2625)
Eugenio Torre (2535)
232
Hamburg 1892

The Filipino GM played 47... g6? .

White replied 48.f4 , and Black could, with a clear conscience,
already admit defeat here. He faced a sad choice: either exchange on
f4 himself, or allow the exchange of pawns on the e5-square. In the
first case, the outcome of the game would have been decided by the
passed white pawn on the e-file: due to the stupid position of the
black pawn on f6, the e4-e5 breakthrough could not be prevented. If
the black kingside pawns were on the f7, g6 and h5-squares, then
White would not have such an opportunity.
After 48... f7 49.fxe5 fxe5 50. f1 Spassky approached the e5-
pawn, won it, and advanced the e4-pawn.
Black’s only practical chance in the diagram position was 47...g5! ,
preventing the advance f3-f4. Now White would have had to play
48. f8! and exchange the a7-pawn for the f6-pawn. We have
already observed all this in the game Shuvalova-Mammadzada.
After 48... xa7 49. xf6 gxh4 50.gxh4 White could place the rook
on f5 and move the king to c4, keeping good winning chances.

233
But isn’t it possible to win this position somehow more easily, using
tactics? After all, in response to the move 47...g5, we can exchange
with 48.hxg5 fxg5 and play 49.f4.

analysis diagram
This looks lovely, but after the correct reaction from the defender,
White loses all of his advantage.
How does Black make a draw?
We are helped in this by an association with a variation from the
game Polugaevsky-Vasiukov.
Polugaevsky-Vasiukov

234
 4... a2+ 5. g3 a3+!
Spassky - Torre

49...exf4! 50.gxf4 a1+

235
49...exf4! 50.gxf4 a1+!. We drive the white king away from the g-
pawn ! On the kingside, he cannot escape from checks and after a
line such as 51. f2 a2+ 52. e3 a3+ 53. d4 g4!? (or 53...gxf4 ),
White is the one who must concern himself with securing a draw.
Associative thinking in operation. Remember this device!
Let us return to the game Demchenko - Aronian
Instead of the natural but inaccurate move 43.h4?, it was logical to
play 43.g3!, retaining the option of carrying out Plan 2 and
organizing a passed pawn on the f-file.

analysis diagram
Analysing further, let us assume that Black plays 43...g5, with the
positional threat 44... a1+ 45. g2 g4, shutting the white king out
of the game. Then White has two winning paths. The most
consistent continuation is 44.f3!. However, White also wins after
44. f1!. There could follow 44... f8 45.a6.

236
 analysis diagram
A) In the event of 45... e8, the simplest is 46. g7, although it is
also possible to play 46. a8+!? d7 ( 46... f7 47.a7 g7 takes
play into a position we have already examined) 47.a7 c7 48. g8
xa7 49. g7+ b8 50. xa7 xa7 51.f4 g4 52.hxg4 hxg4 53.f5!
exf5 54. f2, winning;
B) 45...g4 is the best practical chance: 46.hxg4 hxg4 47. a8+
g7 48.a7. Next, White moves his king to c6 and, using zugzwang,
wins the e6-pawn. It is best for Black to leave the pawn on e6. As
soon as it advances to e5 or e4, it will immediately protect the white
king from vertical checks along the e-file, and any pawn endgame
will become hopeless for Black.
Let’s imagine this position:

237
White plays 1. d8!. After 1... a6+ 2. d6 xa7 3. d7+ xd7
4. xd7 it is time for Black to resign.
But what happens if White takes the e6-pawn with his king and
Black checks from e2?

238
White wins this position as follows: 1. d5 d2+ 2. e4 e2+
3. d3 a2 4. e3!?.

analysis diagram
Zugzwang! The black rook has to leave the second rank, or else the
black king must go to h7 and allow the white rook to reach f8.
A) 4... a4 5. e8 xa7 6. e4 a3+ 7. e2;
B) 4... h7 5. f8 xa7 6. f4 g7 7. f5.
Why does White win the game relatively easily by destroying the
black e-pawn with his king? Because the black pawn on g4 has
moved far away from its king, which occupies safe squares on g7 or
h7.
Looking ahead, I note that if the black pawn were on g5, and the
white pawns were on f3 and g4,

239
then the capture 1. xe6? would lead immediately to a drawn
position after 1... e3+ or 1... a6+.
We have convinced ourselves that after the moves 43.g3! and
44. f1!, to win the game, White must employ schematic thinking
and display a certain degree of accuracy.
Now let us analyse the move 44.f3!, which is logical and also leads
to success for White.

240
 analysis diagram
So let’s begin! 44... a3! 45. f2! (it is important not to fall for the
trap 45.a6? g4! ) 45... f8 46.a6.
Now let us have a bit of analytical training. Assume that Black plays
46... g8 47. a8+ g7 48.a7 e5.

241
 analysis diagram
The analysis of this line is very useful. By studying it carefully, we
can pick up a number of valuable skills, widen our technical arsenal
and develop our associative thinking.
Try to describe White’s plan in words. Have a little think and only
then look at the answer.
Answer: White should immediately play 49.g4! hxg4 50.hxg4, and
then, without touching the e5-pawn, transfer his king to c6 and go
into a pawn ending with the move c8 or d8.
And now we return to our analysis. Instead of 46... g8, Black plays
46... e8 :

242
 analysis diagram
47. a8+ f7 ( 47... d7 48.a7 c7 49. g8 xa7 50. g7+ b6
51. xa7 xa7 52. e3 ) 48.a7 g7. Here the win is quite simple.
White will win the isolated pawn on e6 with the aid of zugzwang
and then will create a passed pawn on the f-file. The most precise
way is with the move 49.h4!, after which the picture becomes
completely clear:

243
However, I am interested in what happens if, instead of the correct
moves 49.h4 or 49. e2, White plays 49.g4?! hxg4 50.hxg4.
This position (diagram A) differs fundamentally from the one we
looked at recently (diagram B), in which the white pawns on f3 and
g4, and Black pawn on g5 are each moved down one rank.

244
 diagram A

diagram B
The main difference between these positions is that in diagram B the
black pawn on g4 is two ranks away from the safe squares for the
black king h7 and g7 , but in diagram A it is only one rank away.
As we have already seen, in the position in diagram B, White’s win
is quite simple. In the position in diagram A, it is not at all clear
whether it can be achieved at all. I had to puzzle over this question
for a long time and use the engine to solve a genuine study.
Looking ahead, I will say: winning is indeed possible, and it is very
useful for studying. If you can understand it and explain it to
yourself, you will certainly make serious progress in understanding
rook and pawn endings.
To solve our study, let’s immediately start using schematic thinking
and remember the truths we have derived.
Let's imagine this position:

245
1... d3+ 2. xe6? e3+ and despite White’s two extra pawns, 3. f5
xf3+ 4. xg5 a3 leads to a theoretical draw.
Here is another important position:

White wins after 1. d8! ( 1. xe5?? e3+! ) 1... a6+ 2. d6 xa7

3. d7+.
246
So how do we win if Black does not advance his pawn from e6? We
have to try to put the opponent into zugzwang with the pawn on e6.
Imagine the following position:

1. b6!. White does not go after the e6-pawn but triangulates with
the king to transfer the move to Black.
1... b3+ 2. a5 a3+ (the threat was 3. b8) 3. b5!!. Black is in
zugzwang. His rook must leave the third rank, since he can’t play
3... b3+ because of 4. a4 and 5. g8+. Otherwise, the black king
must go to h7.
Let us consider both possibilities:
1) If the rook leaves the third rank: 3... a1, there follows 4. b6
b1+ 5. c6 a1 ( 5... c1+ 6. d6 d1+ 7. xe6 e1+ 8. f5, and
Black cannot take on f3 with check) 6. c8! xa7 7. c7+ xc7+
8. xc7.

247
 analysis diagram
The position is similar to that in Andersson-Hübner. The white king
takes control of the three key squares around the e6-pawn – d6, d7
and d8. Black loses whatever move he plays, but with White to
move it is a draw.
There could follow 8... f6 ( 8... g6 9. d8! ) 9. d6 f7 10. d7
f6 11. e8 e5 ( 11...e5 12. f8 e4 13.fxe4 e5 14. f7 xe4
15. g6 f4 16. h5 ) 12. f7. White wins;
2) Now let us look at 3... h7.

248
 analysis diagram
In this case, the white king advances decisively: 4. b6 b3+ 5. c6
c3+ 6. d7 d3+ 7. e7! a3 8.f4! gxf4 9. f6!. The black pawn
on f4 shelters White’s king from checks along the file and the black
king comes under a mating attack.
9... a5 10.g5 f5+ 11. xe6 a5 12. f6 a6+ 13. f7, winning.
It only remains to see what happens if the black rook returns to the
third rank at move 6: 6... a3!.

249
 analysis diagram
7. d6!. Another zugzwang. Now, 7... g7 is not possible because of
8. c8, whilst after 7... a1 White wins with 8. xe6. 7... a6+ 8. e7!
g7 9. d8! (threatening 10. c8; not 9. d8? xa7+ 10. d7 a6! or
immediately 10... a3 ) 9... h7 10. c8! (threat: 11. b7) 10... a3
11. b8! xa7 12. b7+ xb7 13. xb7.

250
 analysis diagram
White has the distant opposition. On 13... g6, there follows
14. c6. Correspondingly, on 13... g7 there is 14. c7, whilst after
13... g8 – 14. c8. The computer promises that the black king will
be mated in no more than 25 moves.
We have made a detailed analysis of this interesting endgame and
described two precise paths to the win, with the help of schematic
thinking. It remains only to see how events developed in the game
Demchenko-Aronian

251
Black has pinned his opponent down on the kingside and launched a
psychological attack. To win, White needs to show exact play.
51. c1?
He wins with 51. a8+ c7 52. e1, and Black finds himself in a
curious zugzwang. On 52... b6, the reply 53. g8 decides.
The game continued:
51... c8 52. e7 xa6 53. d2

252
White has squandered a great part of his advantage. Now Aronian
could have set up a fortress by means of 53... d8! 54. xe4 d7
55. e3 a5. Instead, he played
53... f6? 54. e2 d8 55. xe4 d7 56. e3
and the black rook is in an unfortunate position on f6. White again
had chances to win, but was unable to realize them. In the
subsequent struggle, both sides made mistakes and the game ended
in a draw, but from a methodological point of view this is no longer
of much interest to us.
We saw how White could win the game relatively easily in two
ways using schematic thinking. But the strong grandmaster, 2021
European champion Anton Demchenko, relaxed a little, made
several superficial moves and missed a well-deserved victory.
Alireza Firouzja (2770)
Shakhriyar Mamedyarov (2765)
European Team Championship, Terme Catez 2021

253
Playing for France on the first board in the European Team
Championship, the young French-Iranian super-talent achieved an
excellent result. And his victory in the last round against Shakhriyar
Mamedyarov took him to second place in the world rankings. In this
game, the Azeri team needed a draw for an overall victory in the
tournament. The sporting significance of the battle was enormous.
White is faced with a choice. He can continue the game in the rook
ending with 31. h4 or bring the king to the centre with 31. f1. But
then the game will most likely go into a pawn ending after 31... h7
32. xh7 xh7.
White had to solve the exchange problem intuitively, since there
was no time to think. But even if there was time, you would still
have to trust your intuition or act at random, since calculating the
consequences of going into a pawn ending is completely unrealistic
here.
In the game, 31. h4 was played. I do not even know how to
punctuate this move. It probably throws away the win, but
subjectively this rook move brought White victory. At the same
time, 31. f1 would probably have won the pawn ending, but to find
254
this win, especially in time trouble, is practically impossible. The
winner is always right! So we should give the move an exclamation
mark:
31. h4! d7
The site ChessPro made an interesting observation on this move:
‘Now we get a festival of who can eat pawns faster’.
32. xg4 xd3 33. xg6+ f7 34. c6 d2 35.c4 xb2 36. xc7+
f6 37. c6+ f5 38.cxb5 b3 39.b6 xa3 40.b7 b3 41. a6
xb7 42. xa4

The board has emptied. With correct defence, Black should make a
draw. My computer assesses White’s advantage as 0.46. But Black
must play carefully and accurately.
42...e4?
A mistake in principle, which complicates what should be a
relatively simple path to the draw. Ironically, Firouzja himself
demonstrated the correct method of play six months later against
another top Azeri GM at the Candidates tournament.

255
Teimour Radjabov (2753)
Alireza Firouzja (2804)
Madrid Candidates 2022

41... d6 42. f2 e5 43.g4 e6 44. f5 a3 45. f8 e7 46. b8

f7 47. c8 e7 48. c2 f6 49. e2 b3 50. e3 b2+ 51. g3
c2 52. a3 b2 53. a6+ f7 54. h6 b3 55. f2 a3 56. b6
e7 57. e2 c3 58. f2 a3 59. b2 f6 60. e2 b3 61. e3
b2+ 62. g3 c2 63. b3 a2 64. b6+ f7 65.g5 a4 66. f6+
e7 67. h6 f7 68. h4 a5 69. g4 a4+ 70. h5 a3 71.f4
exf4 ½-½
Why do we mark 42...e4 as a mistake in principle? It is because this
move increases the scope of the white rook on the eighth rank. At
the same time, the defensive capabilities of the black king are
narrowed, for he wants to control the f8 and e8-squares and also to
be able to defend his e-pawn. And for Black’s king, the distance
from the square f7 to e4 is doubled compared to e5.

256
With the help of associative thinking, we recall the endgame
Demchenko - Aronian , which we just analysed. In one version of
the analysis, two positions emerged:

diagram A

257
 diagram B
Both of these positions are winning for White. But if the second is a
relatively easy win, the first involves a great deal of difficulty.
Now let us compare two positions from Firouzja-Mamedyarov.

diagram C

258
 diagram D
Both are drawn, but whereas the second is relatively easy to hold,
the first is much harder.
Returning to Firouzja - Mamedyarov after the move 42...e4?

259
The consequences of the move 42... e4 are clear. After the transfer
of the king to g2, the advance of the passed pawn on g3 will decide
the game, if White’s rook remains on the f-file. Black can prevent
this with the move 48... f3!! – again a pawn endgame, this time
drawn. If 49. xf3 ( 49. e8+ d5 50. g2 f7 51.g4 f4 ), then
49...exf3 50. g1 f5 51. h1 g5 with a draw.
48... d4 49. g2 b7 50. h3 e5 51.g4 1-0
Now let us look at the pawn endgame which could have arisen after
31. f1 h7 32. xh7 xh7.

analysis diagram
Play could proceed as follows: 33. e2 g7 34.c4 bxc4 (not 34...c6
35.cxb5 cxb5 36. d2 ) 35.dxc4 f6 36. d2 f5.

260
 analysis diagram
White can create a passed pawn on the a-file and seemingly win the
game. But, as we have seen many times, pawn endings can be very
difficult. And the most obvious plan is not always the right one.
Does the most straightforward move 37.b3 lead to White’s victory
in this position? In this case, the game develops in the following
direction: 37...axb3 38. c3 ( 38.a4 e6 39. c3 d7 40. xb3 c6
41.a5 b7 42. b4 a6= ) 38... e6, reducing the matter to the
previous variation with 38.a4.
But instead of the move 37.b3?! White has the subtle move
37. d3!. The idea is that White is trying to push the enemy king
away from the e4-square. After Black’s natural response 37... f6,
we get a pawn race and a queen ending the game can continue as
follows in which the white queen takes the pawn on c7 but Black
may be able to draw after all. However, Black can avoid all this by
playing 37...c5! instead of 37... f6, giving back the tempo to White
and preventing the later capture on c7. After 38.b3 axb3 39. c3 (
39.a4 e6 40. c3 d6 41. xb3 c6 ) 39... e4 40.a4 f3 31.a5
261
 xf2 32.a6 e4 33.a7 b2! (to deflect the king) 34. xb2 e3 35.a8 e2
there appears to be no win.
However, one should not underestimate the opportunities that open
up with very deep and accurate calculations of variations for modern
computer programs. The modern Stockfish, having absorbed some
of the capabilities of AlphaZero, has added a ‘human’ perception of
the position.
For example, in the elementary position in the following diagram:

Just recently, the computer score could have been +3.0, but today’s
engine instantly draws zeros. With the breakthrough in computer
technology, the chess player faces another problem. To explain the
engine’s recommendations, you must have decent chess
qualifications.
Modern chess programs sometimes help us to find interesting, I
would say unexpected, logical solutions.
Remember the training exercise we went through recently?

262
Analysing this position using Stockfish 14, I reasoned as follows:
‘White will still have to carry out Plan No. 1 to play for a win, since
there are no isolated pawns in Black’s position...’
Quite recently, my student, FIDE Master Kirill Shoshin, showed me
a completely new opportunity to play for a win, which he found
using a more modern engine. As it turns out, White can isolate the
black f7-pawn for a short time and then attack it! Such an idea is
simply inaccessible to human logic.
Kirill Shoshin proved that in the position in the following diagram,

263
besides playing 1.f3 and bringing up the king, White can also win
with a paradoxical move: 1.a7!!, utilizing several beautiful and
paradoxical manoeuvres.
Let us follow the young master’s analysis, with my explanations.
In order to understand what is paradoxical about the move 1.a7!!, it
is useful to remember the ending of the game McShane -
Kacharava (Game 33 in this chapter), where the two sides each had
three pawns on the kingside:

264
Without a pawn on the e-file, the move 1.a7? makes no sense and
deprives White of any winning ideas.
From a human viewpoint, the presence of the pawns on e4 and e5
does not prompt one to consider 1.a7!! , - it seems just as pointless
as in McShane-Kacharava. However, in reality, the move wins the
game, and in order to convince ourselves of this, we continue to
follow Shoshin’s analysis:
1... a3 (if 1...f5, White replies 2.f3 and with the help of zugzwang
wins the e5-pawn by bringing up his king) 2.f3! (an important move,
essential to White in all variations) 2... a2+ 3. f1

265
 analysis diagram
3... h7 (the move 3...f6 loses here after 4.f4!, by analogy with
Spassky-Torre) 4. e1 (the white king heads towards the key square
d5; Black cannot stop this) 4... g7 5. d1 h7 6. c1 g7 7. b1
a3 8. b2 a5 9. b3 a1 10. c4 a5 11. b4 a1 12. c5.

266
 analysis diagram
In this position, Black can either put the rook behind the enemy
position with 12... a2, or make a waiting move with his king:
12... h7 or 12... f6. Suppose he plays 12... h7. White continues
his plan: 13. d5 a5+ 14. d6 g7 15.g4!.

267
 analysis diagram
What should Black do? He is caught in a curious zugzwang. The
rook cannot leave the defence of the e5-pawn. He only has pawn or
king moves.
Let us examine three pawn moves:
A) 15...hxg4 16.fxg4 f6 17.h5! – White advances his pawn to
h6 and ‘exchanges’ it for the enemy king. After 17...gxh5 18.gxh5
g7 19.h6+ h7 20. c6 a1 21. d5 a5+ 22. d6 Black falls into
another zugzwang. He runs out of moves: 22...f6 23. e6 a6+
24. f5;
B) 15...g5 16.hxg5 h4 17. b8 xa7 18. xe5 h3 19. b1 a2
20. f4 g2 21. b6 h2 22. h6 f6 23. h4;
C) 15...f6 16.gxh5 gxh5 17. c8 xa7 18. c7+ xc7 19. xc7.

268
 analysis diagram
A position of reciprocal zugzwang. If it were White to move, the
game would end in a draw after 20. d7 h7!, but because it is
Black’s move, he loses.
It only remains for us to analyse the king move to f6. Thus,
D) 15... f6 16.g5+ g7.

269
 analysis diagram
At first glance, White has achieved nothing. The black pawn on e5
has signs of being isolated, but in reality, it is not, since the f7-pawn
controls the important square e6. If there were no pawns on the f-
file, White would easily win by moving the king to e6. But the
square e6 is inaccessible to him, and the e5-pawn can only be
removed from the board by exchanging it for the a7-pawn, which
would lead to a draw.
What should White do? Attack the f7-pawn, threatening to move to
a won pawn endgame! The g5-pawn, squeezing the opponent’s
pawn structure on the kingside, allows the black king to be driven
back to the h7-square: 17. c6! a1 (the threat was 18. b6 )
18. d5 a5+ 19. d6.

270
 analysis diagram
Zugzwang!
19... h7 20. c6 a1 (there was a threat of 21. b6 ) 21. d8! xa7
22. d7! a1 (the pawn endgame was hopeless for Black, whilst
after 22... a6+ there is the decisive 23. b5 ) 23. xf7+ g8 24. f6
and White’s position is easily won. The assessment of the computer
is +12.
Now we should examine the situation in which Black takes his king
to e7, trying to prevent the opponent’s winning plan.
Let us examine the move 12... f6 (instead of 12... h7).

271
ANALYSIS DIAGRAM
13.g4 a2 14.g5+ e7.
The position after 14... g7 is one we have just seen. On e7 the
black king defends the square d6, but it is defenceless against the
breakthrough of the white pawn to e6. White only needs to bring his
king closer and guard the passed pawn created by Black on the f-
file. The variation suggested by Kirill Shoshin is sufficiently
convincing:
15. b4 a1 16. c3 a3+ 17. b2 a4 18. b1! a3

272
 analysis diagram
19.f4! exf4 20. c1 f3 21. d2 f2 22. e2 a2+ 23. f1.
The f2-pawn is stopped and will be destroyed. The advance of the
white pawn to e6 decides the outcome of the game. The white king
again turns out to be stronger than the opponent’s rook and pawn.
After seeing this analysis, the question involuntarily arises: ‘Does
Black have any counterplay or does
Shoshin
The young player also answered this question in his analysis.
If instead of the correct move 1.a7!! White plays the waiting move
1. h2?, then Black immediately develops active play.

273
 analysis diagram
1... a2! 2. g2 f5! 3.exf5 gxf5 4.a7 a1! 5. f3 (on 5.f3 there would
follow 5...e4! 6.fxe4 fxe4.

analysis diagram
274
The position is drawn even without the black pawn on e4) 5... a2!
6. e3

analysis diagram
6...f4+! 7.gxf4 exf4+ 8. e4 a4+!. The only move. If White
manages to advance his pawn to f3, he will win.
9. f5 f3. Again the only move, but enough for a draw.

275
 analysis diagram
The position is drawn. Black easily copes with the task of protecting
his pawns.
In this interesting analysis, we observed many of the techniques we
already know about playing rook endgames. A valuable find for us
is the possibility of attacking a disguised target, the f7-pawn, instead
of the obvious target, the e5-pawn.
With the help of associative thinking, I was able to find an analogy
to this technique in Gelfand’s analysis of the game Gelfand -
Kasimdzhanov .

276
This position arose in one of the variations of the analysis. With
Black to move, he loses after 1... f8 2. a8+ g7 ( 2... e7 changes
nothing) 3. d5 b6 4. c5 e6 5. b4!.

analysis diagram

277
Black is in zugzwang. Moves of the black rook on the sixth rank,
such as 5... b6+ or 5... c6, lose after 6. a5 or 6. b5 respectively.
That leaves only 5... h7. But then White exchanges the passed a-
pawn for the pawn on f7 with the move 6. f8 and wins easily.
In the starting position of Gelfand’s analysis, it was White’s move
and he gives the following winning variation:
1. d5 b6 2. c4 e6 3. b4 ( 3. d4! ) 3... f8 4. a8+ g7 5. c4
b6 ( 5... xe5 loses after 6. b8 a5 7. b6 xg5 8. b5 g1 9. a5
c1+ 10. b5 c8 11.a7 a8 12. b6 f6 13. b7 xa7+ 14. xa7
g5 15. b6 g4 16. d5 g3 17. c5 ) 6. c5 e6 7. b4 h7 8. f8,
and White wins.
I recommend you reinforce the knowledge you have acquired with
the following exercise. Let us return to the starting position.

We have already convinced ourselves that after the moves 1.a7 a3

2.f3 a2+ 3. f1 h7 4. e1 g7 5. d1 h7 6. c1 g7 7. b1
a3 8. b2 a5 9. b3 a1 10. c4 a5 11. b4 a1 12. c5 h7
13. d5 a5+ 14. d6 g7 15.g4 hxg4 16.fxg4 f6

278
 analysis diagram
White wins the game with 17.h5!, ‘trading’ his h6-pawn for the
enemy king.
We all know the proverb ‘The best is the enemy of the good’, but
let’s still check whether White can win the game with 17.g5+. In
this case, White intends to implement Shoshin’s plan, which
involves pushing the black king onto the h7-square and exchanging
his passed pawn for the f7-pawn.
Thus, 17.g5+ g7 18. c6 (threat: 19. b6) 18... a1 19. d5 a5+
20. d6.

279
 analysis diagram
Black is in zugzwang and has to move his king away from the
defence of the f7-pawn: 20... h7 21. c6. Again threatening
22. b6.
But let us compare these two positions.

280
after 21. c6 from the training exercise

after 20. c6 in Shoshin’s analysis

281
In both cases, Black is forced to move the rook from a5 to a1. Now
White will exchange his passed pawn a7 for the f7-pawn, using the
manoeuvre a8-d8-d7.
But in the position in the second diagram, the white pawn on e4 is
protected by the pawn on f3, and White wins easily, whereas in the
first diagram, the black rook has the opportunity to attack the pawn
on e4.
Continuing the variation:
21... a1 22. d8 xa7 23. d7 a4! 24. xf7+ g8 25. f6 xe4
26. xg6+ f7 27. f6+ g8.

analysis diagram
Black’s counterplay is sufficient for a draw: 28.h5 h4 29.h6 e4
with equality.
By an irony of fate, a few days after I received Shoshin’s analysis, a
similar endgame occurred in Goryachkina-Harika
We have seen that Black’s counterplay should be connected with the
move ...f7-f5. This is exactly how the Indian player needed to play

282
now, and on each of the two subsequent moves. In this case, White’s
chances of winning would have been slim.
Aleksandra Goryachkina (2576)
Harika Dronavalli (2507)
Douglas 2023

39... b3+?
A mistake just before the time control. With this check, Black
misses the last chance to create counterplay. After the correct
39...f5! 40.exf5 gxf5 41.b6 (White might still try 41. e3 ) 41...f4!
42.b7 ( 42.gxf4 exf4 43.b7 h7 44. e2 b2+ 45. d3 f3! (
45... b3+?? loses to 46. c4! ) the position is drawn also without the
black e-pawn.)
40. e2 b2+ 41. e3 b3+ 42. d2 b2+ 43. c3 xf2

283
44.b6
The most natural move. In the opinion of the engine, 44. e8! wins
relatively easily.
44... f1 45. c4 b1 46. d5 b3 47. xe5?! xg3

284
A sharp rook endgame has arisen, in which the following rule
applies: the king must advance with his passed pawn, and the rook
must fight the enemy’s. This rule correspondeds to the move
48. d6!, so as to play 49. c8! only in response to 48... g5. The
engine here is stronger than the human.
Let’s take a look at the first line of Stockfish 15: 48. d6! g5
49. c8! b3 50. c7 gxh4 51.b7 f6 52.b8 xb8 53. xb8

analysis diagram
53... e5 54. e8+ f4 55. f8 xe4 56. xf7 h3 57. d6 h2 58. f1
h4 59. e6 h3 60. f6 e3 61. g5 e2 62. b1! f3 63. h4 g2
64. b2+ g1 65. g3 h1 + 66. f3.

285
 analysis diagram
And White wins.
Calculating such a variation on increment time is beyond anyone in
the world. It is not surprising that instead of the move 48. d6
Goryachkina played:
48. c8?!
and missed the win.
After
48... b3 49. c6 g5 50.hxg5 h4 51. d5 h3 52. h6 h2 53. c6
c3+ 54. b7 c2 55.e5 b2 56. a7 a2+ 57. b8 c2 58.b7 b2
59. c7 c2+ 60. b6 b2+ 61. c6 c2+ 62. d5 b2 63. xh2
xb7
the players agreed a draw.
The question immediately arises: Was the winning path really so
surprising or did White herself make things more difficult?
In my opinion, White committed a significant inaccuracy at move
47. Let’s take a look at that position after Black’s 46th move.
286
Here, capturing the e5-pawn with 47. xe5?! is a diversion from the
target. In a position with an extra passed pawn on the queenside, it
would be useful for White to remember winning Plan No. 1: break
away from the pawns with the king and free the rook along the
seventh rank.
White’s king has broken away from the f2-pawn and approached his
passed pawn, but the rook on the seventh rank is missing. Such
reasoning suggests the move 47. b7!, after which the win is
relatively simple.
What should Black do after this move?
A) 47... xg3 is hopeless due to 48. c7 b3 49.b7 g5 50. c6
gxh4 51. c8 with White winning;
B) In case of 47...g5, the simplest solution is to play 48. c4! b1
49.hxg5 g6 50. c5, and the outcome of the game is obvious;
C) Finally, after 47... f6 48. c5 xg3 49. a7 b3 50.b7 g7
51. c6 all questions are removed. White wins easily.
Incidentally, apart from 47. b7!, the triangulation idea with 47. c6
c3+ 48. d6 b3 49. d5 wins as well.

287
Using the example of this instructive endgame, we were convinced
that in sharp rook endings it is important to maintain a balance
between calculation of the variations and logical reasoning.
Sometimes, knowing general fighting methods allows you to find a
way to win with a minimum number of difficult decisions.
And now we will move on to consider another type of rook ending
with an extra passed pawn for one side.
Remember the famous expression of grandmaster Siegbert
Tarrasch? He argued that in rook endings the rook should always be
behind the passed pawn – both one’s own and the enemy’s. This
was the case in all the endings discussed earlier in this chapter.
However, many examples have appeared in tournament practice that
prove the effectiveness of placing the rook on the same rank as its
‘strong’ passed pawn. If you can then move your ‘weak’ pawn on
the other flank to the same rank, the king will be freed from the task
of protecting it and come to the aid of his passed pawn. I call this
interaction of rook and pawns ‘playing in a line’.
Let’s see how it works with the help of a fragment of a game from
the Tata Steel Chess Tournament, three years ago.
Polina Shuvalova (2516)
Marc’Andria Maurizzi (2502)
Wijk aan Zee 2022

288
White has an extra pawn on the kingside, but the outcome of the
game is still unclear. Much depends on two factors.
Firstly, who will be able to activate his rook on the queenside and
‘press’ the opponent’s rook? If Black manages to play ...b6-b5 and
exchange pawns on this square, then the black rook on b5 will be
more active than the white one, and the game will end in a draw. If
White exchanges the pawn with b3-b4, then White’s rook on b4 will
force the opponent’s rook to move to the sixth rank.
Secondly, how will events develop on the kingside? If White
manages to line up her strong a4-pawn and his weak kingside pawn
with his rook on b4 along the fourth rank, then she will win. If
Black manages to bring his king to h5 and prevent the white pawns
from moving from the third to the fourth rank, then he will achieve a
draw.
A sample variation, pointed out by Evgeny Gleizerov in the
magazine 64 2 of 2022, runs as follows: 46... d3 47. f4+? ( 47.b4!
wins) 47... g7 48.b4 axb4 49. xb4 h6 50. f4 h5!.

289
 analysis diagram
The position is drawn after 51.h4 a3.
In the event of 46... d3?, the correct thing to do would be to play
47.b4! axb4 48. xb4. If now Black tries to push his king onto h5 by
48... g7!?, then 49.h4?? would be a blunder on White’s part
because of 49... h6 with a draw. The correct option would be
49. f4! h6 50.g4!, preventing the opponent’s king from reaching
the h5-square. White lines up her kingside pawns along the fourth
rank and wins easily.
It seems that everything is clear. Black cannot allow the move 46.b4
and must try to advance his pawn to b5. Can’t this be done right
away? The French player played like that, thereby making the
decisive mistake:
46...b5?
There followed:
47. f4+ g7
47... e6 48.axb5 d5 49.b6 b5 50. f3 xb6 51. g5.
290
48.b4!!

Technique is the command of short-range tactics!

48...g5 49. e4 axb4 50.axb5 b3 51. b4 b2 52. f3
52.b6?? d4+.
52... d3+ 53. f2
Black resigned.
Let us return to the starting position, with which we began our
analysis.

291
On the basis of the above analysis, it is clear that Black cannot allow
the move 47.b4. This means that he needs to play 46... b2!. White
has no choice: 47. e3. Now the white king is ready to begin the
advance towards the enemy e6-pawn. However, Black has the
magnificent tactical resource 47...b5!! (nor does he lose after
47... d2 with the idea 48...b5) 48.axb5 d2!. On the next move, the
black rook will attack the front white pawn on the b-file, resulting in
a draw.
We have seen that the winning plan for the stronger side involves
lining up the queenside and kingside pawns – most often along the
fourth or third rank. The purpose of this technique is to free the king
from defensive functions and direct it either to attack the opponent's
weaknesses or to help the passed pawn. In this case, the active rook
of the stronger side supports its pawns on both flanks.
This technique is often used in rook endings. Let’s give a relatively
simple example.
Ljubomir Ljubojevic (2625)
Xu Jun (2534)
Novi Sad Olympiad 1990
292
50... f5! 51. e3 e5 52.b3 f6 53. e4 g6 54. g1 f4+
55. e5 b4 56. g3 g4 57. d5 f4 58. c5 b4 59. b5 g5
60. g1 h4 61. h1+ g3 62. c5 g2
White resigned.
Let us now study a more complicated ending.
Boris Gulko (2590)
Boris Gelfand (2585)
Amsterdam 1988

293
White has an extra pawn on the kingside. On the queenside, the a-
pawns cannot be exchanged because they are on the same file. The
outcome of the game depends in large measure on the activity of the
rooks.
37. a3!
An interesting moment. My engine, at a depth of 50, recommends
37. c4 a5 38.a4. Nothing like that happened in the game.
The move 37. a3! is the computer’s second line, promising Black
more hopes of a successful defence. Boris Gulko, in his interesting
book Lessons with a Grandmaster: Enhance Your Chess Strategy
and Psychology, considers this move to be the strongest. It seems to
me that he is right.
37...a5 38. f1 e5 39. g2!? g6
Now White transfers his rook behind the enemy pawns with the
manoeuvre:
40. e3! c5 41. e6+

294
Where should the black king retreat to: f7 or f5?
41... f7!
Gelfand plays correctly.
If 41... f5? 42. a6, a position arises in which the black king would
be pinned down to the defence of the h5-pawn or the g6-square. He
will not be able to prevent the white king from passing to the
queenside. Gulko gives the following nice variation: 42...h5 43.a3
g6 44. f1 e5 45.a4 c5 46. e2, and Black gradually loses, since
46...g5 is impossible due to 47. h6 g4 48. g6. All this is closely
intertwined with the motif of White’s play in the game Andersson-
Hübner.
To improve associative thinking in such situations, let’s look at a
few examples.
Pavel Eljanov (2762)
Magnus Carlsen (2840)
Wijk aan Zee 2017

295
Both kings are cut off by the enemy rooks and unable to retreat, but
Black has two extra pawns.
56...a3!
White is in zugzwang. His rook must control g6 and his king must
defend the h3-pawn.
57. f3
In the event of 57. a8, the black king breaks free. In this case White
could win back one of the pawns, for example after 57... g6
58. a6+ f5 59. xh6 b2 60. a6.

296
 analysis diagram
Here, only minimal accuracy is required from Black: 60... b3+!.
The stereotyped continuation 60...a2? would lead to a draw. Here,
unlike in positions in which the opponents have three pawns on the
kingside, there is only one pawn on this part of the board.
As we noted earlier, with an extra passed pawn on the queenside,
the stronger side has two winning plans:
Plan No. 1: Break away with the king from the kingside pawns and
free the rook along the seventh rank.
In this situation, the black king has already broken away from the
h6-pawn, and his rook has freed itself along the second (seventh)
rank.
What is there to think about here? We advance the passed pawn to
the second rank. But then the opponent will obtain counterplay,
since we have only one g5-pawn on the kingside, and the white king
will easily take it.
Therefore, the main thing for Black is to deprive the opponent of
counterplay with the move 60... c3+!, that is, to free the rook along
297
the third (and not the second) rank. We release the rook on this rank
only when we are not able to do this on the second (seventh) rank.
In the game, Pavel Eljanov did not allow the enemy king onto the
sixth rank. But after his move 57. f3, Black freed his rook with the
move:
57... b2!

Technique is the command of short-range tactics! White cannot take

the a3-pawn because of the black king’s move to h4.
58. g3 a2 59. f3 a1 ! 0-1
The queen must be taken – 60. xa1, and after 60... h4 White
resigns.
An analogous situation arose in the final classical game of the
penultimate World Championship Match.
Ding Liren (2811)
Ian Nepomniachtchi (2793)
World Championship Match, Astana 2023

298
Nepomniachtchi withdrew his king from the f6-pawn:
61... d7 62. xf6 b2 63. a6

and freed the rook along the third (sixth) rank with the move:
63... b3+!

299
Why not the second rank? The reason is the same as in the game
Eljanov-Carlsen. Black cannot give freedom of action to the enemy
king. After 63...a2? 64. g3 White obtains instant counterplay with
65.f4, which would immediately lead to a draw.
In the game, the Chinese grandmaster had to demonstrate
considerable accuracy:
64. g2 c7 65.f4! exf4 66.e5 b7 67. a4 c6

68. a6+!
Not being tempted by the bait of the f4-pawn. White’s counterplay
consists in the e5-pawn, and he is prepared to exchange it for the
pawn on a3. After 68. xf4?? d5 69. a4 xe5 Black wins.
68... b5 69. a7 b6 70. a8 c5 71. a6 b5 72. a7 b6
73. a8 c6 74. a6+ d7 75. f2 e7 76. g2 e3 77. f2 g3
78. f1 c3 79. f2 e3 80. g2 d7 81. f2 c7 82.e6 d8
83. a7 e8 84. g2 xe6 85. xa3 g6+ 86. f2 g4 87. a5 xh4
88. f3 e7 89. f5 e6 90. xf4 xf4+ ½-½

300
By the way, if Ian Nepomniachtchi had used Plan No. 2 in the
position of the initial diagram and played 61... a1 and 62...a2, then
the endgame from the following rapid game could well have arisen.
In terms of its ideological content, it is also close to the end of the
game between Andersson and Hübner.
Ernesto Inarkiev (2693)
Sergey Karjakin (2748)
Nazran rapid match 2019

This game continued:

64. b8?
64. d8+! e3 65. b8 d1 66. xb2 d2+ 67. xd2 xd2 68. h2!=.
64... e1! 65. xb2 e2+ 66. xe2 xe2 67. g3 f1 68.f4 e2
69.fxe5 fxe5 70. g2 e3 71. f1 xe4 72. e2 f4 73. f2 g4
0-1
Alexandra Kosteniuk (2517)
Humpy Koneru (2558)
Skolkovo 2019

301
To win, White needs only to put her rook on b8 and the pawn on b7.
Black will put her king on g7 or h7 and the pawn on h4.

analysis diagram

302
Then White exchanges the b7-pawn for that on h4 with 1. c8 xb7
2. c4, with an easy win.
But how do we put the rook behind the h-pawn? After all, on
45. b5? there would follow the tactical blow 45... xg2+!. For help,
we turn to the mechanism 45. a6!! xb4 46.g3+ g5 47.h4+ f5
48. h6 g4 49. g6+ f3 50. g5 b2+ 51. h3 b1 52. f5+ and
53. xh5, winning. Kosteniuk did not find this difficult winning path
and after 45.b5? the game ended in a draw.
Returning to the game Gulko -Gelfand
42. a6 e7 43.a4 f5 44.f3 c5 45. f2 e5

46. c6
Here is Gulko’s commentary on this move: ‘Black defends deftly.
White cannot activate the king without removing the rook from the
active position on a6.’
Associative thinking prompted me to the following question: why
not give a logical check from the square a7? After all, if Black
retreats with the king to f6, then the syndrome of the game
Andersson-Hübner will occur; that is, the black king will be one
303
rank above the white rook. This kind of horizontal ‘cutting off’ is
the most dangerous. For example: 46. a7+ f6 47.g4 h5 ( 47... c5
48.h5 e5 49. c7 ) 48.g5+ g6 49. c7 f5 ( 49... h7 50. g3 f5
51.f4 ) 50. e3 e5+ 51. f4 e8 52. c5 a8 53. c6+ h7 54. e5
e8+ 55. e6 with a decisive advantage.
The remaining move is 46... f8. But then White will play 47. b7! (
47. c7 ), and the game, by analogy with the analysis of the
Andersson-Hübner game, moves in the direction of a complex pawn
ending: 47... c5 48. b5 xb5 49.axb5 e7 50. e3 d6 51. d4 a4
52. c4 a3 53. b3 c5 54. xa3 xb5

analysis diagram
55. b2!!. After this, the engine gives White +60, whereas
55. b3?? leads to a draw after 56...h5.
The point of this endgame is this: with sensible play by both sides,
we reach the following position of reciprocal zugzwang.

304
 analysis diagram
With White to move, the game should end in a draw. But if it is
Black to move, then after 1... f6 2.f5 gxf5+ 3. f4 White wins.
It is similar if we take another position of reciprocal zugzwang:

305
With White to move it is a draw, whilst Black to move loses after
1... f6 2. d5 ( 2. f4?? g5+ ) 2... f7 3.g5.
Here are the sample variations after 55. b2!! c5:

analysis diagram
56. c2!! h5 57. c3 d5 58. d3 e5 59. e3 g6 60.g4! hxg4 (
60... f6 61.gxh5+ ) 61.fxg4 f6 62. d4 e6 63. e4 with the
situation we just studied.
It only remains to understand what could happen in the event of the
incautious move 55. b3?.

306
 analysis diagram
Black must find the only saving move 55...h5! ( 55... c5 loses to
56. c2!! as does 55...g6? because of 56.f4! ) 56. c3 c5 57. d3
d5 58. e2!? e6! 59. f2 f5 60. e3 e5 61. f2 f6 62.g4
hxg4 63.fxg4 g6 64.g5+ e6 65. e3 d7 66. f4 e6 67. e4 d6
and Black holds.
We have once again plunged into an interesting pawn endgame. I
advise you to carefully analyse this instructive ending yourself,
checking your analyses with the engine’s opinion. Pawn endings can
be very complex. They are relatively poorly studied, since they are
rarely encountered in practical games. The engine is significantly
superior to a human in such endings, since by searching through
variations it can bring the game to a victory or a draw without
making any mistakes. It does not need to find corresponding
squares, transfer the move to the opponent using king triangulation,
or generally make logical conclusions. This is the prerogative of
man. But without checking the engine, a chess player can come to

307
inaccurate conclusions. In general, this applies not only to pawn
endings.
Studying the game Andersson-Hübner will help you improve your
understanding of complex pawn endings. From the examples given,
we are convinced that moving to the pawn endgame can be an
effective method of realizing an advantage. This idea can be clearly
seen in the games Demchenko-Aronian and Gulko-Gelfand.
By the way, we need to see the last game to the end. Instead of the
manoeuvre 46. a7! f8 47. b7!, with the help of which it was
possible to win a complex pawn endgame, Gulko played a strong
quiet move:
46. c6!?

46... f7 47. c4 f6 48. e4?!

White has played well so far. The move 48. e4?! to some extent
embodies the principle ‘Do not hurry’. But it makes sense to apply
this principle only when you are sure that your opponent has no
counterplay. Instead, White could have immediately begun active
operations by 48.f4! d5 49. e3, because the rook is located very
308
well on the c4-square. It can always be transferred to a6 or b6 with
tempo, but for now White threatens 50. e4.

analysis diagram
Black can prevent the appearance of the white king on e4 only with
the move 49... f5, but then the white rook moves to the seventh
rank – 50. c7!. Now Black can try with the move 50...g5 to break
out of the insufferable embrace of the white rook along the sixth
rank, which would occur after the moves 50...g6 51.h5! (or also
51. f7+ e6 52. h7 h5 53. c7 f5 54. c6 ) 51...gxh5 52. c6!.

309
 analysis diagram
However, the move 50...g5 does not help Black in view of 51.hxg5
hxg5 52. b7!.

analysis diagram
310
After the tame move 48. e4?!, played by Gulko in the game, Black
has some hopes of saving himself. After all, from e4 the white rook
cannot reach the seventh rank with tempo. There followed:
48... c5 49. f4+ e6 50. e3

50... c3+?!
A waste of time, probably based on an underestimation of White’s
next move. It would have been better, as Gulko pointed out, to
immediately play 50...g5! with chances of a draw.
I will not bore you with a deep analysis of this position – I just want
to give one unforced, but instructive variation: 51. e4+ f6
52.hxg5+ hxg5 53. d3 c1 54. c4 a1 55. d4 a3 56. e4 e6
57.g4 f7 58. f5 xf3+ 59. xg5 a3.

311
 analysis diagram
White has everything in line. But the black king closely monitors
the movements of his colleague. If White managed to cut off the
black king along the f-file from his passed pawn, then he would
easily win by analogy with the game Ljubojevic-Xie Jun. For
example, 60. f4+ e6?? 61. g6. But Black’s king, of course, will
escape the check by moving to the g7-square.
From the last diagram, let’s make a few more moves: 60. f5 a2
61. e5 e7 62. d5 b2 63. f4 b1 64. e5 b2 65. e4 a2
66. f3.

312
 analysis diagram
The white king, sheltering behind his rook, creeps up to the key
square g3. If Black allows the move 67. g3, he will lose, because
his king is cut off from the white passed pawn along the f-file. But
after 66... a3+! 67. g2 c3 68.g5 c5 the game ends in a draw.
Also, 66... b2! draws with the idea 67. g3 b4!.
The direction of play just discussed is in no way intended to be an
analysis of the position, but it demonstrates Black’s defensive
capabilities after the strong move 50...g5!.
Let’s watch the game to the end. We can recall that Gelfand gave an
erroneous check with the rook from c3.

313
51. d2!
An unpleasant surprise. Gelfand had probably banked only on
51. d4 a3 52. c5 g5!, and Black would save himself.
51... c5?
This passive backward move with the rook is the decisive mistake.
Probably, there was a psychological effect from the move 51. d2!,
and perhaps Black was in time trouble. The active move 51... a3
was correct!
In his comments to the game, Boris Gulko pointed out that after
52.h5! (in response to the rook move to a3) White wins. Indeed,
Black’s position looks dangerous, since after 52.h5 the white rook
on the f4-square is very strong. It protects the f3- and a4-pawns, and
after the advance of the g3-pawn, it also has the beautiful f5-square
at its disposal. In such situations, with insufficient time to think, a
strong chess player thinks in diagrams, stopping the calculation of
variations.

314
 analysis diagram
An approximate line of schematic thinking for White could be as
follows: ‘White will bring the king to b2 and displace the black rook
from a3. At some point there will be an exchange of the pawns on f3
and a5 with the white rook on f5 and the g-pawn on g4. White will
then place his rook on g6 and take his passed pawn to a6. This is
where the game ends.’
One of the variations given in Gulko’s comments is this: 52... e5
53. c2 e6 54.g4 e7? ( 54... e5 55. f5+ d4 56. xa5 xf3
57. a6 ) 55. b2 d3 56. f5 e6 57. xa5 xf3 58. a6+.

315
 analysis diagram
And 59. g6 next wins the game.
However, all the variations just given are suitable for thinking in
schemes under time pressure, but they raise many specific questions.
Why in the position after the moves 51. d2! a3 52.h5 e5
53. c2 did Black voluntarily take the king to e6, rather than playing
53... e3, making it difficult for White to regroup his forces? Indeed,
in the variation 54. f7 d4 55. b2 e2+, the activity of the black
pieces provides him with sufficient counterplay for a draw.
But let’s assume that Black followed the path indicated in the
comments and played 53... e6 and White replied 54.g4.

316
 analysis diagram
As Gulko points out, the move 54... e5 loses after 55. f5+ d4
56. xa5 xf3 57. a6. This is not quite correct, although the idea is
right. The winning move is 57. a7!, whereas after 57. a6 Black
saves himself with 57... g3! 58. g6 a3! 59.g5 hxg5 60. xg5 h3!.
But for a human, it is scarcely possible to find this line without a
computer, and I don’t wish to regale you with impossible computer
lines which cannot be found by logical thought.

317
 analysis diagram
Instead of the active move 55... d4?!, the modest king retreat
55... d6!! saves Black. Without the help of the engine, it is
psychologically impossible for a human player to find this, since
with the previous move he was in the mood for active counterplay –
54... e5! instead of 54... e7, and then turns the king back halfway
– 55... d6!! instead of 55... d4. But this switch can be explained
quite easily. After the forced 56. xa5 xf3 57. a7, with the king on
d6, the answer 57... f4 comes to the rescue! The difference is
obvious. With the king on d6, the black rook targets the a4-pawn.
The position is drawn.
If we use associative thinking, we will be able to find an analogy for
the sharp reversal of the black king. Let’s remember Plan No. 1 for
playing for a win from the analysis of the game Grischuk -
Vachier-Lagrave :

318
1. f3 f6 2. e3 e6?! 3. d4 xf2 4. c7 f5 5. c4! g4
6. b3! f6 / f1 7. c4+ xg3 8. a4. The white king’s zigzag,
which leads to a win here, reminds one somewhat of the endgame
we are analysing.
Let us see how the game Gulko-Gelfand ended after the mistake
51... c5? .

319
52. d3 g5 53. e4+ f6 54. c4 d5+ 55. e4 e5+ 56. d4 e6
The black rook has to quit the fifth rank, since 56...gxh4 is
impossible because of 57. c6+ f5 58.g4+.
57.hxg5+ hxg5 58. c5 b6 59. b5
Of course, 59. xa5 also wins.
59... d6+ 60. d5 b6 61. b5 a6 62. c5 f5 63. b6 a8
64.g4+ f4 65. f6+ g3 66. b5 f2 67. f5 1-0
Santosh Gujrathi Vidit (2727)
Amin Tabatabaei (2643)
Chennai Olympiad 2022

320
The position is drawn. Black can simply take with 41... xg5, and
after 42. xh4+ g8 43. b4 the most accurate is to play 43... a5!
44. e3 g5!, not allowing White to set up his pawns along the fourth
rank. But the Iranian GM was probably still short of time and
quickly played:
41... h3?
Allowing a strong tactical blow:
42.g6+!
It is remarkable how many mistakes get made the move after the
time control! One could write whole books on this, but probably
nothing would change.
Events then developed by force:
42... h6
if 42... xg6? White has to exchange rooks: 43. g4+! f6 44. g2
g3+ 45. xg3 hxg3 46.a5 and wins.
43. g2 g3+ 44. h2 h5 45.a5 xg6 46. f5+ h6 47. b5!

321
Now another serious mistake followed from Black. After
47... g3? 48. b3 g5 49. a3 c5 50.a6 c8 51.a7 a8 52.f4!
he acknowledged defeat.
Why have we shown this brief fragment? Not merely to show
Black’s blunder at move 41. I am more interested in the position
which could have arisen after 47... h7!, instead of 47... g3?. The
game might then have continued thus: 48. h5+ g8 49. xh4 a6.

322
 analysis diagram
Visually, Black’s position looks lost after the natural move 50. a4.
According to Tarrasch, the rook should stand behind its passed
pawn. Commenting on this position in 64 No. 9 for 2022,
grandmaster Gleizerov writes: ‘... and the rook manages to stand
behind its passed pawn. That’s pretty much what happened in the
game.’
Probably, Tabatabaei also assessed this position as losing and
played 47... g3 using the method of elimination. I tried to continue
analysing the position in the last diagram and was surprised that I
could not find a win for White after the obvious move 50. a4?!. In
the game, the white rook also stood behind its passed pawn, but the
pawn managed to reach a7, leaving no chance of salvation for the
opponent. Here the pawn is stopped on the fifth rank. Here is one of
Black’s main defensive resources: 50... f7 51. g3 e6 52. g4
f6 53.f4 g6 54. f3 g5! 55.fxg5+ xg5 56. e4 f6 57. d5 e7
58. c5 d7 59. b5.

323
 analysis diagram
It looks as though White is almost there. After 59... a8? 60.a6 c7
61. c4+, he wins. But Black escapes with the beautiful tactical trick
59... h6! 60.a6 c7!! 61.a7 h5+ with a draw.
The move 50. a4?! does not seem to lead to the goal, but the
beautiful ‘linear play’ associated with the move 50. h5 wins! The
evaluation of my computer after 50. h5! is somewhere around +5,
whereas the move 50. a4?! is given +1. Usually a chess player
places his rook behind a passed pawn automatically, without
thinking, by reflex, and most often this is correct. But here we are
dealing with a clear exception to the rule. As they say, live and
learn!
Let’s analyse the position after the move 50. h5! and let's try to
verbally describe the essence of what is happening.

324
 analysis diagram
Let us assume that the sides make the natural moves 50... f7
51. g3 e6 52. f4 f6 53. e4 e6 54. d4 d6 55.f4 g6.

analysis diagram
325
56. d5+!.
It is the possibility of such a check on the fifth rank that favours the
placement of the white rook alongside rather than behind its passed
pawn. Now the black king will be forced to concede the opposition
to the white king, who will move in the opposite direction.
You may ask, ‘What does the line have to do with it?’ Of course, we
are not talking about a line in its pure form here. The white rook
does not simultaneously control its passed pawn and the vulnerable
f4-pawn. But it controls the important squares e5, f5 and g5, which
become crucial for active play in many variations.
My engine assessment is already +13. White wins. Let’s look at a
few variations:
A) 56... c7 57. b5 f6 58. e4 a6 59. f3 d7 60. e5! c6
61. g4 c1 62. e3 g1+ 63. g3 a1 64. g5 xa5+ 65. xg6 e7
66.f5 f8 67. b3;
B) 56... e6 57. g5 f7 58. c4 f6 59. b5! xf4 60.a6 e6
61.a7! f8 62. b6! f6 63. a5.
But now let us imagine that in the position of the last diagram, the
white rook is on a4 rather than h5:

326
Here White does not manage to force the enemy king to concede the
opposition and the game should end in a draw. Here is a sample
variation: 1. a2 d7 2. d5 d8! 3. a4 d7 4. e5 e6+ 5. d4
a6 6. e4 e6+ 7. f3 a6 8. g4 e6 9. a1 f6 10. a4 g7
11. g5 f8 12. h6 g8 13. c4 g5+ 14. xg5 xa5+ 15.f5 a6.
Draw.
It remains to see what could happen if Black tries not to allow the
enemy king into the centre with the move 50... g6 after the splendid
move 51. e5!? f7 52. f5+!? g8 (if 52... e7, then 53. b5 f7
54. b7+ e8 55.f4!. With his king cut off on the eighth rank Black
cannot attack the f4-pawn with 55... g4 because of 56.a6. There
could follow 55... a6 56. b5 g6 57. g5! a6 58. g3, and the
game follows familiar motifs)
53. h5! (White transfers the move to the opponent) 53... f7 54. b5
g8 55.f4 h7 56. b6 g4 57. b4 g6 58. a4 a6.

327
 analysis diagram
Black has managed to force the white rook to occupy the square
behind his passed pawn, ‘according to Tarrasch’, which is often an
unattainable dream of the stronger side. Then subtle play begins, in
which White uses the distance of the black king from the a5-pawn to
move his rook back to the fifth rank:
59. g2 g8 60. f3 f8 61. e3! (a grave mistake would be
61. e4? e7 62. d4 d6 63. c4 c6 64. a2 g6 with a draw)
61... e8.

328
 analysis diagram
62. e4+! d7 63. e5! g6 64. f3 c6 65. g4 c1 66. e3, and we
have already seen all of this. White wins.
An instructive and very interesting ending, demonstrating that
playing ‘alongside’ can be more effective for the stronger side than
the standard placement of the rook behind its passed pawn.
There is an analogy to the advantage of placing the rook ‘alongside’
its passed pawn compared to placing it behind. For example, the
position that immediately comes to mind is ‘rook and pawn against
rook’, in which the weaker side uses a frontal attack.
Let's imagine the position:

329
1. c4 c8+ 2. b5 d8 3. c5 c8+ 4. b6 d8

analysis diagram
5. e4! (alongside! 5. d1? e6 would lead to a draw) 5... f6 6. c7
d5 7. c6 a5 8.d5 and White wins easily.

330
We continue our analysis of rook endings.
Oliwia Kiolbasa (2388)
Rameshbabu Vaishali (2401)
Chennai Olympiad Women 2022

This position arose in the meeting between the women’s teams of

India and Poland, who were in the leading group at the Chess
Olympiad in Chennai. The Polish chess player Oliwia Kiolbasa had
a clean score at that moment – seven victories in seven games, but
this rook endgame did not promise White any special chances of
success.
40... c1+ 41. h2 a1
This move offers White extra chances, which could have been
avoided by means of 41... c3! 42.a4 a3 and 43...g5.
42. a5
The standard path. White had the possibility of 42. b3 with the aim
of establishing the kingside pawns on the same rank as her passed
pawn after the natural enemy response 42...g5. Then there could

331
have followed, for example, 43. g3 a2 44. f3 h7 45.g4 h6
46. e3 g6 47.f3.

analysis diagram
White has realized her plan, and if Black is late with counterplay,
she will lose after the enemy king makes contact with the passed
pawn. The following manoeuvre helps Black out: 47... h2 48.a4
a2!. The meaning of this move is approximately the same as the
move we proposed earlier, 41... c3!? instead of 41. a1: Black
needs to move the enemy rook from the third rank – a line on which
White’s kingside pawns could be located . I advise you to try to
remember this important technique. After 49. b4 f5 Black has
enough counterplay for a draw due to the vulnerability of White’s f3
and h3-pawns.
Capturing the h3-pawn – 48... xh3? (instead of 48... a2! ) would
lead to White’s victory after 49.a5 f5 50. b8! or 50. b4!.
We see that the move 41... a1 did not lose the game after 42... b3,
but why, in a drawn position, provide the opponent with additional
opportunities to play for a win? This is dangerous, because taking
332
into account all the defensive resources of the weaker side after a
series of moves is not easy at all. The possibility of error is high.
This is, to say the least, an unprofessional approach to defending a
drawn position. I would like you to remember: placing rook and
pawns in a line is very beneficial to the stronger side. Your task
when defending inferior positions is to try to prevent your opponent
from reaching such a position.
To see how dangerous such an arrangement can be, let's look at a
few examples.
In the first fragment, Anatoly Karpov found the only way to ask his
opponent some difficult questions:
Anatoly Karpov (2715)
Rainer Knaak (2525)
Baden-Baden 1992

39. a3!
In the event of 39.a4 c2 or 39... c3, Black takes play into standard
drawing paths.
39...g5?!
333
I see no point in going deeper into the analysis of this complex
ending. In my opinion, it is much more important to explain to the
reader in words the essence of the events taking place than to go into
complex specific variations, behind which the overall picture of the
game may turn out to be blurred.
Rainer Knaak made a natural pawn move, capturing space on the
kingside. Why does this move deserve the dubious sign?
As always, you need to ask yourself the question: ‘What does your
opponent want if it were his move?’ and try to think in schemes.
The answer will be something like this: ‘Put the king on d2, the rook
on f3 or e3, the pawn on a4. After this, the white rook will move to
a3, and the pawn on a4 will advance to promote at the speed of an
aeroplane.’ This means that if Black doesn’t find a way to get
counterplay on the kingside, he will lose.
Let’s not blame the German grandmaster. After all, it was the 39th
move, and he probably lacked time to think about it.
However, even if you have enough time to think about it, finding a
way to obtain counterplay for Black is not at all easy. His main
problem is the symmetrical placement of the kingside pawns.
Mark Dvoretsky suggested playing 39...h5!? with the idea of
advancing the pawn to h4 and attacking the g2-pawn by ... g5.
39... c2 40. f3+ e6 41. e3+ d6 42. e2 c3 or 42... c4 and
then advancing the kingside pawns in search of counterplay was
also worthy of attention. But all this requires a detailed and specific
analysis, which, I repeat, is not our purpose in studying this ending.
Let’s instead observe the magnificent technique of the twelfth
World Champion in carrying out his plan.
40. d2 g6 41. c3 a5 42.a3 h5
42...f5 may still hold, as it creates possibilities for counterplay by
fixing the white f-pawn; for example, 43. c2 g4! 44. b3 b5+
45. c4 b2 and Black may save the game.

334
43. c2 a8 44. b3 b8+ 45. a2

45... a8
If 45... d8 46. c2 d3, GM Karsten Müller suggested the variation
47.a4 f6 48. b2! h4 49.a5 with an easy win for White.
46. c4 f5 47.a4
Karpov launches a mechanism for advancing his passed pawn,
alternating with moves of his rook, king and pawn.
47... f6 48. a3 e5 49. c5+ e4 50.a5 h4 51. a4 f4
Black has to spend time on this move, since the immediate 51...g4
loses because of 52.hxg4 fxg4 53. h5.
52. c4+!
A technical move. Black is deprived of all counterplay.
52... e5 53. b4!
Continuing on the same lines.
53... d5 54. b5+ e4 55. b6 f4

335
We already know that 55...g4 is bad because of 56.hxg4 fxg4
57. h6!.
56.a6 g4 57. a5!
Technique is the mastery of short-range tactics. On 57...gxh3 White
would reply 58... b4+.
57...g3 58. b4+! e5 59.f3 f4 60. e4+ f5 61. e2 f6 62. b6
1-0
After watching this instructive ending, let’s once again emphasize
Black’s main problem – the symmetrical placement of the pawns on
the kingside . The opportunities for creating counterplay on this part
of the board were insignificant for Black, and Anatoly Karpov
skilfully stopped them.
By turning on associative thinking, we can find a successful
example of how the weaker side broke the pawn symmetry on the
kingside. To do this, we again turn to the Maryasin-Korbuzov

1...f6!
An excellent move, breaking the pawn symmetry on the kingside.
He grasped the point of the position. After 2.exf6 gxf6 3. e2 e5
336
Black had obtained a potential passed pawn on the e-file, which
helped him save this game in a fluctuating battle.
A deep analysis of a similar ending was given by Boris Gelfand
in his introduction to the book we already mentioned, Decision
Making in Heavy Piece Endings .
Let us look at another interesting example from Gelfand
Boris Gelfand (2748)
Rustam Kasimdzhanov (2706)
Baku 2014

Think about what you would play here.

White played the active move
38. c7?
but after
38... f8 39. e4 a2 40.f4 xh2
Black managed to save the game. Nor does 38. e4? work, with
similar variations.

337
If you have become familiar with the ideas of ‘playing in a line’,
then I hope that associative thinking suggested to you the strongest
move 38. b2!!. Now, after 38... a4 White would respond with
39. c2, and in case of 38... a1 39. e4 d1 40. a2 it would be
difficult for Black to save himself. Gelfand gives as the main line
38... a1 39. e4 a5 40. d5 a4 41. xc5 a3 42. d2 f8 43.h4 e7
44. b4, and White wins.
The young talented grandmaster Aydin Suleymanli and some girls
from the Azerbaijani women’s team found the move 38. b2 in
seconds during individual lessons!! Although visually this move is
not exactly ‘line play,’ it reflects its essence. After all, from c2 the
white rook defends its strong pawn on c4 and potentially weak
pawns on the kingside, and the white king goes to attack Black’s
weak pawn on c5.
I would like to complete the demonstration of ‘playing in a line with
the ending of a game played quite recently at the World Cup in
Baku.
Arjun Erigaisi (2722)
Javokhir Sindarov (2654)
Baku 2023

338
Black needs to play accurately to draw. If the white king gets to the
centre quickly, then problems can arise.
32... c8!
The young Azeri GM correctly sacrifices a pawn to activate his
rook.
33. xa6 c5 34.bxc5 xc5 35.e6 fxe6 36. xe6

339
White has a weak pawn on h3. Black needs to continue activating
his rook with the move 36... d5!, with an attack on h3. After, for
example, 37. f2 d2+ 38. e2 d3 39. g2 g3+ 40. h2 f3
41. c2 h7,

analysis diagram
340
it was possible to agree a draw.
But why does Black not simplify his task and play 36...b4 instead of
36... d5 ...? After all, 37.cxb4 c2 38. e4 h7 39.g5 g6 40. xh4
xg5 41. g4+ f6 42. g2 c4 is an obvious draw. Sindarov
indeed played
36...b4??
,
Missing the move
37. e3!
.

Line play! White’s strong passed pawn on c3 and the weak pawn on
h3 are on the same rank, which is controlled by the white rook.
Black’s position instantly turned from a draw to a loss.
To be fair, it should be noted that the Uzbek grandmaster thought
over the correct move 32... c8! for twenty-six minutes out of the
twenty-nine he had left before the control at move 40, and then
followed the forced continuation he had calculated. The move
36...b4?? was played almost without thinking. Most likely, this
341
move was part of Black’s plan. But here Sindarov lost sight of the
line placement of the white rook and pawns along the third rank.
At this moment, for the remaining five moves before the control,
Black already had 4 minutes and 46 seconds, taking into account the
accumulated time. If Sindarov had given himself a minimal rest
here, calmed down and looked at the position with an ‘uncluttered’
look, he would have seen his opponent’s response here: 37. e3! and
instead of 36...b4?? would have played 36... d5. But it’s easy to
give such advice in a calm atmosphere, and not in the nervous
atmosphere of a crucial game of a World Cup mini-match.
37...bxc3
The move 37... d5 is already not a cure. After 38.cxb4! d1+
39. f2 d2+ 40. e2 d3,

analysis diagram
White wins with the help of 41.g5!, threatening 42. g6. Then
possible is 41...g6 42. e3 d2+ 43. f3 xb2 44. e4, and Black
loses the pawn on h4.

342
38.bxc3 h7 39. f2 g6 40. e2 f6

The time control is passed. The black king is cut off from the
queenside, whilst there seems to be no counterplay on the other
wing. White’s position is winning. My engine assesses it as +9 at a
depth of 40. We give the remainder of the game without comment:
41. d3 d5+ 42. c4 d8 43. c5 c8+ 44. d5 d8+ 45. c6
c8+ 46. d7 c4 47. d6 a4 48. f3+ g5 49. d5 a3 50. f5+
g6 51. d4 a4+ 52. d3 a8 53. h5 f6 54. xh4 g5 55. h7
d8+ 56. c2 e5 57. f7 e4 58. f5 h8 59. xg5 xh3 60. b3
1-0
Returning to the game Kiolbasa-Vaishali after 42. a5

343
42...g5!
Then, for the time being, events developed according to a standard
scenario, which can be left without comment.
43. g3 a2 44. f3 h7 45. e3 g6 46. a7 f5 47.g3 e5
48. f3 f5 49.a4 a3+ 50. g2 g6 51.a5 h6 52.a6 g6
53. a8 f5 54. f1 a2
White made a feeble attempt to carry out Plan No. 1 – to break away
with the king from the kingside pawns and free the rook along the
seventh rank. But it was not a success at all. The Polish player now
decides to move to Plan No. 2 and advance her passed pawn:
55.a7

344
Here Black should think and ask our favourite question: ‘What does
the opponent want with this move?’ It is clear that White will not
achieve anything with one move. Therefore, the question needs to
be a little more complicated: ‘What scheme of action can White use
to play for a win?’ And the most specific and rational version of the
question goes something like this: ‘What needs to happen for
Black’s position to become dangerous or losing?’
Let’s try to think logically now. It is always more difficult to defend
than to develop the initiative. You have to do double work. First,
you need to find the right scheme of action for your opponent, and
then figure out how to prevent it.
Let’s start our reasoning by studying Black’s passive stance. Let’s
assume that she moves her king to h7 and gives White complete
freedom of action, only moving her rook up and down along the a-
file. In this case, White can move her king to f7, place the f2-pawn
on the f3-square and play h3-h4.
After the exchange of pawns on h4, the following position will
arise:

345
Then White’s winning plan looks simple. The h4-pawn goes to h5
and the f3-pawn to f5, and White ends the game with the
breakthrough h5-h6. But with Black to play, she places the rook on
a6.

346
But then after e7, Black is in zugzwang. On the next move, the
breakthrough h5-h6 will become inevitable.
From our consideration, the following conclusion emerges: the path
to the draw for Black requires her to avoid a few boulders. How
does she do this? Ideally, she would like to obtain two safe squares
for the king – g7 and h7. To achieve this, she needs first to transfer
the king to h7 and then play ...g7-g6 and put the king on g7.

In this position, it is already possible to agree a draw. Black needs

only minimal accuracy. In positions like this,

347
there is no point in calculating the consequences of going into the
pawn endgame, for example after 1... a1 2. d8 xa7 3. d7+ xd7
4. xd7 f5 5. e6 g4. It is simpler to give a vertical check from e2 –
1... e2+!.
Now we return to the position after 55.a7
So, we know what we need for a draw and it only remains to check
the assessment of the position after 55... g6 56. e1 h7! 57. d1
g6 58. f8 xa7 59. xf6.

348
 analysis diagram
Obviously, Black should achieve a draw without any problems.
As I mentioned in my previous works, Nimzowitsch aptly compared
the strength and weakness of doubled pawns to ‘the lameness of a
sitting man’. They are very good for defence, as they control a lot of
important squares. But when they move, they give away these
squares, so it’s better for them to stand still.
In the position in the last diagram, Black can immediately solve all
his problems after 59... a1+ 60. e2 g7 61. d6 h1, with a draw.
In the game, the Indian player first made the correct move
55... g6
but after
56. e1
she turned her king a little in the other direction:
56... f7

349
This move does not throw away the draw, but it does give the
opponent the opportunity to ask some questions. The black king has
taken up a dangerous position. If we now look for a winning plan
for White, then the march of the white pawn to the h6-square
catches our eye. But it doesn’t seem to hold much hope for success,
since in the following position,

Black has ... f7-g6 with a draw.

But what happens if instead of h5-h6 White first advances his pawn
to f5 after f3-f4...?

350
Obviously, in this case, Vaishali would have had to interfere with
her opponent’s plan and play ...f6-f5 herself.
Now let's turn on the associations and try to remember what this
reminds us of.
When we talked about the two plans for winning in rook endings
with an extra passed pawn on the flank, the wording of Plan No. 2
sounded like this: ‘The rook of the stronger side is placed on the
eighth rank, the pawn on the seventh, and with the help of
zugzwang, the opponent’s isolated pawn is won’ (diagram A).

351
diagram А

diagram B

352
In the Kiolbasa-Vaishali game, logically and with the help of
schematic thinking, we arrived at the position in diagram B.
This position is absolutely hopeless for Black. After the king has
captured the f5-pawn, White has the resource h5-h6, which removes
the g7-pawn from the path of her passed pawn along the f-file. To
win, she won’t even need the help of the king, she just needs to
advance the f4-pawn to f5 and play h5-h6.
What should Black do? Her main problem is the dangerous position
of the king . That’s why she needs to try to move the g7-pawn to g6,
as g7 would be an ideal square for the king. But after 57.h4!? the
moment to move the king to h7 has already passed, since if
57... g6? 58.h5+! h7 59.f4 White wins using the method already
discussed. The move 58... f5 instead of 58... h7 does not help
Black either, due to 59.f3! g4 60.f4, and the white king makes a
victorious march to the f7-square.

analysis diagram
Let us consider our logical analysis for Black after the move
57.h4!?.
353
 analysis diagram
We have managed to formulate the main problem for Black – the
dangerous position of the king. Now we need to find a safe place for
him. We have just seen that if each side has three pawns, the
problem is unlikely to have a solution. But what happens if, after the
exchange on h4, the king is transferred to f5?
Let us test the variations: 57...gxh4 58.gxh4 g6 59.h5+ f5.

354
 analysis diagram
It is easy to see that it is ‘mission accomplished!’. The position on
the board is a dead draw.
However, the Indian player did not carry out this logical analysis
and played 57... g6? without a preliminary exchange of pawns on
h4. This proved to be Black’s decisive mistake.

355
58.h5+! f7
Logical. Just now, we discussed the moves 58... h7 and 58... f5.
Black prevents the white king transfer to f7 in the most cardinal
fashion. However, there was already no saving the game.
59.f4 gxf4 60.gxf4
Threatening 61.f5 and 62.h6.
60...f5

356
A familiar position. What is the easiest way for White to win here?
That’s right, 61.h6! g6 62.hxg7 xg7, and we come to the
standard situation of Plan No. 2. It seems that the Indian player
simply did not know it. Kiolbasa took a slightly different path, also
leading to victory:
61. d1 a1+ 62. e2 a2+ 63. f3 a3+ 64. g2 a2+ 65. g3
a4 66. h3 a3+ 67. h4 a4 68. g5 a5 69.h6 gxh6+ 70. xh6
a6+ 71. g5 g7 72. xf5 a1 73. e5 e1+ 74. d4 d1+
75. e3 e1+ 76. d2 a1 77.f5 a2+ 78. c3 a3+ 79. b2 a6
80.f6+ 1-0
Hot on the heels of this, I would like to invite you to solve a
beautiful study on the topic of ‘playing in a line', for better
assimilation of the material you have just covered.
Vladimir Korolkhov
Study, 1954

357
 White to play and win
1. d8+ g7 2.e7 d1 +! 3. xd1 f7 4. e1 e8 5.a4

5... b3+!

358
White threatens with 6. e4 to defend both pawns and then quietly
bring his king to the aid of the a-pawn.
6. g2! b2+ 7. h1!!
A high level of chess culture sometimes makes it difficult for strong
grandmasters to ‘humiliate’ their king in the endgame like this. But
for White, this is the only way to get rid of the annoying checks of
the black rook and make the highly important square g1 available
for his rook.
7... b4 8.a5 b5 9.a6 b6 10.a7 a6

11. g1! xe7 12. g8 xa7 13. g7+ 1-0

In the study we have just analysed, we witnessed how the stronger
side was forced to place its king into a corner. This is an exception
to the rule, because the main difference between the endgame and
the middlegame is the change in the role of the king. We observed a
clear manifestation of the principle of ‘centralization (activation of
the king)’ in the game Shereshevsky-Veremeichik. Now I would

359
like to give two more examples demonstrating the enormous power
of the king in the endgame.
In issue 6 of 64 for 2020, International Master Alexey Yarovinsky
analyses an interesting rook endgame that is somewhat reminiscent
of the ending of the game Andersson-Hübner. Here is the position
that arose inYarovinsky’s analysis after White

Black’s winning plan is for the king to conquer the white f2-pawn
and then advance the e4-pawn to queen. For this purpose, Black will
have to sacrifice the g5-pawn with the push ...g5-g4. Black has no
other way to make progress, despite his two extra pawns.
So: 9...g4! 10.hxg4 h3+ 11. h2 e7.

360
 analysis diagram
Let’s remember the wise remark of grandmaster Boris Gelfand: In
the endgame, the king is stronger than a minor piece and is
approximately equal in strength to a rook, if its safety is guaranteed.
Gelfand’s correctness is proven by the analysis of the ending of
Jakovenko-Dubov, Satka 2018, analysed in my book Endgame
Strategy. In one of the analysis variations, the following position
could arise:

361
The white king is not playing (moves along the second rank change
nothing), whilst Black has an immobile rook. The black king easily
holds the position in the face of the white rook and the d5-pawn.
In his wonderful article called ‘Brilliant king work’, Alexey
Yarovinsky went further. He demonstrated the triumph of the black
king over the white rook and the passed g-pawn:
12.g5 f7 13. f6+

362
 analysis diagram
13... e8!! 14. e6+ (if 14. a6, then 14... e7! 15. a7+ f8 16. a4
f7! 17. a6 g7 – Yarovinsky) 14... d7!! 15. a6 e7 16. a7+ (
16.g6 f8 17. a7 g8 ) 16... f8 17. a5 f7 18. a6 g7 19.g6
h6.

363
 analysis diagram
White is in zugzwang. The g6-pawn is lost, after which the black
king calmly moves towards the f2-pawn and destroys it.
On the pages of this book, we have already repeatedly observed how
complex pawn endings can be. Let's look at two more examples in
which the game moved from a rook ending to a pawn ending.
Daniil Dubov (2699)
Vladislav Artemiev (2708)
Sochi 2020

364
White’s position is difficult, as is clear to a human at first sight. By
computer standards, it is simply hopeless. My engine at a depth of
30 gives -7.69. White loses after 55.f4 d3+ 56. e2 ( 56. e4 d4+
) 56...gxf4 57. c6+ d5 58. b6 e3+. Likewise, 55. e4 d4+
56. f5 f4+ 57. xg5 xf3 58.h4 xa3 is no better, with an easy
win for Black.
Daniil Dubov tried his last chance:
55. c6+!?
He presented the opponent with a difficult exchanging problem.
55... c5
Black decided to go into the pawn endgame. I assume that the
players no longer had enough time to accurately calculate the
variations. The pawn endgame is won for Black, but it is much more
difficult than Vladislav Artemiev expected.
From a practical point of view, the move 55... b3 should have been
preferred, but ambitious and confident chess players usually trust
their calculations. There is a whole chapter on this topic in the book
Endgame Strategy , called ‘Between Assessment and Calculation’.
365
Here is an approximate line that could lead Black to victory after the
move 55... b3!? – 56. e4 d8 57. b6 ( 57. c5 b8 ) 57... a4
58.f4 ( 58. f5 d5+ 59. e4 c5 60. d4 c4+ 61. e5 h4 )
58...gxf4 59. xf4 d5 60. a6 d3, and Black’s two connected
passed pawns on the queenside decide the outcome.
56. xc5+ xc5 57.f4 gxf4+ 58. xf4 c4

59. e3!
I suspect that this move came as a surprise to Black. His calculations
were probably based on the variation 59.h4? b4 60.axb4 a4, and
from a1 the black queen controls the promotion square of the white
pawn on h8. Black is still winning, but with just a couple of minutes
left and tired from a long session, Artemiev did not find it. After the
trivial move
59... b3
the draw was agreed following:
60.h4 xa3 61.h5 b4 62.h6 b3 63.h7 b2 64.h8 b1 65. c3+
b3 66. d2! a4 67. a1+ b4 68. d4+

366
Now let us see how Black could have won on move 59:

59...a4! 60.h4 b4 61. d2 and what follows is a nice ‘endgame

study’ in a practical game:

analysis diagram

367
61... b3! 62. c1 ( 62.h5 bxa3 63.h6 a2 ) 62... xa3 63.h5 b3 64.h6
b2+ 65. b1 b3 66.h7 a3 67.h8 a2#.

analysis diagram
Polina Shuvalova (2489)
Salome Melia (2371)
European Women’s Team Championship, Terme Catez 2021

368
In this rook ending, White is a pawn up, but Black’s pieces are
active. Black wants to advance the pawn to h4, after which White’s
extra pawn on the kingside will be neutralized. For example,
48. b2 d1! 49. b3 b1+ 50. c4 h4. If after 48. b2 d1 White
plays 49. e4, then the computer offers to answer 49... d2!? (
49... g1 50.g4 h4 is also possible) 50.g4 b3! 51. xb3 h2 with an
instant draw.
But what happens if White moves into a pawn endgame with
48. e4
...?
Now or never. Polina Shuvalova played this. The following moves
are strictly mandatory for both players:
48... xe4 49.fxe4 e5 50. b2 xe4 51. b3

369
Here we have a good training exercise in the calculation of
variations.
51... f4?
The Georgian player commits the decisive mistake. A draw results
from 51...g4. The variations are as follows: 52.hxg4 h4! (in the
battle against the enemy queen, Black needs the h-pawn) 53. xb4
f4 54.c4 xg4 55.c5 g3 56.c6 xg2 57.c7 h3 58.c8 h2. The
position is drawn.
Or 52. xb4 g3 53.c4 e3 54.c5 f2 55.c6 xg2 56.c7 h2!
57.c8 g2.

370
 analysis diagram
The pawns on h3 and h5 defend the black king against checks on the
h-file and render the position drawn.
After
52. xb4
another mistake followed:
52... g3?
With the move 51... f4? Black lost a vital tempo, but here it was
better to place her hopes on the g-pawn. She would have had
definite practical chances with the move 52...g4!?.
There could have followed 53.c4 g3!?

371
 analysis diagram
54.c5 e3 55.c6 f2 56.c7 xg2 57.c8 h2 58. b8 h4! 59. c3
h1 60. b1+ h2 61. d4 g2 62. e3 g1 + 63. xg1+ xg1

analysis diagram
372
64. f3, and White wins, but in the event of 64. f4?? f2 we get a
draw. It is a childish trap, but when playing on increment, such
things happen.
The move 52... g3? was the line of least resistance and after
53.c4 xg2 54.c5 xh3 55.c6 g4 56.c7 h4 57.c8 g3 58. c7+
g2 59. f4 h3 60. f1+ h2 61. f2+ h3 62. f5 g3
63. c3 h3 64. d2 h4 65. e3
Black resigned.
Why do games so often change from a rook ending to a pawn
ending? In my opinion, there is a direct connection between these
two types of endings. As we have seen many times, the stronger side
most often wins rook endings by applying the ‘schematic thinking’
principle. Using this principle, you can look far into the future and
outline a winning plan of action. And if in this scheme the key link is
to transfer the game to a pawn ending, then the opponent will not be
able to prevent this. In this case, as they say, the weaker side is not
asked. This is a distinctive feature of rook endings. In other types of
endgames, the opponent usually has a choice when solving the
exchanging problem.
Let’s look at two modern examples of knight endings in which
grandmasters had a chance to avoid going into an unfavorable pawn
endgame.
Let’s start with the position that arose in the fifth round game of the
World Cup between two young grandmasters from Armenia and
Iran. Hayk Martirosyan won the first game, and to reach the
quarterfinals he only had to not lose with White.
Haik Martirosyan (2622)
Amin Tabatabaei (2629)
Krasnaya Polyana 2021

373
In the diagram position, Black stands more actively, but hardly any
material remains and the position is drawn.
58. xf3??
He should not have gone into this version of the pawn ending. After
58. d1, the exchange on d2 leads to a draw. Also good was
58. c4. The remaining events can be left without comment.
58...gxf3+ 59. d2 d3 60. e1 e5 61. d1 f5 62. d2 g4
63. d1 h3 64. e1

374
64... g2! 65.g4 d2+ 66. xd2 xf2 67.g5 g1!
White resigned. After 68.g6 f2 69.g7 f1 70.g8 + g2+ the queen
ending goes into a pawn ending that is hopeless for White.
Richard Rapport (2763)
Jorden van Foreest (2702)
Wijk aan Zee 2022

375
White threatens 36. e5+, after which Black will at least lose the h7-
pawn. It can only be saved by 35...fxe4. In this case, Black needed
to evaluate the advisability of moving to a pawn endgame. The
Dutch grandmaster correctly understood the intricacies of knight
exchanges and realized that this pawn endgame was unprofitable for
him.
After 36. xe4+ xe4? 37. xe4 h6 38. d4 d7 ( 38...b6 39.b5
d7 40. e5 e7 41.h4 h5 42.f5 ) 39. e5 e7 40.h4 f7 41. d6
f6 42. c7 f5 43. xb7 xf4 44. xa7 e5 45.b5 the white pawn
is promoted with check.
However, he had another opportunity to offer his opponent a
transition to the pawn ending. If Black answers correctly 36... e7
37. xf6 xf6, the game could have ended in a draw.

376
 analysis diagram
White will not be able to make progress after 38. e4 h6. If it were
Black’s move now, he would lose. But it’s White’s move, and the
position is a draw.
It is possible that Black was in time trouble; he played:
35... d7??
White took the h7-pawn
36. xh7
and won easily.
This extensive section on associative thinking is designed
to help you improve your understanding of rook endings and
minimize the amount of rote learning required.
To use a metaphor from a famous parable, I am giving you a fishing
rod, not a fish. If you, in addition, work through the ‘Schematic
Thinking’ section in my book Endgame Strategy , then your fishing
rod will turn into an e xpensive modern rod. And if you add to this
the exact positions from Dvoretsky’s Endgame Manual or at least
377
from the book Rook Endings by Grigory Levenfish and Vasily
Smyslov, then you will have a veritable trawler at your disposal, for
industrial fishing for half-points and full points in tournament
games.

378
Chapter 8
The problem of
exchanges in inferior
positions
In the final games of the previous chapter, we observed how the
weaker side made mistakes when solving the exchanging problem.
In this chapter, we will delve into the topic of exchanging in inferior
positions.
Let’s return to the Andreikin - Fedoseev game. In it, grandmaster
Dmitry Andreikin heroically saved a difficult rook endgame by
instantly switching from active to passive defence.
He obtained a chance of salvation due to Vladimir Fedoseev’s
mistake in solving the exchanging problem a little earlier.
In the notes, we pointed out that in the position of the diagram,

379
instead of the move 37... f7? Black should have simplified the
position by exchanging rooks . After 37... xf3+ 38.gxf3 exd5
39. xd5 c1+ 40. d1 xd1+ 41. xd1 f4! 42.exf4 g7 43. e1
a3! 44. e2 h6, Black’s position would have been relatively
easily winning.
Exchanging rooks in positions with an extra exchange is almost
always favourable to the stronger side . We saw this in the game
Shereshevsky-Veremeichik.
This unspoken rule is also seen in the following example. The
diagram position in the top-level game Carlsen-Karjakin arose after
White’s 43rd move.
Magnus Carlsen (2847)
Sergey Karjakin (2757)
Stavanger 2021

380
The advantage is on White’s side, and it is not easy for Black to
defend. Sergey Karjakin played 43...b5?. In the magazine 64 No. 10
for 2021, grandmaster Evgeny Gleizerov commented on this
moment in the game: ‘This looks logical – Black ensures the
exchange of a pair of pawns, which is usually beneficial for the
weaker side in such endings. But it was this move that the World
Champion at a press conference called a conceptual mistake . In his
opinion, Black was obliged to exchange a couple of minor pieces,
and he considers the position after 43... f4 44. xf4 xf4 45. f2 h4
to be completely defensible.’ (Gleizerov)
Carlsen’s comment immediately after the game made a strong
impression on me. He managed to instantly grasp the essence of the
position and explain it so accurately with a short remark. Black
needs to exchange the knight for the bishop with 43... f4, and, by
advancing the h-pawn to the h3-square, quickly increase the
security of his king. But the position was really difficult if Sergey
Karjakin, an excellent defender with excellent ‘masonry’ qualities,
failed to grasp its essence and lost almost without resistance.

381
Let’s dig a little deeper and describe Black’s problems in more
detail:
1. Black will lose if White manages to exchange one pair of rooks.
In this case, he will not be able to repel the attack on the king on the
dark squares.
2. Black needs to exchange his knight for the opponent’s bishop
and quickly expand the bridgehead for his king, in order to increase
his security. If White manages to place his pawn on the h4-square,
then Black will lose due to lack of space on the kingside.
3. Black needs to build a fortification in which his f5-bishop will
support the g6- and h3-pawns. There is a high probability that the
game will go into a drawn endgame with White’s king, rook and h-
and f-pawns against Black’s king and rook.
Now let us look at the computer’s first line, if Black follows the
path recommended by Carlsen: 43... f4! 44. xf4 xf4 45. f2 c4!
(Black will need to advance the rook’s pawn on his next move, since
the immediate 45...h4 is bad because of 46. c1! d7 47. e3 f6
48. d2 e8 49. c4 with a win for White) 46.b5 h4! 47.bxc6 xc6
48. e7 h6 49. b4 h3 50. b5 g6 51. g3 d6 52. f7 d3 53. xb6
h5.

382
 analysis diagram
This is roughly the kind of position that the sixteenth World
Champion was envisaging when he pointed out Karjakin’s
conceptual mistake with the move 43...b5? .
The assessment of my engine at a depth of 58 is +1.27. However, it
fluctuates for a long time between +1.2 and 1.3.
Let us continue the analysis of the game:
43...b5? 44. f2 c5 45.bxc5 xc5 46. c1!

383
Only three moves have passed and Black’s position is already lost.
The engine assessment is +3.77 at a depth of 40.
46... d5 47. d2! xd2+
If 47... e5, the simplest reply is 48.h4 and 49. c7.
48. xd2
With the exchange of rooks, White wins easily by attacking Black’s
weaknesses. The engine assessment is up to +4.25.
48... e5 49. c7 b4 50. b7 e6
A blunder in a hopeless position.
51. e7 d3+ 52. e3 c4 53. d4
Black resigned.
Associations involving the trade of a pair of rooks in positions with
an extra exchange also arise in this fragment from another game
between the same opponents.
Magnus Carlsen (2842)
Sergey Karjakin (2773)
Saint Louis 2018
384
In Endgame Strategy I gave the following note:
‘The time control has passed, and in this position it is useful to
outline a defensive plan. Trading rooks is advantageous for Black.
After this, he can easily hold the position if his king moves to the
centre. This means that the threat of exchange can push the white
rook away from active positions.
Then we need to understand what plan White will outline to
strengthen his position and an approximate scheme of his action.
Most likely, he will begin to push the kingside pawns, capturing
space and, if possible, creating threats to the enemy king. The most
dangerous attacks by the white rook on the black king are on the
sixth, seventh and eighth ranks, if we imagine that the white
kingside pawns are on the squares e4, f5 and g4. Therefore, it makes
sense to place our rooks on c8 and c6. These squares are not
controlled by the white bishop, and the white rook from the seventh
rank can always be driven away by moving one of the rooks to c7.
I suggest the following approximate variation: 41... e8 42.e4 ee6
43. c7 bc6 44. a7 c8 followed by 45... ec6. Black has nothing

385
to fear from 45.g5 in view of 45... ec6 46.gxf6 gxf6 47. d7 8c7
48. d8 c8 49. d2 f5, and he has almost achieved a draw.’
In the game, Sergey Karjakin came to this arrangement a little later.
He played 41... h1 42. c7 c1 , giving White several tempi to
improve his position. However, when the black rooks occupied c6
and c8, the position was almost equalized. Only after Black himself
violated the correct defensive formation did his chances of a
successful defence decrease sharply. After mutual errors in time
trouble, Carlsen won.
In Magnus Carlsen’s next game with Teimour Radjabov, the result
depended equally on both the defending side’s solution to the
exchanging problem and its choice between active and passive
defence. Both directions of the game were equivalent according to
the computer’s assessment, but with active defence associated with
a pawn sacrifice, it would be easier to play for Radjabov. We have
already mentioned that when it is easier for one of the sides to play
in an equal position, this is already an advantage. We will be
convinced of this more than once.
English Opening
Magnus Carlsen (2872)
Teimour Radjabov (2793)
Stavanger 2013
1.c4 c5 2. c3 f6 3. f3 c6 4.e3 e6 5.d4 d5 6.cxd5 exd5 7. b5
d6 8.0-0 0-0 9.dxc5 xc5 10.b3 g4 11. b2 a6 12. xc6 bxc6
13. c1 a7 14. e2 d6 15. e5 e7 16. ed4 xf3 17. xf3 fc8
18. d3 a5

386
In a quiet variation of the English Opening, Carlsen used an
interesting novelty, 14. e2!?. Let’s try to understand which
exchange of minor pieces is beneficial for each side.
You are absolutely right if you have determined the following: in
order to seize the initiative and complicate the situation for the
opponent, White needs to exchange his bishop for the knight. In
positions with hanging pawns, simplifications benefit White. But if
White tries to exchange his knight for the black bishop, then after a
line such as 19. b2 e4! 20. d4 xd4 21. xd4 f6 22. fd1 a4
23.b4 a3 24. c3 d6, the position is almost equal, and playing with
Black is quite simple.
19. xf6!
Now Black’s task is more difficult.
19... xf6 20. c2 d8 21. fc1 c5 22.e4
White had the nice tactical manoeuvre 22. b5 b6 23. e5, but
after 23... d6 24. c6 dc8 25. e7+ xe7 26. xb6 a4 27.b4 c4
28.a3 e6, it is easier for Black to defend than in the game.

387
22... g6 23. e1!?
Why did White defend the pawn with this rook? The move 23. e2
looks more natural. Maybe there is no great difference between the
two moves, but I would venture to suggest that Carlsen made his
choice for a reason. The game is close to going into an endgame and
nothing is ever too trivial for him at this stage of the game.
23...dxe4 24. xe4 xe4 25. xe4

In this position, Black should aim for the move ...a5-a4. White, in
order to count on something, will have to take this pawn and then
move the knight to c4. The value of the potential possibility of
approaching the blockade square c4 with the white king also
increases sharply. We have already mentioned that if there are four
rooks on the board, walking around with the king in the centre is
quite dangerous.
With the move 23. e1!? , Carlsen persuaded Radjabov to play
25... d1+
with the exchange of one pair of rooks. The engine approves of this
move, because with correct play Black can hold the resulting
388
position. But for White, with only one pair of rooks, it is easier to
play . The engine today cannot take into account such subtleties of
human thinking.
A game with four rooks on the board could become more unpleasant
for White, although this option for Black seems less strong to the
computer. After 25...a4!? 26.bxa4 f5 27. e1 b6 28.g3 ( 28. c4
a5 ) 28... xa4 29. e6 a5! 30. xc5 xa2 31. xf5 f8 32. xf8+
xf8 33. c6 d8 34. c8 e7

analysis diagram
White does not have great practical chances of winning, although
the engine assesses the advantage as greater than after the exchange
of rooks in the game.
26. e1 xe1+ 27. xe1 d8 28. f1 a4 29.bxa4 d4 30.a5 a4

389
After a series of natural moves, the position has become simpler.
White retains a slight positional advantage and seeks to move his
knight to c4, blockading the c5-pawn. If he succeeds, then Black’s
bishop will become passive, and White’s passed pawn on the
queenside could turn out to be quite dangerous. White’s main
advantage will be the lack of counterplay from his opponent.
31. d2
Magnus acts inventively, using short-range tactics. In case of the
straightforward 31. d3, White would have to reckon with the move
31...f6!?, with a threat of 32... c4. The rook ending after 32. xc5
xc5 33. xc5 xa2, as we know, is a draw.
However, it was easier to deprive the opponent of active counterplay
associated with the sacrifice of the c-pawn. White could have played
simply 31. f3!?. Now an attempt to become more active with the
move 31...c4? could have ended badly for Black due to the answer
32. e5!. After 32... xa5, White has the attack 33. c6 at his
disposal! The rook endgame after 33... a6 34. xa7 xa7 35. xc4

390
g6 36.a4 is bad for Black due to the passive position of the rook on
a7.
The play in this case would most likely have proceeded
approximately as in the game: 31... xa5 32. e5 f8 33. c4. FIDE
master Artem Tjurin pointed out this possibility during a training
session.
Let’s continue the analysis of this interesting fight.
31... f8 32. d3! f6
White’s idea is the move 33. b2 in reply to 32...c4. Black remains a
pawn down. With 32...f6 , Teimour keeps the pawn, but he does not
manage to hold the position.
We can only wonder if Black would have held the draw had he
sacrificed the c-pawn but activated all of his pieces. A sample
variation: 32...c4!? 33. b2 xa5 34. xc4 a4 35. c2 e7 36. e2
e6 37. d3 d5.

analysis diagram

391
My Stockfish 15 evaluates the position +0.07 at depth 47. It offers
38. e3+ xe3 39. xe3 h5!.
We have before us an analogue of the position from the game
Karpov - Knaak (see Chapter 7),

only in our game, the pawn is on g7, not g5. This will allow Black
to advance his pawn to h4 and create enough counterplay on the
kingside for a draw.
The engine fully approves of Radjabov’s 32...f6 . But it is much
easier for a human to play a rook endgame a pawn down with active
pieces, than to hold an almost equal but passive position with the
same material, especially if there is not enough time to think. In
addition, the exchange of bishop for knight with 38. e3+ xe3 is
not necessary. And defending the position after 38... e6 instead of
38... xe3 for Black is, once again, easier than what happened in the
game.
In our example, the margin of safety in Black’s position is quite
large, and he can defend successfully after any of the listed
continuations. The key word, which we have emphasized several
times, is easiest . Which line is easiest to defend? In an active
392
position with limited material a pawn down or in a passive position
with an equal number of pawns? In both cases, if played correctly,
Black’s position is defensible. Each chess player must make his own
choice in such situations.
Radjabov kept the pawn, but could not cope with the difficulties of
the defence.
33. b2

33... xa5
If Black has already taken the path of passive defence, then why not
study the exchange of the second pair of rooks after 33... xa2
34. c4 xd2 35. xd2 e7...? Visually, this method of action looks
very stupid. But in the age of computerization of chess, it often
happens that such play is the most effective way to hold a position.
All this looks very dangerous for Black, but the game is greatly
simplified. My engine promises White a minimal advantage in a
drawn position. Here is one of its approximate variations: 36. e2
e6 37.f3 e5 38. c4+ d5 39. d3 c6 40. a3 b8 41.h3 c7
42.a6 b6 43. b5 b8 44. c4 xa6 45. xc5 b7 46. d5 (
393
46. d6+ c7 47. e8+ d7 48. xg7 e7 still gives White some
edge) 46... e5 47. a3 c7 48. e6 c6 49. c4 f4 50. b2 g3
51. d3 f5 52. xf5 d7.

analysis diagram
It is easy to quote long computer lines ending in a draw, despite the
fact that from a distance, the positions look very dangerous to the
human player, especially with limited thinking time.
34. c4 a4 35. c2 e7 36. e2 e6 37. d3 d5

394
The opponents have brought their kings into the centre, and now
Carlsen asked himself his favourite prophylactic question: ‘What
will be Black’s next move?’ The answer is obvious – 38... b4.
Therefore:
38.a3!
Black is caught in a kind of zugzwang. 38... b8 is impossible due to
the fork on b6. And in the event of 38... a6 or 38... c6, the
pressure is removed from White’s knight, and he gets a choice
between activating the rook by 39. e2 and transferring the king to
b3 via c3. Black’s position remains defensible and my engine shows
zeros.
Now, for clarity, let’s remove the c5-pawn from the board.

395
Here too, after the natural move ... a7-c5 for Black, the computer
gives zeros. But ask yourself which position is easier for Black to
play, especially with little time on the clock.
We return to the game.
38...h5 39.h3 h4 40. c1 g6 41. c2 g5 42. c1 a6 43. e1 b8
44. e7 f4
If 44... e5 White has the possibility of 45.a4, and now 45... xa4? is
bad because of 46. xe5+!.
45. c3

396
45...f5
The engine slightly condemns this move, considering that after
45... e5+ 46. b3 d4 the position is completely balanced.
The Norwegian super tournaments are famous for having very tight
time controls. It is not surprising that the Azeri grandmaster did not
want to send his king into the ‘lion’s mouth’. If, say, the white f-
pawn stood on the f3-square, such a journey would immediately end
with the move 47. d7+. My engine produces the following unforced
variation: 47. d7+ e4 48. d2 d4 49.a4 f5 50.a5 g4 51. a4 gxh3
52.gxh3 f3 53. b5 a8 54.a6 g2 55. b6 a7 with equality.
Calculating such lines is not the most enticing activity when you
don’t have enough time to think. It is not surprising that Teimour
tried to get counterplay, leaving the king in the centre.
46. b3 g4 47.a4 gxh3 48.gxh3 g6 49.a5 g1 50.a6 b1+

397
The game has become sharper but the computer assessment remains
zero. However, Black’s position is now much harder to play! Any
player would choose White here.
51. c3
After 51. a2 b4 52. e3+ xe3 53. xe3 b8 54.a7 a8 55. a3
d4 56. b2 c4 57. a2 b4 the engine proves that the position is
a draw. But in order for a person to understand this, he needs to
continue the analysis. We will not engage in analysis now. But I
would like to ask you a naive, childish question. What is the purpose
of any chess analysis? I once found the answer to this in the book of
Isaac Lipnitsky: the purpose of analysis is to bring us from a
completely unclear position to one that is absolutely clear, which
can be assessed at first glance.
Compare these two diagrams:

398
diagram A

diagram B

399
In order to understand the situation in the first diagram, we need to
continue the analysis. But to assess the second position with the
white knight on f3, one glance is enough for us to confidently
declare: it’s time for Black to surrender.
As we have already noted, the engine assessed the position in the
first diagram as a draw. But Black holds it, being on the verge of
losing. Let’s imagine a little more. Let’s move the black pawn from
f5 to f6 in the diagram after Radjabov’s move 50... b1+.

Here White wins elegantly after 51. a2 b4 52.a7! a4+ 53. b3

b4+ 54. a3 c1+ 55. a2 a4+ 56. b1.

400
 analysis diagram
Now on 56... b4+ there follows 57. b2!! xb2+ 58. a1. The
move 56... c6 does not save Black because of 57. e6+ b7
58. e4, whilst after taking the knight with 56... xc4, White plays
57. e4+, winning. Here we see how the position of the pawn on f5
instead of f6 has a decisive effect. Thus Black’s defences hang on
what appears to be such a trifle.
We have paused at this moment so that we may see how some
positions are easier to defend and some are more difficult.
We return to the game after Carlsen’s move 51. c3.

401
51... c1+ 52. d3 d1+ 53. e2 a1 54. b6+ d6
According to the computer, the strongest move was the king’s
retreat to c6, but then the game would take on a concrete, more
calculational character. Radjabov had less time remaining and the
game becomes sharper.
55. g7!?
Technique is the mastery of short-range tactics. The a6-pawn is
immune because of the check on g6.
55... c6 56. g6+ b5 57. d5 e5?
57... a2+ 58. f3 d2 was best, with equality.
58. b6+ c4 59. e3+ c3 60.f4!
Magnus presses the whole time.
60... d4 61. xf5 c4 62. c6

402
After the strong move 62... a5!, the assessment of my engine is
+0.46. But a man is not a computer and to play such a position after
60-odd moves of very hard struggle is extremely difficult.
Teimour commits the decisive mistake, by attacking the h3-pawn:
62... h1? 63. d6!
Enemy number one must be eliminated. White also wins after
63. xd4 xd4 64.f5, but this continuation requires serious and
accurate calculation of variations.
63... h2+ 64. f3 d3 65. xc4 xh3+ 66. g4 h1 67. a4 f2
68. a3+! 1-0
The variation 68... e2 69. a2+ f1 70. a1+ g2 71. xh1 xh1
72. e4 or 72. h3 is convincing enough.
In this interesting example, we see that positions which are equal by
computer standards are not equal from a human point of view, if
they are easier for one of the sides to play.
This endgame is also interesting because Magnus skilfully held his
position while standing still. He moved the rook from c2 to c1 and
back, thereby causing the opponent’s position to weaken. Such
403
manoeuvres are often found in Carlsen’s games, and they arise from
our favourite prophylactic question: ‘What does the opponent want
if it were his move?’
Associative thinking immediately reminds me of a similar style of
play by Carlsen in his game with Anand, analysed in my book
Endgame Strategy .
Magnus Carlsen (2776)
Viswanathan Anand (2791)
Linares 2009

Anand was the World Champion at that time and the young Carlsen,
one of the strongest GMs in the world, was just starting to earn a
reputation for being able to get blood out of a stone.
Black’s position looks solid, but by subtle manoeuvres, Magnus
finds a way to breach it:
30. g7 h8 31. c2 c8 32. b3 h8 33. g1 c8 34. g7 h8
35. g2 c8 36. g1!
Doesn’t all this, dear reader, remind you of White’s manoeuvres
from the previous game? There is a kind of zugzwang on the board.
404
The move 36... c4 suggests itself, but after 37. xc4! xc4 (
37...dxc4 38.e4 ) 38. g7 e7 39. h7 a4 40.b4 xb4 41. xh6
a4 42. h8 White should win the rook endgame. And if 36...b5
37.axb6 ( 37.a6!? ) 37...axb6, White captures the a-file with the
move 38. a1, though Black myay still defend with 38... e8! with
the idea ... e8-f6xh5.
36... e8 37.e4! fxe4 38.fxe4 f6 39.e5 e4 40. e3!

40...b6?!
Black is haunted by mirages of zugzwang. With the last move of the
time control, Anand avoided an unexpected tactical trap: 40... g5!
41. f1 e4 42. xf7! xf7 43. xd5+. The computer, however,
finds a fantastic flaw in White’s plan. After 41. f1 Black is saved
by an inhuman solution: 41...f5!! 42.exf6 h7!! 43.f7 f6!, and the
worst is behind him.
Chess programs have made giant leaps forward recently. To explain
the computer’s recommendations, you must have decent chess
qualifications. And from the many directions that it evaluates as
equivalent, choose the one in which it is easier for a person to play.
405
In reply to Anand’s 40th move, Carlsen carried out the natural pawn
exchange:
41.axb6
According to the modern engines, this is a serious inaccuracy. He
could have won with the unexpected move 41. d1!!.

analysis diagram
The threat is 42. g4+. Black’s rook is forced to leave c8. Let’s first
analyse the moves along the eighth rank. For example, 41... f8.
Then White exchanges pawns with 42.axb6 axb6 and calmly
strengthens the position with 43. e2! with the terrible threat 44.
c1. The black rook does not control the c-file, and my engine gives
an evaluation of +4.5.
After, instead, 41... c4, correct is not 42.axb6 but 42. g8!. For
example, 42... c1 43. b3!! with the threat 44. d8. After 43... e1+
44. d3 bxa5 45. d8, White wins. But what happens if Black plays
42... b4...?

406
 analysis diagram
That is when White needs the pawn on a5. He will answer 43.a6!
xb2 44. g4+ e7 45. a8 with a win. The move 42.a6 would have
won even after 41. d1!! b8.
All these subtleties are not easy to understand even for a qualified
specialist, and the game itself goes beyond the level of human
capabilities.
Let’s return to the game after Black’s move 41...axb6 .

407
Here too, the engine recommends 42. d1, but after 42... c4! 43. g8
b4 44. e8+ d7 ( 44... f5? loses to 45.b3 ) 45. h8 g3 46. g4+
e7 47. xh6 f1+ 48. d3 b3+ 49. c2 g3,

analysis diagram

408
Black holds the position with the help of some tactical trickery.
But in the same typical style he demonstrated against Radjabov,
Magnus puts Anand in zugzwang, by transferring the move to him.
42. d3!? f2+ 43. e2 e4 44. e3
Now Black cannot concede the c-file to his opponent or lose control
of g8. On 44... g5 there is the unpleasant 45. a1. Therefore,
44...f6
is forced.
45. g6 c1

Now the tiredness from the extreme tension started to tell and the
quality of play declined. Magnus played:
46. xh6?!
And now:
46... h1 47. c2 h3+?!

409
The move 47... e1+! would have led to a draw. Eventually, though,
Anand didn’t manage to hold and the game ended in a win for
Carlsen.
The following would have been a more logical course for the game:
46.exf6 xf6 47. xd5+ e7! (everything else is bad for Black).

analysis diagram
The modern program easily holds Black’s position; the earlier
versions had difficulty doing this.
In both roughly equal positions, Carlsen showed persistence and
patience, asking his opponents difficult questions one after another.
It was precisely by this manner of play that he gained the reputation
of a chess player who ‘squeezes blood out of a stone’. Watching
how delicately Carlsen realizes a minimal advantage, many ask the
question: ‘Is it even possible to get close to such a strength of play
in such endings?’
Yes, you can learn to take care of your smallest chances of winning
with the help of special training exercises. Their essence is to play
positions in which neither side has an advantage, but it is easier for
410
one of the opponents to play. To model such situations, it is good to
select examples from tournament practice and invite the two sides to
play them with a time control of 40 minutes for the game with an
increment of 10 seconds after each move.
I actively use this method of training with my grandmaster
colleague Evgeny Tomashevsky, who acts as a sparring partner for
our players. I select positions that are objectively equal, but are
easier for one of the sides to play. Of course, before starting the
game we make sure that these examples are unknown to both
players.
For one of these training sessions, I picked up the following
fragment and set the position in front of the players:
Boris Gelfand (2714)
Yannick Pelletier (2531)
Biel 2001

After
26... d5?! 27. xd5 xd5 28. e4 b4?! 29. d2!
Black fell into serious difficulties.
411
In his book Technical Decision Making in Chess , Boris Gelfand
writes about 26... d5?!: ‘The strongest move was 26... b5!, and I
do not see how White can exert any pressure. Winning a pawn with
27. xa7 xa7 28. xa7 makes no sense, since Black has 28... xc4
29.bxc4 b1+ 30. f1 c1 with an immediate draw.’
Now let us see how this position was played out in a training game
between two grandmasters, Evgeny Tomashevsky with White and
Aydin Suleymanli with Black:
Interestingly, although unfamiliar with Gelfand’s recommendation,
Aydin did play:
26... b5!
The game continued:
27. a5 xc4 28.bxc4 b4 29. xe5 xc4

The position is completely equal, but it is a little early to agree a

draw. White’s position is a fraction easier to play.
30.h3 a6
A micro-inaccuracy. Better 30...h5 .

412
31. ea5

An interesting moment has arisen. Black should stick to a precise

defensive line here with 31... e4!. In this case, it would be difficult
for White to refrain from the exchange of knights, for example:
32. e2 c2 33. d4 d2 34.f3 d6 35. xa6 xa6 36. xa6 c4
37. a8+ g7 38. d8

413
 analysis diagram
38... d1+ 39. f2 d2+ with a draw ( 40. g3? xe3!, and the d2
is indirectly protected due to the fork on f1).
From the point of view of the fight for a draw, the move 31... e4!
had to be made. But a chess player is prevented from playing this
way by a heightened sense of contradiction: moving to defend a
drawn endgame with three pawns on the kingside against four
means recognizing the opponent’s superiority. Surely, many in such
a situation have the thought of active counterplay with the passed a-
pawn, and at the same time dream of seizing the initiative.
Sometimes such dreams come true, but more often than not such
play leads to problems.
There are other situations – a chess player is afraid to move into a
theoretically drawn endgame a pawn down, because he simply does
not know the method of defence.
Of course, the young Azeri grandmaster did not give up the pawn
and played instead:
31... c6 32. 1a4!? b6 33. f1 g7?!
414
If one move ago it was desirable to play 33...h5! instead of
33... g7?!, then here it was absolutely obligatory.

34.g4!
Of course! Such a space gain has been seen several times in this
book. According to the computer, the position is still equal, but
White’s position is already easier to play.
34...h6?
Up to now, Aydin has committed only micro-imprecisions, but this
move is already a serious mistake, weakening the g6-pawn. It was
better to play immediately 34... e8, keeping the h-pawn on h7.
35. g3 e8 36. g2
Even more accurate was 36.h4!.
36... c7
Black’s position is already not easy to play, taking into account time
pressure. It was stronger to transfer the knight to d6 with the idea of
37... c8.
37.h4! b5 38.h5 ab8 39.hxg6
415
Black faces a complicated exchanging problem, and he had to solve
it ‘by sight’, as he had little time to think.
39... xg6?!
The engine taunts the human player by suggesting 39... xa5
40. xa5 b5! 41. a1 fxg6! with a probable draw.
The game continued:
40.f4!

416
Only now does White’s advantage assume real proportions,
although my engine still assesses it as only +0.3 on a depth of 50.
40... xa5 41. xa5 b5 42. a4 a5 43. f3 d5?!
In Tomashevsky’s opinion, 43... e8!? was stronger.
44. d4!

417
Such a position is difficult to defend even with enough time to think.
In a game against a technically strong and motivated opponent
under time pressure, this task seems impossible. It is completely
unrealistic for Black to execute the series of precise computer
moves 44... b4 45. d6+ h7 46. d7 b6! 47. a7! a6 48. xf7+
g8 49. b7 c6 50. e4 a4 51. c3 a3 52. a2, with good chances
of a draw.
Suleymanli replied:
44... f6 45. f5 h5?

418
Black’s last move was the decisive mistake. His only chance was to
sacrifice his passed pawn with 45...a4! . After a line such as 46. xa4
h5 47. a6+ b6 48. a5 hxg4+ 49. xg4 c3 50. g3, my computer
assesses White’s advantage as +0.8 at a depth of 35.
46.gxh5!
Technique is the mastery of short-range tactics. In the rook ending,
Tomashevsky confidently conducts the game to victory.
46... xf5 47.e4+ f6 48.exd5 b3+ 49. e4 e7 50. a4 b5
51.h6 b6 52. xa5 xh6 53. a7+ e8 54. e5 f8 55.d6 h5+
56.f5 e8 57. f6

419
Black resigned on account of 57... h6+ 58. g7 xd6 59.f6.
With his technical and persistent play, Evgeny Tomashevsky
showed Aydin Suleymanli how important it is to make the most of
your insignificant chances in playing for a win. By regularly training
in such endings, a chess player can significantly improve both the
technique of defence in inferior positions and the technique of
realizing a minimal advantage.
Let’s look at an example from a game in which the stronger side
passionately sought victory, but the opponent put up a stubborn
defence. One of the most important factors in it was the problem of
exchanges.
Alireza Firouzja (2749)
David Anton Guijarro (2675)
Wijk aan Zee 2021

420
We are witnessing a complex middlegame without queens. The
position is approximately equal. Black has a deformed pawn
structure on the queenside, and on the half-open e-file, there is a
backward pawn on e6. The b5-pawn, which will strengthen the
black knight on the important point c4, is impossible for White to
get close to. The e6-pawn, although backward, has dynamic
potential. White will have to watch this pawn’s advance all the time.
18. e1 f7 19. c1
The knight transfers to the ideal square d3. It is time for Black to
decide how he intends to arrange his forces.
19... f8?!
The Spanish grandmaster takes the knight to g6, where it will take
control of the f4- and e5-squares, but the entire arrangement of the
black pieces on the kingside will become somewhat awkward and
inharmonious. Whether White will be able to take advantage of this
circumstance is another question.
In my opinion, the place for this knight is not on g6, but on c4, and
on the g6-square there should be a black pawn. I would suggest
421
approximately the following move order for Black: 19... he8
20. d3 f8 21.g3 b6 22. g2 c4 23. e2 g6.

analysis diagram
Now the black pieces are arranged very harmoniously. They are
ready to repel the pressure of the opponent’s piece regroupings and
at the same time are ready for counterplay in the centre, associated
with the advance ...e6-e5. I admit that I only managed to come up
with such an arrangement in retrospect. And frankly, the engine
does not consider this a significant improvement.
I would like to dwell on the issue of the correct placement of pieces
in such a situation in more detail. Chess today has become very
much a sport, and one must spend one’s thinking time rationally.
The calculation of variations after White’s 19th move plays a
secondary role. Here it is important to understand whether the
position is critical and whether the fate of the game will depend on
the next move related to the location of the pieces .

422
Let’s look at two examples of critical positions from the game
Tomashevsky-Sjugirov which was discussed in detail in my book
Endgame Strategy .
Evgeny Tomashevsky (2708)
Sanan Sjugirov (2612)
Moscow 2009

25. c2!
Threatening to attack the b4-pawn with the aid of the knight transfer
to a2 or d3 via the square c1. Black faces a choice between active or
passive defence. Sjugirov chose the first plan – wrongly.
25... a5?! 26. c1 c5 27. d3 b6 28. hc1 d6 29.e4! dxe4
30.fxe4 cxd4 31. c4
White won a pawn, which he gradually realized.
This happened in large measure due to the poor position of the
knight on f6. Instead of this, Black should have used his worst piece
to meet the enemy threats. After 25... e8!? 26. c1 c7 27. a2

423
 a6 28. f1 db8, White retains some initiative, but he faces
stubborn resistance.
But most positions in a game are not critical, and errors in the
placement of pieces do not often lead to such dire consequences.
The formation can be improved during the game. In this case, the
player is faced with a choice:
1. Decide the best arrangement of the pieces, by spending a certain
amount of time on this.
2. Play by eye, saving precious time for situations when the pieces
come into contact and hand-to-hand combat begins.
Of course, there are chess players who have a great sense of the
interaction of their pieces, and they manage to find the best
arrangement in the minimum amount of time. This is an indicator of
excellent ‘masonry’, and among our contemporaries in terms of
average speed, Carlsen, Anand, Kramnik and Karpov stand out.
More often than not, the first move that comes to their mind during
the game is the best.
I remember the old saying, ‘It is better to be rich and healthy than
poor and sick.’ But in chess, not everyone is able to quickly find the
strongest moves, so sometimes you have to choose. At the same
time, it is worth remembering that in a concrete calculating battle,
the cost of an error increases sharply, and you will probably regret
the time spent on general reasoning. The choice is up to everyone.
Let us return to the game Firouzja-Anton Guijarro. From our
discussion above it is easy to see that the position after 19. c1 is
not critical. This means that the move 19... f8 was perfectly
reasonable for Black.
The game continued as follows:
20. d3 g6

424
An interesting moment. The players have to constantly ask
themselves the questions ‘Where? Who? Why?’ – and find answers
to them.
It is White’s move. What should he do with the bishop on f4? There
are three options: 1) leave it in place; 2) keep it on the h2-b8
diagonal, avoiding the exchange with the move 21. g3; 3) play
21. c1, leaving it on the c1-h6 diagonal.
It’s time to focus on solving the problem of exchanges in this
position. In my opinion, Black benefits from exchanging one pair of
minor pieces, since he has little space on the kingside. The main
events will most likely unfold there, since on the queenside the
pawn structure is fixed due to the doubled pawns on b7 and b5. The
last thing Black needs is an exchange of bishops, because the white
knights are more manoeuvrable than the black ones, and the b4-
square will become more accessible to the white king.
Black will benefit from both the exchange of the knight on g6 for
the bishop and the exchange of the bishop for White’s knight. In the
first case, he will position his minor pieces perfectly: the bishop will
stand on d6, and the knight on c4. In the second case, White’s
425
pressure on the e6-pawn will weaken. White needs to keep all three
minor pieces for the time being, or exchange bishops.
21. g3
Both bishop retreats have their pros and cons. On g3 the bishop
further controls e5, but blocks the g2-pawn. On c1 the bishop
reduces its control of the centre.
21... a5 22. g1
In general, the best square for the king is somewhere around c2, but
White must watch his opponent’s counterplay. If 22. e2, so as to
march to c2 behind the cover of the rook, my computer offers the
following crazy variation: 22... xa3!? 23.bxa3 xc3 24. e3! hc8
25. h3! (not a move for ‘average minds’!) 25... e7 26. d6! f5
27. de5+ g8 28. xc3 xc3 29. b4 c1+ 30. e2 c4 31.g4!
hxg4 32. xg4.

analysis diagram

426
The standard score is +0.13. The sides have equal chances. If
someone were to find all these moves, then probably he would
immediately be accused of cheating.
Instead of 22. g1 or 22. e2, the ugly move 22. e2!? is much
stronger, but high chess culture rarely allows great chess players to
seriously consider such moves with the current strict time control.
After 22... c4 23. d1 the white king safely reaches the c2-square.
White could then double his rooks on the e-file and begin moving
the knight from f3 to h3, strengthening his position on the kingside.
But let’s continue to monitor the events in the game.
22... c4 23. e2 he8 24. h2 f8 25. he1 b6?!
It’s not very clear why this pawn was advanced, but let’s not
quibble. I wrote these comments a year ago. I did not like the move
25...b6?!, but I couldn’t clearly formulate why it was bad. The
engine was indifferent to the move. After some time, I studied
Gelfand’s book Technical Decision-Making in Chess , in which
Boris was able to find the exact definition of such moves: ‘unforced
concession’!
In a game with Harikrishna, Gelfand
Boris Gelfand (2777)
Pentala Harikrishna (2706)
Wijk aan Zee 2014

427
25...f6?!
Gelfand gave this commentary on Black’s last move: ‘An unforced
concession, firmly linked to the next move. I think the correct
method here is to change the direction of the game, retreat 25... e7!
and maintain a solid position. I don’t see how White could achieve
anything significant.’
The key point to understand about unforced concessions is this:
‘Sometimes we make them and everything seems fine, but 10-15
moves later we have to be a little more precise to avoid getting into
a position with real problems. The computer will claim that the
mistake was made later, since precise moves still equalize the
position, but the grandmaster is already experiencing discomfort.’
(Gelfand; emphasis added)
In the game Gelfand-Harikrishna we note that after 26. c4 , a new
series of concessions followed:
26... xc4?! 27. xc4 e7?

428
It was essential to seek counterplay with the raid 28... h2 after
27... d6!. It is easiest to acquaint yourself with Gelfand’s deep and
instructive analysis of the game in his book.
Also, in the game Firouzja-Anton , the position of the pawn on b6
only caused Black difficulties on move 44, as referred to in the
italicized section above.
That game continued:
26. g1 a8
Black waits.

27. f4
The problem of exchanges! It seems as though Black cannot
exchange on f4 because of the double threat to the pawns on h5 and
e6. But after 27... xf4 28. xf4 g6! 29. xe6 d6! ( 29... xb2??
30. eg5+ ) White cannot defend his queenside and the game heads
for a draw.
The aforementioned king transfer to c2 was rather more unpleasant
for Black, for example 27. f1 ad8 28. a1!? a8 29. e1 ad8
30. d1 a8 31. c2 ad8 32. ae1.
429
 analysis diagram
Then White would have the opportunity to move the knight to h3,
and Black’s position would become alarming.
Of course, Black did not have to stand still, moving the rook from
a8 to d8 and back. He should have improved the position of the
knight on g6 and put a pawn in its place. But probably neither
opponent took into account Black’s tactical possibilities after the
capture 27... xf4!.
27... ac8 28.g3 c6
Now the capture on f4 becomes impossible because of the
undefended d5-pawn.
29. g2 e7 30. g1 f8 31. h3 e7
Black has successfully held his defences. If the bishop retreats to c1,
the reply is 32... d6. It is clear that Black’s defences cannot be
broken without his help.
32. g1 f8 33. f3 e7 34. h3 a8 35. c1 e8
Let us compare two positions:
430
diagram A

diagram B

431
In Diagram B, White begins to calmly open up the game on the
kingside with g3-g4, while his king watches from the safe square c2.
In the game (diagram A), the white king himself will have to take
part in the battle when the board is full of enemy pieces. In this case,
no one can guarantee the king’s safety.
36.g4!
The moment to begin active operations was chosen well. There are
still five moves to the time control, and Black, under time pressure,
needs to switch from positional schemes to calculating specific
variations. White conducts the game in a ragged rhythm. This style
of play is very unpleasant for the opponent when he has little time
left to think.
36...hxg4+ 37. xg4 h8! 38.h5 f8 39. f4 d8! 40. h1!
The king can only go in for such dashing rides deep in the endgame.
After the direct 40. h4? f5+ 41. g3 c7 White is in trouble.

40... g8?

432
Time trouble. Correct was 40... c7! 41. g6 f5+ 42. g5 d8+
43. f4 c7+, and White should settle for a draw by repeating
moves with 44. g5, since if 44. fe5+ xe5+ 45. xe5+ xe5
46. xe5 ( 46. xe5?? xh5 47. xh5 g6# ) 46...g6 47.h6 f6, he
already needs to show accuracy so as not to fall into an inferior
position.
41. h4!
Now the white king has a relatively quiet post on f3, whilst Black
comes under a press on the kingside.
41... c7 42. fg6 d7
Black has an unpleasant position after 42... xg6 43.hxg6+ e7
44.f4.
43. he1 e8 44. f4!

Discussing the problem of exchanges in this game, we came to the

conclusion that exchanging bishops is beneficial for White. Black
was satisfied with the exchange of his bishop for a knight, but on the
f8-square without the white pawn h5 moving to g6.

433
At this moment, Black’s ‘unforced concession’ associated with the
move 25...b6?! starts to be felt. If twenty moves ago Black had left
his pawn on b7, he could now have easily retreated with the bishop
to a5 and maintained a reliable defence by moving the knight from
d7 to h7 via the f8-square . As Boris Gelfand wisely noted, an
‘unforced concession’ does not show immediately, but after many
moves – in this case, after twenty.
How to recognize in time an ‘unforced concession’, i.e. not to make
it? There can be no exact recipe here. Much depends on the
experience and intuition of the player. But if at the time of choosing
a candidate move you do not like it from an aesthetic point of view
and it is not forced by specific circumstances, try not to make it.
This is an indicator of high-level masonry. It is interesting that the
computer does not notice such ‘little things’, since their
consequences appear only after many moves.
Let us return to the consequences of the move 25...b6?!. The reply
44... d6 is now impossible. The exchange of the white bishop for
the knight suits Black very well in principle, but White wins with
the help of a beautiful tactic: 45. xe6! xe6 46. h8+ g8 47. xe6
xh8 48. e7! f8 ( 48... b8 49. xd6 ) 49. xc7 xc7 50. xd6.
Instead of going for the combinational complications, White can
also simply play 45.f3, depriving the black pieces of coordination.
After the move 44. f4!, Black faces a difficult choice, which is
often the case when defending inferior positions. Does he need to
agree to the exchange of bishops and patiently suffer in a bad
position? Or should he try to change the course of the game by
means of 44...f5+ , but risk losing quickly?
The Spanish GM played:
44... xf4?!
– wrongly. My computer suggests this variation: 44...f5+!? 45. f3
f6 46. xc7 xc7 47. f4 xh5! 48. xh5 h8 49. f4!? xh5

434
50. f3 g8 51. xe6 xb2 52. e8+ f7 53. g5+ xg5 54. xg5
d3 55. 1e2 xf2 56. 8e3 e4+ 57. xf5 c6.

analysis diagram
Black, as they say, ‘never had it so good’. But seeing a tactical blow
like 47... xh5! is far from easy for a tired human player.
45. xf4 d6

435
It seems as though things are not so bad for Black. Everything is
defended. As well as the pawn on e6, White needs to find another
weakness in the enemy position.
46. h3!
The pawn on g7! It is quite hard to come up with the idea of moving
the rooks off the e-file to the g-file. The weakness of the g7-pawn is
not obvious at first sight.
46... f8 47. fg6 h7
Bad is 47... xg6 48.hxg6+ followed by 49. f5 and a mating attack
along the h-file.
48.f4! dd8
The rook cannot come to d7: 48... d7 49. xe6 xe6 50. h8+.
49. g1

436
Black to move. With a lack of time to think or with only increment,
almost everyone would play:
49... d6?
This move was the decisive mistake! A computer, unlike a human,
has no associations with the previous move. For it, after each move,
a new position arises. But after the rook g1 moved to e1, it is
imprinted in a human player’s memory that it was impossible to
move the rook to d7 due to the answer h4-f5. In time pressure, it is
very difficult for a chess player to realize that circumstances have
changed, and now moving the rook to the d7-square was not only
possible, but also necessary – especially if the player is already
determined to transfer the knight from square c4 to e4. After
49... d7! Black would have held the line. Most likely, he would
have lost this game anyway, since it is very difficult to defend such
a position under time pressure.
After Black’s mistake, there was swift action:
50. eg2 g8
The threat was 51. e5+.
437
51. e7!
Technique is the mastery of short-range tactics.
51... g5+ 52.fxg5 xe7 53.gxf6+ xf6 54. g6+ f7 55. f3
df8 56. e5+ e7 57. xg7+ xg7 58. xg7+ f6 59.h6 h8
60. g4 xh6 61. d7 1-0
In all of the following examples in this chapter, the problem of
exchanges will play a key role for the defender. The result of the
following game depended on a correct solution to this problem.
Radoslaw Wojtaszek (2686)
Richard Rapport (2763)
Berlin 2022

White has an extra pawn, but Black has an advantage in space and
his pieces are very active. First of all, he needs to decide: should he
preserve his pawn leverage on the kingside, maintaining the tension
there, say, with 31... 3b5, or eliminate it? If Black decides to relieve
the tension on the kingside, how should he do this? With 31...hxg4
or 31...h4...?

438
All these very difficult questions are related to solving the
exchanging problem and require time to think. But I doubt that the
Hungarian grandmaster had this.
31...hxg4?!
Sadly, this is the worst possible choice. The move 31... 3b5 (or
31... 8b5 ), preventing the strengthening of the white position by
c4-e5-d3, required assessing the position after 32. d6 5b6
33. c5 xb2 34. xg5+ f8 35. c4. But the best option was to fix
the white pawn on h3 with 31...h4! and take the knight to the square
g6. In this case, Black retains sufficient counterplay to maintain
equality. For example: 32. e5 3b5 33. d3 d7 34. c4 b3!
35. xa4 xd3 36. xd3 xb2+ and 37... c5. After the exchange on
g4, Black’s temporary activity on the h-file can easily be
neutralized, whilst White’s extra pawn becomes a more significant
advantage.
32.hxg4 h8 33. e5 b5?!
It was more logical first to harry the white king with the check
33... h2+.
34. d3 h2+ 35. f1?!
Why not 35. e3...? Obviously, the players were in time-trouble.
35... xd2 36. xd2 d7

439
Once again, it is useful to discuss the problem of exchanges. Black
needs to strive for the exchange of knights. With them off the board,
White has weak pawns on b2 and f3 and the black rook on b3
promises sufficient counterplay. White, on the other hand, would
like to go into a knight endgame.
37. e2 f6 38. c2 b8
Not 38... e5 because of 39. c5.
39. c4 e5 40. b4 xb4
After 40... a8?! it is hard for Black to count on saving the game
with his rook passive.
41. xb4

440
41... c4?!
A tactical mistake. He should have started with the move 41... e7.
42. d3?!
Technique is short-range tactics!
White wins beautifully with 42. c6!! xb2 43.e5+ g6 44. a5! f6
45. d2, and the black knight is trapped and lost within a couple of
moves.
42... e7 43. d1 d6?
The computer suggests 43...f5! 44.exf5 exf5 45.gxf5 f6 46. c2
e3+! 47. c3 xf5 48. b4 d4 49. e1 e5 50. xa4 f4
51. b4 xf3 52. xf3 xf3, and White has to play the endgame of
queen and pawn against queen. After Rapport’s move, the game can
already not be saved. The Polish GM conducts the game to victory
convincingly.
44. c2 c6 45. c3 b5 46. d4 f6 47.e5 f5 48.gxf5 exf5 49.e6
g4 50.e7 d6 51. d5 e8 52.f4! g3 53. e1 g7 54. g2 b6
55. d6 b5 56. e3!
441
A very serious mistake was 56. d7?? c4 57. e3+ b3 58. xf5
g2, although even here after 59. d4+ c4 60. f3 White retains
winning chances. Black resigned.
Alireza Firouzja (2723)
Levon Aronian (2773)
Internet 2020

We are witnessing a complex middlegame without queens. Which

plan should White choose? My engine offers a choice between
bland exchanges along the e-file associated with the moves 23. fe1
or 23. ae1. But this is almost equivalent to offering a draw. After
the exchange of rooks along the e-file, the position will be
emasculated.
Is there anything more interesting? Associative thinking suggested
to me one interesting method of playing the most drawish variation
of the Petroff Defence, used by the talented young grandmaster
Vladimir Onischuk.
Petroff Defence
Vladimir Onischuk (2623)
442
Wang Yue (2689)
Riyadh blitz 2017
1.e4 e5 2. f3 f6 3. xe5 d6 4. f3 xe4 5. e2 e7 6.d3 f6
7. g5 xe2+ 8. xe2 e7 9. c3 c6 10.0-0-0 d5 11. de1 0-0
12. d1 d8

13.h3 bd7 14.g4 c5 15. d2 c7 16. h4 d7 17.f4 fe8

18. ef1 h8

443
The regrouping carried out by the Ukrainian GM looks very logical.
White refrains from the exchange of rooks on the e-file, conceding it
to his opponent. But all of the entry squares are controlled by the
white bishops, whilst at the same time, the white rooks excellently
support a pawn storm on the enemy king.
Vladimir Onischuk played
19. e2
but the continuation suggested by GM Alexey Kuzmin was even
more energetic: 19.g5! g8 20. h5 g6 21. g4 with a serious
initiative for White.
Of course, a good reply was soon found to Onischuk’s plan. For
example, in the Russian youth championship of 2020, Artem Tjurin
won the following game.
Petroff Defence
Anton Kirillov
Artem Tjurin
Russian Junior Championship, Loo 2020

444
1.e4 e5 2. f3 f6 3. xe5 d6 4. f3 xe4 5. e2 e7 6.d3 f6
7. g5 bd7 8. c3 xe2+ 9. xe2 h6 10. h4
On 10. d2, the simplest is 10... c5 followed by ...c7-c6.
10...g5! 11. g3 b6 12.0-0-0 g7 13. de1 d8
It was also possible to ignore the discovered check with 13... d7.
14.h4 g4 15. d2 h5

The development of the dark-squared bishop at g7, together with the

seizure of space on the kingside, has allowed Black to solve his
opening problems successfully. After completing his development,
he began an attack on the enemy king with his a- and b-pawns. He
won convincingly.
Let us return to the game Firouzja-Aronian
23. fb1!
White gives the opponent the e-file, on which he controls the
invasion squares in his camp, by analogy with the game Onischuk-
Wang Yue. Next, he prepares the knight retreat, the advance of the
g2-pawn and the knight’s return to the g3-square. At the same time,

445
he protects the important d3 from exchange, preventing the move
... c4-b2. All this is very similar to Onischuk’s plan in the Petroff
Defence.
What should Black do? Unlike Aronian, who was playing rapid, we
have time to think about this. Let’s ask ourselves the question:
‘Which enemy piece is most unpleasant for us?’ In general, all of
White’s minor pieces are quite active, but the light-squared bishop is
perhaps endowed with the most useful functions. It not only looks
lustfully at the enemy king, but also cements White’s queenside.
What should you do with enemy number one? That’s right:
exchange it off. In general, when the opponent has the advantage of
two bishops, it is always a good idea to think about trading one of
them.
Let us look at 23... e8!?. Then there could follow 24. e2 g6
25.g4 ae8 26. g3 xd3 27.cxd3 a5.

analysis diagram
With the exchange of light-squared bishops, White’s attacking
potential is reduced to zero. His position is a fraction more pleasant,
446
but a draw is the most likely outcome. When playing rapid, the
simplification would be a considerable plus for Black and make his
subsequent decisions relatively easy.
Let us return to events in the game. Aronian preferred the move
23...g6 .
There followed:
24. e2 ae8 25.g4 6e7 26. g3 e6 27. h1

White is ready for the attack. What should Black do? Nothing! He
should just keep calm and wait. But it’s easy to say this when you’re
sitting calmly in front of a computer, enjoying the comfort of home.
In tournament conditions against a really strong opponent, with
extremely limited thinking time, things look different.
27... f7?
The wrong square! He should have retreated the bishop to c8,
leaving the f7-square free for the knight or rook in the event of
28.g5.

447
For example, 27... c8! 28.g5?! ( 28.a5 ; 28. hg1 ) 28...fxg5
29. xg5 e6 ( 29... f7!? ) 30.h5 f7. Now, however, White’s
attack develops without hindrance.
28.g5! fxg5 29. xg5 d7 30.h5! b2 31. e2 e6?! 32.hxg6 hxg6
33.a5 b5 34. h6

34... g7?
This move loses quickly. The computer recommends 34... f7 and
suggests the following beautiful variation: 35. xg6+ h7 36. xe6
xe6 37. c1! c4 38. d3+ g8 39. f5 with an extra pawn for
White.
35. ah1 g8 36. c1! bc4 37. d3
The light-squared bishop returns to d3, to crown the attack.
37... e6 38. g1! f7 39. xg6 c7 40. h5+ f8 41. h8
Black resigned.
Levon Aronian (2772)
Santosh Gujrathi Vidit (2727)
Berlin 2022

448
Black’s position is worse. The mutually isolated pawns in the centre
limit the bishop on b7, and Black faces the unpleasant threat of 36.
f5 on the kingside. Considering that Black had six moves left before
the time control, it is very likely that the Indian grandmaster did not
have time to think for long. And he was faced with a difficult
exchanging problem.
Black could mechanically prevent the movement of the white f-
pawn by 35...f5, but then it was necessary to evaluate the bishop
endgame with a pawn minus after 36.g4 fxg4 37.fxg4 hxg4 38. h7+
f6 39. xe7 xe7 40. xg6. In this case, Black would lose the g4-
pawn, but there would be very few pawns left on the board.
He can also play 40...a5!. After 41. f5 axb4 42.axb4 a6 it is not
clear how White can play for a win.
Another path for Black was to give up the pawn on h5 and seek
counterplay by activating the rook:
35... c7?! 36.f5 g5 37. xh5 g7
This is what Vidit played, but the assessment of the resulting
position turns out to be in White’s favour.

449
There was also a third way: 35... c8!? 36.g4 ( 36. c1 f5! 37. xf5
gxf5, and the rook ending, like the pawn ending, is probably
defensible) 36... c7!? 37.gxh5 f5, and in order to understand how
great White’s winning chances are, we need to perform a serious
analysis of this position:

analysis diagram
So why did Vidit reject the natural move 35...f5!, which leads to a
relatively simple position in a drawn bishop ending? I would
venture to suggest that psychological factors played a considerable
role. In the last round of the prestigious Wijk aan Zee tournament,
which had finished a few days before the event in Berlin, he had lost
an endgame a pawn down with roughly the same material.
Sergey Karjakin (2743)
Santosh Gujrathi Vidit (2727)
Wijk aan Zee 2022

450
The Russian GM took play into a winning bishop ending with:
44. f5! xf5 45.gxf5 e7 46.c5 f6 47. d3 e5 48. e4 c8
49.c6 h5 50.c7 a5 51.a4 f6

52. f2!

451
The final part of the plan involves the transfer of the bishop to h3.
52...h4 53. g2 e5 54. h2 f6 55. d3 e5 56. f1 f6 57. h3
d7 58. g4
And now White’s king heads off on a long journey to b8.
58... c8 59. g2 d7 60. f2 c8 61. e3 d7 62. d4 c8
63. c5 f7 64. b6 1-0
The position which could have arisen in Aronian-Vidit after the
moves 35...f5! 36.g4 fxg4 37.fxg4 hxg4 38. h7+ f6 39. xe7
xe7 40. xg6 a5! is drawn:

analysis diagram
What does this position have in common with the trend of the
Karjakin-Vidit game? Nothing concrete. The only thing they have in
common is that in both cases the Indian grandmaster would be faced
with a bishop endgame with White having an extra pawn.
Remembering the negative experience of the earlier game, Vidit
refrained from the correct decision to go into a drawn bishop

452
endgame a pawn down. We see that even with a strong grandmaster,
emotions may prevail over logic.
Another explanation is possible. With a chronic lack of time to
think, the defending side understands that it is almost impossible to
hold a position only with the help of passive defence. Therefore,
strong chess players instinctively strive for dynamics in order to
create problems for their opponents in conditions of mutual time
pressure.
But in the game against Aronian, only Black had problems.

Black has sacrificed a pawn, to seize the c-file and, apparently,

obtain counterplay.
38. h1!
This manoeuvre may have come as a surprise to his opponent. When
calculating variations, retreating moves are often overlooked.
The retreat of the rook with the intention of bringing it to the
queenside destroys all Black’s illusions. It reminded me of a
fragment from a classic game.
Bobby Fischer (2780)
453
Samuel Reshevsky (2430)
US Championship, New York 1962

White threatens to give the opponent a weakness on the queenside

with 26.b5. Reshevsky prevented this in a mechanical way and
played
25...b5
himself. In reply, Fischer immediately transferred the rook, so
splendidly placed on the strong outpost at d5, to the queenside, with
the manoeuvre d5-d1-a1.
Now we return to the game Aronian-Vidit after 38.Rh1!. White has
a healthy extra pawn and the game goes to the technical stage.
38... c3 39. e2 c6
It is bad to take the pawn with 39... xa3 because of 40. c1.
40. a1?! d7?!
Mutual mistakes on the final move of the time control. For White, it
was more precise to fix the backward pawn with 40.b5! and only
then defend the a3-pawn with his rook. And Black, instead of
454
40... d7?!, should have first played 40...a5, and then retreated the
bishop.
41. d2 c8 42.b5 h8 43.g4

White has seized space, restricting the enemy bishop with pawns on
the light squares. Now it is time for the advance of the a3-pawn.
43... c8 44.a4 f8 45.a5 e7 46.axb6 axb6 47. a7 b8 48. e3
d6 49.f4 e8+ 50. f3 gxf4 51. xf4

455
51... xb5!?
Beautiful! This blow is easily overlooked. But no miracle happened.
The white king can escape the attentions of the enemy rook.
52. xb5 e4+ 53. f3 f4+ 54. e3 e4+ 55. f2 f4+ 56. e1
xd4 57. a4 xa4 58. xa4 e5 59. d2 f4 60. d1 d4 61. d3
e5 62. c4 e4 63. c2+ e3 64. b1 1-0
Daniil Dubov (2720)
Nils Grandelius (2672)
Wijk aan Zee 2022

456
White faces a complicated choice. Should he go into an inferior
rook ending with 57. xa6 or muddy the waters by means of
57. c2?! a1 + 58. xa1 xa1 59.f5 ... ?
Most likely, Daniil Dubov made a mistake in solving the problem of
exchanges because he was in poor form at the tournament. Later it
transpired that he was suffering from Covid and had to withdraw.
After 57. xa6! xa1 58. b2 f1 59. xa2 xf4 60. b3 xg4
61. xe6 h4 62. c3 xh5 63. d4 g5 64. f6 g7 65. xd5 g4,

457
 analysis diagram
White draws easily with 66. e4!, although he also has other
drawing lines. But Dubov chose the second line and ended up losing
a piece.

458
59...exf5 60.gxf5 c1+ 61. d2 f1 62.f6 c5!
Perhaps from afar White had reckoned only with 62...gxf6?
63. a7+! g8 64. xa6 fxe5 65. xh6 (or 65. a5 f7 66. xd5 f5
67. d6 xh5 68. e3 ) with a draw.
63. a7 g8
White would also stand badly after 63... g1 64.f7 ( 64. c3 g4 )
64... f1 65. e7 d4! 66.e6 f2+ 67. e1 f5 68. e8 xe6 69. xe6
xf7.
64. xg7+ f8 65. c7 e6 66. b7 g5 67. e3
There was no sense in 67. b8+ f7 68. b7+ e6 69. e7+ f5.

Black’s position looks very good. White’s pawns are blocked, and it
seems that Black’s main task will be not to let the enemy king into
his camp. If he comes to the aid of the pawns, say, on the g6- or d6-
square, then White’s activity will increase sharply.
Grandelius made a seemingly logical move:
67... f5?

459
This throws away the win. The black rook has taken away the
important square f5 from his own king, which allowed him to hide
from the horizontal checks of the white rook. Black should have
played concretely and not by general considerations. To do this, it
was necessary to calculate the forced variation, in which the white
king was driven onto g6. After 67... e1+ 68. f4 ( 68. d4 f3+ )
68... e6+ 69. f5 d4+ 70. g6 c6!!, it is time for White to
resign.
As we have already pointed out, retreating moves by any piece are
often overlooked in the calculation of variations, and especially
knight moves!
68. b8+ f7 69. b7+ f8
69... e6 70. e7#.
70. b8+ f7 71. b7+ e8 72. e7+ f8 73. d4 f7 74. xd5
xh5 75. e4 g5

This is what playing by general considerations has led to. The most
Black can achieve is transpose into an endgame with rook and
knight against rook.
460
76. f4?
A reflex move made without calculating variations. It was possible
to move the king to d4, but the most precise was the concrete
continuation 76.e6 d6+ 77. d3 g3+ 78. c2. Intuitively one
does not want to move the king away from his pawns, but here this
leads to a draw after a line such as 78... e3 79. h7. Dubov’s move
loses because of a far-from-obvious knight move.
76... h8!!
The Swedish player does not miss a second chance to win the game
by retreating the knight into the corner. The white king turns out to
be in the wrong place at the wrong time. The rest is elementary.
77. e4 g6 78. h7 xe5+ 79. d4 h5 80. g7 g5 81. h7 e5
82. g7 f7 0-1
At the end of the last century, when, under a time control of 2½
hours, games were adjourned after move 40, players used the
‘method of elimination’ to defend inferior positions. How did this
work? Let’s assume that the opponent has created a serious threat.
You have two defensive moves at your disposal. You begin
calculating the variations for the first possible move. You find that it
leads to defeat. Then, to save time, you make the second move
without calculating the variations. It can’t get any worse. In his book
Recognising Your Opponent’s Resources , Mark Dvoretsky has a
large section on this topic.
But today this method of defence hardly works. The reason is banal.
With today’s time controls, the player does not have enough time to
correctly calculate the first variation. Most often, he reaches some
position that at first glance looks lost, but is actually defensible. But
he has already stopped calculating and without thinking he chooses
the second option – and loses. This results in a ‘ reverse elimination
method ’, which is presented in more detail in the work Endgame
Strategy !

461
I would like to conclude our chapter on the problem of exchanges
with a modern example on this topic. In theory the position in the
following diagram should be a draw.
Anish Giri (2776)
Maxime Vachier-Lagrave (2758)
Bucharest 2021

When defending an opposite-coloured bishops ending a pawn down,

the weaker side must be guided, first of all, by the ‘schematic
thinking’ principle. You first need to determine the scheme of the
opponent’s winning plan, and then think about how to counteract it.
In general, this is what you should do when defending inferior
positions with any other material balance.
However, it is in the ‘opp B’s’ ending that drawing chances tend to
be greatest. In endings with a different material balance, the picture
is often more blurred.
So, on the one hand, it is relatively easy to find a drawing scheme in
opposite-coloured bishop endings. On the other hand, the cost of an
error in thinking in schemes increases. If you have assessed the

462
position incorrectly, then it is no longer possible to correct anything.
You will reduce a won position to a draw, or a drawn position to a
loss.
What winning plan does White have? He can play f2-f4, force the
exchange of the g5-pawn on the f4-square and try to keep his pawn
on the f-file. Black must not be allowed to exchange the f4-pawn for
the two black e- and f-pawns. In this case, the black king will
occupy h8, and the bishop will eliminate the white passed pawn on
the a7-square.
This means that Black’s task is to retain the g5-pawn at all costs or
to exchange it for the e3-pawn. This goal is met by the move 42...f6!
. After the answer 43.f4 there is the resource 43... c1! 44.fxg5 (on
44. f3, both 44... a3 and 44... d6 are possible, and White will not
make progress) 44...fxg5 45. xg5 xe3+ 46. g6 f8 47. h7
(otherwise, the black king will come to h8).

analysis diagram
Black’s defensive plan is quite simple – 47... f7!. Black puts his
king on f7 and then moves the bishop to and fro on the a7-g1
463
diagonal. He awaits the appearance of the white pawn on h5,
whereupon he plays ... f6.

analysis diagram
After the moves h5-h6 and f6-g5, the position is drawn.
Instead, Vachier-Lagrave defended the g5-pawn with his king:
42... f6?
I do not know what the French GM intended after 43.f4!?. In my
opinion, the position after 43...gxf4 44.a6 c5 45.exf4 ...

464
 analysis diagram
is winning for White. I do not see how Black can exchange his two
pawns for the f4-pawn.
To better understand the battle in endings with opposite-coloured
bishops, let us look at some variations from the analysis of this
interesting endgame:
A) 45... g7 ( 45...e5 46.f5 ) 46.h4 h6 ( 46...f5+ 47. xf5 )
47. b5 f5+ 48. f3 g6 49.h5+ xh5 50. d7 e5 51.fxe5 g5
52. b5 g1 53. d3 c5 54.e6.

465
 analysis diagram
B) 45... a7 46.h4 g7 47. b5.

analysis diagram

466
The white bishop on b5 is very well placed. On the one hand, he has
the opportunity to get to d7 or e8 in order to prevent the opponent’s
counterplay with the scheme ...f7-f5, ... f6, ...e6-e5. By attacking
the opponent’s pawns standing on light squares, White tries to move
them to dark squares in order to blockade them. On the other hand,
the bishop can always return to d3 or e2 to protect his pawns.
If we imagine such a position, then White’s winning plan will look
like this:

sample position
1. g4 b6 2. d5 a7 3. e6! e4 4.h6 xh6 5. xf6 e3 6. e2 c5
7. f7 g5 8.f6 f5 9. d3+ e5 10. g7, and one of the white
pawns promotes.
After 47. b5 g6 48. f3 b8 49. d7 h5 50. e8 g6 51. e4
g7 52.h5 f6 53. a4 a7 54. b3 e5 55.f5, we obtain a position we
have just examined. The situation is quite complex and subtle.
Without careful analysis, assessing the position would be
problematic.

467
In the game, the French grandmaster played 42... f6?. It is quite
likely that, for general reasons, Vachier-Lagrave thought that he
would somehow be able to exchange his two pawns for the f4-pawn.
43.a6 c5

Now he got an unpleasant surprise:

44. b5!
It turns out that Black has no defence against the move 45. e8!,
followed by the bishop sacrifice on f7. The game ended quickly:
44... a7 45. e8! e7 46. xf7! xf7 47. xg5 g7 48.h4 e5
49. f5 b8 50. e4 1-0
This example splendidly demonstrates that careless thinking in
schemes when defending opposite-coloured bishop endings can
quickly lead one to defeat.
We saw that Vachier-Lagrave refrained from the move 42...f6! in
favour of 42... f6?. He probably calculated the variation 42...f6!
43.f4 c1! 44.fxg5 fxg5 45. xg5 xe3+ 46. g6 f8 47. h7

468
 analysis diagram
and reasoned something like this: ‘White will bring his outside
pawns to a6 and h6, with his bishop he will take control of the g8-
square along the a2-g8 diagonal and win after h7-g6.’ He did not
study the position further and played 42... f6?. But with today’s
tight time controls, the method of elimination most often works the
other way around.
Let’s look at an interesting example from the meeting between the
leaders of the women’s teams of Azerbaijan and Georgia in the last
round of the 2022 Olympiad. This game was of great sporting
importance. Both teams were competing for medals. The match was
very tense. If they won, the Azeri girls would win the Olympics, a
draw would bring them the silver medals, and a defeat would
relegate them to seventh place.
Gunay Mammadzada (2470)
Nana Dzagnidze (2531)
Chennai Olympiad Women 2022

469
In the Catalan Opening, White managed to outplay her opponent,
and in the position of the diagram she is winning a pawn.
24. xa6?!
From a human perspective, Gunay’s decision is easy to understand.
In such a tense situation, you always want to stock up on material,
but you still need to be able to realize the extra pawn.
Black responded with
24... d5
and established interaction between her pieces, which immediately
became active.
We will dwell on the subject of realizing an advantage in detail in
the next chapter, but for now, we will simply note: the most
important element in this process is limiting the opponent’s
possibilities.
When realizing your advantage, you must constantly ask yourself
logical questions and find the right answers to them. One of the
main questions is: ‘Which of the opponent’s pieces annoys me the
most? Or – who is my number 1 enemy?’
470
If Gunay had asked herself this question, she would certainly have
found the answer. Of course, it is the black rook on b4. After its
exchange with 24. xb4 xb4 and the move 25. a1, White would
have won the pawn in a much better version. It’s bad for Black to
defend it with 25... a8 due to 26. c6! f8 27. b6 with complete
dominance by the white knights.
And in case of 25... d5 26. xa6 f6 27. b6 xb6 28. xb6,

analysis diagram
White has the reply 29. c6 after any bishop retreat.
In the game, after 24... d5, White could have retained pressure with
accurate play: 25. c1! f8 26. c6, but it looks very frightening to
walk into the pin on the c-file.
25.e4?! e7 26. d6 xd6 27. axd6 f8 28. d8 g6

471
And the game ended in a draw:
29. g4 xd8 30. xd8+ g7 31. d7 c6 32. c7 h5 33. xc6 hxg4
34.e5 xb2 35. c4 e2 ½-½

472
Chapter 9
Associative thinking
in superior positions
In this chapter, we will look at how associative thinking can help
you outplay your opponent and turn a better position into a winning
one. We will learn the technique of realizing an advantage in the
games of World Champions Kramnik, Karpov, Petrosian and
Smyslov
Let’s turn to the work of the ninth World Champion Tigran
Petrosian.
Tigran Petrosian (2605)
Boris Ivkov (2480)
Bugojno 1982

473
19. b1!
When my students and I analyse their games and such a structure
appears on the board, I ask them to define the essence of the
position in two words. Answer: the knight on c4!
To carry out a knight manoeuvre along such a route was as simple
for Petrosian as solving an example from the multiplication table for
a maths teacher. At the same time, the question of the safety of the
white king after the exchange of rooks on the c-file is removed.
Black’s position begins to deteriorate rapidly.
19... xc1 20. xc1 c8 21. a3 xc1 22. xc1 h5 23. c4 c8
24. d2!
With a space advantage, the stronger side will only agree to
exchanges if they are extremely favourable to him. The need for the
move h2-h4 has disappeared. Now, after 24... h6 White replies
25. b4.
24... f6 25. a5 b6 26. c6 g7 27. e3 f7 28. d3!

474
White only needs to force the advance of the enemy pawn to a5 and
then a path will open up for the white king towards the opponent’s
queenside pawn weaknesses.
28... e8 29.a4 d8 30. db4 a5 31. a2 f4 32. f2 g5 33. d3 f8
34.h3 g6 35. c3 f6 36. c4 d7 37. b5 c7 38. a6 g4
39.hxg4 hxg4 40.fxg4 h4

The last interesting moment in this game. White’s position is

completely won. A simple calculation of the variations shows that
after 41. xb6+ xb6 42. b5+ d7 43. xb6 g5 44. b8+ c8
45. a7, Black’s counterplay associated with the attack on the g2-
pawn is clearly too late. But why calculate variations if you can,
‘Gelfand-fashion’, minimize the number of difficult decisions with
41. d1!
...?
After all, if Black retreats with 41... f6, then White will adjourn the
game, sealing the move 42. c3. After home analysis, in response to
42... h4 he will choose between the moves 43. xb6+, 43. b5+,

475
43. d1 and 43. g1. All of them win for White, so Black is forced
to exchange bishops on f2.
41... xf2 42. xf2 ce7
42... h4 43.g5.
43. xe7 xe7 44.g5 g6 45. g4 h4 46. a7 g6 47. h2 h4
48. f3 g6 49.b3 f8
And without waiting for 50. h4, Black resigned.
Nowadays there are no adjournments, but the device of d1
remains relevant in our day. After all, it presents the opponent with
a choice and hence, also the chance to go wrong! We will soon see a
similar example in the game Petrosian-Gufeld, Moscow 1961.
Why was this knight on c4 so strong that it determined the essence
of the position? Because it was correctly positioned one square
behind and to the side of the key white pawn on d5, which had
wedged itself into the opponent’s camp.
I propose to consider in more detail the situations associated with
the knight on c4. We will start with a classic game.
English Opening
Akiba Rubinstein
Oldrich Duras
Carlsbad 1911
1.c4 e5 2. c3 f6 3.g3 b4 4. g2 0-0 5. f3 e8 6.0-0 c6
7. d5 f8 8.d3 h6 9.b3 d6 10. b2 xd5 11.cxd5 e7 12.e4 c5
13.dxc6 xc6 14.d4 g4 15.d5

476
15... e7?
In his book Positional Decision Making in Chess , Boris Gelfand
pointed out that in order to avoid getting into a strategically lost
position, Black had to sacrifice a pawn with 15... d4. The game
could then develop like this: 16. xd4 exd4 17. xd4 g6 18. ac1.
Black needs to exchange the light-squared bishop for the knight and
strive to exchange both rooks on the c-file. In this case, he would
have a great chance of a draw.
How could Oldrich Duras know more than a hundred years ago that
in this King’s Indian/Spanish pawn structure, White’s spatial
advantage and the unfortunate position of the e8, f8 and e7
give White an almost decisive positional advantage? In those distant
times, players had only a vague idea about the King’s Indian
Defence, and no one had studied the pawn structure of this opening.
In the game Petrosian-Ivkov, we succinctly formulated White’s task
– he needed to move his knight to c4. Akiba Rubinstein goes for this
and gets rid of the knight pin:
16. d3!
477
Black responds by seeking to exchange light-squared bishops:
16... d7?!
As Gelfand rightly noted, Black’s queen’s bishop was needed to
protect the queenside and should have been moved to the rear –
16... d7, in order to be able to respond to the move 17. d2 with
17...b5.
I am often asked about the advisability of the exchange 16... xf3. I
should note that in the King’s Indian Defence, trading Black’s light-
squared bishop for a knight is almost always bad for Black. And our
example is no exception to the rule.
There is only one open file on the board. To capture it, the white
bishop needs to take control of the square c8. After a line such as
17. xf3 d7 18.h4! ac8 19. e2 a6 20. h2, Black’s position
would become strategically lost.
17. d2 h3 18.a4 xg2 19. xg2 eb8 20. c4 b5 21.axb5 xb5
22. a3 g6 23. fa1 a6

478
The white knight has safely occupied the most important point c4,
and Black’s position is strategically very bad. Let’s try to simulate
Rubinstein’s train of thought from this moment:
‘We will start asking ourselves questions. What does the opponent
want to do if he were to move? Perhaps he wants to give the game a
concrete character and play 24... f5. This idea looks very risky,
because Black could immediately get a very bad position, although
perhaps not so much worse than in the game. It’s clear that Black
stands badly, but his opponent isn’t quite there yet. Knowing
Oldrich, I am sure that he will prefer passive defence.
Next question. Which of my pieces is not working? The bishop on
b2. Where should it go? To e3. So, let’s check the move 24. c1.
Let’s say he still decides to play 24...f5 25.exf5 e7. Here I can
play at least 26. a5 xb3 27. xb3 xb3 28. xd6. I will somehow
use the extra pawn in the endgame against Duras. There are no other
active opportunities for him. So, let’s play:
24. c1 b7 25. e3 f6
Why was this move made? What does the opponent want, if it were
his move? Most likely, he is simply strengthening his position, and
is going to wait and react only to my active operations.
How can I take this fragile fortress? On what, or rather, on whom,
does Black’s defence rest? Who is my number one enemy? Of
course, the queen! What should you do with the enemy? Exchange it
off! How do we exchange it and still protect the b3-pawn? We can,
of course, bring the king to e2, but that seems a bit exotic. This
means that it is necessary to move the queen to f1, and then make
the move c4-d2!.
But if I immediately move the queen to f1, then he will figure out
my plan. I will actually force him into complications after 26. f1
f5. Do I need to rush? Maybe I can strengthen the e4-pawn first? I
guess that’s what I’ll do.
26.f3
479
Yes, I am weakening the second rank but it is not obvious how he
can exploit this.
26... e7 27. f1! c8 28. d2!
The plan has been completed. The queens are exchanged, after
which the a6-pawn is lost. The realization of the extra pawn begins.’
We have tried to model the train of thought of Akiba Rubinstein, the
brightest representative of the classical school of Steinitz. A
hundred years ago, everything was different. But the questions we
have listed came to the mind of the white player, at least intuitively,
and the plan was brought to life with iron logic and crystal purity.
What’s good about classics? In those distant times, the plans of the
strongest players did not meet worthy resistance. This circumstance
allows us to see the great chess player’s plan in its pure form,
unclouded by time pressure and other mistakes. After all, he was
many times stronger than his opponent, and the latter only began to
understand what was going on when it was too late to prevent it.
Please note that Black’s position collapsed after White, the side with
the space advantage, forced him to exchange queens. We will look
at this factor of superiority in space a little later.

480
28... b4
28...a5 is no help because of 29. xb5 xb5 30. c4.
29. c4! xc4 30. xc4!
White’s extra pawn should be passed on the b-file.
30... ab8 31. d2 c7 32. xa6 c2 33. 6a2 xa2 34. xa2 e7
35. f2 f7 36. e2 e8 37. d3 d7 38. c3 d8 39. c4 c7

481
White’s position is absolutely won. He could go for the scheme:
a6, b4, b3, a4, b5, sweeping the opponent off the board on the
queenside. In addition to the extra pawn, White has an
overwhelming advantage in space, which can be considered the
second weakness in Black’s position. But Rubinstein made the
move:
40.g4 ,
creating a classic second weakness for the opponent – the g7-pawn,
far removed from the extra b3-pawn.
I once analysed this game while working with the talented Azeri
grandmaster Rauf Mamedov. When we reached the move 40.g4,
Rauf surprised me with a witty question: ‘Do you think Rubinstein
was a sadist?’ I hesitated a little and objected that all great endgame
players are, to some extent, sadistic in realizing their advantage,
preventing their opponent from freeing himself from the grip.
Perhaps this is not a very attractive figurative comparison, but it is
very memorable. We will talk separately about the technique of

482
realizing advantages. For now, let’s finish watching Rubinstein’s
game.
40... d8 41. a6 c7 42.h4 d8 43.h5 c7 44.b4 b7 45. a8 d8
46. b3 b8 47. xb8 xb8 48.b5 e7 49.b6 f5 50.gxf5 g8
51. f2! c8 52. h4
Black resigned.
Now for another example of a knight on c4.
Peter Leko (2732)
Fabiano Caruana (2782)
Wijk aan Zee 2013

A typical position from the Chigorin Variation of the Spanish.

17.d5!
Leko closes the centre and obtains a space advantage.
17... c7
Black is thinking about the move 18...c4. Now how should White
play? We can say that a critical moment in the game has arisen. The
further course of the game depends on what White plays now.
483
18. c4!
Not an easy decision at all. Firstly, one does not want to exchange
off an enemy knight that is poorly positioned on the edge of the
board. Secondly, if the stronger side has an advantage in space, it is
better not to exchange minor pieces. But the Hungarian grandmaster
delved deeply into the position and subtly noticed that there would
be time to occupy the c4-square with the other knight. And this
circumstance will provide him with a great, and perhaps even
decisive, positional advantage.
Leko rejected the natural move 18. e2 due to the sharp answer
18... h5!. The play could then have continued with 19. xa6 f4
20. f1 c4! 21. xc4 xc4 22. xc4 ec8! 23. xc7 xc7, winning
back one pawn and exerting pressure on the queenside.

analysis diagram
In this way, the character of the game would have changed radically
and this would not have suited White. White’s advantage in this line
is only minimal, so from a practical viewpoint Leko’s decision was
perfectly correct.
484
18... xc4 19. xc4 eb8 20. d2 e8 21. d3 d7 22.a5! f6
23. c4 b5
If Black manages to exchange the white knight, then he will have
hopes of a successful defence, but Leko had calculated from afar a
few more moves ahead.
24. b6! a7 25. e2
It turns out that Black cannot play 25... d7 because of 26.c4.
25... xd3 26. xd3 d7 27. c4

This is the position for which the Hungarian GM was striving when
he played 18. c4. We can see fully the power of the knight on c4.
Together with the pawn on d5, it hangs over the enemy centre and
queenside, exerting pressure.
27... ab7 28. d2
White has no need to hurry. The position is one of those where one
side plays and the other spectates.
28... b5 29. f1 e7 30.f4 f6 31.g3 8b7 32. g2 f8 33. f2
g6 34. af1 e7
485
What should White do now? It is tempting to set up a pawn bind on
the kingside with 35.f5. In this case, Black transfers the knight to h7
and evacuates his king to the centre or queenside. Leko finds a far-
from-obvious idea to strengthen his position before advances his
pawns.
35. h2!?
Leko called this the first move of the ‘finishing plan’. More accurate
was probably 35.f5, intending h3-h4 is Black plays his knight to h7.
Now Black could have equalized with the ingenious engine line
35...exf4 36.gxf4 d8 with the plan 37... b8 and ... b1.
35... b8 36. g2 c7 37. fg1!
Doubling rooks behind his g3-pawn! One does not often see such a
thing. White is prepared for the decisive attack.
37... b3 38.f5 f8 39.h4!
White wants to play 40. h3 and then 41.g4.
39... d7

486
What does the opponent want? In reply to 40. h3 he strengthens
things with 40...g5. Therefore:
40.g4!
Now after 40... xh4 there follows 41.g5 hxg5 42. xg5 xg5
43. xg5 g6 44. h3 with the idea of 45. h5. Caruana played
40...g5
but after
41.fxg6 fxg6 42.g5 hxg5 43. xg5 g7 44.h5 8b4 45.hxg6
acknowledged defeat.

Prophylaxis – limiting the opponent’s possibilities

We continue to replenish our baggage of associations with
characteristic playing techniques. Now we will bring to your
attention the ‘hymn of prophylaxis’ – the famous game Petrosian-
Gufeld, Moscow 1961. In this game, I would like to draw your
attention to two subtle points that are usually missed by young chess
players unless they are specifically explained.
King’s Indian Defence
Tigran Petrosian (2580)
Eduard Gufeld (2434)
Soviet Championship, Moscow 1961
1.c4 g6 2.d4 g7 3. c3 f6 4.e4 0-0 5. g5 d6 6. d2 c5 7.d5
a5 8. d3 a6 9. ge2

487
We will not discuss the opening phase. The theory was very
different in the far-off days of 1961.
9...e5?
A positional mistake. Now the game assumes a closed character.
White gets a serious space advantage, whilst Black’s four minor
pieces are very cramped on the first two ranks. A modern player
would opt for 9...b5 either immediately or after 9... bd7.
10.0-0 bd7 11.a3
Petrosian sets about the standard plan of play on the queenside.
11... h5
In this position, White could have played the programmed move
12.b4!?, after which Black cannot take the pawn with 12...cxb4?
13.axb4 xb4?? because of 14. e3!, and his queen would be in
trouble.
Gufeld would have retreated the queen 12... c7 with an unpleasant
position. But instead of playing 12.b4!?, Petrosian preferred to ask
himself what his opponent would do if it were his move. The answer

488
is clear – 12...f5. It was possible to prevent this with 12. c2, of
course, but then what? If the opponent wants to do something
stupid, one should not interfere.
12.f3!
It is as if White is saying ‘Come on then, make my day!’ But after
the ‘attack’ 12...f5? 13.exf5 gxf5 14. c2! Black stands very badly.
12... f6

The first interesting moment, on which we should concentrate our

attention. White has a space advantage, and the exchange of pieces
is usually unfavourable to the stronger side in such circumstances.
But on the contrary, he should strive for the exchange of queens!
We will see this from the next example.
How should White best react to the bishop move 12... f6...? The
reply 13. e3 is tempting, but Petrosian plays differently.
13. h6!?
He gives his opponent the right to choose and hence also the chance
to go wrong.

489
Now, in reply to 13... g7, I am sure that White would have played
14. g5! f6 15. e3, but Gufeld made a mistake and blocked the
attack on his rook with his knight:
13... g7?
Remember this device, involving the repetition of moves to test the
opponent’s actions. This was why Petrosian played 40. d1!? in his
game against Ivkov.
Incidentally, this was a technique used a great deal by Anatoly
Karpov, and, among the current elite, Anish Giri.
What would you play in this position? Naturally, 14.b4. But
Petrosian would not have been a genius of prophylaxis if he had
played so straightforwardly. He presents his opponent with the
chance to play actively in an inferior position, so as to speed up his
defeat.
14.g3!!?
A true Petrosian move.
14... b8 15. h1!
‘Be my guest, Eddie – go for it!’ says Petrosian, provoking his
opponent to play the move 15...b5. Against this, he had prepared
16.cxb5 axb5 17.b4! a6 18. xb5 xb5 19. c3 or 19.a4.
15... c7

490
But at this moment, it is better not to allow 16...b5, since after
17.cxb5 could follow 17...c4 and 18...axb5, seizing space on the
queenside.
So what did Petrosian play here? Looking at White’s preceding
play, guessing his next move should not prove difficult.
16.b3!
White does not give Black any play at all.
At the Kramnik chess school, Vladimir often explains to young
players that in such situations, the opponent tends to get nervous,
think for a long time and play below his usual level.
16... e7 17. ab1 h8
What does the opponent want? Probably to transfer the d7 via f6
to g8. And then, after the retreat of White’s bishop from h6, Black
will seek to start counterplay with the move ...f7-f5.
So what is the drawback of bringing the knight to g8? Yes, it
weakens Black’s control of c5. So that means White should play b3-
b4. But what would be a useful preparation for this move? Indeed:
double rooks on the b-file!
491
18. b2!
Hardly anyone has ever managed to find this move of Petrosian.
Here is the device we saw with Leko – White doubles rooks behind
the b3-pawn .
18... f6 19.b4 g8 20. e3 f5 21.bxc5 dxc5 22. fb1 f6 23. b6
d6
Not 23... d7?? because of 24.d6!, winning.

Black’s pawn structure is compromised and his pieces are scattered

around the board, cooperating poorly, but the battle is not yet over.
After all, only one pair of pawns has been exchanged. Petrosian
again asks his favourite question: what does my opponent want?
It is obvious that Black is prepared to play 24...fxe4 and if the pawn
recaptures then 25... g4, whilst after 25. xe4 he will exchange
knights on e4 and activate the g7 with 27... f5. Petrosian reacts
with:
24. h6!
Now on 24...fxe4 there follows 25. xe4 xe4 26. xe4 f7 27.g4!,
and Black has no play at all.
492
24... f7 25. g1!?
Yet another move which strengthens the white position before the
mutual time-trouble. Now, after 25...fxe4!? White has the choice
between the simple exchange 26. xg7+ xg7 27. xe4 xe4
28. xe4 with a strategically winning position, and the more
complicated variation 26. xe4 f5 27. xd6 xd6 28.g4 with a
large positional advantage.
But Gufeld’s patience is exhausted, and he launches himself into a
desperate counterattack:
25...f4?! 26.gxf4 d7 27.fxe5 xe5
It is clear that things would be bad for Black after 27... xb6
28.exd6, but after the text his position is lost.

28. e6!
Tigran Petrosian was a wonderful tactician, like all the world
champions who took decisions intuitively.
28...b5?!

493
The computer recommends 28... g8, but I doubt that a time-
troubled Gufeld even looked at this move.
29.cxb5 c4
Continuing to throw wood onto the fire.
30. c6 d8 31. xc4 h4 32. c1!
A typical Petrosianesque move in time-trouble.
32... h5 33. g5 g3+ 34. g2 xe4 35. xe4 xh2+ 36. f1
xf3+ 37. xf3 h1+ 38. f2
Here the agony ended, as Gufeld resigned.
In this beautiful game, we observed excellent prophylaxis and drew
attention to a very good technique – the doubling of the white rooks
behind the knight’s pawn on the third rank, only on the queenside,
and not on the kingside, as was the case in the game Leko-Caruana.
In the next game, we will look into the problem of exchanges in
positions with an advantage in space.
To do this, it is important to understand and be able to clearly
explain to yourself what an advantage in space is. Boris Gelfand
covered this topic well in his work Positional Decision-Making in
Chess . I will try to summarize his explanation of spatial superiority.
Let’s assume that the French Defence was played, and the game
developed according to a scenario that was not very successful for
Black:
1.e4 e6 2.d4 d5 3. d2 f6 4.e5 fd7 5. d3 c5 6.c3 c6 7. df3
e7 8. e2 0-0 9.0-0 b6 10. f4 b7 11. e1
My engine confidently gives White a large advantage of +2.3.

494
If we ask ourselves the question: ‘Does White have a space
superiority here?’, the answer will be unequivocal: ‘Yes, of course.’
How did it arise and how is it expressed?
A superficial answer will concern the pawn structure. Yes, indeed,
the e5-pawn is constricting the black position. But the point lies
rather in the different levels of activity of the minor pieces of the
two sides. While the white knights and bishops occupy active
positions and can easily manoeuvre, their black opponents pitifully
huddle on four nondescript squares and are simply cramped on the
board.
Now let’s remove one light piece from the board on each side. First,
let’s remove the white knight on f4 and the black knight on d7.

495
The engine immediately becomes kinder; White’s advantage falls to
+1.5.
Now let us remove the bad bishops on b7 and c1.

Now does White even have a space advantage? The engine gives
him a minimal advantage of +0.5 on account of his more active
496
bishop. With opposite-coloured bishops, there is only an
insignificant chance of developing activity against the black king.
And, finally, let us remove the final pair of bishops from the board.

The position is equal – the engine gives 0.1.

We have just seen that the side with less space finds his position
easier to play as the minor pieces disappear from the board. This
means that the weaker side should strive to exchange minor pieces.
The task of the strongest side is usually the opposite. He must keep
the minor pieces and strive to exchange the heavy ones, especially
the queens ! As Gelfand aptly puts it, the side that has pushed its
pawns forward may have ‘cracks’ in his position through which the
enemy queen can leak into its camp. With the exchange of queens,
the chances of uninvited guests entering will be minimized.
In the game with Ivkov, Petrosian exchanged all the rooks on the c-
file, and Black’s position immediately became lost. And in the game
with Gufeld, he did not allow any of the minor pieces to be
exchanged.
Let’s now look at another example.

497
Benoni Wall
Mario Bertok (2282)
Efim Geller (2455)
Kyiv 1959
1.d4 f6 2.c4 c5 3.d5 e5 4. c3 d6 5.e4 e7 6. ge2 0-0 7. g3
e8 8.h4 a6 9. d3 g6 10.h5 g5 11. d2

White’s last move does not spoil anything, but it does not look very
natural. It was simplest to exchange bishops with 11. xg5 and after
11... xg5 to offer the exchange of queens with 12. d2!.
11... f6 12. c1 xd2+ 13. xd2 f4 14. ge2
One gets the impression that the players are slightly flirting with
each other.
14... xd2+ 15. xd2

498
Finally, the queens have been exchanged. Look at the black pieces –
all of them are on the eighth rank. There is only one positive aspect
to this: if Black wants to stop this dull game and start a new one, it
will not take much effort to set the pieces back up.
But joking aside, it can be argued that Black’s position is already
strategically lost, although my Stockfish at a depth of 40 gives
White just a little more than +1. Black’s main problem is that he has
no active counterplay. As soon as he plays 15...f5, for example, his
position will immediately begin to deteriorate sharply after 16.hxg6
hxg6 17.exf5 gxf5 18.f3! or 18. g3!.
If we try to find at least some plus in the strength of Black’s ‘good’
bishop, then, upon closer examination, this plus immediately turns
into a minus. Black’s bishop has only visual signs of being ‘good.’
It is weaker than its white opponent, as it is limited by the enemy’s
pawns. It has only a single square, d7, at his disposal, but if it
occupies it, the knight on b8 will cry.
White can calmly concentrate his forces and prepare an attack on the
queenside, without fearing any counterplay from the opponent.

499
15... g7 16. g3 d7 17.a3 f6 18.hxg6 fxg6 19.b4 d7 20.f3
b8 21. ab1 e8 22. b2 c7 23. hb1 a8 24. ge2 cxb4
25.axb4 a5 26.bxa5 xa5 27. c1 f7

Up to here, the sides have played well. Of course, certain moves can
be improved, but there is no denying the logic in them. White needs
to find a way to develop his queenside initiative. The Yugoslav
player plays a natural quiet move:
28. c2
Today’s engine recommends a brilliant concrete decision: 28. b3
a7 29.c5! xc5 (or 29...dxc5 30. b5 ) 30. xc5 dxc5 31. c2!!
with the idea of 32. a4. But such a decision is beyond a normal
human, of course.
The game continued:
28... c5 29. b3 xb3+ 30. xb3 e7 31. b6

500
Things have improved markedly for Black. He has managed to
exchange one minor piece and gain control of the c5-square, the
most important point of the entire position. True, he has two
vulnerable pawns on b7 and d6. However, the distance between
them is small, and so they can be considered as only one weakness.
That is, Black’s position remains worse, but quite defensible. Thus,
Black’s main task was to prevent the formation of a second
weakness in his position. This goal would have been answered by
the move 31...h5!, aimed against White’s 33rd move. Instead of this,
Geller, in time-trouble, made the decisive mistake:
31... a6?
Bertok replied
32. b5! d8
32... f6 33. a7.
33.g4!

501
Now a second weakness, on h7, is fixed in the black position. The
tactical solution was equally good: 33. h1 h5 34.g4!, but the fate of
the game is already decided.
33... a2 34.g5 f8 35. c3 a3 36. h1 f7 37. b5 a2 38. xd6
xf3 39. xc8+ d8 40. f6 g3 41. f8+ c7 42. xh7+
Black resigned.
If in those days anyone had told you that a game involving Bertok
and Geller ended in a White victory, then everyone would have
assumed that the outstanding grandmaster Efim Geller had been
White. But it was Bertok who won, and the game was one-sided.
The reason for this amazing result lies not in the class of the players,
but in the properties of the position itself.
If we pass over the move repetition with 11. d2 f6 12. c1 and
play instead the direct 1.d4 f6 2.c4 c5 3.d5 e5 4. c3 d6 5.e4 e7
6. ge2 0-0 7. g3 e8 8.h4 a6 9. d3 g6 10.h5 g5 11. xg5 xg5
12. d2 xd2 13. xd2, we reach a position in which White has a
space advantage and Black has no counterplay. If 13...f5? 14.hxg6
hxg6 15.exf5! (Black was threatening 15...f4 to seize back part of
White’s space advantage) 15... xf5 16. h6! followed by ah1,
when White would have a decisive positional advantage. My engine
assesses the white advantage as +6.
If we draw conclusions about Black’s plans in the pawn structure
shown in this diagram...

502
... then it can be argued that in such positions, White’s main
strategic plan is to play on the queenside, associated with the
movement of the b-pawn to b4. And a middlegame without queens
is most often bad for Black due to the lack of space for his minor
pieces. There are a couple of examples on this topic in the book The
Shereshevsky Method : O’Kelly-Ulvestad, Malaga 1966 and
Gligoric-Quinteros, Manila 1973, but we will not go deeper in this
direction.

Prophylactic techniques
Now we return once again to Tigran Petrosian’s play and analyse a
game of his in a similar structure.
Benoni Wall
Tigran Petrosian (2580)
Herman Pilnik (2126)
Candidates Tournament, Amsterdam 1956
1.d4 f6 2.c4 c5 3.d5 e5 4. c3 d6 5.e4 g6 6. f3 g7 7. g5 a6
8. e2 c7 9. d2 d7
503
Black has a cramped and unpleasant position, but we will not be too
harsh about the opening play of the 1950s.
10.a4
A committal move. Petrosian very much loved space advantages,
but with the misplaced knight on c7 it was much more natural to
place this pawn on a3 and then play b2-b4.
10...b6?
A strange move. Black had the chance to close the queenside with
10...a5!! 11. b5 c8. But there are few joys in such positions.
After, for example, 12. b3 h6 13. e3 xb5 14.cxb5 d7, White
gets a promising attack with g2-g4, 0-0-0 and h2-h4.
11. b5 xb5?
This move is already a serious mistake. Here, 11... xb5 or 11...0-0
were better.
12.cxb5 0-0 13.b4!
An energetic move.
13...h6 14. xf6!?
504
A typical Petrosian decision, based on prophylactic thinking. White
puts pressure on the opponent’s position with the help of leverage,
which is the pawn exchange on c5. If Black himself takes on b4,
then he will have to reckon with the penetration of the white rook
into his camp via the weakened square c6.
Black’s only defensive resource is to move the knight to the c5-
square, covering the c-file from White’s invasion. Such protection
does not guarantee a quiet life, but Petrosian immediately deprives
Pilnik of even this opportunity, although it was also possible not to
hurry with the exchange on f6 and to retreat with the bishop to h4.
14... xf6 15.0-0 fd8 16. c4 f8

A curious position. White has a large positional advantage due to his

advantage in space, better pawn structure and the active placement
of his minor pieces. The difference in the strength of the steed and
‘master of the position’ knight on c4 and the pitiful nag on c7 is
especially noticeable.
What is the difference between good positions and bad ones?
Sometimes it is useful to ask yourself such naive questions. There

505
are, of course, many differences. In my opinion, the main thing is
that in a good position, the stronger side has several tempting moves
that can take the game in different directions. By contrast, in a
worse position, the weaker side often has to make a choice between
a losing continuation and a move that allows him to stay afloat.
Such moves are usually the only ones.
Try to guess how Petrosian played in this position. You will
definitely be able to find the next move if you have carefully studied
his game with Eduard Gufeld.
My engine suggests immediately taking the bull by the horns and
sacrificing a pawn by 17.bxc5 dxc5 18.f4 exf4 19.e5 g5 20.d6
e6 21. f3 ab8 22. d3. This gives White an advantage in the
region of +3. This assessment is undoubtedly close to the truth. But
this is a computer whose playing strength in human rating terms
would be above 3000. And it is programmed to destroy the
opponent’s position in the shortest possible way. The human player,
on the other hand, is afraid of making mistakes and prefers to be in
control. He does not like to abruptly change the nature of the
struggle in positions with a large positional advantage, unless he
immediately obtains a concrete, tangible gain.
Psychological factors also come into play. Why go into a concrete
calculating battle if you are playing, for example, against a talented
young tactician? Or if you don’t have enough time to think, or if you
simply didn’t get enough sleep the night before? And you can come
up with a lot of such ‘ifs’.
Petrosian’s move
17.g3!
clearly demonstrated to the Argentine grandmaster that it was time
for him to determine the position and get rid of his opponent’s
leverage on the queenside. Otherwise, it will be too late. White has
many more position-improving moves like 17.g3!.
17...cxb4 18. b3 g7 19. fc1!
506
Petrosian continually keeps a close eye on his opponent’s
counterplay. It would be a mistake to take the b4-pawn. After
19. xb4? e6! 20.dxe6 d5 21. b3 dxc4 22. xc4 fxe6 23. xe6
d4 24. d5 c8, Black’s affairs improve significantly.
19...h5 20. e3 e8 21. xb4 dc8 22. c6 d8

Black has been outplayed and White has a decisive positional

advantage. As always, Petrosian asks himself what Black wants. Of
course, he wants to bring the knight via d7 to c5.
23. ac1 f6 24. f1!
Now we see the value of the move 17.g3!.
24... cb8 25. h3 a6

507
A famous position. White needs to return the knight to c4, and then
attack the b6-pawn, moving the queen to the e3-square. But with the
knight on c4, the e4-pawn will be left unprotected.
It would seem that there is nothing to think about. Let’s reinforce
the e4-pawn with the f3-pawn, and then carry out the planned
arrangement. But Petrosian rejected the move f2-f3, and I doubt that
he even looked at it. After 26.f3? axb5 27.axb5 h4 with the idea of
28... h5 and 29... g5, clouds could gather over the white king.
Petrosian defended the pawn:
26. e1!
Moving the rook away from the open c-file.
26...axb5 27.axb5 h7 28. c4 a2
The position after Black’s 25th move has made the rounds of many
chess books and periodicals. Indeed, Petrosian’s unobvious move
26. e1! is very impressive. The position after the 28th move, on the
other hand, has received very little attention. No one has dwelt on it
in chess literature – at least, I have not seen it. But it was largely for
its sake that I started analysing this fairly well-known game.
508
Let’s remember that after the game Bertok-Geller we decided to talk
about the techniques that chess players use to implement their
prophylactic ideas. We know that the main prophylactic question is:
‘What does the opponent want, were it his move?’ If the game is not
concrete, then you need to mentally let your opponent make two,
three or four moves in a row. This is necessary in order to
understand the immediate pattern of his actions and, if necessary, to
prevent it.
But quite often, a strong and unpleasant action for an opponent is to
play against his next move. Withdraw a piece or defend a pawn that
has not yet been attacked, but which the opponent wants to attack
with his next move. In short, we play against the opponent’s next
move.
This is how this technique can be described verbally. Now let’s try
to simulate the thought process of the ninth World Champion.
Although, as grandmaster Lev Psakhis wittily noted, the algorithm
for finding a move is very different between genius chess players
and mere mortals.

509
‘So, what does Black want to do next move? He wants to play
roughly according to the scheme 29... g5 30. g2 f6 31. f1 f3+
32. xf3 xf3 and then 33...h4, with the idea 34...h3. Maybe instead
of 31... f3+ Black does even better to advance the pawn to h4.
What should I do?
Answer: transfer the queen to e3 and win the b6-pawn. I can
immediately play 29. b3 but then we have to reckon with 29... ba8
30. xb6 2a3 or 30... a1. This is rather messy, there is little time
left to think – is it possible to avoid all this? What will happen if I
immediately put the bishop on g2?
29. g2!
Let’s check the lines:
29... f6 30. f1 g5 31. b3! ba8
Only here will I ‘delight’ him with the move
32.h4!
The end. The black pieces retreat. I take the b6-pawn and win the
game.’
For a modern computer, the strongest move is still 29. b3, and it is
not at all impressed by Black’s play after 29... ba8. It proves
White’s advantage in the complications that arise.
We recently noted that in superior positions, as opposed to bad ones,
the stronger side often has many good decisions. Of course, you can
dive into complications under time pressure and emerge victorious
from them with honour. But in that case, you are taking a risk. And
as we remember, the goal of the Gelfand advantage-realization
process is to ‘minimize the number of difficult decisions’. The
choice is yours. I only see my task as helping you to verbally
formulate and consolidate in your associative thinking Petrosian’s
method of action: ‘We play against the opponent’s next move!’
It only remains to see the end of this instructive game.

510
32... h7 33. xb6 a1 34. c6 8a2 35. e3 d8 36. xa1 xa1+
37. h2 f6 38.f3 b8 39. b3 d7 40.b6 c5 41. b2 a4
42. b5 a2 43. c7 g5 44. e3!
GM Yuri Razuvaev used to say that ‘With a knight on f5, there is
always a mate!’. Petrosian concludes the game with a direct attack
on the black king.
44...gxh4 45. f5+ g8 46.gxh4 a6 47.b7 a7 48. c8 xb7
49. e8 d7 50. xd6 1-0
Now we continue to examine the ninth World Champion’s use of
prophylaxis.
English Opening
Tigran Petrosian (2640)
Alexander Beliavsky (2450)
Leningrad 1974
1.c4 c5 2.b3
This game was played over 50 years ago. Alexander Beliavsky had
qualified for the USSR Championship final for the first time at 19
511
years of age. Tigran Petrosian selected the modest move 2.b3, not
wishing to enter a theoretical dispute with him.
2... c6 3. b2 f6 4.e3 e6 5. f3 d5 6.cxd5 exd5 7. e2 a6

Black plays the opening competently. He anticipates that White will

play d2-d4. After the exchange of pawns, a position will arise with
the isolated d5-pawn in Black’s position. The most active plan for
Black in this case would be to build a battery: bishop on c7 and
queen on d6. To protect himself from the attack c3-b5, Black
plays his pawn to a6.
The strength of such a battery can be assessed in the next two
examples:
Tarrasch Defence
Tigran Petrosian (2640)
Yuri Balashov (2540)
Leningrad 1974

512
1.c4 f6 2. c3 e6 3.d4 b4 4.e3 c5 5. d3 d5 6. f3 0-0 7.0-0
dxc4 8. xc4 c6 9. d3 cxd4 10.exd4 e7 11. e1 b6 12.a3 b7
13. c2 c8 14. d3

White has played straightforwardly and even artlessly in the

opening. But with White, one can allow oneself some liberties. The
light-squared bishop made four moves to get to c2. Black completed
his development during this time, but he is required to have some
knowledge of standard (by today’s knowledge) techniques. After all,
an attacking battery aimed at the black king cannot be
underestimated.
14... e8?
This natural move can be regarded as Black’s decisive mistake. It
was essential to close the battery line with 14...g6 or to play
14... a5.
15.d5! exd5 16. g5!
How could Yuri Balashov have known about this dagger attack back
in 1974? I also doubt that Tigran Petrosian had prepared this at
home. He never had a reputation as a serious student of the opening
513
and had acted too toothlessly in the initial stage of the game. Most
likely, he found this idea at the board.
16... e4
The only move. 16...g6 loses to 17. xe7!.
17. xe4 dxe4 18. xe4 g6

As a result of the tactical operation, we have a symmetrical position,

but there is not a single defender around the black king. White has
enough time to launch a winning attack.
19. h4! c7
According to my engine, the only way not to lose at once was the
inhuman move 19... d7!!. Here is the variation suggested by the
computer, ending with an assessment of +1.50: 20. b3 f8 21. ed1
f5 22. f6 c7 23. d5 c8 24. g5 h6 25. d3 with a strong attack
for White.
Returning to the game, after the natural 19... c7 Petrosian
demonstrated a forced win:
20. b3!
514
Threatening 21. xf7+.
20...h5
If 20... f8 White wins with 21. f4, whilst after 20... d6 there is the
decisive 21. f6.
21. e4! g7

Here White landed the decisive blow:

22. xf7! xf7 23. h6!! d6
On 23... d6 White mates by 24. g5+ f6 25. h7+.
24. c4+ f6 25. ad1 d4 26. xd4+ xd4 27. xd4 c5 28.h4!
Black resigned.
From this brilliant miniature, we were able to see that Petrosian was
not only a genius of prophylaxis, but also a mighty master of attack.
True, he did not attack too often, since he was an excellent defender
and did not like unprepared attacks.
In the next game, the talented Uzbek grandmaster Georgy Agzamov
defeated the eighth world champion Mikhail Tal, using Petrosian

515
Tarrasch Defence
Mikhail Tal (2605)
Georgy Agzamov (2570)
Sochi 1984
1.c4 c5 2. f3 f6 3. c3 e6 4.e3 c6 5.d4 d5 6.cxd5 exd5 7. e2
cxd4 8. xd4 d6 9.0-0 0-0 10. f3 a6 11.b3 e8 12. b2 c7
13. d3 d6

Black has emerged from the opening with an active attacking

position and White needs to play very carefully.
14. fd1?
The losing move. It was essential to play 14.g3.
14...d4! 15.exd4 g4
In terms of ideas, this and the Petrosian-Balashov games are simply
copies.
16.g3 b4 17. a3
The move 17. d2 is also hopeless because of 17... xe2! 18. xe2
e8.
516
17... xd3 18. xd6 xd6 19. xd3 b4
The white pieces are overloaded with protecting each other. After
20. f1 f5 Black won an exchange, which he successfully
converted.
We were able to see how dangerous a bishop-queen battery can be
for a side playing against an isolated pawn.
Now let’s return to the Petrosian-Beliavsky

8.d4 cxd4 9. xd4 b4+ 10. c3 d6 11. d2 0-0 12.0-0 c7

517
Black has played the opening well and he needs only to establish the
battery towards the white king with 13... d6. Petrosian understands
that in playing against the IQP his chances of success are not great
and so he changes the structure:
13. xc6 bxc6
Black has a pair of hanging pawns on c6 and d5. If he manages to
advance the pawn from c6 to c5, then his opportunities for active
counterplay, as a rule, increase. The black pawns on c5 and d5 can
only be supported by pieces. White’s task is to prevent the black c-
pawn from advancing to c5.
14. c2!
Petrosian finds a way.
14... e8
If 14... d6 15.g3 c5, White has 16. xf6, compromising the black
pawn structure on the kingside.
15. d4! d6 16.g3

518
Let’s now speculate about the strength of the bishops in this position
for both colours. Visually, it can be noted that both pairs of bishops
are good. But if we start to be guided by such a formal sign as the
pawn structure, then the white bishop on e2 can be considered more
valuable than the bishop on d4, and the black bishop on c7 should
be given preference over the bishop on c8. But in the example of the
Bertok-Geller game we have already seen how deceptive these
formal signs can be.
Beliavsky played the move
16... g4?
and thereby committed a serious mistake in solving the problem of
exchanges. The Rumanian GM Mihai Suba once pointed out
shrewdly that ‘bad’ bishops defend ‘good’ pawns. After the
exchange
17. xg4! xg4
it turns out the pawn on c6 can no longer be defended. Instead,
Black should have played 16...a5 , when the game might have

519
proceeded as follows: 17. ac1 d7 18. fd1 h5 19. c3 h4 20. xf6
xf6 21. xf6 gxf6 with dynamic equality.

What scheme is Black going to follow next? The answer is obvious:

18... h6, 19... e6, 20... h3, 21... h6 with an attack. Petrosian, of
course, immediately discovered this, by asking himself his favourite
prophylactic question.
The next question goes something like this: ‘How to prevent Black’s
play?’ The ninth World Champion found a brilliant prophylactic
solution with elements of tactics. He was a great tactician.
18. f3!
We play against the opponent’s next move. We defend the h2-pawn,
which at present is not attacked. Beliavsky naively replied
18... h6?
and ran into a knockout blow:
19. f5!
Technique is the mastery of short-range tactics! It transpires that
19... h3? is bad due to 20. xf7+!.

520
19... e6
He is forced to retreat and go into a difficult ending.
20. xe6 xe6 21. ac1 f6 22. c2

22... e5?
After this move, the battle ends at once. It was essential to play
22...a5! 23. fc1 a6, and Black is still somehow hanging on.
23. xe5! xe5 24. fc1 c8 25. c5 d6 26. 1c2
Petrosian takes control of b2 and continues to restrict his opponent’s
possibilities. White’s position is winning and he has many good
lines, For example, he could win a pawn immediately with 26. a5
a8 27. cc5 f7 28. a4 and, at the right moment, ca5. The
computer suggests breaking up the enemy position in the shortest
way possible with the energetic move 26. e1!. Petrosian uses this
recipe slightly later.
26... f7 27. f1 e6 28. e1 d4
Beliavsky goes for a desperate counterattack, but he is already
hopelessly placed.
521
29.f4! d3 30. d2 b2 31. xd3 a8 32. xd6+ xd6 33. d3 a5
34. c4 a3 35. a4 c5 36. xc5 xc5 37.b4+ c4 38. xa5 b8
39.a3 d3 40. f2 b7 41. c5 a7 42. xc6 xa3 43. f3 1-0
Now for some more prophylactic ideas.
Queen’s Gambit Declined
Georgy Agzamov (2570)
Efim Geller (2445)
Sochi 1984
1.d4 f6 2. f3 d5 3.c4 e6 4. c3 e7 5. g5 0-0 6.e3 h6 7. xf6
xf6 8. d2
This move is directed against the reply 8...c5. It is not very popular
today because of 8...dxc4.
8...c5
Geller chooses the most principled continuation.
9.cxd5 cxd4 10. xd4 exd5

522
Now, Georgy Agzamov switches to prophylactic thinking. What
would Black do if it were his move? Of course, bring his knight to
c6.
11. b5! d6
Geller could not know that 40 years later, it would turn out that
Black can bring the knight to c6 all the same. This is one of the
possible variations demonstrated by the computer: 11... c6!!
12. xc6 bxc6 13. xc6 d6 14. xd5 xc3+ 15.bxc3 c7 16. e7+
h8 17. c6 e5 18.0-0 h3 19.gxh3 xe7. In this case, Black can
confidently hold the position.
12.0-0 d8 13. ac1 c6 14. xc6 bxc6 15. a4
White has obtained a stable positional advantage. Black could have
played better in this phase of the game, but we are more interested
in the further course of the play.
15... d7 16. c5 dc8 17.b4!
Black wanted to play 17...a5.
17... e8 18. c2 cb8 19. fc1 d8

523
White is exerting serious pressure on his opponent’s position and
has a comfortable positional advantage. But Black is again ready to
play 20... a5.
The Uzbek GM could have retreated the knight to d3 and after
20. d3 c8 advanced the pawn to a4. Then after 21.a4 c7!? 22.g3
b6 23.a5 xd4 24.exd4 Black would face an unpleasant fight for a
draw.
However, Agzamov found another way to prevent ...a7-a5 and
played the original:
20. a6!? b6 21. e2 g6
Now what does Black want to play? Of course, 21...c5.
22.a3! d7

Once again, we ask ourselves what Black wants to do next: attack

the g2-pawn with 23... h3.
23. f1!

524
Playing against the opponent’s next move . Agzamov defends the
pawn before it is attacked. The move 23... h3 is now impossible
because of 24. xc6!.
23... f6?! 24. c5
As Kramnik sagely observed, this style of play exhausts the
opponent. When you do not allow your opponent to carry out his
plans, take preventive measures against them and simply do not let
him play, then he begins to spend too much time thinking, gets
nervous, and plays below his usual level. So, Geller here made a
serious mistake:
24... f5?!
It was possible simply to retreat with 24... c8, in order after
25. cb3 a6 26. d1 to exchange the dark-squared bishop with
26... xd4 27.exd4 e8, followed by 28... e6.
25. xf5! xf5 26. b3
As in Petrosian-Beliavsky, the exchange of the light-squared bishop
makes it harder to defend the c6-pawn.
26... d8?
The decisive mistake. Geller’s nerves did not hold out and he
decided to offload the pawn in order to create counterplay. He
should have put up with the pressure with 26... c8 27. a5 e6. If
White now plays 28.b5, then after 28...c5 29. xc5 xc5 30. xc5
d4!, active play by Black would be more appropriate.

525
 analysis diagram
For example, 31. c4 dxe3! 32. xb6 axb6 33. c2 exf2+ with equal
play. But understanding this without the help of the computer is
practically impossible.
In the game, Black faced an unpleasant surprise:
27. a5!
After 27. xc6 b2! 28. d4 e4, Black retains definite hopes of a
positive outcome to the game.
27...d4 28. xc6 d3 29. d2 d7 30. a5

526
White has won a pawn and Black’s only source of pride, the d3-
pawn, is securely blockaded. White can attack it with four pieces
whilst Black can defend it with only three. To add to everything
else, Geller was in time-trouble.
30... bd6?
30... a6! was no help either in view of 31. c5 .
31. b3 d8 32. c5 a5
In a hopeless position, Black missed the move
33. b7
and resigned a few moves later.
I would like to conclude this section on Petrosian and prophylaxis
with one more game.
English Opening
Tigran Petrosian (2580)
Florin Gheorghiu (2375)
Moscow 1967

527
1.c4 e5 2. c3 c6 3. f3 f6 4.g3 b4 5. g2 0-0 6.0-0 e8 7.d3
h6 8. d5 f8
So far the play has repeated what we saw earlier in the game
Rubinstein-Duras.

9. xf6+
The great Akiba played 9.b3.
9... xf6 10. d2 d6 11. e4 d8 12. c3 d7
An interesting moment. In this position, my engine suggests for
Black the prophylactic move 12... b8!? instead.
13.b4
In Soviet books and periodicals, this move was always accompanied
by an exclamation mark. After all, in the event of 13... xb4, 14. b1
followed by 15. xb7 looks very strong. But the engine finds an
amazing resource here: 14...d5!! 15.cxd5 c6!, with a completely
acceptable position for Black. And after 13... xb4 14. xb7 b8
15. g2 c6 White’s advantage is not so significant. I am simply

528
demonstrating how much the modern computer has changed chess:
moves that once seemed very bad now look quite acceptable.
13... c8 14. b1!
I give an exclamation mark to this move for Petrosian’s ability to
look far into the future. Obviously, Black’s last move prepared the
exchange of bishops on h3. White does not mind this. In response,
he is ready to occupy the e4-square with the e-pawn and gain a long-
term advantage associated with the passivity of the f8-bishop. But
when the white central pawn advances to e4, the black knight is, as
it were, invited to the alluring square d4, weakened by White. In
reality, though, the knight does not attack anything from there and it
will be exchanged for the remaining white bishop. As a result, a
situation will arise in which the white knight will be much stronger
than the black bishop.
Petrosian does not want to scare off his opponent with his activity
on the queenside and refuses to play the strongest move from a
computer point of view, 14.b5. Unlike his opponent, he understood
that e7 was a better square for the black knight than d4, even if there
is a white pawn on e4.
The game continued:
14... h3 15.e4! xg2 16. xg2 g6

529
Events are developing according to Petrosian’s scenario. White
already has a decent positional advantage. My engine estimates it at
+1.0.
But how can he develop his initiative? The computer suggests the
resolute 17.f4 exf4 18. d5 g7 19. b2 fxg3 20. f3 f5 21. xg3
e7, with unclear complications, as one of the main paths for
White.
But this manner of playing did not impress the ninth World
Champion at all. We have seen how strong he was in attack. But in
strategically better positions, he almost always preferred to limit his
opponent’s options through prophylactic thinking rather than enter
into unclear complications.
17.h4!
A typical Petrosian move. What should Black do now? He must
either allow the white pawn to advance further or weaken his
kingside with 17...h5. In the second case, the white bishop gets the
important square g5 after 18. d5 g7 19. g5, and White’s
initiative on the kingside looks threatening. But the engine is not

530
very impressed with this initiative, and it suggests the quiet move
19... d4.
Instead of 18. d5, the computer finds an amazing idea: 18.f3!?.
This move does not occur to a human at all. White’s next move is
going to be 19.g4!, and in case of 18...f5 19. d5! g7 20.exf5 gxf5
21.f4, White’s position is close to winning. If Petrosian had won
after the moves 17...h5 18.f3 and 19.g4, then this game would have
been in line with other calling cards of the ninth World Champion.
But chess history, like any other form of history, does not recognize
the subjunctive mood.
17... g7 18.h5 g5 19. d5 d4

I would like to draw your attention to one more important point.

White has achieved a decent positional advantage. All he has to do
is drive away the enemy knight on d4 or exchange it for his bishop,
and place his own knight on the f5-square. To do this, he can either
immediately play 20. e3 in order to meet 20...f5 with 21. b2, or
start with 20. b2, and in response to 20...f5 make the move 21. e3.
There is no real difference, but habit is second nature:

531
20. e3!
Of course – Petrosian was used to playing against the opponent’s
next move !
20...f5 21. b2 fxe4 22.dxe4 e6

The Rumanian GM prevents White’s idea by activating his queen,

which takes aim at the white pawns on c4 and e4, and if necessary
the h5-pawn from the square f7.
Guess Petrosian’s next move.
Answer: 23. c3!?
He is in no hurry and just prevents the move 23...a5. Probably, his
next move will be 24. g2, and then 25.b5 or 25. g4, depending on
the opponent’s actions.
23...b5?
The decisive mistake. Black had to hold his inferior position and
wait for the right moment for active counterplay. This is where the
vivid description of Petrosian’s playing style from Fischer’s book
My 60 Memorable Games fits perfectly: ‘Petrosian loves to play cat
and mouse, hoping that his opponents will make a mistake when
532
there is no direct threat. The most amazing thing is that they usually
do!’
Instead of the move 23...b5?, Black could have played 23... f8, and
the game would have gone roughly in this direction: 24.a4 a6 25.a5
followed by 26.b5!?. My engine estimates White’s positional
advantage here at +1.44.
Can we consider that Petrosian applied the ‘Do not hurry’ principle
when playing 23. c3!?...? More likely no than yes! He was simply
engaged in prophylaxis and monitored the possible counterplay of
his opponent. The ‘Do not hurry’ principle is applied at the moment
when the opponent no longer has counterplay. And in order to
deprive him of counterplay, sometimes you need to hurry, and hurry
a lot!
24.cxb5! xa2
24... xb5 25. d3 xc3 26. xc3 xa2 27. xc7 is also bad for
Black.
25. d3! e2
Otherwise White would have played 26. xd4.
26. xe2 xe2 27. d5 ab8 28. fe1 d4 29. xd4 exd4

533
The endgame is winning for White. Black’s pawn structure is
broken, and the advantage of the knight over the bishop is
overwhelming. The subsequent play probably took place under
mutual time pressure, but White easily and confidently brought the
game to victory:
30. bc1 b7 31. xc7 e5 32. c6 g4 33. d5 xb5 34. xd6 b7
35. g6 h7 36. xg4 d7 37. h1! e6 38. d1 c6 39. d2 e5
40.f4 h8 41.f5 1-0

The process of realizing an advantage

Now I would like to invite you to delve deeper into the topic of
techniques for realizing advantages. Realizing an advantage is a sore
point for most young chess players. From my experience as a coach,
I can say that approximately 80% of the chess players I have worked
with find the technique of realizing an advantage in better or even
won positions one of the main problems of their game. And only
20% are satisfactory in this regard. What is the difference between
these two groups? It seems to me that the best techniques for

534
realizing an advantage are mastered by players who have a good
sense of the rhythm of the game and feel the moment when they need
to move from statics to dynamics.
At one of the online masterclasses of the Kramnik school for the
strongest young chess players in Russia, Vladimir said that there is
no general methodology for realizing an advantage. Every strong
chess player approaches this process according to their style.
Vladimir Kramnik shared how he personally approached realizing
an advantage.
For a certain time, the fourteenth World Champion would limit his
opponent’s possibilities, outplaying him in a calm positional fight.
When he realized that his advantage had become decisive, he
abruptly switched to dynamics, increasing the number of calculated
variations and finishing off his opponent in the shortest possible
way, without avoiding any complications.
I would like to emphasize that before switching to dynamics,
Kramnik, having excellent ‘masonry’ skills, literally would not
allow his opponent to play. As Boris Gelfand noted at one of the
school sessions: ‘It’s very difficult to play against Kramnik. He
doesn’t let you have much fun.’ As for the final stage of the game,
Kramnik himself described it as follows: ‘The killer instinct
awakens and I smell blood.’
Let’s look at a specific example.
Queen’s Gambit Declined
Vladimir Kramnik (2730)
Jan Timman (2590)
Belgrade 1995
1. f3 f6 2.c4 e6 3. c3 d5 4.d4 bd7 5.cxd5 exd5 6. g5 c6
7.e3 e7 8. d3 h5 9. xe7 xe7 10.0-0 0-0

535
We leave the opening stage of the game without commentary. In
almost 30 years since the game was played, theory has made great
leaps forward. We will make use only of Kramnik’s brief comments
in his best games book to his next move:
11. b1!
‘And it turned out that, after thinking for about 15 minutes in this
theoretical position, I came up with a significant improvement.
Instead of the usual 11. b1 White prepares his minority attack with
tempo.’
11... hf6 12.b4 e8
Try from this moment on to guess Kramnik’s moves.
Black is ready to play 13... e4, and will meet 13.b5 with 13...c5
14.dxc5 xc5 with good play.
13. c1!
A prophylactic against 13... e4. In this case, White will play
14. xe4 dxe4 15. d2 f6 16.b5, and the c1 exerts serious
pressure against the opponent’s queenside.

536
13...a6 14.a4 g6 15. b2!

White ‘holds’ the position. Now a psychological battle starts. Very

few players are able to stand still and hold a position. Let us first see
another example of this.
Queen’s Gambit Declined
Peter Leko (2741)
Vladimir Kramnik (2788)
Internet 2020
1.d4 f6 2.c4 e6 3. f3 d5 4. c3 bd7 5.cxd5 exd5 6. g5 c6
7. c2 h6 8. h4 e7 9.e3 0-0 10. d3 e8 11.0-0 e4 12. xe4
xh4 13. h7+ h8 14. xh4 xh4 15. d3 e7

537
In this game, Kramnik has to defend against his opponent’s activity
on the queenside.
16. ab1 a5 17. a4
After the direct 17.a3 g8 18.b4 axb4, Black answers 19.axb4 b5,
followed by bringing his knight via b6 to c4.
17... g8 18. fc1
What should Black do? Hold! But how does one play with the rooks
not coordinated and a passive bishop on c8? The answer is simple,
although it seems complicated – stand and hold! Chess is a battle of
characters. Sometimes one needs to be able just to hold on and not
concede anything to the opponent, however difficult this may seem.
18... d6 19.a3 g6 20.h3 g7
Black is not at all bothered by the move 21.b4. After 21...axb4
22.axb4 b5! 23. c5 xc5 24. xc5 xc5 25. xc5 d7 he has only
one weakness, on c6, and, as is well-known, this is not enough to
lose.
21. c3 h5 22. a1 f6! 23.b4 axb4 24.axb4 g5!
538
White has managed to advance the b-pawn, but Black has organized
counterplay on the kingside.
25. f1
To demonstrate the seriousness of Black’s intentions, we can offer
the following beautiful variation: 25.b5 f6 26. h2 h4 ( 26... g4+
27.hxg4 h8 28. g1 hxg4 29.bxc6 bxc6 30. xc6 h5 31. f1 f5
32. xf5 xf5 33. d6 h1+ 34. e2 xa4 35. xa4 xc1 36. e5+
xe5 37.dxe5 c2+ 38. f1 g3 with a draw) 27.bxc6 bxc6
28. xc6? xh3 29.gxh3 xe3 with a decisive attack for Black. I
don’t usually like to quote long computer variations, but in this case,
I wish to show that White needs to play very carefully, so as not to
fall under a crushing attack. In this rapid game, Peter Leko did not
manage this.
25... f6 26. c2?!
26. c5.
26... h4! 27. b6?
27. e2 h7 with a strong attack.
27... xa1 28. xa1 xh3! 29.gxh3 xh3+ 30. e1 h1+ 31. f1
h4
The h-pawn cannot be stopped. Black has a decisive advantage,
which my engine assesses at -8.2. Kramnik scored a convincing
victory in a few more moves.
Now we return to Kramnik-Timman
15...a5
The Dutch GM does not want to run on the spot and decides to
change the picture of the game. By way of alternatives, one can
suggest trying the move 15... b6, so as to meet 16.b5 with
16...cxb5 17.axb5 a5, although the white position deserves
preference in this case. In addition, he can go over to a modern-day
plan with 16.a5.

539
16.bxa5 xa5 17. d2 g4 18. b3 d6 19.g3 a7

As Kramnik pointed out in his notes, it was best to move the rook to
a8, but this only became clear after the post-mortem analysis
between the players. Now what should White do?
20.e4!
Kramnik decides to open the game, so as to exploit the certain lack
of harmony between the enemy pieces.
20...dxe4 21. xe4 f8

540
I would like from this moment to switch off the engine (if it has
been on) and try to ‘become Kramnik’.
First, we ask ourselves our favourite question: what would Black
play if it were his move? Of course – 22... df6.
Which of my pieces is the worst placed? The answer is: the c1.
Can we combine two objectives – preventing the opponent’s desired
move and improving the position of the rook?
22. e1
The engine does not approve of this move because of the reply
22... b4!.
But the queen has only just retreated from d6 to f8! 22... b4 looks
dangerous because of 23. ab1, and on 23... xa4 White plays
24. e2 with the threats of 25. c5 and 25. xg4. The human player
will generally stop calculating at this point, but the engine goes on
and pronounces equality after 24... df6. Humans just do not play
like this!
22...b6?!

541
Black would like to play 22... df6, but was afraid of the reply
23. bc5, coming under a press after 23... xe4 24. xe4 f6
25. b6 a8 26. f3.
Again, White asks himself: which of his pieces is the worst placed?
It turns out that with the pawn on b6, it is now the b3. Where
should we put it? On c4.
23. bd2! a6
The next question is whether it suits White to exchange light-
squared bishops when he has a space advantage. The answer is no.
24. c2!
The place for the bishop is b3, from where it attacks f7.
24... b7 25. b3 gf6
Question: can we improve the position of any more of our pieces?
Which of them is not working now? The answer is: the a1.
26. ac1! c8

Over the last five moves, the future World Champion outplayed a
super-grandmaster from the world elite with apparent ease, asking
542
himself the same question: ‘Which of my pieces is not active
enough?’ In fact, everything was not so simple. Kramnik, of course,
was calculating variations as well, but he felt that not everything
was ready for the decisive blow, and it was better not to go into
hand-to-hand combat yet, but to limit the opponent’s possibilities .
White’s advantage has now become decisive, and it is time for him
to look for a way to win by calculation. He can, of course, take a
pawn with 27. g5 d5 28. xc6, but here he still has to deal with
the conversion. Perhaps this would be too sluggish a solution.
We need to look for something tougher. Is it possible not to win a
pawn here, but, on the contrary, to sacrifice one? Kramnik’s move
search algorithm was, of course, different, but his solution now
coincides with the first line of the computer.
27. xf6+! xf6 28.d5!! xd5 29. e4!
With this, the game ends. The engine assessment is +4.76. The
move 27. g5 gets less than half that. In time-trouble, Timman lost
very quickly:
29... d8 30. xc6 b4 31. f6+ f8 32. xh7+ g8 33. f6+ f8
34. e4
Black resigned.

Varying the rhythm of the game

I learned about the ragged rhythm of the game in the early 80s of the
last century. At one time I also had a problem with the realization of
advantages. I turned to my coach, grandmaster Isaac Boleslavsky,
for advice. He recommended that I look at Smyslov’s games.
I played through the openings of Smyslov’s games, and then began
to guess his moves. I was already a decent master, and I was amazed
by the rhythm of the play of the seventh World Champion. He made
strong moves and gradually outplayed such powerful opponents as

543
Keres, Bronstein, Reshevsky, and Geller. There seemed to be no end
to his calm and high-quality moves.
But suddenly, at a completely unexpected moment for me, Smyslov
would abruptly change the nature of the struggle and plunge into
complications, which for some reason worked out in his favour. He
precisely felt the moment when the quantity of his excellent
positional moves turned into quality, which means the moment had
come for a sharp aggressive game, requiring concrete calculation of
variations. If such a moment is missed, then the opponent can
gradually strengthen his position, and the advantage will begin to
evaporate.
Vladimir Kramnik was also very impressed with Smyslov’s playing
style. In 2004, Vladimir gave a long interview dedicated to his
predecessor world champions, and he characterized Smyslov’s work
as follows:
‘Smyslov is the truth in chess! Smyslov played the endgame
brilliantly, and in general his game flowed like a song. You look at
his games, and there is a feeling of some kind of lightness, as if the
hand itself is making a move, and the person is not straining at all,
as if he is drinking coffee or reading a newspaper! A feeling of some
kind of Mozartian lightness! There is no strain, no tension;
everything is simple, but brilliant. This makes Smyslov very
impressive to me, I really love his games.’
Let’s look at two games from the seventh World Champion.
Double Fianchetto
Vasily Smyslov (2494)
Lev Polugaevsky (2585)
Interzonal tournament, Palma de Mallorca 1970
1. f3 f6 2.g3 g6 3.b3 g7 4. b2 d6 5.d4 0-0 6. g2 e5?!

544
Black would do better not to hurry with this move, as Smyslov’s set-
up is specifically directed against the opponent’s King’s Indian
Defence.
7.dxe5 g4 8.h3 xe5 9. xe5 dxe5 10. xd8 xd8 11. d2 d7
12.0-0-0 e8 13. c4 b6 14. a5 b8

White has a certain initiative. His fianchetted bishops rake the

opponent’s queenside and centre along the long diagonals. But
Black is ready to make two obvious moves: 15... c6 and 16... e6,
after which White’s pressure will be gradually neutralized.
How can White prevent this? 15. a3 is useless because of the
answer 15...e4! with the idea 16... c3. Nor does the move 15.c4
promise anything in view of 15... f8.
But Smyslov finds what is at first glance a very simple solution.
15. d2! c6 16. hd1
If you look through this game quickly, then you are unlikely to pay
attention to these last two moves. What could be more natural than
to double rooks on the only open file? But if you are a trainer and
you ask your pupils, in the position of the last diagram, how to
545
prevent the opponent playing 15...c6 and 16... e6, then, take my
word for it, even a strong young GM in the 2550 range will need
prompting with the motto ‘ Technique is the mastery of short-range
tactics ’.
In fact, after 16. hd1 the move 16... e6? is impossible because of
the blow 17. xb7!. Continuing the variation: 17... xb7 18. xc6
bb8 19. xe8 xe8 20. d8 xd8 21. xd8+ f8 22. a3, winning
for White. Let us remember what Kramnik wrote about Smyslov:
‘There is no strain, no tension; everything is simple, but brilliant.’
If we seek the advice of the computer, it shows that instead of
15...c6 Black would have done better to play 15...e4! 16. hd1 f5!
17.c4 xb2+ 18. xb2 e5 19.b4 g7 20.c5 a4+ 21. b3 e3
22.fxe3 b5.

analysis diagram
No human can come up with such a line and even explaining it to
oneself with hindsight is not easy.
16...f6?

546
A natural move which turned out to be the decisive mistake. It was
essential to advance the f-pawn two squares, reserving f6 for the
bishop.

Again, we have a great training exercise before us. Try to find a way
to obtain a decisive positional advantage.
Answer: 17. d8! f7 18. xe8! xe8 19. a3!
Young players tend immediately to start calculating long variations
such as 17. d8 f7 18. a3 f8 19. xf8 xf8 20.c4 e7 21. 8d6
f7 22. xc6 bxc6 23. xc6 b7 24. d8+ xd8 25. xd8.

547
 analysis diagram
White stands better, but after, say, 25...h5 with the threat of
26... xh3, Black can offer decent resistance. The engine score is
about +1 in favour of White.
It is likely that with closer analysis, one can find a way for White in
these complications to gain a great advantage. However, the path
chosen by Smyslov is elegant and does not require complex
calculations. It is significantly less energy-consuming.

548
The most difficult move in this series was the exchange of rooks on
e8. After all, it gives the black king control of the square d8. But the
white rook ends up commanding the 8th rank even so, after the
transfer of the bishop via a3 to c7. It is ‘all simple, but brilliant’. The
computer assessment after 19. a3 is +2.6.
19... d5 20. d6 a8 21.c4 e7 22. c7 f5 23. d8+
White has carried out his plan and has a decisive positional
advantage on his side. Smyslov was always good when it came to
realization. I advise you to continue guessing White’s moves, most
of which look very simple on the surface.
23... f7 24.c5 d5 25. d6
The computer suggests the exchange sacrifice 25. xd5!?, but
Smyslov does not hurry to force events.
25... f6 26. f8+ g7

549
27. c4?!
The first and only blemish on White’s play in this game. After the
strong and human move 27.e4!? the following could have occurred:
27...fxe4 28. xe4 c3 29. d3 xa2+ 30. d2! b4 31. c4 d5
32.f3

550
 analysis diagram
and Black is quite defenceless.
But the computer suggests 27. xd5!! cxd5 28.b4!. It is hard for the
human player to play this, as one has to overcome many shibboleths
and habits of thought. The engine assessment is +4.17. Black’s
counterplay falls short, which is far from obvious. For example:
28...b6 29. c6 bxc5 30.bxc5 a5 31. a7!, and it is time for Black to
resign.
27...e4 28.f3 exf3?!
Again, not obvious, but also a mistake, like White’s 27. c4?!. It is
hard even to explain in words why. The computer gives a mass of
very complicated variations, beginning with the move 28... g5+!?.
We will give one of these: 29. d1 b5 30. e5 e3+ 31. e1.

551
 analysis diagram
And here Black is saved by the inhuman move 31...f4!. All other
moves lose. It is absolutely unrealistic for any player to see such a
line during the game, so I do not advise you to pay much attention to
such errors.
29. xf3 g5+ 30. d1 c3+ 31. e1

552
31... b5
After 31... xa2, White could have started a decisive attack on the
defenceless black king with 32. e5. White’s next move will be
33. d7, and then 34. e5+ and 35. g8. Therefore Lev Polugaevsky
returns the knight to his own camp and attacks the enemy bishop.
32. e8 f7 33. h8 f6 34. e5+ g7 35. e8 h5 36.e3
White’s position is absolutely won. Smyslov plays a quiet and
useful move, which takes the square d4 from the enemy knight and
prepares the transfer of his bishop to c4.
36... c3

553
37. xc6!
It is time. There follows a short tactical skirmish, as a result of
which the game comes to its logical conclusion:
37...bxc6 38. xc6 b7 39. xa8 xc6 40. xa7+ g8 41. a6 d5
42.c6 e4 43.c7 b7 44. a8+ f7 45. b4 1-0
The following game was played almost a quarter of a century later,
when Smyslov
Dutch Defence
Vasily Smyslov (2520)
Lembit Oll (2595)
Rostov-on-Don 1993
1.d4 f5 2.g3 f6 3. g2 g6 4. h3 g7 5. f4 c6 6.h4 d6 7. c3 e5
The opening has not gone well for Black. White is ahead in
development and is striving to seize the initiative. Black has to take
into account the advance of the white d- or h-pawns, as well as the
move e2-e4. With his last move, Oll simplifies the position in order
to extinguish the opponent’s nascent initiative. From a purely chess
554
point of view, this decision is correct. But playing in the endgame,
and indeed in positions without queens, was not the Estonian
grandmaster’s strong suit.
8.dxe5 dxe5 9. xd8+ xd8 10. d3

The pawn structure is rather similar to the previous game. And, just
as in the game between Smyslov and Polugaevsky, playing ‘by eye’
leads Black into trouble.
Out of general considerations, Black played:
10... bd7?!
But he should have proceeded more energetically: 10...e4! 11. f4
a6!. This is needed, so as after 12.f3 to have the possibility of
12... b4!. Then the game could continue 13. d1 e8 14.fxe4 fxe4
15.a3 bd5 when Black may even have a slight advantage. And if
12. d2, White needs to be prepared for complications after
12... g4 13.f3 e3 14. c1 f2 15.0-0 e8.
In the game, Smyslov immediately stopped any possibility of active
counterplay, gaining a noticeable positional advantage:

555
11.e4! e8
Now a question for you, dear reader. How would you play if you
were Smyslov?
The engine recommends the sharp move 12.h5!?. But a person does
not always want to leave their comfort zone in a better position and
take on increased obligations with a pawn sacrifice after 12...fxe4
13. xe4 xh5, although the move 14. d6 gives White a
formidable initiative. The engine assessment in this position is close
to +1.00. Smyslov plays the computer’s second (human!) line:
12. g5!
Before bringing out the bishop to d2, it is very important for White
to weaken the opponent’s kingside pawn structure and try to force
the advance of the rook’s pawn to h6.
12...h6?!
The decisive positional mistake. Black should have left the bishop
alone and played 12... c7.
13. d2 b6 14.0-0-0 c7
After 14... c4 White would have replied 15. xe5.
15.b3 fxe4 16. xe4 xe4 17. xe4 f5 18. de1

556
Black’s defence is not easy. He can only meet the direct threats and
otherwise watch as his opponent strengthens his position.
18... ad8 19. c3 d7 20. e2 e7 21.h5!
White had anticipated this situation when he provoked 12...h6 by
means of 12. g5!.
21... xe4 22. xe4 gxh5
The reply 22...g5 does not save Black because of 23.f4!. Then play
could develop something like this: 23...gxf4 24.gxf4 de8 25. he1
d6 and now 26. d1 or 26. d2 will lead to a decisive advantage
for White.
23. xh5 de8

557
Let’s study this position more closely. Black has two weak pawns
on e5 and h6. Does this remind you of a classic game?
Bobby Fischer
Tigran Petrosian (2580)
Candidates Match, Buenos Aires 1971

558
In both diagrams we see weak isolated black pawns, only on
different flanks. In the position in the second diagram, the future
World Champion Bobby Fischer made his legendary exchange of
the excellent knight on c5 for the main defender of Black’s pawn
weaknesses – the ‘bad’ bishop on d7 – with 22. xd7!! – a move, by
the way, which the latest AI engines do not even regard as the
strongest! And in the game Smyslov-Oll, Black’s position would
quickly fall apart if the white knight and black bishop were removed
from the board. But White’s knight on d3 is attacking the e5-pawn,
and there is no need to move it to somewhere such as f5 now.
Smyslov played:
24. h1
Again, I cannot refrain from drawing an analogy with Fischer’s
play, this time the game we have already seen when we studied
Aronian-Vidit in Chapter 8: the game Fischer - Reshevsky , New
York 1963:

Reshevsky, in defending against 25.b5, put his own pawn on that

square with 25...b5, giving himself a weak pawn on a6. In reply, the

559
excellently-placed rook on d5 was immediately redeployed by
Fischer with 26. d5 d1 27. a1. White won quickly, by alternately
attacking the two weak pawns.
In Smyslov-Oll , in reply to 24... f6

Again, an interesting moment. White has the possibility of winning

the h6-pawn, but in doing so he would permit his opponent to
coordinate his forces and simplify the position a good deal. After the
move 25. d2 the game would continue as follows: 25... g5 26.f4
exf4 27. xe7 xe7 28.gxf4 f6 29. xh6 g7. White has an extra
pawn, but realizing it would not be easy, because very little material
remains on the board.
Vasily Smyslov preferred a different line of play, involving the
transfer of the white king towards the kingside, which is the
computer’s first line:
25. d1!
White understands that the win of the pawn is only a matter of time
and the opponent is hardly likely to be able to prevent it. Therefore

560
he need not hurry with this and should first improve the position of
his forces.
It is very hard for Black to hold the position, and with one nervous
move he can accelerate his demise. Therefore Smyslov offers Oll
some rope with which he can hang himself.
25... g5 26. e2 d6?!
The Estonian GM undertakes some unjustifiable activity. He tries to
bring his king to f5 under the fire of the enemy pieces.
27. d1! e6?
He should have returned the king with 27... c7.

28.f4! f5 29.g4+!!
Technique is the mastery of short-range tactics.
29... xe4 30. f2+ xf4 31. g1!
Black resigned because of inevitable mate next move.

The art of manoeuvring

561
In these two examples of Smyslov’s play, we saw splendid
‘masonry’ and a brilliant feel for the rhythm of the game. Now I
would like to go on to look at the technique for realizing advantages
of the 14th World Champion, Vladimir Kramnik.
Vladimir Kramnik (2760)
Anish Giri (2758)
Doha 2014

36. c1
Kramnik gave the following comment: ‘Another pensioner’s move.
Nothing was threatened, but I like such moves. I learnt this from
Karpov, who was the master of unhurried prophylactic play. In
winning positions, when your opponent has few possibilities, you
should just make useful moves and give him the chance to lose the
game by himself.
36...a4
Desperation. If 36... b4, I could win by force: 37.c7 c8 38. ac2
(threatening 39. f5) 38... d6 39. g5 g6 40. d5+ g7 41. c6, and

562
Black cannot take the c7-pawn because of the knight fork on e6. The
game is over.
37.bxa4
Anish resigned in view of 37... xa4 38. xa4 xa4 39.c7 c8
40. f5.’ (Kramnik)
And here is a fragment from an interview Kramnik gave in 2004:
‘Karpov continues the line of universalism?
Yes, sure. But there is a certain mystery in his game, he knew how
to do what no one else can do. It’s easier for me to talk about him
than about other champions, because my first chess book was a
collection of Karpov’s games. I studied it in my childhood, and then
played many games with him. He was a universal chess player, a
good tactician, great at calculating variations, positionally very
strong, but he had one distinctive feature. Speaking half-jokingly,
half-seriously, he refuted Steinitz ’s postulate, which states: "He
who has an advantage is obliged to attack under the threat of losing
this advantage." Karpov , when having an advantage, began to
stand still, and his advantage increased! I don’t understand how this
is possible. This component of his playing has always amazed and
delighted me. This moment is difficult to describe in words, but there
is something elusive in it, something specifically Karpovian. ’
(italics mine)
I really like Vladimir’s deep thought that Karpov’s playing style to
some extent contradicted Steinitz’s postulate. However, I would like
to develop and complement this idea a little with you. And the
interesting book The Method in Chess by Iosif Dorfman will help us
with this. In this work, the grandmaster notes that advantages should
be divided into dynamic and static. In my opinion, when it comes to
a dynamic advantage, which can evaporate if one plays quietly,
Steinitz’s idea is certainly correct – the one with the advantage is
obliged to attack. But when we are talking about a static, long-term
positional advantage, various ways of exploiting it are possible.
563
As we have already noted, Kramnik’s technique for realizing his
advantage is as follows: for the time being, he limits the opponent’s
options, and then, when he feels that the position is ripe for
delivering a decisive blow, he changes the rhythm of the game and
switches to dynamics. In doing so, he acts as resolutely and
aggressively as possible in order to complete the game in the
shortest possible way.
Karpov usually preferred to choose the path that led to minimizing
risks. As grandmaster Yuri Razuvaev aptly noted: ‘Karpov prefers
manoeuvring, avoiding positions in which one specific move could
be decisive .’ The advantages that Karpov obtained in a
manoeuvring battle were most often of a static nature.
At the same time, Karpov created masterpieces of prophylactic play
and was a great master of the art of manoeuvring. This was noted
even at the beginning of his career. So, for example, in his article
‘How to beat Anatoly Karpov’, Mark Dvoretsky wrote about his
impressions of blitz games with the future twelfth World Champion:
‘Sometimes it was possible to outplay him in the opening or catch
him with an unexpected tactical blow, but in the art of manoeuvring
it was completely impossible to compete with him – Karpov sensed
too subtly the hidden spring of interaction between the pieces. You
get an unclear or slightly better position in the opening, and you
make seemingly reasonable moves, but after 10-15 moves, for some
reason you encounter insurmountable difficulties. His playing made
a strong impression on me even then.’
And I remember my conversations with grandmaster Gennady
Kuzmin, who had played with Karpov as a member of the student
team of the Soviet Union. Kuzmin asked Karpov to show his game
after the end of each round. And he was amazed by the playing style
of the future World Champion. First, it demonstrated complete
coverage of the entire board. As soon as the opponent moved some
pawn on the flank, this detail could not escape Karpov’s watchful

564
eye. He could instantly change the entire previously planned scheme
of action to reflect the changed circumstances. Secondly, his every
action was logical and thought out – he did not make a single move
without an idea. There were, of course, third, and fourth, and fifth
features, but Kuzmin called these two distinctive features in
Karpov’s game the main ones.
In general, Karpov himself rarely explained his play, and it is
difficult to truly understand. Some features of his play were even
difficult to formulate verbally, and only many years later have
people managed to find good definitions for them. Thus, in one of
the programmes on the YouTube channel ‘Levitov Chess’, Vladimir
Kramnik perfectly commented on the famous 9th game Karpov-
Spassky, Leningrad 1974 from the Candidates semi-final match. In
these comments, I first heard the deep thought that Karpov’s
prophylactic moves were ‘dual-purpose moves’. They not only
limited the opponent’s possibilities, but also prepared a future
offensive or counter-offensive by the twelfth World Champion. And
at the moment when the game could move from positional struggle
to hand-to-hand tactical combat, Karpov’s pieces found themselves
in the right place at the right time.
Let us see Karpov’s technique for realizing an advantage.
We will begin by analysing the well-known game Karpov-Timman,
Montreal 1979. Whereas I called the game Petrosian-Gufeld,
Moscow 1961, ‘a hymn to prophylaxis’, the Karpov-Timman game
can best be described as a ‘ballad of prophylaxis’. This game has
been annotated in many sources, but I would like to do so with an
emphasis on Karpov’s technique for realizing an advantage.
Pirc Defence
Anatoly Karpov (2705)
Jan Timman (2625)
Montreal 1979

565
1.e4 d6 2.d4 f6 3. c3 g6 4.g3 g7 5. g2 0-0 6. ge2 e5 7.0-0
a6 8. e1 c6 9.h3 e8 10. g5 h6 11. e3
We will not analyse the opening stage. I would only like to point out
that Black’s move 7... a6 should have been connected with further
pressure on e4 by means of 11...exd4 12. xd4 c5, although here
too Black’s position is not impressive. After Timman’s move
11... c7?!
my engine’s assessment immediately jumps from +0.3 to +1 in
White’s favour.
12. d2 h7 13. ad1 d7

The sides have completed the mobilization of their forces and it is

now time for White to decide on a future plan. The move 14.f4 is
tempting, to seize space on the kingside. But Karpov prefers first to
strengthen the e4-pawn further and transfer his knight to g3:
14.g4!
The first line of the modern computer. This is a dual-purpose move.
With a pawn on g4 and a knight on g3, it is easier for White to

566
become active on the kingside, whilst the overprotection of the e4-
pawn makes it harder for the opponent to develop counterplay in the
centre.
14... ad8 15. g3 c8 16.f4 b5 17.a3 b4 18.axb4 xb4
It is customary to respond to play on the flank with a counterattack
in the centre. Here, everything happens the other way around: in
response to White’s play in the centre, Timman started counterplay
on the queenside. From general considerations, such an idea for
Black should not be good, but it is not at all easy to question it.
Karpov managed to do this in a relaxed, prophylactic manner. As we
have already noted, he rarely explained his train of thought. An
interesting quote from a collection of his selected games goes:
‘Black’s idea was to somehow position his pieces after playing a7-
a5, c8-a6, e5xd4 and c6-c5.’ (Karpov)
In this position, many players in White’s position would think about
active play. For example, it is worth considering 19.fxe5 dxe5
20.d5, which is quite liked by the computer. But the twelfth World
Champion chose to continue limiting his opponent’s options:
19. ce2
This is again the first line of the modern computer.
19...exd4 20. xd4 a5 21.c3 a6

567
With skilful prophylactic play, White has managed to completely
outplay his opponent. He has an advantage in space, a formidable
attacking group on the kingside, pressure in the centre, and excellent
interaction between all his pieces. And on the queenside, where
Black tried to be active, White’s position is better.
It seems it’s time to think about doing something decisive. The
engine, at a depth of 50, estimates White’s advantage to be around
+2 and suggests starting an attack on the opponent’s king with
22.g5!? hxg5 23.f5!. But let us recall Yuri Razuvaev’s words:
‘Karpov prefers manoeuvring, avoiding positions in which one
specific move may be decisive .’
In the art of manoeuvring, Karpov was far superior not only to
Timman, but also to all his other rivals. Most likely, he did not even
consider such attacking continuations, and played:
22. c2!
Here is his comment: ‘A subtle move, which, firstly, does not allow
the black knight to jump out to c5 (due to the answer b2-b4), and

568
secondly, continues the previous strategic line – strengthening the
e4-pawn .’ (italics mine)
Once again we see a wonderful, multi-purpose move. On the one
hand, Karpov played against his opponent’s intended next move (
22... c5? 23.b4! ). On the other, he overprotects the e4-pawn and
obtains additional possibilities to manoeuvre his forces in the centre
and on the kingside.
22... d7?!
If the bishop must move, it would be better to put it on b7, so the
a6 is not left undefended.
23. f3!
White has this possibility thanks to the fact that the e4-pawn is
defended.
23... e7?!

And now it’s time to remember Vladimir Kramnik: ‘Karpov, having

an advantage, began to stand still, and his advantage increased!’

569
If we look at the position in the last diagram abstracted from the
events in the game, we see an incredible concentration of white
pieces on the kingside. By all indications, they are ready for a
decisive assault on the opponent’s king. At such moments, in
Kramnik there awoke ‘the instinct of a killer who smelled blood.’
He turned on his powerful calculation ability and brutally finished
off his opponent, as was the case in his game with the same Jan
Timman.
In the diagram position, White could have played 24.g5!, with an
engine assessment of +5.8! Here is a sample variation: 24...hxg5
25.e5 dxe5 26.fxe5 e8 27. f2 c8 28. b6 and White wins.
But Karpov played differently. He was diverted by prophylactic
thoughts and asked himself this question: ‘My opponent played
23... e7. What is that for? What is his next move going to be?’ The
answer was something like: ‘Most likely, he wants to play 24... e8
or 24... e8. How can we prevent this?’
24. f2!?
Again, over-protection of the e4-pawn. Now after 24... de8 White
replies 25. d3 with a double attack on the knight and the d6-pawn.
Timman played
24... e8?!
against which Karpov had prepared
25. d3! b7
Now, an unexpected trick:
26. a1!

570
A surprising position! On a full board, Black loses the a5-pawn.
Meanwhile, with the single exception of the move 10. g5 , White
has not once crossed the demarcation line on the fourth rank.
‘Running on the spot’ within one’s own camp and by doing so
totally outplaying the opponent, as Karpov has done here, is the
peak of the art of manoeuvring. After losing the a5-pawn, the battle
effectively ends. We give the remaining precise realization stage
without comment.
26... c7 27. xa5 dd7 28.b4 e6 29. e3 c5 30.f5 d8 31.b5
h8 32. f2 c7 33. a4 b8 34.c4 a7 35. xa7 xa7 36.e5 dxe5
37. xe5 a2 38. xc5 1-0
To demonstrate the magnificent technique of the 12th World
Champion, we present another example.
Boris Spassky (2650)
Anatoly Karpov (2700)
Candidates Match, Leningrad 1974

571
Try, dear reader, to outline a scheme of action for Black over the
next few moves. Let’s start with the question: ‘What does the
opponent want?’ Most likely, to play 27. e4 with the threat of 28.
xf6+.
What should we do? The logical answer is 26...g6 27. e4 g7.
Such a line is, in principle, possible, but White would have some
initiative after the move 28.b4 or 28.g4. Karpov finds a more
flexible arrangement of his pieces, in which the king also plays an
important role:
26... c8! 27. e4 d8!
The main difference between Karpov’s formation and the variation
just mentioned, 26...g6 27. e4 g7, is that it is not purely
defensive. Black is preparing to take active action at any moment
and try to seize the initiative. Karpov’s plan becomes clearer. His
bishop is transferred to the b6-square, a pawn column e5/f6/g7 is
lined up against the white bishop, and the black king is preparing to
take a convenient place on the e6-square, controlling the enemy
passed pawn. The moves 26... c8! and 27... d8! were of a

572
defensive nature. But now Black’s pieces are ready to become active
at any moment. We have another example of ‘dual-purpose moves’.
The game continued:
28.g4 f6 29. g2 f7

30. c1?!
The problem of exchanges! We have repeatedly seen that if there are
two pairs of rooks on the board, the position can have a middlegame
character. And each side needs to carefully evaluate whether it is
worth bringing the king to the centre. After exchanging at least one
pair of rooks, the king’s position in the centre is almost always safe.
To demonstrate the dangers of walking the king around the board
when there are two pairs of rooks, I would like to give the ending of
the Aronian-Nakamura rapid game played in the final match of the
Grand Prix series tournament, Berlin 2022, with brief explanations.
Levon Aronian (2772)
Hikaru Nakamura (2736)
Berlin rapid 2022

573
In this game, Aronian needed a victory, since he had lost the first
game of the tie-breaker. White’s position is preferable. He should
aim to move the king to the centre on d4 and then try to exchange
one pair of rooks. For example, 39. e2. The engine finds ‘tricks’
that are unthinkable for rapid play: 39...a5! 40. xa5 f6 41. d3
e8 42. c4 bb8 43. e1 xe4 44. d4 f6 45. xe8+ xe8
46. c4 e4+ 47. d3 f3! 48.gxf3 f4 with equality. Such a
development of events on move 39 is difficult to imagine in a game
with a classical time control.
Most likely, the game would have proceeded in this vein: 39... f8
40. d3 h5 41. d4 h4!. An important move. Black’s kingside
threatened to become vulnerable, and he would need to protect
himself from the h3-h4 attack.

574
 analysis diagram
The situation that we predicted and tried to describe in words has
arisen. White needs to protect the b4-pawn by moving the knight to
c6, and then try to exchange one pair of rooks. Black will strive for
counterplay against the e4-pawn. The game could develop
something like this: 42. 3b2 e7 43. a5 f7 44. c6 a8 45. a2
f6 46. a5 g6 47. e1 xa5 48. xa5 b8 49. c4 xb4 50. c3
a5 51. xd6 a4 52. c4 f7, and Black is holding on somehow.
All of these moves are optional, of course, but we wanted to
consider a possible direction for the game.
Instead, Levon Aronian immediately began sharp play:
39.h4?! c8 40. a5 f6 41.hxg5 hxg5 42. c6 e8! 43. d4 g4+
44. xf4 xe4+ 45. f5 xd4 46. xf6

575
46... dxd5 47. f1 h5 48. f4?
Now White gets mated:
48... h6+ 49. e7 b7+ 50. d8 h8! 0-1
Now we return to Spassky -Karpov

576
Spassky’s move 30. c1?! was not a mistake and the position has not
yet moved beyond the bounds of equality. But it has now become
easier for Black to play than for his opponent. And this, as we know,
is a definite advantage. White should have broken up the opponent’s
central pawn group with f2-f4. After, say, 30. f1!? b6 31.b4 c6
32.f4 exf4 33. xf4 ec8 34. g3!? xd6 35. fe4 ( 35. f5 d1 )
35... d1 36. e7+ f8 37. f5 g6 38. h7 gxf5 39. h8+ f7
40. xc8 g1+ 41. h2 xg4, with active play White would maintain
the balance and it would more likely be Black who would have to
defend.
Now the initiative passes to Karpov.
30... b6 31. ec2 xc2 32. xc2 e6 33.a4

Now it would have been good for Black to hassle the enemy knight
with 33...g6 and 34...f5. But Karpov very often defers such active
operations or even refrains altogether from them, so as instead to
limit his opponent’s possibilities.
33...a5! 34. a3

577
Try to guess Karpov’s next move.
Answer: 34... b8!!
This mysterious rook move is the computer’s first line.
Nimzowitsch would be delighted! Black prepares the bishop transfer
to d4, followed by the advance ...b7-b5.
The moves 26... c8! and 34... b8!! are again ‘dual-purpose moves’.
On the one hand, they strengthen the position and counteract the
opponent’s play; on the other hand, they prepare to seize the
initiative in a tactical fight. Kramnik himself is very fond of such
multifunctional moves. In classes with young chess players in
Russia, he recommended making them at the first opportunity,
without wasting time looking for something better.
35. c4! d4
Spassky maintains a high level of play. He prepares to simplify the
position:
36.f4
Karpov replied:

578
36...g6 37. g3 exf4 38. xd4 fxg3

There are two moves left to the time control. White, without
thinking, regained the pawn:
39. xg3?!
To use Evgeny Gleizerov’s colourful expression, the computer,
‘laughing its circuits off’, suggests 39. e4+ d5 40. e7 c6 41.g5!
fxg5 42. e1!.

579
 analysis diagram
I suspect that if the game had been adjourned at this point, the
players would have agreed a draw without resuming play. But in
time-trouble, it is not realistic to see a move like 41.g5!.
39... c8 40. d3 g5!
Black has fixed the white g4-pawn. Two pawn weaknesses have
emerged in White’s position, but there is little material left on the
board. White’s chances of a draw are very high, since his pieces are
active. And worldly wisdom (according to Barsky) suggests that a
rook and bishop, with pawns on two flanks, are stronger than a rook
and knight. The assessment of a modern computer is -0.5.
41. b2 b6

580
42. d4?!
I do not know when exactly the game was adjourned or who sealed,
but it seems likely that this was the sealed move. Otherwise, it is
hard to explain why White did not play 42. c3, since after the
exchange of rooks with 42... xc3+ 43. xc3 the move 43... xd6 can
be met by 44.b4, whilst in the event of 43... c5 44. f3 xb3,
White is saved by 45.d7 with a draw.
42... c6 43. c3 c5
The following variation is striking: 43... xd6 44. xd6+ xd6 45.b4
b5!?.

581
 analysis diagram
Associative thinking puts us in mind of the game Shuvalova -
Maurizzi from Chapter 7!

582
However, in Spassky - Karpov this lovely pawn trick only leads to a
draw: 46.bxa5 bxa4 47. b4+ c6 48. e7 b5 49. f3 e5+
50. g3 d7 51. f3 xa5 52. e4 b5 53. d5.

analysis diagram
The white bishop dominates the knight. The position is drawn,
despite Black’s two extra pawns.
44. g2 c8 45. g3 e5!?
There seems to be no other way for Black to play for a win.
46. xe5 fxe5

583
By all indications, the rook ending looks absolutely drawn. In those
distant times, there were no computers and analysis had to be done
by hand. It is quite likely that Spassky’s team, when analysing the
adjourned game, focussed on the positions in which the bishop was
opposed to the knight. They did not analyse the position in this
diagram, but limited themselves to just an assessment.
Indeed, the position of the diagram is easily defensible for White,
but Spassky immediately started making mistakes here and lost it
with two inaccurate moves. Today the engine immediately notes
that after 47. g2 d8 48. f3! xd6 49. f5 d3 50. xg5 xb3
51. g6+ d5 52.g5 b4 53. g3 b3+ 54. g2 e4 55. f6 e5
56. f8 the game ends in a draw.
White has other ways to achieve his goal. But today games are not
adjourned, and with only increment time to think, players often
make mistakes in simple situations. We will not judge Boris
Spassky too harshly and will continue to analyse this interesting
ending.
47.b4?!

584
Commentators at the time unanimously condemned this move as the
decisive mistake, but it was not.
From the point of view of the method of defence in the rook ending,
White’s move is quite logical. Sacrificing a pawn to activate the
rook is the most common technique. Now, after 47...axb4 48.d7 d8
49. b3 xd7 50. xb4, White will easily achieve a draw. The move
47... d8 doesn’t promise much for Black either. The position after
48.bxa5 bxa5 49. b3 xd6 50. b5 d3+ 51. f2 d4 52. xa5 xg4
53. a8 is clearly drawn. Unfortunately for the tenth World
Champion, he missed a zwischenzug:
47...e4!

Now White’s task is already difficult. The tempting 48. b3?, by

analogy with the previous variation, no longer saves White. Black
wins instructively after 48... xd6 49.bxa5 bxa5 50. f2 d5
51. e3 c5 (linear play!) 52. a3 ( 52. b8 c3+ 53. d2 d3+
54. e2 e5 ) 52... e5 53. b3 d5 54. a3 d3+.
Spassky panicked and played:
48. d4?
585
In fact, only now did White make the decisive mistake.
There was only one possibilty to escape: with the move 48. d1!.
But what happens if Black responds here with 48... d8...? At first
glance, it is difficult to discern the difference between the position
after Black’s move 48... d8 and the variations just discussed. White
still loses the d6-pawn, and the black rook will establish a ‘line’ by
... d5.
However, there is still one subtle detail here. After 49.bxa5 bxa5,

analysis diagram
White will direct his rook not to the open b-file with 50. b1?, which
loses, but with the move 50. f1!! he will occupy the open f-file,
which leads to a draw. It is unimaginable for a person guided by
logical thinking to grasp this difference, but the computer will see
this, because its move search algorithm is based on enumerating
variations.
Let’s figure out what’s going on. After 50. f1!! the game will most
likely develop this way: 50... xd6 51. f5 d3+ 52. g2 d5 53. f8.
Let’s remember this position.
586
 analysis diagram
Now let us see what happens after 50. b1? xd6 51. b5 d5
52. b8 d3+ 53. g2 e5 54. f8 and ‘feel the difference’!

analysis diagram
587
It lies in the black king’s ability to conquer the f4-square. Black
easily wins with 54... f3! followed by 55... f4 after any retreat of
the white rook. In the previous case, Black’s king was on the e6-
square, and White could prevent it from getting to f4. After
53... d2+ 54. g3 d3+ 55. g2 f3 56. e8+ d5 57. g8, it is a
dead draw. And after 53... e5 White responds with 54. f2 d2+ (
54... d4 55. e2 e3 56. e8 ) 55. f1 d4 56. f5 d3 57. d5+
e3 58. xg5, and again manages to keep the balance.
In the game, with 48. d4? , Spassky provided Black with a vitally
important tempo, which allowed him to bring the king to the key
square e5:
48... e5!

49. d1
If 49.d7 d8 50. d1 axb4 51. d2, Black wins beautifully by means
of 51...e3! 52. d1 e4! 53. d6 b5!. However, he can also dispense
with all the beauty and just play 51...b3. The game did not last much
longer:

588
49...axb4 50. b1 c3+ 51. f2 d3 52.d7 xd7 53. xb4 d6
54. e3 d3+ 55. e2 a3 0-1
As we have seen, Anatoly Karpov limited his opponents’
possibilities with prophylactic play, involving them in a
manoeuvring fight. Having achieved an advantage in a manoeuvring
battle, where he had no equal, he was in no hurry to change the
rhythm of the game in order to finish off his opponent with a direct
attack. Instead, he continued to maintain the tension in the position
and ask his opponent difficult questions. As a rule, his opponents,
having less strong ‘masonry’ skills, could not withstand this tension
and made positional errors. This is how Karpov managed to obtain
technically won positions, which he brought to victory with minimal
risk. With this manner of realizing his advantage, Anatoly Karpov

The ability to do everything

Of course, since the late 1970s and early 1980s, which marked the
peak of Karpov’s career, chess has changed significantly. A large
number of strong players have appeared in the world elite, whose
level of resistance has increased, and the time to think about moves
has decreased significantly.
Now it has become impossible to avoid a sharp transition to
dynamics, as Karpov often did. Today, when realizing an advantage,
a chess player needs to be able to do everything: limit the
opponent’s possibilities, hold a position without making sudden
movements, change the rhythm of the game in time, win better
endgames, or finish off the opponent in a tactical fight with precise
moves.
In my opinion, all these elements of the game can be clearly seen in
the games of Vladimir Kramnik
Slav Defence
Vladimir Kramnik (2809)

589
Alexander Morozevich (2742)
Dortmund 2001
1.d4 d5 2.c4 c6 3. f3 f6 4.e3 f5 5. c3 a6 6. d3 xd3 7. xd3
e6 8.0-0 e7 9.e4 0-0 10. d1 b5 11.c5 dxe4 12. xe4 xe4
13. xe4 d5 14. xd5 cxd5
White has played the opening without much ambition to gain an
advantage, and the game, having passed the middlegame phase, has
moved into an endgame-type position.

From White’s point of view, let’s start asking ourselves important

prophylactic questions and strive to limit our opponent’s options.
The first question is: ‘What does the opponent want?’
Answer: To bring the knight to the ideal blockade square c6.
Second question: ‘Which enemy annoys us the most, or who is our
enemy No. 1?’
Answer: The knight on c6.
Third question: ‘What should we do with it’?
Answer: Exchange it for our knight.
Such logical reasoning leads us to the move
590
15. f4!
with the threat 16. xb8. Black responds with
15... c6
and we play
16. e5!
Now, in the event of an exchange of knights on e5 with 16... xe5
17. xe5, White’s position should be preferred. His bishop will be
more active than the opponent’s, and on the queenside White will
have the opportunity to create pawn leverage with the moves b2-b4
and a2-a4. Let’s imagine that the pawns are located exactly like this.
In this case, doubling White’s rooks along the a-file will put Black
in an unpleasant situation.
With this development of events, play will be for two results. This
does not suit Morozevich, and he retreats his knight:
16... b8!?

Again we ask ourselves our favourite question: ‘What does the

opponent want if it were his move?’ Obviously, Black wants to

591
either exchange the e5-knight with the move 17... f6, or drive it
away with 17...f6 and still install his knight on c6. How should
White react?
17. d3!
Prophylaxis! A wonderful move by Kramnik, limiting the
opponent’s options.
Let’s try to logically justify this manoeuvre. The move 17... f6
cannot be prevented, but what are its disadvantages? It allows the
answer 18. g4!, and either White will exchange the bishop for a
knight, or after 18... c6 he will make the reverse exchange of
knight for bishop on the f6-square, deforming the opponent’s pawn
structure on the kingside.
Is it possible for Black to first play 17...h5, taking away the g4-
square from the white knight? In this case, he will have to reckon
with the possibility of 18. h3 h4 ( 18...g6 19.c6 ) 19.g3, and White
increases his activity on the kingside.
17... c8 18. e1! f6?
The correct idea of exchanging a knight for a bishop, but in the
wrong form. Black should have insisted on his move 18... d8!,
moving the bishop to c7. In this case, he could well count on
equalizing after something like 19.g4 h6 20.a3 c7 21.h4 xe5
22. xe5 c6 23. d6 e8 24.h5 f6. White has some initiative, but
breaking through Black’s defensive formation will be very difficult.
The computer offers another concrete solution for Black. Instead of
17... c8, Black can immediately play 17...f6!?.

592
 analysis diagram
Then the computer defends with 18. f3 f7 19. e1 c6 20. de3
xd4! 21. xd4 xc5 22. xe6 xe3 23. xe3 fc8! 24. c7 d4
25. a3 g5!.

593
 analysis diagram
With his last move, Black deprives the white bishop of control of
the c1-square, and the computer begins to draw bold zeros. But let’s
be objective: a human player is simply not able to calculate all these
variations at the board.
Let’s return to the events in the game. After Morozevich’s move
18... f6? the game developed as follows:
19. g4! c6 20. xf6+ gxf6

Black has finally positioned his knight on the excellent blockading

square c6, but his kingside pawn structure has become vulnerable to
attack by the white pieces. As we already know, doubled pawns are
weak in mobility. Kramnik finds a strong human manoeuvre, the
purpose of which is to provoke the movement of the black pawn on
f6:
21. g3!
The engine prefers 21. ee3. The point of White’s manoeuvre is to
transfer the bishop to h4.

594
21... g7?!
An inaccuracy not obvious to the human eye. The computer offers
here the solution 21...h6!!, which contradicts the human instinct of
self-preservation. Black, it turns out, needs to further weaken his
pawn position on the kingside. As they say, ‘the barn is burning
down, so let’s burn the house down too!’
It is surprising that after 22. h4 f5 23. g3+ h7, the black king
actively enters the game and neutralizes the activity of the white
bishop. And in the case of 22. f4 h7 23. ee3, Black is saved by
23...e5!. Most likely, White would have to play 23. d2 g8 (
23...e5? 24.dxe5 fxe5 25. ee3 ) 24. ee3 g6, and Black’s position
would be quite defensible, but the initiative is on White’s side.
But such an anti-positional move as 21...h6!! can only occur to a
person if he is sure that other moves lose. This example can be used
as an exercise to develop unconventional thinking.
22. h4 e7 23.g4 g6?!
Black misses his last real chance to create counterplay by 23...h5!.
24. g3 a5?
This fairly useless move is the decisive mistake. It was essential to
urgently return the knight to e7.
25. d6!
Now, Black is too late. The white bishop cuts off the knight and
keeps it out of play. The white pawn threatens to advance to f5.
25...h5
Black needed to do this two moves ago, but now the move cannot
save him.

595
26.gxh5!
Kramnik is faithful to his approach to realizing an advantage. Thus
far, he has ‘held’ the position and limited his opponent’s options
with precisely calibrated prophylactic moves. And at this point in
the game, he felt that it was time to look for a winning continuation.
White could have simply played 26.f3, but in such positions the
fourteenth World Champion was already moving from prophylaxis
to finishing off his opponent, regardless of sacrifices.
All this is very reminiscent of his actions in the game with Jan
Timman we saw earlier, although the positions themselves do not
have the slightest similarity.
26... h4 27. g3+ h6 28. e7!! f5 29. xf6 xg3 30.fxg3!

596
Specifically with this pawn! White needs the half-open f-file to
support the position of the dominant f6-bishop. The g- and h-files
are covered by white pawns which deprive the opponent of any
counterplay. Black’s position is hopeless, and the game lasted only
eight more moves:
30... g8
What does Black want? To transfer the rook to e4 via g4.
31. g2! xh5
If 31... g4 then 32. f3 is decisive.
32.h3! g6 33. f1 h7?!
The only way to prolong resistance was 33...e5.
34.g4 b4 35. g3 b3 36.a3 ac8 37.h4 g6 38.h5
Black resigned because White’s next few moves are going to be
39.g5, 40. g4, 41. e5 and 42. f6.
English Opening
Vladimir Kramnik (2785)

597
Georg Meier (2650)
Dortmund 2011
1. f3 f6 2.c4 e6 3.g3 a6 4. c3
Kramnik feels superior to his opponent in practical play and does
not want to check his computer analyses after 4. g2 b5. He is ready
to give up the fight for an advantage in the opening.
Kramnik is quite happy with an equal position in which the players
have to make decisions ‘with their own heads’ without the help of a
computer.
4...d5 5.cxd5 exd5 6. g2 e7
6...d4 7. a4+.
7.0-0 0-0 8.d4 c6 9. e5 a5 10. f4 f5 11. b3 a7 12. fd1 bd7
13. ac1 h6 14.a3 b6 15. xb6 xb6
White has not gained anything from the opening. On the board we
have a complicated middlegame without queens, with equal
chances. As Kramnik shrewdly pointed out in his masterclass, he
would not be particularly upset if he had to play this position as
Black.
16.f3
The only way for White to develop activity.
16... e8 17.e4 h7

598
Try to explain the essence of this position in words and guess the
move played by the 14th World Champion.
Answer: In his masterclass, Kramnik summarized the essence of the
position very succinctly: White's only trump is his centre and it is
essential to strengthen it.
18. e3!
When I give this problem to students in class, most often they rashly
rush forward, pushing the kingside pawns with 18.g4 and 19.h4. But
usually nothing good comes out of this idea for White.
I advise you to carefully follow the following three moves by
Kramnik:
18... aa8 19. f2! ad8 20. e1! f8 21.h3!
Almost all of these moves are the engine’s first choices.

599
With his last move, White prevented the enemy knight from coming
to g4, which was possible if the white knight left the square e5 for
d3 and pawns were exchanged on e4.
Kramnik’s masonry looks more powerful than that of his opponent.
The position is easier for White to play, but Black needs to carefully
check his moves. Although we are not yet beyond equilibrium,
Black is already close to the line beyond which problems begin. The
guidelines for Black are very blurry; he wants some kind of
certainty, and time to think about it is running out.
My Stockfish suggests that Black make two sharp pawn moves here:
21...g5 and answer 22.h4 with 22...gxh4 with equal play according
to the engine. But in order to play so sharply in a relatively calm
position with Kramnik in 2011, you had to be Anand (then World
Champion), Carlsen (the next) or at least Caruana.
21...dxe4 22.fxe4 fd7 23. d3
As we already know, with a space advantage it is better not to
exchange minor pieces.
23... c4 24. e2!
600
It was for this move that Kramnik gave the useful explanation that
we already know: ‘When you see such a useful, multifunctional
move that has only advantages, play it and don’t look for anything
better.’
Meanwhile, my engine again offers Black two sharp pawn attacks
on opposite flanks. Its first line is the move 24...g5, the second is
24...b5 . The choice is large, but as we remember, the right to
choose also means the right to make a mistake. Kramnik himself
pointed out the reasonable possibility of continuing to hold the
position with the move 24...f6, planning to transfer the bishop to f7
via g6.
Instead of the move 24...f6, which is difficult to think of, the
German grandmaster got carried away with his plans. Perhaps he
even asked himself the question of which of the pieces would be
useful for him to improve. His choice was the knight d7, and he
played:
24... b8?
The idea is reasonable. If Black had managed to play 25... a6,
everything would have been fine. But it takes two to tango.
25. a4!
With this knight jump to the edge of the board, White tips the scales
sharply in his favour. Black comes under a bind.
25... d6 26. dc5

601
Kramnik commented on this position: ‘ All White has done is
strengthen his position and monitor his opponent’s counterplay.
This is often enough.’
For the time being, Black, like White, stood still and held his
position. At the moment when Meier started actively transferring the
knight to b4 by 24... b8?, he made a serious mistake. I am
reminded of Mark Twain’s humorous expression: ‘It is better to
remain silent and be thought a fool than to open your mouth and
completely dispel any doubts.’ If we paraphrase it in relation to
chess, it will turn out: ‘Sometimes it is better to stand than to take
active measures, worsening your position.’
Kramnik’s final conclusion to the position on the diagram: ‘ Black
didn’t manage to stand firm .’
Let’s continue analysing the game.
26... e7 27. b6 c8 28. c4 b6 29. b3 a4 30. bd2
The white cavalry have separated the enemy’s queenside infantry
from the heavy guns deep in the rear.
30... b7
602
And now a rapid attack begins in the centre, which cuts the
opponent’s position into two parts. In our terrible times, I would
prefer to use military terminology only on the chessboard.
31.e5
The engine recommends first playing 31.g4! to take away the f5-
square from the enemy bishop and only then advance the central
pawn.
It is likely that the opponents had little time to think about their
moves.
31... a7?
It was necessary to first play 31... d3 and after 32. e1 respond with
32... f5, transferring the bishop to e6, and keeping defensive
chances. After the move 31... a7?. Black’s position immediately
becomes lost.
32.d5! c5?! 33. xc5 bxc5 34.e6! f8 35. a5
35. e5 looks even stronger but this is just a matter of taste.
35... c7 36. xc5 fxe6 37.dxc6 e7 38. dc4 d1+ 39. h2 d3
40. e3 xc4 41. xc4 d2
The time control is past and White ends the game with a direct
attack on the king.
42. ce4 bxc6 43. xe6+ f7 44. c4 c2 45. d6+ g8 46.h4
g5 47. f5 gxh4 48. h3 c5 49. g6+
Black resigned.
In the game Kramnik-Meier, we were convinced of how important it
is to be able to limit the opponent’s possibilities, ‘ run on the spot ’
and at the same time maintain a position, constantly strengthening it.
In the next two examples from the work of the fourteenth World
Champion, we will see how you can limit your opponent’s options
when carrying out active, wide actions across the entire board
Vladimir Kramnik (2625)

603
Mikhail Ulibin (2570)
Chalkidiki 1992

We are watching the beginning of a fight between two bishops and

two knights in a position in which each side has a reliable pawn
structure, without any weaknesses. When I gave this position to my
students, asking them to outline the immediate scheme of action for
White, most of them wanted to start a pawn attack in the centre.
Such a plan does not promise any special achievements for White,
as the e4-pawn often becomes vulnerable.
The young Vladimir Kramnik came up with a much more
interesting plan involving ‘hugs’: the side with the advantage of two
bishops ‘embraces’ the opponent’s position with pawns from both
flanks, conquering space.
21. fd1 h6 22.e3 d7 23. f1!
A quite unexpected move. If the opponent has a pair of rooks, it is
not always a good idea to bring one’s king to the centre.
23... e7 24. e2 de8

604
25.g4!?
The ‘hugs’ begin.
25... e6?!
An imperceptible error.
Black adheres to wait-and-see tactics, but in a game against two
bishops, this rarely leads to success. It would have been better to
start counterplay by 25... ec5!? and 26... e5. Black would thus
immediately begin to harass the white pawn that had moved
forward, and, if appropriate, the opponent’s king by moving the
knight to g6.
26.h4 6e7
Now guess Kramnik’s next move.
Answer: 27. f3!
Vladimir Kramnik’s masonry skills were already at the highest level
at 17. This multifunctional move with the bishop, once it has been
reproduced on the board, seems quite obvious. But finding it during

605
a game is not an easy task. Moreover, chess technique consists of
just such moves.
27... dc5 28. d5!

With his last move, White indicated his intention to advance the
pawn to g5 and open the game on the kingside. Of course, it made
no sense to investigate the consequences of the complications
arising after 28. g1?! xf2!.
28... a6?!
It was safer to reinforce his position with 28...f6. The engine even
suggests 28/29...g6, though it looks awful.
29.a3 ac5?!
This is already playing with fire and here too, he should have
preferred 29...f6.
30.b4 axb4 31.axb4 a6 32.b5 ac5 33. a1 f6
In this position, the computer suggests an exchange on f6. Of
course, after 34. xf6 gxf6 White has a clear positional advantage,
but taking Black’s fortress would not be easy, since the doubled
606
pawns on the kingside are good for defence. The future World
Champion preferred a more active plan:
34. dd1! fe4 35. a7 f8

36.g5!!
‘Hugging’ has turned into a breach of the enemy defences, with a
wonderful positional pawn sacrifice.
Associative thinking immediately reminded me of an excellent
analogy in the following game by Magnus Carlsen
Magnus Carlsen (2845)
Alexander Grischuk (2771)
Shamkir 2019

607
29. e3!
This pawn sacrifice is much more complicated than the one in the
Kramnik game, although the two are similar.
After 29...exf4 30.gxf4 xe4 31. b1 e7 32. fe1, the light-squared
bishop on b1 breaks through to g6 with great effect.
Now back to Kramnik -Ulibin
36...hxg5 37.hxg5 xg5 38. c6 c8 39. h1!

608
The two white pawns on the queenside completely paralyse the
three black pawns, whilst White can start a direct attack on the black
king on the kingside.
39... g8 40. aa1!
Taking advantage of his surplus in space, Kramnik quickly involves
the second rook in the attack. Five moves ago, this rook from a1
went forward to a7 to take control of the black pawn on c7. Now, it
is coming back and decisively strengthening White’s attack. Such
backward moves produce strong impressions and often fall outside
the opponent’s field of vision, as we saw in the games Andreikin-
Fedoseev, Aronian-Vidit, Böhm-Timman and Shuvalova-Maurizzi.
Now the black position is already defenceless and White
comfortably plays out the decisive attack.
40... ce6 41. h4 f6 42. g1 f7 43. d5 e8 44. h8+ f8 45.f4
ge6 46. f3 f5 47. g6 c5 48. xg7 f7 49. xf8
Black resigned.
Vladimir Kramnik (2772)
Nigel Short (2642)
609
Dresden Olympiad 2008

In this position, the incompleteness of Black’s Carlsbad pawn

structure immediately catches the eye. If he managed to move the
knight from c6 to e6, and put the pawn from c7 in the knight’s
place, then Black’s position would become stable and solid. But
now it’s White’s move, and Kramnik deprives his opponent of the
opportunity to carry out this manoeuvre:
13. g3!
Now, the exchange of queens on the g3-square is unfavourable for
Black, since after 13... xg3? 14.hxg3 White will cramp his
opponent on the kingside by playing g3-g4.
13... fd8 14.b4! g6 15. a4 c8 16. c1 a7
Black’s position is unpleasant. He is under pressure from the white
pieces on the queenside. Nigel Short reveals a clear intention to
advance his pawn to b5 and then direct his knight to c8-b6-c4.
17.0-0!?

610
Kramnik decides not to stop his opponent’s plan, having noticed a
tactical operation which transforms one type of advantage into
another.
17...b5 18. xd6 xd6

19. c3!
Technique is the mastery of short-range tactics! This is the
manoeuvre Kramnik had prepared when castling two moves ago,
rather than the tempting move 19. c5 – although even after 19. c5
Black would have faced an unpleasant fight to equalize. Now White
threatens to take the d5-pawn with 20. xd5.
19... 6e7

611
20. xb5!?
A curious situation. A highly experienced chess player with
excellent endgame technique, who at that time was not inferior to
Magnus Carlsen in the ability to squeeze blood out of a stone,
abandons a pleasant technical game for two results after a line like
20.a4!? c6 21. a1 dd7 22.g4!. Instead, he decides to dramatically
change the pattern of the game.
The modern computer, on the contrary, prefers the move 20.a4!?,
estimating White’s advantage in this case to be almost twice as large
as after 20. xb5!?.
20...axb5 21. xb5 da6 22. xa7 xa7 23. c5! b6
23... xa3 was more tenacious.
24. a1 f8

612
25.a4!
Again, a short-range tactic!
25... e8
Black cannot take the pawn with 25... xa4? because of 26. a5.
26.a5 c4 27. a2
NIn the endgame, one must not forget about centralizing the king.
But the immediate 27. f1? is bad because of 27... d2+ and
28... b3.
Better, however, was the forceful 28.b5 or 28.a6, since after the text,
27...c6, now or on one of the next moves, almost equalizes.
27...f5 28. f1 d7 29. e2 d6 30.b5!
White finally exploits a simple tactic to deprive his opponent of the
possibility of ...c7-c6. The move 30... b7 would be bad because of
31. xc7+! xc7 32.b6+ b8 33.bxa7+ xa7 34. c2.
30... c4 31. a4!

613
Kramnik realizes his advantage with relentless precision. Now
Black cannot take the a5-pawn with the rook because of 32. cxc4,
nor with the knight in view of 32. c5-c2-a2.
31... d8 32.a6
Black was now threatening to take on a5 with the knight, so as to
answer 33. c2 with 33... 5c6.
32... b6 33. a1 ec8

On the queenside, Black has fought back somehow. It’s time for
White to expand his play.
According to the principle of two weaknesses, Kramnik proceeds to
open a second front on the kingside:
34.h3!
It is clear that White has already outplayed his opponent, and his
advantage is not far from decisive. But even in this position,
Kramnik vigilantly monitored his opponent’s counterplay.
At the next online session of his school, he suggested that young
chess players find opportunities for counterplay for Black. Nobody

614
managed to find the idea 34... d6! 35.g4 d7!? 36. xd5 e7! and
37... e6 with the capture of the rook on d5. It is another thing that
it no longer ‘heals’ Black much, since after 37.gxf5 gxf5 38. c1
e6 39. xd6+ xd6 40. c6+ e7 41. d3, he is absolutely lost:

analysis diagram
This idea for Black does not appear in my engine’s top three lines.
34...h5 35.g4! hxg4 36.hxg4 fxg4 37. g1 e7 38. xg4 a4

615
39. h4!
Again, short-range tactics in the service of technique! The move
39. c1? could throw away White’s advantage after 39...c6! 40.bxc6
xa6 41.c7+ c8 42. h4 c6 and Black is close to a draw.
39...c6
If 39... xc5 40. h8+ d7 41.dxc5, Black’s best practical chance is
41... c6! (41... f5 42.b6 ), but after 42. d2! e6 43. c3 e7
44. b4 c6+ 45.bxc6 xa6 46. h7! the game ends in a victory for
White.
40. h8+ c7 41. c2 b6 42. h7 b8
Or 42... d6 43. xe7 xe7 44. xc6.
43.bxc6 c7 44. f3! a7 45. a2 bc8 46. b2 xa6

616
47. h1!
A beautiful finish to an excellent game, and a mirror-image of the
manoeuvre 40. a7-a1 in Kramnik- Ulibin.
Should we play widely or narrow the game?
In the games Kramnik-Ulibin and Kramnik-Short, we saw how the
fourteenth World Champion seized space, limited his opponent’s
options across the board, and, having achieved a decisive advantage,
launched a direct attack on the king. The final chord in both cases
was a beautiful transfer of the white rook from one flank to the
other. As we have already noted, in these games Kramnik played
widely all over the board .
But other situations arise when you have a serious advantage on one
of the flanks. In this case, it is very important to ask yourself the
following non-standard question: ‘ Is my advantage sufficient to win
the game on this flank? ’ If the answer is positive, you do not have
to conduct an offensive on another part of the board. If we draw an
analogy with the art of war, then a side with an overwhelming
advantage in one of the sectors of the front has no reason to launch

617
an offensive on the other at the same time . It is logical to first win
one victory, and then, with combined forces, achieve the second –
the final one.
The same thing happens in chess. If you are confident of victory on
one of the flanks, narrow the front! The key word here is ‘ narrow ’.
I really like this term, which I first heard from grandmaster Evgeny
Tomashevsky, with whom I work in Azerbaijan. When analysing
one of Aydin’s Suleymanli games, Evgeny told the young
grandmaster: ‘ Here you needed to narrow the game! ’ Let’s take a
look at a fragment from this game.
Vasif Durarbeyli (2618)
Aydin Suleymanli (2543)
Azerbaijan Championship, Baku 2023

Black has a large positional advantage due to the better placement of

his pieces and White’s doubled isolated pawns along the e-file.
Black needs to outline a course of action to realize his advantage.
Briefly, this scheme can be described as follows: Black needs to
exchange rooks and at least one pawn on the queenside . And this

618
needs to be done quickly so that the white king and knight do not
arrive in time to defend. That is, Black should ‘ narrow the game ’
to the size of the queenside, and not pay attention to the kingside for
now.
Here is an illustrative possible variation of such a course of action:
29... d3! 30. e2 c5 ( 30... d6! ) 31. xd8 xd8 32.c4 d7 33. c3
c6! 34. f1 b5 35.cxb5 ( 35. e2 e5 ) 35...axb5 36. e2 e5
37. d2 d6 38.a3 c4 39.b4 d3 40.a4 bxa4 41. xa4 e5 42. b6
xb4 43. xc4+ xe4 44. e2 c5

analysis diagram
with an easily winning position for Black.
Instead of this, Aydin played a less active move:
29... e6
The game continued as follows:
30. xd8 xd8 31. f1 d6 32. h1!
Black did not foresee this defensive resource. Such moves often
escape a player’s attention.
619
32... e5
A natural move. Black could have tried to transfer the knight to c5
with the move 32... f7!?, so as to direct play along the
aforementioned route. His aim would be to exchange a pair of
pawns on the queenside. After 33. f2 e5, any human player
would play 34. e2?!. The modern computer shows that in this case,
White would lose after 34...b5.
I do not wish to delve any deeper into these variations. The game
ended in a prosaic draw:
33. f2 h5?! 34. e2 e6 35.h4 g5 36.g4 g7 37.gxh5 xh5
38.hxg5 fxg5 39. f3 f6 40. d3+ e6

620
The two players’ knights occupy ideal positions behind and to the
side of their passed pawns.
41. c5+ e5 42. d3+ ½-½
I would ask you to pay special attention to the position which arises
after White’s move 33. f2.

621
Evgeny Tomashevsky did not like Suleymanli’s ‘wide’ move
33...h5?! , since Black’s play should be on the queenside. After
Black’s straightforward play with 33... e6 34. e2 c5 35. f3 a5
36.h4 b5 37.b4 axb4 38.cxb4 a4 39. d3+ d6 40. f4 c5
41.bxc5+ xc5 42. xc5 xc5 43.e5 d5 44.exf6 e6 45.f7 xf7
46. e5, a pawn endgame arises with a distant passed pawn for
Black and a strange -1.44 assessment from my computer at depth
40. Instead of the move 36...b5 , Black could also play 36...a4 37.b4
d7, preserving all chances of winning. Tomashevsky’s expression
‘ Narrow the game !’ related primarily to this moment.
It is important to note the following: the main goal of narrowing the
game is to make it easier for yourself to realize an advantage. And
the means to achieve this goal may be different. When choosing a
method, you should keep in mind Capablanca’s catchphrase: ‘ The
most beautiful way to win is the simplest !’
Let’s take a look at a couple of ways to narrow the game.
So, for example, the end of the game Flohr-Bondarevsky , which
we discussed at the beginning of this book, was preceded by the
following position:
622
If we compare the activity of the rooks, then there cannot be any
argument – the white rook is the more active. But Flohr exchanged
it for its passive opposite number:
62. xe8!
By narrowing the play, he reduced to a minimum the possibilities
for active play by his opponent. The simplification of the position
led to a winning minor-piece endgame for White.
In the next position, Boris Gelfand
Boris Gelfand (2693)
Daniel Campora (2549)
Izmir 2004

623
He narrowed the game and thereby acted in full accordance with the
requirements for solving the problem of exchange for the stronger
side, who has an advantage in space:
26. xc7 xc7 27. xc7 xc7 28. c1 xc1 29. xc1
Gelfand’s detailed comments on the ending of this game can be
found in his book.
As we have seen, depending on the specific situation, you can
narrow a game in different ways: lock one of the flanks, exchange
your active piece for a passive opponent, or give up an open file that
you control, in order to exchange heavy pieces, etc. The main thing
is that narrowing the game helps make it easier to realize an
advantage. The final phase of realizing an advantage most often
occurs in the endgame.
Let’s now look at two examples on the topic of narrowing the game
from the practice of the endgame virtuoso, Swedish grandmaster Ulf
Andersson.
English Opening

624
Ulf Andersson (2560)
Michael Stean (2540)
Sao Paulo 1979
1.c4 f6 2. f3 c5 3.g3 d5 4.cxd5 xd5 5. g2 c6 6.d4 cxd4
7. xd4 db4 8. xc6 xd1+ 9. xd1 xc6 10. xc6+ bxc6
11. c3 e5 12. e3 h5 13.h4 b4 14. d2 e7 15. hc1 d8+
16. e1 e6 17.a3 xc3+ 18. xc3

From the opening, the game has moved into a pleasant endgame for
White. His advantage is quite insignificant, and it lies in the better
pawn position on the queenside. With opposite-coloured bishops on
the board, this is a small advantage, and in those days in such
positions players often agreed to a draw. Today, with strong opening
preparation among grandmasters, even a slight advantage in an
unprepared position is considered a success on emerging from the
opening. Not trying to make use of it is an unaffordable luxury. To
realize such an advantage, in my opinion, it is necessary to have
strong masonry, patience, psychological stability and strong-willed

625
qualities. Ulf Andersson had all this, which is why he liked to play
such positions 45 years ago.
18... d5 19.b4!

If Black succeeds in pushing his a-pawn to a5 (or, even better, a4),

or exchanging it for the white b-pawn, then White’s winning
chances will be sharply reduced. Therefore White’s first task is to
prevent this freeing move.
19... db8
Black banks on advancing the a-pawn, which does not work
immediately because of 19...a5? 20.bxa5 xa5?? 21. b6.
Now what should White play? On 20. c5 there follows 20... b5!,
and the exchange of one pair of rooks can be counted a success for
Black.
20. d2!
Prophylaxis! If 20...a5?, White replies 21.bxa5 xa5 22. e3,
winning a pawn.
20... e6
626
What should White do next? He could, of course, simply withdraw
the rook from c3. But Andersson seeks to deprive his opponent of
counterplay on the queenside in the shortest possible way. To do
this, White needs to place the rook on c5 and take away the square
b5 from the enemy with the move a3-a4. Therefore he plays:
21. c5! b5 22. cc1!
A nice manoeuvre! Now the white pawn on a3 will advance with
tempo. As we see, imagination can be shown even in such ‘bland’
situations. The Swedish endgame virtuoso was probably not at all
bored thinking about the position; he fought with full concentration,
and his opponent soon had cause to feel sad.
22...f6 23.a4 b7 24. c5 g8?! 25.f3!

627
White has tactically prevented the opponent’s counterplay
associated with the move 25...g5. Black is threatened by 26.e4 b3
27. a3 with the capture of the bishop.
Why did I decide to give this position a diagram? Because it reflects
a critical moment in the game, which arose completely
unexpectedly. Black could not allow the move e2-e4, and the
English grandmaster defended against it with the move 25...f5?!,
after which his position began to noticeably deteriorate.
As we know, when defending a worse position, the most difficult
thing for the weaker side is to choose the moment when it is
necessary to part with material, most often a pawn, in order to
activate its forces.
The engine believes that such a moment has arrived here and that
Stean needed to sacrifice a pawn with the move 25...e4, with the
idea of 26... e3. Then the game could have continued 26.fxe4 xe4
27. xh5 d5. It is much easier for Black to defend such a position
than the one with equal material that occurred in the game. My
engine at a depth of 45 estimates White’s advantage +0.2, a very

628
modest assessment, although in such positions it does not matter
much. It all depends on the quality of the opponents’ further moves.
25...f5? 26. f2 d8 27. c3 e4
The threat was 28.e4.

White has managed to weaken Black’s strong pawn column e5/f6/g7

with the help of the threat of e2-e4. Like a magnet, he has pulled the
enemy kingside pawns into his camp and placed them on light
squares. Now the difference in the strength of the opposite-coloured
bishops is striking.
But how can White realize his great positional advantage? Let’s ask
ourselves the question we already know: ‘ Is my advantage
sufficient to win the game in a narrow sector of the front? ’ In this
case, we are talking about the queenside. To answer this, let’s try to
outline a new course of action. If White managed to place his rooks
on a5 and a6, move the bishop to c5 or d4, and use the king to
protect the b4-pawn from c3 or a3, then Black would not be able to
hold the a7-pawn. With the black king on a8 and the rooks on b7
and d7, Black’s position will collapse after the white pawn

629
advances: b4-b5. The kingside does not participate in this scheme.
In order not to run into the opponent’s counterplay, it is better to
close the kingside by placing the pawns on e3 and f4:

schematic position
After the move b4-b5, White’s position is winning. The assessment
of my computer is +6.2.
In the course of a logical analysis of the position, we have identified
the necessary piece set-up and realized that if it can be achieved,
then we need to narrow the play by closing the kingside.
The Swedish GM, however, played:
28. d1
How should Black play now? The first line of my computer is
28...exf3!? 29.exf3 bd7.
There could be many reasons why Andersson didn’t immediately
play 28.f4!?. Perhaps he had not yet found the final winning plan, or
he was thinking through the details of its implementation. But, most

630
likely, the Swedish grandmaster felt that even after the white pawn
had advanced to the f4-square, Black could find a defensive set-up.
Let’s try to find it out. If Black divines White’s plan and makes the
move ...a7-a6 in time, preventing White’s rooks from being
transferred to the squares a5 and a6, then it will be difficult to break
through his defensive fortifications. To do this, White will again
need the help of his opponent.
So now he needs to try to win the game by cunning, using the ‘Do
not hurry’ principle. He has to disguise the key move f3-f4 by other
insignificant but useful moves, in order to dull the English GM’s
vigilance and divert his attention from the emergence of the main
critical moment of the entire ending.
28... dd7?! 29. e3 g6?!
29...exf3!?.
30.f4!?
Finally!
30... d6?!

631
Another interesting point.
White had the sharp move 31.b5! at his disposal, which Black could
have prevented by playing 30...a6 earlier. The engine likes 31.b5!,
but it is a very responsible decision. After all, White would
immediately rid his opponent of the weak pawn on c6 and abandon
his original plan. But purely intuitively, it is very difficult for a
person to understand whether capturing a6 with one’s rook will be
enough to win. Black’s defence, in this case, would be quite
difficult. For clarity, I will give one unforced variation: 31...cxb5
32.axb5 dd7 33. e5 d8 34. a1 f7 35. a6 e6 36. cc6 e8
37. a5!? c8 38. d4.

analysis diagram
Now the exchange 38... xc6? 39.bxc6 c7 40. c5 is bad for Black,
but after 38... cc7 my engine score is around +2.0, although the
fight is not over yet.
Instead of 31.b5!, Andersson made the planned move:
31. b1!?

632
He judged his opponent well and rightly believed that Stean would
not be able to grasp the critical moment that had just arisen.

31... dd7?
Now the white plan will be implemented without any hindrance. We
can only observe with what precision and artistry Andersson carried
it out. As on the previous move, Black should have played 31...a6 !,
but whereas earlier he had a choice, now it was obligatory.
Why did the English grandmaster make the decisive mistake so
easily? Perhaps he was already under time pressure, but I think the
reason is different. It’s always difficult to do double duty: first find a
winning plan for your opponent and then think about how to
neutralize it. Not everyone has the patience to do this. In addition,
you must have a certain gift of foresight.
The principle ‘Do not hurry!’ has done its job. Black did not
succeed in seeing the point of the moves 28. d1, 29. e3, 30.f4 and
31. b1. In fact, White’s first two moves in this sequence were
useful, but insignificant; their goal was to disguise the main idea,
which was precisely connected with the move 30.f4, which

633
narrowed the game. After Black’s unsuccessful reaction 30... d6,
White could play 31.b5 and get an almost won position. Black’s
31... dd7 was a decisive mistake.
All that remains is to add that after 31...a6! instead of 31... dd7,
Black’s position would have been completely defensible. For
example: 32. a5 a7 33. a1 ( 33.b5 cxb5 34.axb5 b7 ) 33... a8!
34. d4 d7 35. c5 b7. Black holds the position with all his
strength, but his defensive resources are enough to withstand the
opponent’s onslaught. I think that in this case, Andersson would still
have won this game, although the move 31. b1!? itself contained a
considerable amount of risk. After all, if Black had played 31...a6!,
then the computer’s score would have immediately dropped to +0.7.
32. d4!

Now, 32...a6 no longer helps Black, since after 33. a5 c4 34. c5

b8 35. c1 he loses the pawn on a6.
32... f7 33. b2 e6 34. d2 f7 35. c3 e6 36. a5 f7
37. a6 e7 38. c5 ed7 39. d2 bc7 40. e3 b7 41. c5 bc7

634
42. c2!
The final trick. Before transferring the king to a3 to defend the b4-
pawn, White needs to take control of the square c4.
42... b7 43. b2 dc7 44. a3 d7 45.e3 dc7 46. d4 d7
47. c5 bc7 48. b2!
Do not hurry!
48... b7 49. c3 b6 50. ca5 xa6 51. xa6 a2 52. xc6 1-0
Another example by Andersson
Ulf Andersson (2600)
Robert Hübner (2640)
Johannesburg 1981

635
Again, the game goes straight from the opening into an endgame:
15. xc6+ xc6 16. g5 d5 17. xd5 exd5 18. xe7 xe7
19. fc1 xc1+ 20. xc1 d6

The position is equal, but not yet drawn.

636
21.f3!?
Black has a choice between active and passive defence. Active
defence here involves fighting for space on both the kingside and
the queenside. Black has to make active pawn moves, such as ...f7-
f5 or ...c7-c5. At this moment, the German grandmaster might have
asked himself: ‘Do I need to play actively? Isn’t it easier to stand
still without making sudden movements?’
21...c6
Hübner chooses the ‘lazy’ path, which perhaps seemed the most
reliable to him. However, as we have seen many times, passive
defence in a rook endgame is often the riskiest strategy if the
defending side fails to build a fortress.
It would have been better to play 21... c8!? with the idea of 22...
c5!. In response to 22. c5, Black should have been active again by
22...f5! 23. f2 c6 24. c3 b4! 25. c5 c7 with a likely draw.
22. f2 e8
Continuing the same tactics. It was more sensible to make the active
move 22...f5.
23.g4! a5 24.h4 g6 25. c2

637
Andersson starts to ask his opponent some definite questions. White
wants to bring his king to f4, so as then to seize space with e2-e4.
Black needs to react somehow.
25...h6 26. g3 g5?!
Black wrongly fears 27. f4. Because he has hitherto chosen
waiting tactics, he would have done better to stick to them for the
time being with 26... e3. After the possible continuation 27. f4
e6 28.e4 dxe4 29.fxe4 f6+ 30. e3 f1 31.h5 e1+ 32. d3 d1+
33. c3 h1, he would not have been risking anything.

638
Now try to guess Andersson’s next move.
Answer: 27.h5!
Narrowing the play!
Why did White voluntarily close the kingside and give up pawn
leverage? The point is that he would not be able to use this leverage
due to the vulnerability of the e2-pawn. If 27. f2?!, Black manages
to play 27...gxh4! 28. c1 h5 29.g5 g8 30.f4 f6 with a draw. After
the text, a potential weakness has appeared in Black’s position – the
h6-pawn! It’s not entirely obvious yet, but the Swedish grandmaster
has already outlined an active scheme of action to use his advantage
in space.
27...f6?!
An unforced concession. It was better not to occupy this square with
a pawn, but to play 27... e6 or 27... c7.
28. f2

639
Black’s passive play has led him into a dangerous situation. White
has a clear plan to strengthen his position: put the pawn on e3, the
king on d3 and the rook on c5. This is followed by a breakthrough in
the centre with e3-e4. Black can no longer stand still. For example:
28... d7 29.e3 c7 30. e2 d7 (active defence with 30...a4 gave
better chances to hold) 31. d3 c7 32. c5 d6 33.e4 b8

640
 analysis diagram
34.exd5! cxd5 35. c3 e8 36. d2 b8 37.b3, and Black loses
because of zugzwang. The weakness of the h6-pawn plays a key
role in White’s victory, and that was the weakness Andersson
creating with his ‘narrowing’ move 27.h5!.
Black needs urgently to find a defensive stance. At first, Hübner
follows the correct course:
28... c7! 29.e3 b6 30. e2 c8 31. c5 e8 32. d3 e7 33.e4
d7!

641
Black has found the right defensive set-up. The main question in
this position is whether White will be able to reinforce his rook on
c5 with the pawn from b4 . In the event of the exchange 34.exd5,
Black intends to take back with the rook: 34... xd5!. The pawn
ending after 35. e4 xc5 36.dxc5+ xc5 37. f5 d6 38. xf6 c5
39. g6 e6 40. xh6 f6 41. h7 f7 42.h6 b4 43.f4 gxf4 44.g5
f3 45.g6+ e7 46.g7 f2 47.g8 f1 becomes a queen ending
which my engine assesses as a draw.
These are the kind of heroic exertions required of Black to save the
game, which is the consequence of Hübner’s choice of passive
defence.
If we imagine that after 34... xd5! the white rook was defended by
the pawn on b4, then 35. e4 would decide the game. Jumping
ahead, we can say that this is what eventually happens.
34.a3

642
34... d8?
Here, Hübner departs from the correct path. His position at first
seems solid enough. Of course, Black should have played 34...a4!.
35. c1 d7
By now it was too late for 35...a4 due to 36.exd5 xd5 ( 36...cxd5
37. c5 and c8) 37. e4.
36. c2 d8 37. c3 d7 38. c2 d8 39. c1 d7 40. c5?! d8?!
Here, 40...a4 was possible again since White’s rook is immediately
attacked after 41.exd5 xd5.

643
The time control has passed. With time to think, Andersson played
41.b4!
... and conducted the remaining technical phase very confidently.
41...axb4 42.axb4 d7 43.exd5 xd5 44. e4 d8

644
45.d5!
This decisive breakthrough became possible thanks to the fact that
White’s rook is defended by the pawn on b4.
45... e8+ 46. f5 cxd5 47. xd5 e3 48. xf6 xf3+ 49. g7 f4
50. d6+ c7 51. xh6 1-0
In both these games, we observed a successful narrowing of White’s
play on the kingside. At the same time, White set difficult tasks for
his opponents, which they could not cope with. However, it is
important to remember that narrowing the play can backfire on you,
because it will deny you the opportunity to play on the flank you are
covering.
If we compare the games of Kramnik with the games of Andersson,
we discover the main difference between them. Kramnik expanded
the game and won the game using the ‘Two Weaknesses Principle’,
while Andersson narrowed the game, closed one of the flanks and
thereby deprived himself of the opportunity to play according to this
principle. Thus, it may seem that the method of narrowing the game
contradicts the ‘Two Weaknesses Principle’. However, it does not.
Rather, the narrowing method complements the principle of two
weaknesses, because if you correctly assess your ability to achieve
victory on one of the flanks, you will make the win much easier for
yourself.
In this large chapter, we have repeatedly emphasized that the main
means of increasing one’s advantage is to limit the opponent’s
possibilities, or prophylaxis. We analysed the ninth World
Champion
Petrosian’s prophylactic techniques. Importantly, in the process of
increasing his advantage to decisive proportions, a player should
strive to minimize the number of difficult decisions (on the wise
advice of Boris Gelfand). It is very important to ‘feel’ when the
advantage has reached decisive proportions. At this moment, it is
necessary either to narrow the game and realize your advantage
645
technically, or vice versa – to sharply change the rhythm of the
game and brutally finish off the opponent in the arising
complications. Today, a chess player must be able to do everything.

646
Chapter 10
The difference
between associative
thinking and play in
typical positions
Throughout this book, we have gradually replenished the baggage of
images and techniques for associative thinking, which will become a
reliable companion for you during play. With well-developed
associative thinking, it will be easier for you to grasp the essence of
a position, find the necessary analogy, and find the optimal solution.
If you remember, at the beginning of this book we indicated that
associative thinking is not at all the same as playing standard
positions. Associative thinking is based on memories arising from a
familiar image, which contains the logic of actions in a similar
situation. This could be regrouping the pieces, changing the rhythm
of the game, using prophylactic techniques and much more.
Playing in standard positions, in turn, is based on memorizing only
one or two methods of action in positions that are completely similar
in appearance.
How does associative thinking work? When we analysed the game
Réti-Rubinstein, Gothenburg 1920, we turned to Gelfand’s paraary:

647
30... e6?
‘Rubinstein’s only bad move in this game, allowing White to resist.
30... d7! was stronger, fixing the white pawns on the black squares.
Here they will not be the target of an attack from the bishop’s side,
but they will also not be able to prevent the black king from
penetrating the enemy’s camp, which is the biggest problem for
White.’ (Gelfand)
When I first saw this para, I immediately grasped the essence of
Rubinstein’s mistake. Black needed to play 30... d7!, to leave the
opponent’s pawns on the kingside on dark squares, and on the
queenside on light squares.
Associative thinking immediately suggested to me a similar
situation in the game Fischer-Taimanovfrom their Candidates
match in Vancouver, 1971, analysed in the same chapter (No. 2).
With the help of associative thinking, you can better understand the
logic of the actions of great chess players.

648
After all, Fischer literally persuaded Taimanov: ‘Please place the
kingside pawns on the light squares, otherwise I will arrange a pawn
clamp on this part of the board.’
Having recognized the image, we were able to verbally explain what
the stronger side should strive for in positions when the opponent’s
pawns are on squares of the same colour or squares of different
colours on opposite flanks. This association inspired us to analyse
this endgame in detail in Chapter 2, devoted to the interaction of the
knight and pawn.
Now I would like to invite you to consider a little-known game in
the Carlsbad Queen’s Gambit structure, played forty years ago by a
student of mine, international master Alexander Ryskin
Queen’s Gambit Declined
Alexander Ryskin (2440)
Alexey Deiko
Belarus Championship 1984
1.d4 d5 2.c4 e6 3. c3 f6 4.cxd5 exd5 5. g5 c6 6.e3 e7 7. d3
0-0 8. ge2 bd7 9.0-0 e8 10. c2 f8 11.f3
649
The plan involving a central pawn advance was first seen at the top
level in the game Botvinnik-Keres, Moscow 1952. Today it is one of
the most unpleasant lines for Black.
11... h5 12. xe7 xe7 13.e4 e6?!
Black intends to blockade the white pawns in the centre, but this
plan is unfortunate. It was better to exchange on e4.
14.e5 g6 15.f4 g7 16. g3 f5

Black has managed to stop the advance of the enemy pawns on the
kingside. In case of 17.exf6 xf6, his position is even preferable.
But White is able to transfer active operations to the other flank,
utilizing his space advantage. Black’s bishop and knights are
cramped on the last three ranks, with only one good blockade
square, e6, available among them.
17. a4 ac8 18.a3 b6 19.b4 d7 20. c3 fe6 21. ge2

650
Until now, the players have acted logically. Some moves could
probably have been improved, but let’s not quibble. Now Black
faces the threat of 22.b5, after which his pawn structure will
noticeably worsen. We can consider that one of the critical moments
of the game has already arrived.
21... ed8?!
After this natural move, White reacted correctly with the direct:
22.b5!
Black should have first mitigated the negative consequences of this
undermining of his pawn chain with 21... c7!. The game could then
have developed something like this: 22. b3 ge6 23.b5?! c5!
24. xd5 xd5 25. xd5 ed8, and Black’s situation would have
improved noticeably. White, of course, should not have rushed with
23.b5, but then Black would have established the interaction of his
forces by 23... ed8 and 24... e8.
Let’s return to the events in the game. Deiko played:
22... e8 23. b3 b7 24.bxc6 xc6

651
What associations can this position suggest to us? We immediately
recall the endgame Lasker - Capablanca , Havana 1921, examined
in Chapter 3, ‘To hurry or not to hurry’:

But in order to get a similar pawn structure on the queenside (with

reversed colours) in the game Ryskin-Deiko, the white a-pawn must
652
reach a5 and be exchanged for the black pawn on b6. Then Black’s
position would suffer from weak pawns on b6 and d5.
If we draw analogies further, then, to complete the comparison with
the game Lasker-Capablanca, the white pawns would need to be on
the squares e3 and f2. But their advance to the squares e5 and f4 has
its pros and cons, whereas there are no pros in the location of the
black pawns on b6 and d5. Therefore, White’s options were 25.a4 or
25. fb1.
Alexander Ryskin chooses another interesting path:
25. ac1!
If there is an advantage in space, the exchange of heavy pieces can
be advantageous.
25...a6 26. a2! b5 27. b4! a8

Ask yourself our favourite prophylactic question: ‘What is Black’s

next move going to be?’.
I am immediately struck by an association with this game:
Boris Spassky
Tigran Petrosian
653
World Championship Match, Moscow 1969

Petrosian, of course, played

30... c4!
and won beautifully.
This means that Black wants to move the bishop to b7 and sacrifice
the exchange with 29... c4!. And no matter how much Stockfish
tries to convince me that 28. f2 is the strongest move here, I
believe that the human player should play 28. xc6!, as did
Alexander Ryskin
What does this exchange remind you of? I hope you recall the
famous game Fischer-Petrosian

654
Here the future eleventh World Champion carried out his famous
exchange of a handsome knight for the opponent’s passive bishop
with 22. xd7+!.
By the way, while I was writing these lines, my computer changed
its mind and at a depth of 47 elevated Ryskin
28. xc6!
after which Black’s position collapsed in just a few moves:
28... xc6 29. xc6 xc6 30. b4 c8 31. c1 d7 32. xc8+ xc8
33. f2
White must be attentive to blows such as 33... xf4 and 34... c1+.
33... d8 34.g3
Black was threatening the check 34... h4+.
34... f7?
This move leads to immediate defeat, but I find it hard to imagine
that after the engine recommendation 34... c7! 35.a4 ge6 36. e3
e8 37. c3 Black will succeed in holding the position.
The game continued:
655
35.a4!

35... b6 36.axb5 c7 37. c3 ge6 38. a4 xd4+ 39. xd4

xd4 40.b6 ce6 41. xa6 d8 42. c3 g5 43. c8 gxf4 44.gxf4
4e6 45. xd5 1-0
The associations that helped us explain certain decisions of
Alexander Ryskin referred us to positions that had no external
resemblance to the positions in this game . The possibility of
creating pawn weaknesses, exchanging an excellent knight for a bad
bishop, preventing an unpleasant exchange sacrifice – all these
methods of play were remembered primarily for the logic of their
actions .
The situation is completely different when playing in standard
positions, which are similar to each other, like twins. Here, it is
enough to know the correct method of action and apply it in all
cases. Now we will see this with two examples.
King’s Indian Defence
Lev Polugaevsky (2585)

656
Wolfgang Uhlmann (2434)
Amsterdam 1970
1.c4 f6 2. c3 g6 3.e4 d6 4.d4 g7 5. e2 0-0 6. g5 c5 7.d5 e6
8. d2 exd5 9.exd5

Let’s not comment on the opening part of this game of fifty years
ago and just consider the position. White has an obvious space
advantage thanks to the d5-pawn. Black’s trump card is some
advantage in development. But since an advantage in development
is a temporary factor, and an advantage in space is permanent, there
is a strategic risk for Black to get squeezed.
As we know, in a cramped position it is necessary to strive to
exchange minor pieces. If Black makes a series of natural moves
that are not subordinated to this goal, then he will find himself in an
unpleasant situation. For example: 9... e8 10. f3 a6 11.0-0 c7
12.h3 d7 13.a4 e7 14. d3 f8 15.g4 h6 16. f4, and White’s
positional advantage is beyond doubt. Therefore, nowadays Black
strives to exchange the knight on f6 with the help of the attack
... f6-e4. He seeks happiness in complications like 9... b6 10. f3

657
 f5 11. h4 e4 12. xe4 xe4 13.f3 h6! ( 13... xb2? 14. c1
clearly favours White) 14. xh6 xh6 15. xh6 xb2 16. c1 c2,

analysis diagram
and my computer assesses this position as roughly equal. In those
far-off days, when there were no computers and analysis had to be
done by hand, theory did not extend so far.
In our game, Uhlmann played according to the theory of the time:
9... e8 10. f3 g4 11.0-0 bd7 12.h3 xf3 13. xf3 a6 14.a4

658
And the game heads directly to an endgame after the exchange of all
the heavy pieces on the e-file:
14... e7 15. ae1 f8 16. d1 xe1 17. xe1 e8 18. xe8 xe8
19. c2 b6 20.b3 bd7 21. f4 e7 22. e2 f8 23. xe7+
xe7

659
A position has arisen in which White has the two-bishop advantage.
We have already touched on this topic in the game Kramnik-Ulibin
and noted that the main game plan for the stronger side to win is to
seize space with ‘hugs’ – covering the opponent’s position with
extreme pawns from the flanks.
For White now the mandatory move is 24.a5!. If it were Black’s
move, he would himself advance the a6-pawn to this square, after
which the probability of a draw would increase sharply.
We present the rest of this game almost without comment:
24.a5! h5 25. d2 e8 26.g3 d4 27. g2 g7 28.f4 f5 29. d1
h6 30. f3 f5 31. d3 d8 32. e3 e7 33. c2 b2 34. e3 f6
35. e1 d4+ 36. f3 b2 37. g2 d7 38. h4 f6 39. e3 f7
40. c2 a1 41. e2 b2 42. e1

660
Black’s decisive mistake was:
42... a1?
He should have prevented the breakthrough
43.g4!
with the move 42... h6!, but Black’s position would have remained
unpleasant all the same.
After the breakthrough, the game was over within a few moves:
43...hxg4 44.hxg4 fxg4 45. xg6 g7 46. h4 f8 47. f5 f6
48. c8 d8 49. f5 h5 50. d2 d4 51. xd4 1-0
This game impressed the German grandmaster so much that he
included this line in his White repertoire and won two convincing
victories over Svetozar Gligoric and Ulf Andersson
King’s Indian Defence
Wolfgang Uhlmann (2570)
Svetozar Gligoric (2600)
Hastings 1971

661
1.d4 f6 2.c4 g6 3. c3 g7 4.e4 d6 5. e2 0-0 6. g5 c5 7.d5 e6
8. d2 exd5 9.exd5 e8 10. f3 g4 11.0-0 bd7 12.h3 xf3
13. xf3 a6 14.a4 e7 15. ae1 f8 16. d1 xe1 17. xe1 e8
18. xe8 xe8 19. f4 e7 20. e2 f8 21. xe7+ xe7

The position is almost identical to that in the previous game.

22.a5! e8 23. d2 h5 24. f1 d4 25.b3 g7 26. c2 e8
27. e2 b2 28.f3 g7 29. f2 f6 30. c3 d4+ 31. e2 f5 32.f4
e8 33. d3

662
Until now, the game has developed according to the scenario of the
Polugaevsky-Uhlmann game. In this position, the Yugoslav
grandmaster decided to turn aside and exchanged his bishop for a
knight. It is not easy for two knights to fight against bishops without
the help of other pieces. But sometimes this balance of forces is
more favourable for the weaker side than in cases where three minor
pieces (two knights and a bishop) fight against three (two bishops
and a knight). It all depends on the specific features of the position.
33... xc3 34. xc3 ef6 35. e1 f7 36. e3 e7 37. c2 f7

663
38.b4 cxb4 39. xb4 c5 40. d4 fd7 41. d1 e7 42.g4 hxg4
43.hxg4

The German GM attempts to break through the opponent’s defences

with pawn breaks alternately on two flanks.

664
43... f6 44. e3
Here Black played:
44...b6
Here is what I wrote about this move in the original Russian edition
of Endgame Strategy, published in 1988:
‘Waiting tactics no longer save him: 44... f7 loses due to 45.gxf5
gxf5 46. h5+ e7 47. g6 f6 48. h7 e4 49. a3 with the threat
of 50. b2+.’
Today the engine in this variation recommends Black to play
46... g7 (instead of 46... e7 ). Then it ‘wriggles’ with 47. e8 f6
48. c3 ce4

analysis diagram
49. a1 c5 50. c3 ce4 51. d4 c5 52. e2.
This looks like it is all over. Next move the bishop comes to d7 and
the black position collapses. But this is not the case: 52... b3!
53. c3 xa5! 54.c5 f8! 55. a4 xd5 56. xa5 dxc5.
The assessment is +0.13.

665
 analysis diagram
So, how do we win? The computer can trick its way out of even
such positions as this, an ability far in excess of the human player.
Not surprisingly, after the text moves 42.g4 hxg4 43.hxg4 the
engine recommends simply 43...fxg4 44. xg4 b3+ 45. e3 bc5,
and maybe only another monster engine can beat it from this
position.
But back to the game. After 43... f6 44. e3 b6 , White played:
45.gxf5

666
Gligoric automatically replied
45...gxf5?
and this obvious move proved to be Black’s decisive mistake. The
engine, ‘laughing its circuits off’, plays 45...g5!! and confidently
holds the position after 46.fxg5+ xg5 47.axb6 xb6 48. d4
bd7. But, as they say, this line is not for average minds.
After 45...gxf5?, Uhlmann reacted in the best way:
46. xc5!! xc5
If 46...bxc5 47. a4 b8 48. f3, White is practically playing with
an extra piece. Black will lose the f5-pawn because of zugzwang
with the white bishop on e8.
47.axb6 a5 48. c2 e7 49. d2 d8 50. xf5 a4 51.b7 c7
52. c8 c5 53.f5 e4+ 54. c2 b8 55. b3 d2+ 56. a4 xc4
57.f6 e5 58. xa5
Black resigned.
In both games, a typical position emerged with the same pawn
structure and White having the advantage of the two bishops. If

667
White manages to clamp down on the opponent’s position on the
queenside with a4-a5 in the endgame, then the game has two results
– a White victory or a draw. The standard plan for White in this
position – ‘hugs’ – was encountered in many games. Eventually
Black got tired of fighting for a draw in a worse position, and this
line disappeared from tournament practice.
We are convinced that a good knowledge of game plans in standard
positions is useful, and every chess player should have it. But if he
relies only on the skills of playing in standard positions, then when
he finds himself in a situation unfamiliar to him, he is left without
guidelines. A chess player with highly developed associative
thinking, as a rule, understands the logic of the game more deeply,
and in the same unfamiliar situation it is easier for him to find the
right solution.
I would like to end this chapter with one recent example of how
associative thinking could help make the right decision under severe
time pressure.
Consider the following position as a training exercise:

How should Black play?

668
If you have worked through the section on rook endings well, you
know the standard methods of defending for Black. I am sure that
the obvious move 1... a1 will not cause any hesitation. Further, in
response to 2. g2, you will choose between the answers 2... a3!
and 2... a2, by analogy with the game Grischuk-Vachier-Lagrave.
But in the position in the diagram there is another stronger solution:
1... c6!! 2. g2 e6!

Here, a draw can already be agreed. Black’s rook on the e6

stronghold will not allow the white king to reach the queenside.
Black can simply make waiting moves with the king, and as soon as
the white rook moves to a8 and the pawn to a7, the move ... e6-a6
will follow. This method of defence is an exception to the rule, but it
is useful to keep it in one’s collection of images for associative
thinking.
Quite recently, at the Women’s World Cup, this position arose:
Anna Muzychuk (2522)
Mariya Muzychuk (2523)
Baku 2023

669
38... c7!
Black correctly sacrifices a pawn and activates the rook. This
technique was figuratively described by Nimzowitsch in his classic
work My System and he gave the following example:

670
 position from My System
Nimzowitsch compared the active manoeuvre 1... b2! 2. xa5 a2
with the weak move 1... a7?. He commented on Black’s passive
decision as follows: ‘The active rook then experiences true
satisfaction, like a prima donna who, while playing the main role,
sees her rival, languishing in some insignificant role in the same
play. But in this case, one should not be surprised if the ‘humiliated’
rival calls in sick and disrupts the performance.’
If we compare the last two diagrams, we see that the position from
the Muzychuk sisters’ game is more dangerous for Black than the
one given by Nimzowitsch. After all, in the first diagram, in White’s
position on the kingside there is a g5-pawn, which squeezes the
position of the black king. Nevertheless, activating the rook with the
move 38... c7! helps Black out here too.
Let’s continue analysing this ending.
39. xa5 c4 40. a7+ g8 41. g3 g4+ 42. h3 b4 43.a5 b3+
44. g2 b4 45.a6

671
White has broken away from the h4-pawn with her king and in case
of 45... xh4?? would free her rook along the seventh rank with the
move 46. c7 with victory. Mariya Muzychuk was in time-trouble
and made a decisive mistake here:
45... a4?
Remember the defensive method associated with 1... c6 2. g2 e6,
which we discussed in the training exercise (first diagram previous
page)?
If Mariya had known this technique, then her associative thinking
would have told her to play 45... b6!! 46. f3 e6! and thereby
achieve a draw.

analysis diagram
We present the rest of the game without comment:
46. f3 f8 47. e3 e8 48. a8+ f7 49. d3 g7 50. a7+ f8
51. c3 e8 52. a8+ f7 53. b3 a1 54. b2 a4 55. b3 a1
56. c4 e6 57.a7 f5 58. f8+ g4 59.a8 xa8 60. xa8 xh4
61. d3 xg5 62. a4

672
Black resigned.

673
Acknowledgements
As you know, I have been specializing in the endgame for a long
time and have always collected training exercises to improve a chess
player’s technique in the endgame and middlegame. In recent years,
I have acquired a lot of interesting material from modern top-level
tournaments. Analysing them, I began to notice that the techniques
used by leading modern chess players, in their ideological content,
are very reminiscent of the actions of the great champions of the last
century. This is how the idea came to write a book about images and
associations that involuntarily arise in my mind like memories from
the classics.
To tell you the truth, at first, I only had a set of exercises on various
topics. But everything changed when my daughter Maria joined the
work. She doesn’t play chess, she has a linguistic background, and
she used to also look at my books before publication, checking them
for style and spelling. But with this book, everything was different.
Having worked as a copywriter in an IT company, she acquired a
number of valuable skills and much knowledge of how to correctly
present complex material in a simple way, so that it would be
understandable to readers even without technical knowledge.
I was pleasantly surprised that she took the composition of this book
into her own hands. She managed to transform a patchwork of
training exercises into a coherent linear narrative. At the same time,
she showed iron logic and extraordinary strong-willed qualities.
When she made the next round of edits, finding out the essence of
my thoughts, I sometimes almost brought her to tears and called her
‘nitpicking’ and a ‘nuisance’ which, of course, I now regret.
I noticed that as she worked on the book, she became very adept at
prophylactic thinking.

674
Dvoretsky has an excellent aphorism on this topic: ‘Remember the
Opponent’. Mary’s favourite motto was ‘Remember the reader’.
And her main question was the following: ‘Will the reader
understand what you want to say? Put yourself in his place.’ This
was followed by the merciless removal of my abstract arguments,
after which we together replaced them with concrete descriptions of
the events taking place in the games, and new chapters and sub-
chapters appeared in the book. So this time I can fully consider her a
co-author of the book Associative Thinking. The strength of the
average move is ‘masonry’. We hope you liked the structure of the
book.
And now for another discovery. We all talk about how chess has
changed with the advent of powerful computer programs. But it is
best understood by those who play at the highest level and those
coaches who prepare them. Grandmaster Evgeny TomashevskyFor
many years I have been working with chess players, using my
method. We fill gaps in basic knowledge and improve playing
techniques, thanks to which they successfully reach a rating of
2400-2500. I used to think that a chess player’s further progress was
a matter of talent. I had no idea that with the advent of powerful
computer programs, a technique could appear that would allow a
chess player with a 2500 rating to reach a level of 2700+. And the
key place in it is given not only to talent, but to hard work and
correct systematic work with a computer, under the guidance of a
specialist who really knows this technique.It is very interesting for
me to watch the coaching work of Evgeniy, who perfectly
understands the real practical strength of a 2700+ level chess player
and knows how it is achieved. At one time, he himself reached a
rating of 2758, and even now he is an excellent player, in the top 50
according to the FIDE rating. Evgeniy worked twice as a coach in
World Championship matches and twice in Candidates tournaments.
For his encyclopedic knowledge, at a young age he received the
nickname ‘Professor’ in chess circles.I am very grateful to Evgeniy
675
for the knowledge I acquired from communicating with him about
the development of modern chess, which is largely reflected in this
book. Personally, I didn’t know how to help a chess player with a
2500+ rating make tangible progress. Now I know a person who
does this job very well.

676