Fighting Chess Move by Move - Colin Crouch 2012

Colin Crouch
move by move
EVERYMAN CHESS
www.everymanchess.com
First published in 2012 by Gloucester Publishers Limited, North burgh House,
10 Northburgh Street, London EC1V OAT
Copyright© 2012 Colin Crouch
The right of Col in Crouch to be identified as the author of this work has been
asserted in accordance with the Copyrights, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system or transmitted in any form or by any means, electronic,
electrostatic, magnetic tape, photocopying, recording or otherwise,
without prior permission of the publisher.
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
ISBN: 978 1 85744 993 8
Distributed in North America by The Globe Pequot Press, P.O Box 480,
246 Goose Lane, Guilford, CT 06437-0480.
All other sales enquiries should be directed to Everyman Chess, North burgh House,
10 Northburgh Street, London EC1V OAT
tel: 020 7253 7887 fax: 020 7490 3 708
email : info@everymanchess.com; website: www.everymanchess.com
Everyman is the registered trade mark of Random House Inc. and is used in this work under
licence from Random House Inc.
To Audrey Eileen Crouch, my mother, born December 192 7, died July 2011.
With many memories.
Everyman Chess Series

Chief advisor: Byron Jacobs
Commissioning editor: John Emms
Assistant editor: Richard Palliser
Typeset and edited by First Rank Publishing, Brighton.

Cover design by Horatio Monteverde.
About the Author
Dr Colin Crouch is an International Master, a tremendously experienced tournament player

and a highly regarded chess writer. His books have received great acclaim for their thor
oughness and originality.
Also by the Author:
Rate Your Endgame

Chess Secrets: Great Attackers
Modern Chess: Move by Move
Why We Lose at Chess
Analyse Your Chess
Contents
Preface 7
Introduction 9
1 Vladimir Kramnik-Levon Aronian, Zurich, April 2012 13

Game One: Kramnik-Aronian 13
Game Two: Aronian-Kramnik 27
Game Three: Kramnik-Aronian 34
Game Four: Aronian-Kramnik 47
Game Five: Kramnik-Aronian 56
Game Six: Aronian-Kramnik 67
2 World Championship, Anand-Gelfand, Moscow, May 2012 87

Game One: Anand-Gelfand 88
Game Two: Gelfand-Anand 96
Game Three: Anand-Gelfand 104
Game Four: Gelfand-Anand 122
Game Five: Anand-Gelfand 130
Game Six: Gelfand-Anand 138
Game Seven: Gelfand-Anand 147
Game Eight: Anand-Gelfand 157
Game Nine: Gelfand-Anand 165
Game Ten: Anand-Gelfand 1 80
Game Eleven: Gelfand-Anand 187
Game Twelve: Anand-Gelfand 195
The Sequel: Tie-Break Games 206
3 S eventh Tal Memorial, Moscow, June 2012 215
Round One: Carlsen-Kramnik 219
Round One: Radjabov-Tomashevsky 239
Round One: Morozevich-Caruana 24 7
Round Five: Radjabov-Carlsen 261
Round Six: Morozevich-Nakamura 2 75
Round Eight: Caruana-Kramnik 284
Final Notes 293

Index of Openings 295
Index of Games 296
Preface
Many books get written as a result of illness, of being stuck in bed. This is another of these
examples. In the spring of 2012, I found myself with swollen legs and could barely walk. I
could climb upstairs only with hands and feet, and it seems that my hands were taking too
much of the strain, trying to pull myself up the stairs, or out of the bath, or into bed. So my
arms became damaged as well.
When I felt myself strong enough to get up and sit in front of the computer, I wanted to
play through some really top-level games of chess. The timing was fortunate for me. I no
ticed that there was a friendly match between Aronian and Kramnik coming up soon, and
then the World Championship between Anand and Gelfand, just a month later. Although I
was starting to walk again a little, I was still in and out of hospital. To play through world
class games was a lifeline to me to the outside world.
I blogged each of the games of the two matches, without any real thought about pub
lishing in book format. I soon appreciated that, with a little extra effort, fully revising all
my earlier notes and adding further games, this might be of unusual chess interest - not
necessarily because of the strengths and weaknesses of my own writing, but because I had
stumbled on unusually good timing.
This became fully clear after these two matches, and then the next really big tourna
ment, the Tal Memorial. Remember that at the start of the 1960s, Tal beat Botvinnik to be
come world champion at the age of 23, a then unprecedented display of chess youth and
vigour. Tal's health deteriorated very early on and, while he remained a strong and dan
gerous grandmaster through to the end of his life, he was never the dominant force his
admirers had hoped for.
Given this context, the Tal Memorial of Moscow 2012 was a strong reflection of what he
had himself achieved just over fifty years earlier. This time, though, there were three
grandmasters, all in their early twenties or teens, aspiring to show that they too could soon
become world champions. After a tense battle, Carlsen (21}, Radjabov (25) and Caruana
(19) came out first, second and third, against opposition of vastly over 2700 strength.
This must surely herald the switch of the generations. Anand will of course aim to re
main World Champion for as long as possible, but eventually younger players will take
7
Fig h ting Chess: M o ve by M o ve
over. Could this be Carlsen, with his admirable ability to avoid defeat? Or Caruana, a few
years younger? Or Giri, who is even younger? Or one of a small number of other players?
Colin Crouch,
Harrow,
November 2012.
8
Introduction
This book is based on the idea that every move is important, any mistake by either player is
significant, and any mistake by the opponent should be pounced upon. The theme in this
book is based on what can loosely be described as "positional chess", on giving nothing
away to the opponent, and on being alert to opportunity given by the opponent.
I am fascinated that the strongest players avoid losses to a remarkable extent, even
when play appears sharp and double-edged. How, I wondered, do these top grandmasters
keep their balance? The statistics are awe-inspiring . In six games against Kramnik, Aronian
lost only once. In six games against Aronian, Kramnik lost only once. In twelve games
against Anand, Gelfand lost only once. In twelve games against Gelfand, Anand lost only
once. Four losses out of 36. Few club players, playing against opponents of their own
strength, could achieve such a low percentage of losses. What is the secret of the top play
ers?
Personally, in my own games, I find I have wins and losses, very rarely draws, and even
more rarely do I achieve solidly played draws. I would love to know how to turn these losses
into draws, except I suspect that the answer is relatively simple. I am usually good enough
to find wins against players up to about IM strength, but quite often, in declining health, I
get tired, and cannot think clearly enough, and so I lose.
I was also startled, when going through recent games, that players somewhat younger
than me (I am in my mid fifties) can occasionally lose their sharpness, and sometimes
make uninspired mistakes. Just before the World Championship, Anand handled the open
ing dreadfully against Tiviakov, in the German Bundesliga, and was straightforwardly
ground down in a Sicilian, where Anand played ...e7-e5, and lost control of the ds-square
and the files and diagonals nearby. Kramnik, too, in his first game against Aronian, played
almost unrecognisably. Could they, on bad days, play almost as badly as me?
One bad loss happens, but it is important, if at all possible, not to start a string of bad
losses. lt is a question of match survival. In the two matches being examined, all four play
ers lost a game, but they did not lose any further games. The problem is, if anything, more
the opposite, an excess of "animal spirits", a belief that if you have won one game, you can
play whatever you like, and you are immune to mistakes. Both Aronian and Gelfand suf
fered from this.
I felt slightly disappointed with the World Championship match, not because it was
"boring", but rather because there seem to have been several opportunities for both sides
9
Fig h ting C h e s s : M o ve by M o ve
to try for an edge in many lines. Too often, the initiative tended to fizzle out much too
quickly. If your position is clearly level throughout, then you have every right to offer or
accept a fairly quick draw. If, on the other hand, one of the players had even the slightest of
edges, that player should try to make the opponent suffer. A win plus four draws is better
than five draws.
If the reader feels slightly disconcerted that there is such a switch between the first per
son and the third person, the second person - you - can have your point of view in the
analysis. Imagine that I am sitting in my room, with the computer, trying to make sense of
what is going on in a series of difficult positions, while being aware that the two players
involved are vastly stronger than me. I know, however, that they occasionally make mis
takes, as they do occasionally lose games. I am trying to assess, perhaps with the help of
the computer, what is going on in a string of moves; to decide whether the player is seeing
things more clearly than me, and finds a much better move than I was thinking of; or
whether the player has made a mistake in a critical position, which I noticed; or whether, if
the player and I chose different moves, both moves might be equally valid.
Watching live chess games is one of the best ways of sharpening a player's thinking.
There is an immediacy which cannot be achieved just by going through games which have
already been played, recorded and analysed. For a writer, maybe it is a useful prod for the
reader to invite him, or her, to be asked what the player should be thinking, in a new posi
tion, in a book. Hence plenty of questions and answers.
After the live game, I try to analyse further, and I have been blogging it up. All the
games in the Aronian-Kramnik match and the Anand-Gelfand match are written up in my
blog, shak thinking, usually a day or two after the game. I though about deleting my com
ments in these games, before publication, but I decided against it. The shakthinking notes
were merely an earlier draft, with, I have to admit, many typos (it's difficult with only half
of one eye working), and if I am able to find mistakes in my earlier annotations, then of
course I can try to correct these mistakes for a later draft. lt might still be useful for some
readers to compare before and after notes.
Which leaves this to the more detailed questions of the reader. I am asking you, in ef
fect, the same questions as I asked myself in playing through the games live. If I felt that a
player has pushed a pawn too early, for example, I want to re-analyse the position. If I find
that the player's move was, after all, correct, that is fine, and I have learned something. If I
find that the player has got it wrong, and I cannot see any way to disprove the argument I
have made, then this is also knowledge. What I am asking you, the reader, is to go through
the same exercises yourself.
Many of the exercise are open-ended. Remember that most of the games in this book
end up in draws, and so there is no clear-cut winning line, or winning plan, that needs to
be found. Instead, we are dealing far m ore with positional uncertainty. If there are appar
ently four reasonable moves in a given position (and the reader can check out these moves
on the computer}, which of these is the safe equalizing line? Or on a different set of four
possible moves, one might give a fractional edge; another might be about equal; another
10
In tro d u c t i o n
might end up, after some tactics, with a repetition; while another, apparently equally
promising, might end up with a slight disadvantage.
These exercises are based mainly on positional chess, on giving nothing away to the
opponent. More specifically, they are based on fighting positional chess, on recognizing
that your opponent will want to give nothing away, while you yourself do not want to give
anything away. There is no assumption, in fighting positional chess, that everything will
end up with a quick handshake after around a dozen moves; nor even in a quick win after a
blunder by the opponent. No, these games are played out to the end, and well contested.
11
C h a p t e r One
Kramnik-Aronian,
Zurich 2012
Game One: Aronian-Kramnik
21st April 2012

V.Kramnik-l.Aronian
Zurich 2012 (Game 1)
Semi-Slav Defence
1 tt::lf3 ds 2 d4 tt::lf6 3 c4 c6 4 tt::lc3 e6
13
The ultra-solid Semi-Slav Defence, which players have increasingly relied on for Black,
when they want to make sure they have no realistic chances of losing.
s ..igs
Keeping to the main line. 5 e3 is the chief alternative.
Question: For those who are addicts to deep theory, or to gambit chess, or
who quite simply want to know what is happening in the Botvinnik Variation.
What is happening at high levels in these with 5 ... dxc4 - ?
s h6
...
Aronian is happy to play in "normal" Queen's Gambit chess.

He could have chosen the sharp gambit line with 5 ...dxc4 6 e4 b5 7 e5 h6 8 ..ih4 g5 9
lt:Jxg5 hxg5 10 ..ixg5 ttJbd7 - wild stuff, with plenty of opportunities for Black to play for a
win, but recent top-level games suggest that in the main line White, first, has the option to
take an easy perpetual, and second, that White might well be better in complicated play.
The hammered-out main line would be 11 exf6 ..ib7 12 g3 c5 13 d5 �6 14 ..ig2 0-0-0
15 o-o b4 16 ttJa4 'ilt'a6 17 a3 ..ixd5 18 ..ixd5 lt:Je5 19 axb4 !:txd5 20 'i&'e2 cxb4 2 1 lt:Jc3 'i:Vd6.
As I said, wild stuff - but as so often, sharp and accurate play may end up in a quick
perpetual check. For example, 22 :!.xa7 bxc3 2 3 l:!.a8+ {or 23 .l:!.fa1 �4 24l:!.a8+ 'Et>b7, again
with a perpetual, M.Vachier Lagrave-Y.Solodovnichenko, French League 2011) 23 ... Wd7 24
l:!.a7+, soon drawn by perpetual, S.Ganguly-A.Shirov, Spanish League 2011. This was of
course known before.
White could also try for more, with 2 2 ttJxd5 �xd5 2 3 f3 ..ic5+ 24 ..ie3 ttJd3 2 5 ..ixc5
'iUxc5+ 26 'lt>h 1. A recent game, A.Grischuk-A.Shirov, European Team Championship, Porto
Carras 2011, continued 26 ... 'i!Vd4 27 l:ra5 .l:!.d8 28 l1fa1 .l:!.d7 29 h4, and it was not all that
surprising that both players were able to promote a passed pawn, ending up with another
perpetual check a few moves later.
14
Kra m n ik - A ro n ia n, Zurich 2 0 1 2
If, i n critical lines in the opening, the end result i s a known draw some 2 5 moves deep, it
is often best, to try to keep the game open, to play something less highly theoretical. Aro
nian chose s ... h6, and it worked out well for him.
6 i..xf6
White could still continue with gambit play after 6 i..h 4 dxc4 7 g4 g S 8 ..tg3 bS, al
though perhaps slightly less convincingly so. Certainly, many players as Black have swal
lowed the pawn. Alternatively, Black could continue to decline the gambit with, for exam
ple, 6 ... tt:Jbd7.
6 .. .'iVxf6
7 e3
The simplest, and therefore perhaps the best, chance to aim for a slight edge. 7 e4 dxe4
8 lt:Jxe4 i..b4+ 9 'it>e2 'ilif4 10 'ilid3 0-0 11 g3 'ilic7 12 i..g 2 is equal.
7 ...lt:Jd7 8 i.. d 3
Many choices - and, indeed, in game five Kramnik tried 8 .te2. White is not too con
cerned about giving away a tempo with ..td3, ... dsxc4; ..txc4, as this exchanges off Black's
d-pawn for White's less centralized c-pawn.
8 ... dxc4
Black's position is not so cramped as to need to hold on to the dS-pawn indefinitely. He
will want to open up the centre and create lines for his bishop pair with ...c6-cS, or more
likely, ... e6-eS.
9 i..xc4 g6
Aronian turns the game into, in effect, a Grunfeld, the bishop putting pressure on
White's d4-pawn, with the help of a later ... e6-eS or ... c6-cs. Instead:
a) 9 ... gs is aggressive, but probably not so good, as he is loosening pawns in front of the
king (assuming king side castling). White would just castle, 10 o-o, and see what Black does
next.
b) 9 .. es?! would have been tactically inaccurate, in view of 10 dxes tt:Jxes 11 lt:Jxes 'iVxes
.
15
12 '1Wb3, and suddenly Black h as problems with his f7-square.

Black is behind in development, and he needs to be careful not to open up his position
too quickly. Therefore, slower ways are required, quiet development, before then opening
up the pawn structure.
10 0-0 .i.g7
Question: What do you think should White play next?
11 �e1
Ideally, every move in a game of chess should have its justification, and every quiet
move should prove, a few moves later, to be part of a greater plan. lt is not really that clear
why White's rook should be on e1. He already has enough defensive ballast to cover any
pawn advance with e3-e4-e5, so the rook on the e-file does not seem to be genuinely nec
essary.
The immediate 11 e4 is safe enough, in that his pawn moves to a comfortable square,
and he gains some space in the centre. Then perhaps 11 ... es (to prevent White from play
ing e4-e5 himself) 12 dS 0-0, when White can transpose into the game with 13 .l:te1, but
surely he can do something more constructive? Inevitably, we would not be thinking of a
big advantage for White, but keeping a slight edge would be satisfactory. Possibly 13 .i.e2
(into safety) 13 ..Jld8 14 'il¥h3 ! ? and he can still try for a slight edge. The obvious 14.. .<�JC5 is
not as big a gain of tempo as it looks, since the knight would be on a better defensive post
at b6, covering the ds-square, rather than at c6. Perhaps a slight edge for White after 1 5
"i¥c2.
11 0-0
...
lt has to be played sooner or later, and this seems to be a good time.

12 e4
White has played a mixture of �fe1 and e3-e4. Just one of these would have been bet-
16
Kra m n i k - A ro n ian, Zu rich 2 0 1 2
ter, it seems. Here 1 2 iVc2 forces Black to show how h e can find complete equality. Black
may have the bishop pair, but he is slightly behind in development.
12 es
...
At last, there is a genuine pawn.

13 ds
The only sensible move. After 13 dxes? tt::lx es Black is already better, with his bishop pair
starting to take control.
13 ....l:i.d8
14 ... tt::lb 6 also makes good sense, but Aronian wants to keep his options open.
14 l:'i.e3
Question: What is White's idea with this move?
When I saw this on the computer, my instincts were that it surely cannot be good. The
rook is soon on an uncomfortable square and does nothing effective, If l:'i.d3 soon, then
... tt::lc s, and the rook must move again. White is not yet worse even then, but if he makes
another couple of slightly inaccurate moves, then quite certainly, as in the game, he will
genuinely be in trouble.
Instead, one might expect White to play 14 dxc6 bxc6 15 iVc2. This would be an ultra
technical way of handling the position, relying not so much on White creating active play
for his pieces, but rather to mess up Black's queen side pawns slightly.
But what about Kramnik's own rook move? lt seems to be the old story of trying to trick
the opponent with a "Theoretical Novelty", using a sharp move which the opponent will
not have seen before, hoping that he will not be able to find a good reply. Most of the at
tempted big theoretical novelties in the two 2012 matches end up going badly wrong. lt is
better to find good and accurate moves.
14 bs
...
17
Fig h ting Chess: M ove by M ove
Aronian looked at the sharpest reply and decided that the position was safe, and poten
tially good.
15 dxc6
In my blog, written just after the game had finished, and without time to examine the
comments made by others, I noted that this was "another odd m ove, showing that Kram
nik has perhaps lost his sureness of positional touch. Would he really have played like this
a decade ago?"
What in fact happened was that Kramnik had followed a somewhat doubtful new line,
thinking that it was good, and missed the big tactical improvement for Black, which Aro
nian found easily enough over the board. See the comments to Black's 16th.
He needed to keep his bishop. Having two knights versus two bishops is likely to put
himself under pressure. 1 5 i.b3, or maybe 1 5 .i.d3, would still be about equal.
15 . bxc4
..
Question: Which is better? 16 tLld5 or 16 cxd7 - ?
There is no trick question here, no deeply hidden third option. I would suggest, how
ever, that the reader spend some time trying to think about which of these moves is better,
before playing through the rest of the game. Try to answer this with the minimum of hind
sight. Imagine what you would do over the board.
16 tLld s
This was the move that Kramnik played.
16 ...'i!i'e6!
And this was Aronian's reply.
A recent game, in which Kramnik was foll owing, went 16 .. .'iVd6 17 cxd7 .i.xd7 18 tLld2
i.b5 19 'ii'c 2 .l:iab8 20 li.c3 with advantage to White, and later a win in V.Gunina
A.Muzychuk, European Women's Championship, Gaziantep 2012. This may l ook impressive
18
Kra m n ik - A r o n ian, Z u rich 2 0 1 2
and an unusual rook zigzag, but Kramnik trusted Gunina rather more than h e actually
should.
Should we therefore claim that Aronian's win was the result of a massive new theoreti
cal innovation ? Not really. Innovation or no innovation, the only way that a player can lose
a game of chess is by making a mistake, and it is the fault of the loser, not the winner, who
made the mistake. All Aronian had to do was to find a few good moves. At various points of
the game, he had used an hour less on the clock than his opponent.
17 cxd7
17 tt'Jc7? "ifxc6 18 tt'Jxa8 "ifxa8 would have been pointless. Black therefore h ad no need
to cover the c7-square. Black's 16 ...'il'd6? in the earlier game was a significant loss of time.
This was Muzychuk's m istake. And Kramnik's attempt at a novelty had backfired. By now
Kramnik was in danger of being seriously worse, and spent a lot of time thinking.
17...l::tx d7 18 �a4 .ltb7
19
Fig h ting C h e s s : M ove by M ove
We see now the basic problem. The advanced knight on ds is a weakness, rather than a
strength. Kramnik should not have played 16 lt:lds.
lt is a sophisticated example of the beginners' error, moving the knight to the other side
of the board, without being part of a genuine attack, and without having ensured that the
knight is safe, not just on the next m ove, but also on later moves. Sometimes what hap
pens is that the knight has to retreat, with loss of tempi.
19 �XC4
This allows favourable simplification for Black, now that the dS-square folds. Few play
ers would have had the cold discipline to admit that the earlier knight move was wrong,
and then simply retreat it with here 19ctJc3 !?. White is still worse, even so. Black could try
something like 19 ...l:l:ad8 20 .l:i.c1 'it>h7, keeping open the later king side attack with .. .f7-f5.
Of the world championship greats, perhaps Emanuel Lasker over a century ago, and
quite probably Karpov, would have considered such a retreat under pressure, and maybe
also Spassky or Petrosian. lt would, of course, be extremely difficult to second-guess what
others might have played in such a position, and given the history of the previous part of
the game - but one thing that is totally clear is that Kramnik did not play 19.ctJ c3 himself.
19 ...i.xd s
The start of a chopping of pieces and pawns on ds. No general comment is required,
move-by-move, except to note that it is important, even if elementary, to count up the
number of attackers and the number of defenders to ensure who is winning the battle.
20 exds �xd s 21 �xds .l:i.xds
22 .l:i.ae1
Now White is attacking the pawn on es, but it is simple enough to bring a defender into
play.
22 .l:i.e8
•••
Aronian aim s for the most direct plan, a pawn roller with .. .f7-f5 and ... e5-e4, supported
by the rook, and opening up the long diagonal for the bishop. Even so, the plan is not quite
20
Kra m n ik -A ro n ia n, Zurich 2 0 1 2
as straightforward as it looks, and there are more indirect ways of trying to achieve the
same goal.
22 .. .f6 ! ?, the quiet way of protecting the pawn, is to be considered. Black then has the
choice, with the rook, of moving to b8, or c8, or d8, as well as e8. Black's initiative that way
is not based solely on the king side. Also, his king will be more active on f7, rather than on
g8, or even (as in the game) on h7. Black does not have to worry about the "bad bishop",
since the bishop can re-emerge with ...�f8. lt is a different way of handling the position,
and probably not inferior.
White is uncomfortable - perhaps not yet losing, but he would have to play with ex
treme accuracy and hope that his opponent does not have the same level of accuracy. The
main problem is that Black is threatening to roll over the centre with .. .f7-f5 and ... e5-e4,
opening up new possibilities for the bishop and the two rooks. White's two queenside
pawns are then seriously open to attack.
Question: What should White do here?
23 g4!
When in trouble, a top grandmaster will generally find excellent defensive moves. This
pawn push looks extravagant at first, but Kramnik is fully aware the he must not allow the
rolling of the pawns.
23 tt::lh 4 is the main alternative, if White is not content with passive defence. Then after
23 ...�h7 24 f4 l:tb8 2 5 fxe5 (25 b3 exf4 26 .l:!.xe7 g 5 is good for Black) 2 5 ...l:txb2, White's
king side pawns appear to be more united (g2 and h2) than in the main line (f2 and h 2).
What Kramnik has to consider, though, is that if the f-pawn has not been exchanged off,
there is no defence on the second rank, should the black rooks be doubled on that rank. The
f2-pawn is needed, to protect the g2- and h2-squares.
21
Fig h ting C h es s : M o ve by M o ve
Question: What is the most accurate way for Black to maintain a slight edge?
23 'it>h7
...
Aronian played this quickly. The computer consensus was that 23 ...f5 24 lZ'lh4 fxg4 2 5
lZ'lxg6 would have been better, but this is not fully clear. The knight looks exposed, but if
Black were to try to trap the horse with 25 ... .ii.f6, it turns out that his e-pawn is pinned after
26 lZ'lf4. lt is still uncomfortable, but it is not so clear that Black has a definite win.
If, for the moment, we forget about the computer and its analysis, the question is, what
is best in purely positional terms? Computers are not always so good in assessing this.
lt is time to remember the old idea of prophylactic chess, as discussed almost a century
ago (time flies!) by Nimzowitsch. When dressed up in m odern terms, the idea is not that of
direct defence, of covering any immediate attacks by the opponent. Rather, the idea is to
prevent the opponent from making a good move, and to prevent any counterplay by the
opponent, so that the player himself will remain at least comfortably equal, and, if all goes
well, keep a slight advantage.
Black is clearly not scared of White being able to play for an edge with g4-g 5. Therefore,
any attempt by Black to prevent g4-g5 cannot be regarded as defensive. Rather, Black is
aiming for a prophylactic m ove, to prevent White from equalizing.
This suggests 2 3 .. .f6!?, just waiting, and preventing his opponent from doing anything
more than waiting himself. Clearly, Black can try .. .f6-f5 later in many lines, but there is no
need to hurry. Black has the additional option of re-developing with ....ii.f8 and ....ii. c 5, as
well as the more obvious ...'it>f7.
For instance, if 24 lZ'lh4, then 24...Wf7 2 5 f4 l:lb8! (more ambitious than 25 ... l:ld4 26 l:le4)
26 fxe5 l:lxe5 27 l:lxe5 fxe5 28 b3 .Ub4 with a clear plus, due to the passed pawn, and the
bishop being far more active than the opposing knight. Even here, it would be far too early to
claim a decisive advantage, but it would take highly accurate play for White to defend.
22
Kra m n ik - A ro n ian, Zurich 2 0 1 2
24 g5!
This gives White excellent chances of equalizing. lt looks ugly, leaving himself with two
isolated pawns against Black's three united pawns {after the next move from either side),
but it is not so clear, in dynamic terms, that Black can create a serious king side advantage,
and meanwhile White has the extra pawn on the queenside as a counterbalance.
24... hxg5
If it were not for the en pass ant rule, Black would have a probably winning advantage.
As it is, 24.. .fs 25 gxf6 (e.p.) 2S ... i.xf6 26 b4 is only equal.
2 5 lbxg5+ 'it>g8
Kramnik h as made good defence so far, but he still has to play accurately and carefully.
Question: Another difficult decision, this time for White. Kramnik played 26 f4,
opening up the centre. Is this brilliant? Or awful ? In either case, why?
23
Fig hting C h e s s : M ove by M ove
26 f4?
This is the losing move, and a strange one, in that Kramnik was being careful a few
moves earlier to avoid opening up the second rank with f2-f4.
Instead, White needs to ask Black how to proceed. 26 b4 seems the most constructive of
the quiet moves, maybe even adding an escape route for the knight via e4 and cs, if re
quired. Also, in comparison with ideas involving b2-b3, White still has counterplay with
.l:ta3. White is not yet equal, but neither is he clearly losing.
26....l:tb8
An easy way of breaking the pin.
27 fxes
There is not much else, but now it is difficult to defend either of his kingside pawns, be
ing an extra file apart.
2 7 ... .l:txb2
With Black's rook on the seventh, and a possible doubling of rooks on that rank, and re
newed chances of bringing the bishop into play (....�.J8, or ...j_h6, or pressure on the es
pawn), White is in trouble, with his king exposed, and all his pawns being under attack.
Kramnik's f2-f4 idea was disastrous.
28 'Llf3
He could h ave tried 28 a4, intending 28 ...l:tdd2 29 'Llf3 and there is no obvious immedi
ate win . White is still in trouble, even so.
28 ....l:txa2
Black is now a pawn up, furthermore an outside passed pawn up. There is no realistic
possibility of White ever winning the a-pawn, or even trying to sacrifice the knight for it
(after full pawn exchanges on the king side). All that White can hope for is to open up the
centre, and push Black's king around by checks, hoping that, with two rooks and knight on
the board, something might happen.
29 e6
24
Kra m n ik - A r o n i a n , Zu rich 2 0 1 2
So he tries it, but Aronian i s careful with his tactics.

29 fxe6
...
Obviously, he cannot allow White's advanced pawn to survive.

30 Itxe6
Some modest counterplay? Perhaps, but Black's bishop covers any potential mating
threats.
3o . JU s
.
Whereas White's knight is exposed to attack.

31 tt'lh4 .l::tf4 32 .l::t6e4 Itf6 33 Itg4 'itif7
The rest of the game looks fast-moving, with lots of checks, and Black's king being forced
to run. The point is, though, that the king is genuinely able to run, with the help of the other
pieces covering several dangerous squares. Before long, White's knight is unable to join in
the attack, as Black's king is far too far away. White's king, of course, can do nothing.
lt looked complicated in the time scramble, but Black was always in control.
34 Itc1
Staying active for as long as he can.
34 ....i.h6
Black's bishop joins in. Two rooks plus bishop, against an exposed king on the edge of
the board, add to mating threats.
3 5 Itc7+ 'itie8
3 S ...'itie6? 36 Itc6+ makes life far more difficult.
36 Ite4+
More checks. 36 .l:!.xg6? .i.e3+ is an instant collapse.
36 'itid8
••.
Question: Is Aronian bringing his king too far into the open?
25
Everything is fine, White runs out of checks, and Black has a breathing space to coordi
nate his pieces.
37 .l:th7
Attacking the bishop, but it can run to better squares.
37 ... ..if8
Aronian is being careful not to allow White's knight into play, though there are plenty
of winning choices here. For instance, 3 7 ... .l:td6 38 iLlf3 is no doubt good for a win, but then
Black will need to think about the endgame. Easier to win in a queen less middlegame.
38 .l:td4+ 'it>cS 39 .l:tc4+ 'it>b8 40 .l:td7
No more checks to force the king to escape, and indeed the next threat with lld8+ is life
less. Black has time to find the winning punch.
40 g6 41 tLlg6
...
Or 41 tLlg2 .l:tal+, winning the knight.

41 .....id6 0-1
26
it's over. Black i s still attacking the knight, obviously, while .....ixh2+ mating i s an even
stronger threat.
This is what I wrote in my day-after blog:

"Just on the basis of this game, Kramnik appears to have lost much of his positional
grip, as indeed in Anand's loss a week earlier (Black against Tiviakov in the Bundesliga). In
neither case was the issue one of missing tactics in wild and sharp combinations, as one
might suspect when the brain is not quite as quick as in the early twenties. lt seems more
serious than this. Positional play ought, in theory, still be close to the player's peak until
quite a late stage, but there are clearly lapses being made, even at the highest level.
As I write this, it is now a couple of hours before the second game. Maybe first-round
nerves could be a problem. Can Kramnik fight back? Or, at the very least, can he avoid los
ing further ground?"
Now I am writing at the end of July, a couple of days after Morozevich (age 3 5), sud
denly had to drop out, for health reasons, at Biel. At the time of the match against Aronian,
Kramnik was 36, and there were clear weaknesses in his play in the first round - although,
as a top professional, he did not lose any more games. At the Tal Memorial, however, just a
couple of months later, he lost twice in a row, after grinding out a very l ong endgame
against Tomashevsky, inevitably under great time pressure (and Tomashevsky should have
held the draw). At the World Championship in May, few would dare to say that An and (42)
or Gelfand (43) were starting to make more mistakes, even if some people posited the
catch-all argument, unfairly, that play was getting "boring".
Clearly, age is a factor. You cannot be as energetic after 35 than beforehand. For myself,
I now have the perfect excuse for never becoming a grandmaster. I did not make a system
atic attempt to become an IM until I was 34. Too late.
But what about the younger players? Aronian, at the time of his match against Kramnik
was 29, and the second-highest rated player in the world. He would need to do something
quickly, not least because Carlsen is several years younger, and there are already other ex
tremely dangerous younger players too.
Game Two: Aronian-Kramnik

lt was something of a relief that, in the end, this game ended up as a steady draw. First, we
would like to think that, when two of the strongest players of the world play each other,
their current level of technique should be close to flawless. Second, Kramnik was having a
bad off-day in game one, and it would be upsetting if one of the world greats were at the
start of a string of decline.
No long comments on this game. The point of interest at the time, when watching the
game live on the computer, was to see how much Aronian could squeeze out of almost
nothing. The point of interest afterwards is in seeing how Kramnik could hold, with relative
ease, an uncomfortable early endgame. Most of us would have found it troublesome.
27
Fig h ting Ch ess: M ove by M ove
lt is easy enough for the reader to skip through such games. "it's only a draw. I want to
leam how to win games." Very few players, even at a high level, are comfortable just mak
ing absolutely sure that they can hold a position, with draw after draw after draw. If the
position is level, they want to try to win, and if they are not quite level, they feel uncom
fortable, and are often not very good in finding safe ways of holding the balance under
slight pressure.
lt is therefore a good idea for the reader to take note of how Kramnik was able to hol d a
slightly uncomfortable semi-endgame with reasonable ease. Aronian was certainly trying
to press for the win but in the end he was unable to break though.
Another interesting point is that, after his disastrous first-round loss, Kramnik did not
attempt any extreme counterplay. The last thing he would have wanted to do would be to
lose his first two games. He was content to hold the position as Black, and wait to see what
happens in later rounds.
22nd April 2012

L.Aronia n -V.Kramn i k
Zurich 2012 {Game 2)
Ruy Lopez
1 e4 es 2 lbf3 lbc6 3 ..ibs lbf6 4 o-o lbxe4 s d4 lbd6 6 ..ixc6 dxc6 7 dxes lbfs 8 'it'xd8+ r;i;>xd8
Kramnik's famous "Berlin Wall". When he had to play a long match against Kasparov in
2000, his main concern was to cut out any possibility of a loss as Black against his formida
ble opponent. lt worked - two wins with White, and a string of draws, and Kasparov was
unexpectedly the first player since 1921 not to win a game in a world championship
match.
There are some obvious weaknesses in Black's position. He has, for example, the dou-
28
Kra m n ik-A ro n ian, Z u rich 2 0 1 2
bled c-pawns. However, it i s highly unlikely that White can set up an attack against the c
pawns. The problem is more that White has the extra pawn on the king side, so the big
danger is that these pawns might become active. Black has to be fully aware of this prob
lem. Another obvious point is that his king is stuck in the centre - dangerous, but with care
Black can overcome this. Here, the doubled c-pawns are useful in covering squares on the
d-file.
The Berlin Defence is a difficult line for White to try to handle. Indeed, it is difficult for
both players to handle, usually involving heavy m anoeuvring, without any clear tactical
opportunities.
9 tt::lc3
The main choice here. 9 .l:!.d1+ �e8 is natural for White, but Black often wants to play
... �e8 anyway, and maybe White will have better things to do than move the rook.
9 i.e6
...
lt would be slightly unfair to ask the reader what Black should play next, since there are
several possibilities which Black has tried in top grandmaster games. 9 �e8 is the most
...
popular, while 9 ...h6 or 9 ...i.d7 are also fully playable.

Not 9 . i.e7?! though, as this makes it easy for White to eliminate the bishop pair with
..
10 i.gs, when Black's pawn structure suddenly loses its flexibility.

10 .l:f.d1+ �es 11 tt:Jgs
Question: What should Black do here? He does not really want to give
away his bishop pair, but there is no obvious good move for the bishop.
11 i.c8!
...
This simple bishop retreat looks the most secure. Black has lost two tempi, true, but
White's knight is not on such a good square, and a return to f3 will involve the return of
the two tempi. 12 tt::lf3 i.e6 13 tt:J g s is a plausible sequence, and of course a quick draw.
29
Fig hting Chess: M ove by M ove
There are lines in the Berlin where Black could allow the capture on e6, creating an iso
lated pawn there - but the pawn is not easily attacked on e6, since White's own pawn on
es blocks any attacking lines on the e-file. Probably this is a little too early to try, as Black
has made a slight concession.
Instead, 11...il.c4?! is just about possible, but not very good, as Black will need a few ex
tra moves to bring the bishop onto secure squares; while 11...il.d7? is tactically bad, in view
of 12 tt:Jxf7.
12 h3
Hoping to keep his king side pawn majority mobile.
12 tt:Je2 as 13 a4 b6 14 b3 cs is steady and equal, subsequently drawn in V.Topalov
V.Ivanchuk, Wijk aan Zee 2012. The next few moves were 15 il.b2 il.e7 16 tt:Je4 il.e6 17 lLlf4
hS, and some moves later White exchanged on e6 without gaining much. He has given
away tempi with lLlc3-e2-f4xe6 and, as a result, Black's defence is solid.
12 il.e7 13 il.f4!?
...
New? Unusual ? An innovation? Certainly Kramnik thought for a long time on the next
couple of moves.
Question: What are the main choices for Black? Is one better than the others?
13 l2Jh4
...
Yet more manoeuvring with the minor pieces. Kramnik intends to hit the es-pawn with
...tt:Jg6.
13 ...h6 is also good, intending 14 lLlge4 il.e6 or 14 lLlf3 il.e6, with equality in either case,
and he has regained his tempi with the bishop.
Since I have asked my question, I suppose I should give an answer. My own view is to
prefer 13 ...h6, rather than the extra knight manoeuvring, which a potential loss of tempo.
See the next comment.
30
14 e6
With open play and hopes of a microscopic edge. The problem is that the edge is far too
small after reasonable defence by the opponent. 14 tt:'lge4 tt:'lg6 14 .th2 keeps the tension
for longer.
14.. .f6
Kramnik decides that he can hold with reasonable care.
14.. .fxe6 is a more tactical attempt. After 15 tt:Jge4 es 16 .txes .txh 3 17 .tg3 .1g4 18 f3
.1e6 19 .txc7 tt:Jfs, play is equal.
15 tt:'lf7 11gB 16 .txc7 .txe6 17 tt:'ld6+ .txd6 18 .txd6 'li;f7
Question: Isn't this a good time to offer a draw?
Play looks level and, in his earlier calculations, Kramnik would clearly have assumed
that he had equalized. Indeed, a few moves later, he offered a draw, leaving Aronian
slightly irritated. lt is still up to Black to prove that he is fully equal.
There are still some weaknesses in Black's queen side pawn structure. Black has not
been able to keep his pawn on c7, and so he can no longer guard the d6-square. True,
White's bishop on d6 does not attack anything, even though it is well advanced. If, how
ever, he can find a way to bring his knight on d6, that will be more serious. Meanwhile,
Black has to decide what to do with his other queen side pawns. White will probably want
to play .tcs at some stage, forcing Black to push either his a-pawn or his b-pawn (with
...b7-b6}, to allow him to move his rook. Extra pawn pushes, to cover existing weaknesses,
can soon end up with other weaknesses.
I was interested to see how Kramnik was going to handle the position. There is no rea
son to believe that White has any big advantage, but how was Black going to hold the bal
ance?
19 f3
31
Not so much to allow the king to develop to f2, but rather, as we shall soon see, to give
his bishop a retreat square.
19 cufs 20 ..ics b6
...
The ... b7-b6 push is needed, Kramnik decides.

21 ..if2 .l:!.gd8 22 a4
Question: What should Black do next, to defend his position?

General ideas and a specific move here, please.
22 CUe7!
...
Black could just about drift around with perhaps 22 ...h5, but it is starting to be risky.
The computer suggests that White has only a slight edge, but if White plays with clear and
confident moves, it could easily end up with Black's position deteriorating. Black, too,
needs to play with confidence.
Black's plan is not the obvious ... cuds, to try to exchange knights and end up with the
drawish opposite-coloured bishops. lt is in any case difficult to arrange this. What he is try
ing to do is defend with ... c6-cS and ...cuc6. This creates extra weaknesses, particularly on
the light squares, with possibilities for White of trying CUbS and CUd6. lt is uncomfortable,
but it has to be done. Kramnik's defence is the traditional one: that for each attacker, there
needs to be a defender in play. More specifically, a knight on c6 is an excellent defender,
giving counterplay as well as passive defence. White cannot quite break through, and
Black, with care, holds the draw.
23 as cs 24 cubs cuc6
32
Kra m n ik-A ro n ia n, Zurich 2 0 1 2
Everything i s covered, if only just. Black has defended it extremely economically.

25 .l:.xd8
The knight-versus-bishop exchange with 25 0.c7 does not achieve much. Play is level af
ter 2 S .. Jhd1+ 26 .l:Ixdl .l:Ic8 27 axb6 axb6 28 0.xe6 Wxe6 29 .l:Ie1+ Wf7.
25 ... .l:.xd8 26 axb6 axb6 27 l!a6
White has no time to prepare his pieces, before the attack with the rook. If 27 .i.e3?!
then 27...0.b4 and Black already has the initiative.
21 .l:td1+ 28 'it>h2 .l:td2
.••
A lot of pieces and pawns soon drop off on either side. Indeed, the reader might well be
puzzled why the players did not agree a draw a few moves earlier. At the time, I speculated
that perhaps they wanted to avoid having to play an exhibition rapid game if they had
agreed an early draw. One very serious game in a day is enough.
I did not know that Kramnik h ad offered a draw slightly prematurely. Aronian wanted
33
to play it out to its completion, to show that he was the one who had the right to offer the
draw.
29 .i:!.xb6 .i:!.xc2 30 lt:Jd6+ �e7 31 lt:Je4 lt:Jd4 32 .i:!.b7+ 'it>f8 33 l:!.c7 .i:!.xb2 34 .i:!.xcs tt:Jfs 35 lt:Jg3
.i:!.xf2 36 tt:Jxfs .i.xfs 37 .i:!.xfs .i:!.a2 Yz-Yz
Kramnik, like Anand a week earlier (in the Bundesliga), was careful to make as certain
as at all possible not to lose two consecutive games.
Game Three: Kramnik-Aronian

This is the strangest of all the nineteen games covered in the two main matches. lt was
strange at the time, and looks dodgy now. There was no need for Aronian to have sacrificed
his queen, for lesser pieces. He was already a point up against Kramnik, and if he was a
strong enough player to hold out for a few draws, in a six-game match, he would win
against one of the most solid match players in history. Once, Kramnik beat Kasparov in a
World Championship match, with just two wins and a whole string of draws. Aronian had
the perfect opportunity to emulate Kramnik's play a dozen years earlier.
If you go on the basis that if you are a point ahead in a match, the classic response is to
play steadily, and remain a point ahead. This was an exhibition match, not part of the
world championship, and both players wanted to play interesting games, yet did not want
to give away their main opening secrets. This backfired, with first Kramnik, and then Aro
nian, finding interesting but bad novelties.
After this game, points were level. lt is hard to gain an impression as to who has the
psychological edge. The easy answer is that, quite probably, the match will end up level. lt
is indeed a possibility that both players will want to settle down and play three solid and
unexciting draws. Against that, it has to be remembered that, if the players finish with any
quick draws, the match conditions will require them to play a rapidplay finish, after the
main game, to please the spectators. lt is arguably less exhausting for the players to play a
full-length game. If the games are long and intense, it is more than likely that there will be
34
an odd number of wins, rather than an even number, in which case one of the players will
win .
But perhaps we are going ahead of the story. I am writing just before the fourth game,
and speculating about the finish, when I should be writing up the third game.
24th April 2012

V.Kramnik-L.Aronian
Zurich 2012 {Ga me 3)
Four Knights Game
1 e4 es 2 lbf3 ltJc6 3 lbc3

This is one way of avoiding the Marshall Gambit: 3 �bS a6 4 �a4 lbf6 5 o-o �e7 6 l:te1
bS 7 .ib3 o-o 8 c3 dS ! ?, the main line of which is 9 exds tt:lxds 10 tt:lxes tt:lxes 11 .l:i.xes c6,
although Marshall himself played 11...tt:lf6 against Capablanca.
Note the way in which Marshall tried to gain a tempo, in playing ... d7-d5 in one turn, in
stead of ... d7-d6, quietly defending his pawn structure. Sometimes in the main line Ruy,
Black will push the pawn again, with ... d6-dS, after long preparation. other times, Black
merely tries to defend the pawn centre on d6 and es.
In this game, in the Four Knights, Aronian is clearly inspired by Marshall's concept, play
ing a sharp ... d7-dS. There is nothing wrong with this, so long as he gives nothing away to
his opponent.
3 ...ltJf6 4 d4 exd4 5 tt:lxd4 �cs
Unusual, but there are some well-known drawing lines with s ... .ib4 6 tt:lxc6 bxc6 7 .id3
dS. lt is too early to agree a quick draw and have a rapidplay game.
6 �e3 .ib6
7 'ir'd2
35
Question: Kramnik clearly wants to develop his pieces quickly, to try to keep
an initiative, before Black can complete his development. Does 7 t'Llxc6 look
like a good idea? Is it better? Worse? Or different?
lt is difficult to see how White can achieve much, in terms of an edge, by exchanging
knights with 7 t'Llxc6 bxc6. The computer suggests that White might still have a slight plus
after 8 es it..x e3 9 exf6. Play on for an extra move, though, and after 9 .. SL.h6! Black is com
.
fortably equal. His king might by slightly misplaced (after 10 �e2+ 'i£ff8), but this is not a
serious problem, as Black's other pieces will soon be able to find comfortable squares. He is
helped in this by making good use of the bishop pair, versus bishop and knight, where the
knight is not on a particularly good square. Black's pawn structure is slightly unusual, but
none of his pawns is actually weak and, again, if he keeps his pieces active, he should be
fine.
7 o-o 8 o-o-o l:!.eB 9 f3
...
g ds
...
A highly direct double pawn push, in the style of Frank Marshall, close to a century ago,
and of course Levon Aronian. Black plays ... d7-d5 in just one go, the possible disadvantage
being that Black has not yet developed his pieces - necessarily so, in that he has not yet had
the chance to move his queenside bishop. When play suddenly opens up, before all the
main pieces have been developed, tactics can quite often arise. White has not yet fully de
veloped his own pieces either, his light-squared bishop remains unmoved, and his dark
squared bishop being open to attack.
There are other ideas, such as 9 .. d6 and ... SL.d7, perhaps more in the style of Steinitz,
.
rather than Marsh all.

1 0 exds t'Llxds 11 it..g s
36
Kra m n ik - A ro n ia n, Z u rich 2 0 1 2
Question: Tactics time! I s there a safe way for Black of remaining at least equal?
And what are your impressions about Aronian's queen sacrifice, with 11 ...l2Jxc3
12 �xd8 l2Jxd1 - ? Indeed, would you even have considered this yourself?
11 l2Jxc3?
...
This is one of the several examples in these two matches in which big innovations turn
out to be big mistakes. Kramnik (game one) and Aronian (here) got caught by aiming to hit
the opponent with tactics, only to find that their own position has crashed. Later, both An
and and Gelfand got caught themselves. lt is not enough for something to be interesting
and innovative. Against strong opposition, it also has to be good and accurate. Many
games in the 2012 matches failed this test.
Kramnik was clearly under pressure. This is what happens in gambit chess. The oppo
nent sacrifices, usually at some stage in the opening, gaining some sharp attacking ideas,
and with the hope that the defender will not be able to find fully accurate defensive moves.
Unfortunately for Aronian, Kramnik is a strong enough player to calculate vast numbers of
lines in great depth, and he found the best lines, and won.
Was Aronian in a bad position already? Or did he have a safe alternative?
The most straightforward attempt is 11 ...l2Jf6 1 2 l2Jxc6 Wixd2+ 13 l:txd2 bxc6 14 ..tc4.
Black's pieces are fine, but his fractured pawn structure will be of concern, through to the
endgame. An irritating edge for White.
Another possibility is to concede the bishop pair with 11 ...l2Jde7 1 2 l2Jxc6 Wixd2+ 13
..txd2 l2Jxc6 14 l2Jds ..te6 1 S l2Jxb6 axb6 16 a3 ..tfs, but it would be difficult for Black to
demonstrate full equality. 12 l2Jb3 also keeps Black under some pressure: remember that
his pieces are not yet fully developed.
11 ... ..txd4, this time only a temporary queen sacrifice, is not fully acceptable either: af
ter 12 ..txd8 ..te3 13 l2Jxds ..txd2+ 14 �xd2, Black will soon drop a pawn on c7.
37
The best equalizing opportunity is 11 .. .f6!.
Did the reader reach this far? We are only one move after the last test position, but sev
eral other ideas needed to be considered, and usually to be rejected.
Attack is met by Black with counter-attack, with minimal loss of tempo. No unnecessary
retreat, but no wild sacrifices either. Now:
a) 12 .i.c4 looks sharp, but after a few tactics, play fizzles out to a drawn double-rook
endgame: 12 .. .'�:Jxd4 13 lZ:lxds .i.e6 14 .i.xf6 gxf6 15 lZ:lxb6 axb6 16 'ii'xd4 iYxd4 17 J::!.xd4
.i.xc4 18 11xc4 .l:!.xa2 19 'it>d2 l:!.e7.
Perhaps this is too far for m ost of us to calculate in our heads, but the sequence of tacti
cal play is logical enough, without any big surprises.
b) 12 lZ:ldbS looks interesting, one of the main points being that after 12 ... lZ:le3? 13 �xd8
lZ:lxd8 14 .i.xe3 .i.xe3+ 15 'it>b1, the knights for once out-perform the bishop pair. Black has
not yet found time to develop his queen side pieces.
38
Kra m n ik - A ro n ian, Zurich 2 0 1 2
Still, Black seems to equalize comfortably with 1 2 ...lt:Jcb4!.
Black quickly plays ... c7-c6 and White, sooner or later, has to give away a tempo with the
dark-squared bishop.
The pyrotechnics by Aronian therefore seem unnecessary, and worse, bad. Aronian
would seem to have had the possibility of equalizing by normal means.
12 i.xd8 lt:Jxdl
Question: How should Kramnik handle Aronian's piece sacrifice?
13 i.xc7!
Kramnik is attempting to make the absolute maximum out of his position, perhaps al
lowing concessions and various tactical opportunities, but also trying to ensure that he has
queen and pawn versus assorted pieces, rather than playing with just queen versus other
39
pieces. A humble pawn, if working with other pawns, often tips the balance.
White can also try 13 �g S ! ? 'Llxd4 14 'ifxd1, when Black has rook and bishop for the
queen. Possibly White can claim a slight edge, or possibly Black can say that he is fully
equal. lt would be unrealistic, though, to suggest that Black is better.
That being so, whether or not Kramnik's 13 �xc7 can be regarded as a refutation or just
a good move, Aronian's queen sacrifice seems suspect.
13 ... �xc7 14 'Llxc6 'Lle3 15 � bs
Very sharp. Kramnik would h ave seen this before his 13th. Of course, the real tactics
arise in the next few moves. Kramnik, despite his lapse in round one, is still clearly playing
confident chess, believing that, as one of the world's all-time greats, he will have the ana
lytical strength and resourcefulness to combat any hand-to-hand battles by his opponent.
lt is a close call, but it seems that Kramnik has handled it extremely well.
1S ... bxc6
Question: Tactics corner. Aronian grabs the knight, but allows his rook to drop.
Could he have done better by bringing one of his bishops into play?
The obvious 1S ... �f4? does nothing for Black after 16 'Lle7+ Wf8 17 g3 �h6 18 f4, since if
18 ... .l:!.xe7, then 19 'ifd8+ and mate next move.
Instead, 1S ...�f5 is interesting, with a few tactics after 16 'Lld4 �f4 17 �xe8 'Llxg2 18
�xf7+ 'it>xf7 19 Vi'xf4 'Llxf4 20 'Llxfs.
40
Kra m n ik - A r o n i a n , Z u rich 2 0 1 2
The trouble i s that, with the help of the bishop desperado sacrifices on c7 and f7, White
is a pawn up, and should eventually win. This is an example of why it is still important to
count the number of pawns, even when the other pieces on both sides are being hacked
around.
16 kxc6
Question: More tactics to consider, every move. The computer

likes 16 ... ..tf4 here. What is likely to happen next?
There is another chance to set up some tactics with 16 ... kf4, but yet again, with best
play, White ends up a pawn up in the endgame.
One can reasonably assume that Kramnik, some moves in advance, will have noted that
he can play 17 �d4, and that, at the very worst, he can end up with equality with a perpet-
41
ual check. This would happen after 17 .. .<�:Jf1+ 18 'it>d1 {not 18 �xf4?? lle1 mate) 18 .. .<�:Je3+
19 �cl, etc. This is a useful back-stop when trying to analyse during a game. Realizing that
he is at least equal, Kramnik does not have to worry that he might be losing. He can play
the sharp line with confidence.
But can White improve on this?
Time to try 17 'it>b1! l:tb8.
Then 18 ..txe8 l:txb2+ leads to complicated equality after 19 'it>a1 or 19 'it>c1. Or at least
the computer suggests that it is equal.
White can, however, play for much more with 18 'ifd4! ..tes 19 'ifxe3 llxb2+ 20 'it>c1
.l::rb4.
lt would be difficult for most of us to keep track of what is in going on in such a posi
tion, several moves in advance. The basic point, though, is that with forced simplification
White will remain a pawn ahead, after 21 ..txe8 .1l..f4 22 .l::re 1 .1l..x e3+ 2 3 llxe3.
42
lt seems almost unfair that, yet again, after s o many wild tactics on either side, the final
result is a pawn up for White in a joyless endgame for Black. But if a player is better at the
start of a string of tactics, it is logical enough that, with good play by either side, he should
remain better at the end of the tactics.
16 tL:lc4 17 'ikd4
...
Other queen moves look good too. The tension has gone, and Kramnik is on the way to a
win.
17 ....te6 18 .1l.xa8 ..tb6
Aronian sets an extremely deep trap, avoiding the more obvious 18 ....l:Ixa8.
19 ii'd3
Question: Just about any safe queen m ove is good, but which is the
most accurate? Can he force his opponent to resign by around move 30?
I leave this as something of a trick question. The position is not as clear as I thought. The
obvious reply, protecting the queen with 19 'ife4, at first seems to keep a decisive material
advantage, but if we dig closer, Black's mighty minor pieces give him enough for a perpet
ual. After 19 ... ..te3+ 20 'it>d1 tLlb2+! 2 1 �e2 .tb6, there is an unusually convoluted repeti
tion on 22 c;t>d2 tt:lc4+ 23 'it>c1 ..ie3+ 24 'it>d1 tt:lb2+ 25 'iii>e 2 ..ib6, and the roundabout con
tinues.
The l ast try for White would be 25 'it>e1 (or, earlier, 21 c;t>e1), but again there is another
repetition after 25 ... ..td7 26 'iVb7 .tb6+ 27 'it>d2 il.e3+ 28 �el ..tb6+. Plenty of checks and
repetitions, but not quite enough for Black to checkmate. And of course White does not
have quite enough for the win.
This would be an extraordinarily difficult position to try to analyse over the board.
Quite possibly Kramnik's instinct would have made him decide that the minor pieces are
far too difficult to handle in open play. lt seems that Kramnik played the best move here,
43
ignoring his bishop, and keeping his queen and rook developed and centralized.
19 'ifh4 .Uxa8 20 'it>b1 is another good try for the win, with a different defensive ap
proach. lnstead of keeping his queen and rook active, White's approach would be to keep
his king and queen as safe as possible.
Question: The final tactical exercise. What happens if Black tries 19 ....Ud8 ?-
19 .:xa8
...
Maybe Kramnik was hoping for 19 ... l:rd8. Then White has the visually attractive 20
i.ds ! ? .l:i.xds 21 'ii'xc4 nd1+ 22 �xd1 i.xc4 23 b3 and there has been simplification, and the
rook and two pawns, well coordinated, should eventually force a win.
The commentators at the time, looking at their computers, noticed that after wild play
with 20 "i¥e2 .ie3+ 2 1 'it>b1 tba3+ 22 Wa1 .ic4 2 3 .t'td1 l::tx a8 24 l:Id3, White is still winning.
Maybe so, but would Kramnik, still slightly out of form, have tried such an obscure line?
Sometimes it is best to keep things simple.
Anyway, White has two good replies.
20 .l:i.el I!.d8
44
Kra m n ik - A ro n ia n, Zu rich 2 0 1 2
Question: Should White try to speed things up,

with 21 l:txe6 - ? Then 2 1 ...l:txd3 22 l:te8 is mate.
This would indeed speed up the finish, but the wrong player wins. After 2 1 .l:txe6?? fxe6
22 'ifxc4 .i.e3+ 23 �b1 l:td1 it is Black who gives checkmate. Beware of back-rank howlers!
21 'ife4
While writing my blog, I noted that "there is not all that much that needs to be said. The
battle between queen versus bishop and two knights is potentially interesting, but White
has two extra pawns, and once these pawns start to advance, Black's minor pieces start to
be repulsed." I was getting tired!
Clearly White was winning, and clearly there was a big time scramble, with Kramnik in
particular having much to think about. Somehow, Kramnik l ost track with his position, and
there were still a few fighting chances for Aronian.
45
21 ... gs
At the time this appeared to be slightly desperate, creating pawn weaknesses. I had ex
pected 21 ... h6. Still, White is winning anyway.
22 c3 �cs 2 3 .l::i.e 2 h6
24 g3
Perhaps Kramnik is concentrating a little too much on the square-by-square approach.
24 b3 seems more dynamic, forcing Black to retreat his pieces:
a) 24...�a3+ 25 'it>c2 �f5 26 �xf5 .l:!.d2+ 27 .l:!.xd2 tt:le3+ 28 'it>d3 tt:lxf5 29 'it>e4, and after
the flurry of tactics, White has a simple win in the endgame.
b) 24 ... tt:ld6 25 �c6 .l:!.c8 26 �a6 and Black has to disorganize his pieces. No crashing
through, but it is progress.
24... as 25 f4 a4 26 fS �ds 27 �d3 �b6 28 b3 axb3 29 axb3 tt:Jas
30 .l:!.e8+!
Kramnik was never completely out of control, even during his time scramble. In other
46
Kra m n i k - A ro n ia n , Zu rich 2 0 1 2
words, h e i s winning ! That i s all that really matters.

Kramnik's exchange sacrifice looks strange at first, giving away his material advantage.
He is about to have queen and two pawns, versus rook, bishop and knight. However, there
are two points greatly in favour of White: firstly, it is difficult for Black to cover the two
connected passed pawns; and secondly, Black's knight is completely out on the edge.
There may be other ways of forcing through a win, and therefore it would be unfair for
the reader to ask what White should have done next. The simple point, though, is that after
his exchange sacrifice, Kramnik was able to win without difficulties.
30...lixe8 31 'ifxd5 l:!d8 32 'ii'b 5 ltd6 33 'it>c2 'it>g7 34 b4 tZ:lb7 35 c4 l:!f6 36 g4 tt::Jd 8 37 c5 .Yt..c 7
3 8 �d7 ti:Jc6 3 9 b5 tZ:la7 40 'ifxc7 tt::Jx b5 41 'ife5 tt:'la7 42 'it>d3 1-0
Probably both players could h ave handled this slightly better during the time shortage,
but there is no doubt that Kramnik was better throughout.
Game Fou r: Aronian-Kramnik

Not much to think about, one might assume. Play was level and accurate, and the players
agreed a draw at move 25. lt happens. One cannot force an opponent to make a mistake, and
if no mistakes by either side are made, the logical result is a draw. Like it or not, that is chess.
If anything, what is surprising is that only half of top level games end up in draws. Be
tween both players rated over 2700, the approximate draw percentage is around 60%, with
a very small plus score for White. This is based on results (from the beginning of April 2011
to the end of March 2012) in 301 games, as recorded by Live Chess Ratings, 2700. So most
games at the highest level are indeed draws, but not as many as some might suspect. Even
2700+ players do not play perfect chess.
What probably happened was that Kramnik in game one, and Aronian in game three,
have been jolted by their recognition of fallibility in chess, and both players felt they
needed to settle down. No speculative chess, just solid chess. A similar dynamic can be seen
in the World Championship match, a month later.
The tell-tale sign is not so much that they played quiet and accurate moves. Again, ac
curacy is not a problem. Rather, it is that sometimes players are a little too eager to rush to
safety, or, as here, they are happy to agree a relatively quick draw, when there was still in
teresting play in the position.
There is only one question to be asked of the readers in this game, but it is quite a big
question. What might have happened if the game were to be continued?
2 5th April 2012

L.Aronia n -V.Kramnik
Zurich 2012 {Ga me 4)
Ruy Lopez
1 e4 e5 2 tZ:lf3 tt::Jc6 3 j_b5 ti:Jf6 4 0-0 tt::Jxe4 5 d4 ti:Jd6 6 .txc6 dxc6 7 dxe5 tt::Jf5 8 fixd8+ 'it>xd8
47
Another Berlin Defence and another draw in the main line.

lt is, however, a myth to suggest that the line is notoriously drawish, queens off, a bit of
sensible manoeuvring, and a draw agreed. The database I mentioned (2700+ players
against other 2700+ players, early 2011 to 2012) give the score as +5 =5 -2 (from White's
point of view) out of 12 games. The draw proportion is therefore lower than the average
(about 40%, as opposed to about 60%). Players with Black cannot rely on an easy half-point.
They still have to work for it.
The importance of Kramnik's ability to halve out every time against Kasparov, in the
famed 2000 match, was that Kramnik h ad exceptionally deep knowledge of the strategies
of this line as Black. Possibly, who knows, Kasparov did not have equally deep knowledge as
White. So Kramnik was better prepared.
9 h3
Varying from the 9 tt:lc3 of game two.
Kramnik has had a recent loss in this line against Karjakin. Bearing in mind the previous
comments, the young grandmaster managed to do something that Kasparov failed to
achieve. Karjakin was, on this occasion, able to set up a mobile kingside pawn majority,
after 9 tt:lc3 .i.d7 10 h 3 b6 11 a4 as 12 g4 tt:le7 13 tt:Jg s .i.e8 14 f4 h6 1 5 tLlf3. Quite how mo
bile the pawns were is open to question after 1S ... g6 (S.Karjakin-V.Kramnik, Russian Cham
pionship, Moscow 2011). But zoom forwards some moves, and we reach a complicated mi
nor piece endgame.
48
White to play
Rooks have just been exchanged on d8, and one can easily gain the impression that we
are about to head for a quick draw. However, Karjakin has a sacrificial breakthrough in his
mind, and he plays 30 e6! tt:Jxe6 (30 ....i.xc3?? 3 1 e7, queening) 3 1 .i.xg7 tt:Jxe7 32 tt:Jes.
White's knights are dominant; while one of the main problems for Black in the Berlin
Defence is that it can be extremely difficult to create a passed pawn from the crippled
queen side majority. White's king side is far more mobile and, indeed, he was able to create
a decisive passed pawn, after 32 .. .fs 33 Cbf6 gS 34 Cbc6+ �b7 3 5 Cbd8+ �g8 36 Cbc6+ (a
harmless repetition, to help reach the time control} 36 ... �b7 37 tt:Jd8+ �b8, and now 38
fxg s hxg s 39 Cbh 7, regaining the material at last, and then creating a passed pawn, and
ultimately winning.
With modern technology, it is easy enough to find the whole game on the Internet, and
49
Fig hting C h e s s : M ove by M ove
so there is no need to repeat all the moves. The reader might also note a win for Black in
D.Navara-S.Karjakin, Wijk aan Zee 2012.
Clearly Karjakin has excellent understanding of the Berlin. He is, at the time of writing,
one of the five players aged 25 and below in the top ten world list. Carlsen, Radjabov and
Caruana get a full mention, in the later part of this book, being the three leaders in the re
cent Tal Memorial; of the other two Nakamura was not quite at his best in that tourna
ment, whereas Karjakin did not play at all. So it is a good opportunity to pay my respects to
him here.
g 'it>eB 10 tt:lc3 hs 11 ..igs ..ie6 12 b3 ..ie7 13 .Uad1
...
For 13 .l:lfd1!? see the next note.

13 ... h4
This move is part of Kramnik's defensive repertoire. The main problem in Black's set-up
is not so much any weakness of the queenside doubled pawns (it would be extremely diffi
cult for White to attack them), but rather, that Black has fewer pawns on the king side. The
... h S-h4 plan is designed to prevent White from setting up a kingside pawn roller.
The danger then, from Black's point of view, is that his h-pawn can be difficult to de
fend. If, however, he can cover his h-pawn effectively, without his pieces being excessively
tied down to defence, he should be equal.
There is another Karjakin game that can be mentioned here. In V.lvanchuk-S.Karjakin,
Wijk aan Zee 2012, White had the other rook on dl. From the stem line in the opening, play
went 9 tt:lc3 'it>e8 10 h3 h 5 11 l:td1 ..ie7 12 i.g 5 ..ie6 13 b3 h4 (reaching the same position as
in the game, but with 13 .l:.fd1 rather than 13 .Uad1) 14 'lio>fl (slightly more attractive for
White? - perhaps, but it was still only a draw) 14... a5 1 5 a4 l:th 5 16 ..ic1 ..ib4 17 tt:le2 ..id5
18 tt:le1 l:rd8 19 ..ib2 l:th6 20 c4 ..ie6 21 tt:lf3 ke7 22 .l:!.xd8+ 1/2-V2.
50
The computer seems to think that White i s slightly better here. What do you think? I
leave this as an unanswered question.
Back to the Aronian-Kramnik game.
14 .l:lfe1
Apparently a minor innovation, but it is unlikely to have any drastic impact, one way or
the other.
14 ... .l:ld8 15 .l:lxd8+ 'it>xd8
With reference to lvanchuk-Karjakin in the earlier note, it is clear that in such lines
Kramnik prefers to move his king to the queenside, rather than the king side, where it is
possibly still slightly exposed.
16 'be4 b6 17 1Lf4 'it>cS 18 lbegs
This eliminates Black's bishop pair. lt is far too early for White to claim any sort of ad
vantage.
18 ....1Lxgs 19 1Lxgs .ids
51
White's knight on f3 is now pushed away from any attack square.

20 lLlh2 cs 21 l:!.d1
This helps explain why White moved the knight to h2, rather than to d2. He wants to
put pressure on the bishop.
21 ....i.c6 22 c3
To prevent ...lt:Jd4.
22 ... as 23 lt:Jg4 .i.d7 24 f3 a4 25 'it>f2 Yz-Yz
The end of the game? Or should it have been only the beginning? This is the sort of posi
tion where players of equal strength would be happy to take an early half-point, but where
the stronger player, whether as White or as Black, would want to try to grind it out indefi
nitely. Or perhaps, with two players of equal strength, one of the players will want to grind
it out anyway. The position is equal, but it is not a clear draw. One of the players might well
make a minor inaccuracy, maybe pushing a pawn carelessly, or maybe moving a piece
from a better square to a worse square. Then the opponent would have legitimate ways of
trying to play for a win.
The exercise I am setting asks you to make good use of your computer and its chess en
gine.
Exercise
Try to play out this position, with the help of the computer. In most positions, unless we
are in the middle of an exchange or some kind of tactic, there will be something like three
to five choices on each turn. You will not be able to analyse everything, and neither can I.
What I am suggesting is that you carry on trying out apparently good moves, and keep try
ing variations until there is a completely drawn position, or alternatively, where one of the
players suddenly has a slight edge.
The idea of the exercise is to give the reader a better grasp of holding the balance in a
level position. lt is so easy to make a slight stumble. The hope is that your opponent will
52
slip up before you do. You will almost certainly find that, even with the help of computer
suggestions, it will be difficult, unexpectedly so, to keep to complete equality throughout.
According to my own computer, the main choices here are:
a) 2 S .. J.:te8;
b) 2 S ... .l:!.h 5 ;
c ) 2 S . . .axb3 26 axb3, and either 26 ....i.e6, o r one of the rook moves already suggested;
d) 2 S ... c4.
Now it is up to you.
Some suggested a n swers

If Black is determined to head for a drawn position, with as little fuss as possible, the sim-
plest line here is to offer quick simplification with 2S ...axb3 26 axb3 i.e6 27 b4 (27 c4 al-
lows Black to think about playing his knight to d4) 27 ... cxb4 28 cxb4 .l:!.e8. As White, I would
perhaps be slightly worried about allowing Black the possibility of creating a passed pawn
on the queen side, and certainly it would seem sensible to simply further, with a knight ex
change, with 29 tt::le 3 tt::l x e3 30 i.xe3 i.fs 31 i.gs .l:!.xes 32 .l:!.d8+ �b7 33 .i.xh4.
Determined players might well squeeze it out for a few more moves, but it is difficult to
see how, even if one of the players could create a passed pawn ( ... C7-c5 for Black), he would
have much chance of creating a strong one.
In fairness, this might be the sort of thing that both players would have been thinking
about, when agreeing a draw. After all, Kramnik's original intention with 24...a4 would
have been to create the possibility of quick simplification on the queenside.
Suppose, though, Black were to try to tweak it out for m ore than a slight edge, albeit
while allowing an extra degree of risk. Again, trying out equalizing moves with the com
puter, play could continue with 2S ... c4 26 b4 .l:!.e8. This looks good if Black's rook can some
how reach White's side of the board, but how can it get there? White has everything cov
ered on the d-file. Try 27 a3 i.c6 28 .l:!.e1 �d7, all apparently computer-safe moves.
53
But White suddenly has a tactic: 29 e6+ l:txe6 30 lLleS+ �c8 (not 30 .. .'it>e8?? 31 t2Jxc6 and
Black's rook is pinned} 31 t2:lxf7 l:txe1 32 �xe1 and Black is definitely under pressure.
Going back, 2S ... l:te8 seems safe enough, waiting to see what he wants to do with his
queen side pawns. Then perhaps 26 b4 (again, there are other moves) 26 ... cxb4 27 cxb4 ..ie6
28 a3.
Such positions, characteristic of the Berlin, can be positionally disorientating if your in
stincts are for trying to make progress. Looking at the pieces, any king move will end up on
a slightly worse square. Shifting the knights on either side, or White's bishop, would re
lease the pressure on the opponent. Black would also like to keep his rook lined up against
the es-pawn. So what is left, if there is no constructive pawn move? White's rook is mobile,
but cannot attack anything. Black's bishop is mobile, but can do nothing more than prod
the rook. A possible line might be 28 .....ib3 29 .l:i.d3 ..ic4 30 l:td2 ..ibs, and then a rook move,
and then a bishop move, and so on.
54
Finally, and the reason why I feel slightly surprised by the early draw agreement, Black
can prod the bishop with 2S ... .l:!.hs ! ?.
If White's bishop moves, then Black's other pieces can start to move. This is the way in
which Black is starting to explore for a win. For instance:
a) 26 i.f4 will leave White a little uncomfortable after 26 ... axb3 27 axb3 i.e6.
b) 26 i.d2, covering both the c3- and b4-squares, might be a better defensive plan. Black
still seems to keep the initiative after 26 ... axb3 27 axb3 i.e6 28 b4 cxb4 29 cxb4 CDe7 30 f4
(Black threatened ... i.xg4, winning the es-pawn) 30...l:th8, but it still fizzles out to equality
after the unexpected 3 1 .i.e1!.
White may have what is theoretically the bad bishop, but it still creates pressure on the
h4-pawn, preventing Black from gaining full control with his rook and knight. If Black is
now content with equality, he could worry the rook with 3 1...i.b3, and there is little oppor
tunity for White to improve his pieces.
55
Perhaps Aronian has everything covered, but it would have been interesting to see what
would have happened, if Kramnik had declined the draw offer.
Game Five: Kramnik-Aronian

The fifth game ended up in a steady and unexciting draw, but of course there is nothing
wrong with "steady and unexciting" draws. What this means, in top-level play, is that nei
ther player has made a mistake, surely no bad thing in a game of chess. Aronian, in game
five, was content to give nothing away as Black. Looking backwards, he would not want to
have tried overexcited chess, as in game three. Looking forwards, his best chance of win
ning the six-round match is to put some pressure on his opponent as White.
There was perhaps the slightest of suggestions that, at one stage, Aronian could have
taken the initiative with Black. Whether or not this was anything significant, the chance
was missed, and play soon ended up with another draw.
27th April 2012

V.Kramnik-L.Aronian
Semi-Slav Defence
1 tDf3
Back to normal resumption. Kramnik, after his game one loss, switched his openings
around in game three, with 1 e4, and won in a Four Knights. it would be nai·ve to suppose
that Aronian would have tried the queen sacrifice again. He would undoubtedly have var
ied at some stage.
1 ds 2 d4 tt:lf6 3 c4 c6 4 tt:lc3 e6 5 kgs h6 6 .ixf6 �xf6 7 e3 tt:ld7 8 .ie2
...
Kramnik avoids 8 ..id3 this time. The virtue of playing the opening quietly, with i.e2,
56
Kra m n ik-A ro n i a n , Zurich 2 0 1 2
rather than the apparently more aggressive .i.d3, i s that the d4-square i s ultra-safe, with
the queen providing an extra guard on d1 or, perhaps later, one of the rooks reaching d1.
8 .. .'it'd8
Question: Why this queen retreat? Is there nothing else?
Black's idea is not as strange as it looks. He is happy that he has induced his opponent
to concede the bishop pair, with the veiled threat of the Botvinnik Variation (see notes to
game one). Once he has done that, what can he do next? The danger is that the queen and
the dark-squared bishop are in the way of each other. If, for example, Black were to play
...i..e 7, the queen does not have many squares; while if instead ... �d6, there is the danger
of a pawn prod with e3-e4-e5. The argument is that Black h as made a gain with the bishop
pair, but that he will need at some stage to give up a tempo or so with the queen.
There are, of course, credible alternatives: 8 ... dxc4 9 .txc4 g6 transposes into game one.
A computer search suggests that, as well as 8 ... dxc4, Black can try 8 ....i.b4, or even 8 ... i.e7
(though he will then have to take care of the queen).
Perhaps unexpectedly, 8 .. lt:Jb6! ? is another possibility, prodding the c-pawn. This gives
.
away a couple of tempi after 9 cS lt:Jd7, but it is no great loss, seeing that White's c-pawn
no longer adds pressure to the centre, whereas Aronian's own move lost a tempo anyway.
Black can think, in some variations, of opening up the centre with ...e6-es. This line looks
fully playable for Black.
9 o-o i.e7 10 'i'Vc2 o-o
11 a3
Question: The computer suggests that there are more than twenty
reasonable moves for White, each of which gives equality, and very
little more. Why did Kramnik choose this particular one?
57
If no m ove is likely to give any sort of clear advantage, then in practical terms it is often
a good idea to play a quiet waiting move. This can often imply an insignificant-looking
pawn move, slightly improving the pawn structure, while waiting to see exactly what the
opponent is planning to do. Here, the pawn is just slightly better on a3 than on a2, allow
ing White to try b2-b4 in some lines, or maybe merely preventing ... ..tb4, or maybe creating
a hole for the bishop, after Black exchanges with ... dsxc4; ..txc4, and then a retreat to ..ta2.
Who knows? lt is no bad thing to open up possibilities for later, given the opportunity for a
quiet move.
The obvious alternatives are the various rook moves, though it is difficult to predict
which rook is best, and where it should move to. Of the top ten moves suggested by the
computer, two each are given as either of the rooks moving to cl, dl, and el, while 11
.l:f.abl is the seventh. Time perhaps to wait? White would not want to lose a tempo, if he
finds he has not placed the rook the m ost effectively.
The immediate 11 e4 dxe4 12 lLlxe4 �c7 opens up play, but then Black is equal. lt is of
ten better to maintain the tension. So we return to think about the best rook move.
11 .l:i.adl looks in many ways the most natural move, but what then happens after
11...b6 - ? 12 e4 is still too early for an edge, so what else can he try? Attention may focus on
12 cxds cxds, but then we have the problem that the rook is not on the best square. lt
would be better on cl instead.
By a circuitous path, ll l:f.fdl may well be regarded as most accurate, and if 11...b6,
White still has several reasonable choices, maybe 12 .l:f.acl or 12 cxds, or possibly 12 e4 or
12 �3. Clearly this cannot possibly be calculated trying to analyse move-by-move. What is
important is positional judgment, and being able to anticipate what to do after the oppo
nent's ideas. What, for example, would White do if Black were to try ... b6 - ?
11 . b6
..
He has to do something with the "bad bishop", often the biggest positional problem in
the Queen's Gambit.
12 .l:f.fd1
Note that if White had played 11 l:f.fd1 instead, and if Black were still to try 11...b6, then
White might find something m ore constructive to do than 12 a3.
12 ... ..tb7 13 l:f.ac1
Kramnik decides what to do with his rooks, but perhaps very slightly too late. lt needs to
be emphasized of course that he has not done anything seriously wrong, and that he is not
worse, but he has lost some of his opening initiative.
Aronian still has to play carefully to try to prove he has complete equality. His next
move is startling.
58
Kra m n ik - A r o n ia n, Zu rich 2 0 1 2
13 ."il!Vb8
..
Question: Why another queen move? He has already taken on f6, retreated on d8,
and now another queen move, which attacks nothing, and defends not very much.
The first point to note is that the strongest grandmasters, while undoubtedly making
mistakes on occasion, also have the uncanny ability to keep the position in balance, avoid
ing losing, when under pressure. lt looks very much the case that White is better here, but
in a few moves, Aronian escapes with easy equality. How does this happen?
In terms of chess thinking, the central point, which makes stronger players better than
weaker players, is the question of defensive tactics. We all learn about attacking tactics, I
attack his king, he tries to defend, I think about a sacrifice, analyse it, decide that it works,
and so play the sacrifice. Or alternatively, if the sacrifice does not work, then play it posi
tionally, and aim for the next breakthrough.
Defensive tactics are more difficult to grasp, but essentially the idea is that I want to de
fend the position quickly and securely, but I need to calculate. My pieces are not at the best
squares at the moment, but my idea is that in a couple of moves, I will start to be on good
squares. So I need to ensure that, in the meantime, my opponent cannot create an advan
tage. He plays this, then I defend, then he plays this, then I defend. Is my position awful? Or
is it equal? lt requires a l ot of thought.
The queen is out of the way on b8, and White cannot attack it. The basic defensive plan
is to keep the queen on the queenside, and bring both rooks to the queenside as well. Black
could consider 13 ...l:!.c8, when he is not seriously worse, but then he still has to decide what
to do with the rest of the queenside. Another option perhaps is 13 ... a6, followed by ...b7-bS.
In the end, what Black is aiming for is complete equality, not trying to defend a slightly
worse position.
14 cxds
59
Now that Black has given away a tempo, it is time for White to end the quiet manoeu
vring. If his pieces are already on good squares, the only way to make progress is with the
pawns.
14 e4 dxe4 is equal, like a safe Caro-Kann, so instead he exchanges pawns on ds, to see
how Black recaptures. White is hoping for a modest advantage after either recapture.
Question: Which way should Black recapture?
An easy question to answer for players who use the Queen's Gambit for Black and vari
ous other lines, such as the Queen's Indian - and also, of course, for players with White
who start with 1 d4.
Black has the worse pawn structure after 14 ... exds and will usually need to work hard
to try to equalize. The main problem is the "minority attack" - two queenside pawns at
tacking three - White setting up the attack with ideas of b2-b4, quite possibly 'ifh3, then
a2-a4, and at some stage b4-b5, trying to put Black's c6-pawn in great discomfort. If, in
such an attack, Black were to exchange with ... c6xbs, he ends up with an isolated pawn on
ds, which is very weak. There are ways to counteract such an attack, but Black has to be
careful as to which methods he will need to try to defend. If in doubt, avoid.
Therefore:
14 cxds
...
The pawn structure is now symmetrical, but White is slightly ahead in development.
Aronian needed to be fully satisfied that his pieces are active enough to hold the balance.
Watching this on line, I felt confident that Kramnik should be able to squeeze out an
edge, but it seems that Aronian has defended with great accuracy.
15 'il'a4 tt::lf6
A critical position, and potentially the difference between a slight advantage for White
or mere equality. Has White got any genuine edge though? Understandably, Kramnik
60
Kra m n ik - A r o n ia n, Z u rich 2 0 1 2
thought for a long time here.
16 �a6
Question: Was this the best? Does this give White any advantage?
If not, does he have an improvement instead?
The game move allows Black to exchange his supposedly bad light-squared bishop. The
point is that if Black managed to play ... a7-a6 and ... b6-bS, he would have a standard sym
metrical Slav position, where it is difficult for White to claim any sort of edge. Black will
eventually try to open up lines with ... bS-b4.
There is also the tactical problem that if White were to try to exchange off his f3-knight
for one of Black's minor pieces, White's queen could get stuck on the wrong side of the
tracks after a later ... a7-a6 and ...b6-bs. For example, Black is better in lines such as 16 l'Lles
a6 17 l'Llc6? £l.xc6 18 1\Vxc6 bs - or, in this line, 17 l'Lld7? l'Llxd7 18 'i'xd7 .1l.d6 19 h3 bs, and
White is forced to retreat far too quickly.
Instead, Fritz (an old version) suggests doubling rooks with 16 l:tc2 a6 17 .U.dc1, but this
seems cosmetic. After 18 ... bs, Black is equal.
Houdini suggests 16 �3. aiming to prevent Black's plan of ... a7-a6 and ... b6-bS. Cer
tainly this looks better than equality, but only very slightly so after 16 .. J�.c8. Maybe White
can try stonewalling with 17 ctJeS followed by f2-f4, aiming for an edge in the middlegame,
rather than in the endgame, but then he has to think about how to cover the hole on e4.
Kramnik decided to keep things simple. lt is already starting to look drawish, provided
Black finds a few accurate moves.
16...�xa6 17 'ii'xa6 'iic8
61
18 �xc8
Question: Should White have retreated, rather than exchange queens?
lt does not make all that much difference, in that Black is equal either way. If the queen
retreats, Black plays 18 .. 'i!Yb7 with equality. Aronian has demonstrated excellent defensive
.
play.
19 .. J::tfxc8 19 tt:les i.d6 20 tt:ld3
Question: Hasn't Black even got a slight edge now? His bishop covers
both sides of the board, while White's knights do not attack much.
lt is difficult to imagine that this position is anything other than level. If Black had a
62
knight, rather than a bishop, on d6, it would be time to agree a draw. With the bishop
there, he will have good chances of exchanging one of the knights - if, for example, White
were to try ttJes. Alternatively, as in the game, the bishop might give an occasional prod
against one of the pawns, but even this does not amount to much. Bishops are at their best
in open play. With closed or half-closed positions, the bishop is no more effective than the
knight.
20 ... tbe8 21 'it>f1 'it>fS 22 'it>e2 'it>e7 23 h3 �c4
After a few moves of consolidation by both sides, Black's rook advance is so annoying
that White is induced to push a pawn. He is about to weaken his pawn structure, though
not seriously so.
24 b3 �cc8 25 a4
Question: Testing the computers. Of the two chess engines I was using,
one said that that Black was better, while the other implied that the
position was so dead equal they might just as well offer a draw.
Which assessment do you think is more accurate?
2s ....ia3
"Sometimes the battle of the chess engines can be more interesting than the game it
self, from certain points of view. Both computer engines suggest that 2S ... a6 might have
been slightly better. Houdini gives this as 0.00 dead equal, while Fritz (and I have to em
phasize that this was an old version) gives this as a slight edge for Black. Houdini suggests
the sensible consolidation in the centre with 26 'it>d2 tbf6 27 f3, followed by tbe2. it is diffi
cult to imagine any sort of breakthrough by Black. Fritz gives instead 26 tbb1 tbf6 27 tbd2,
and perhaps it is understandable why Black is given a slight edge."
These were the notes I made at the time. Fritz seems to overestimate (or perhaps, used
to overestimate) the relative strength of the bishop over knight. Computer assessments are
63
useful and are the start of any modern analysis, but they cannot be relied in 100%.
The position is equal.
26 �e2 'fle7 27 'flal .i.d6 28 'i?o>d2 aS 29 �eel �ae8
30 f3
If White wanted to finish the game with a draw as quickly as possible, the simplest way
is 30 tt:lbs 'flxc1 3 1 l::tx c1 �xc1 3 2 tt:Jxcl. The computer that preferred the bishop suggested
Black could try to carry on with 32 ... .i.h2?? - a reminder of Fisc her's grotesque blunder in
the first game against Spassky in 1972. Black is, of course, in great trouble after 33 g3 and
the king moving to the kingside, remembering also to play h3-h4 if Black tries to push with
...h6-h s.
3o fs
...
Just as we are about to reach the l ong-discarded statutory 30-move requirement before
the draw (which seems to have re-emerged at the recent Dubai International), something
64
Kra m n ik - A r o n i a n , Z u rich 2 0 1 2
happens, and play opens.
Question (and another computer-generated comment}:

Was 30 h s even better? Can Black claim a slight edge?
...
have to admit that I ignored this suggestion in my initial draft, but, writing this for a
I
book, it is time to draw up some loose ends.

30 ... h s is tempting, until one recognizes that 31 g4 h4 32 e4 dxe4 33 fxe4 quite possibly
favours White. He has three pawn islands, as opposed to two, which is commonly regarded
as a slight weakness; but each of these pawn pairs is strong and compact and, m ost impor
tantly, Black's bishop does nothing effective, while White's two knights work well together,
both with each other and with the pawns. lt is a case of complementarity, rather than of
duplication.
In the game, Aronian made sure that his opponent was not allowed to set up a pawn
centre with d4 and e4.
31 g4
Opening some space on the kingside ...
31 g6 3 2 CLle2 .l:bc1 3 3 CLlexc1 CLlf6 34 CLle2 CLlh7
•••
...while Black tries to attack White's more advanced pawns. The play gets tense, but it is
still close to level.
3 5 CLle5 CLlg5 36 CLlf4 ..tb4+
Suddenly, without an error on either side, both kings are under attack, and each player
has several weak pawns, both for himself and for his opponent, to think about. However,
there are not enough pieces to give checkmate, while the pawns on either side are so weak
that, with general hacking, neither player will find a clear pawn advantage. Therefore, after
a few accurate moves, a draw.
3 7 'it>d1 .l::rc 3
65
Question: We are uncomfortably close to the time control, and

play has become suddenly less than sedate. The choice here is in
challenging the rook on the c-file with 38 l:Ic1, or setting up
a pawn attack on the kingside with 38 h4. Which is better?
38 l:Ic1
Safe.
38 h4? is a punt too far. Never mind that Black can head for a draw with 38 ... .l:!.xe3 ?! 3 9
ttJfxg6+ 'it>e8 40 l:rc1 l:re1+ 4 1 'it>c2 tbf7, when the tactics soon subside. Worse still, the
knight sacrifice 38 ... ttJxf3 ! 39 ttJxf3 .l:!.xe3 40 ttJxg6+ 'it>e8 41 tiJd2 fxg4 42 l:Ic1 �c3 puts
White under great pressure.
Without any pawns supporting them, the knights themselves are powerless, less effec-
66
Kra m n ik -A ro n ia n, Zu rich 2 0 1 2
tive than Black's single bishop. Then, of course, there i s the passed g-pawn.
Gambling during a time scramble is dangerous. The usual rule when short of time is:
don't give anything away. Let your opponent make the mistakes.
38 .. J:txe3
Allowing the exchange of rooks with, for example, 38 ....l::!.x c1+? 39 'it>xcl fxg4 40 fxg4
leaves White much better. Apart from questions about the minor pieces on each side,
White is going to win the g6-pawn.
39 l:tc7+
Blink, and White has the control of the c-file, a reversal of earlier events. Black has
enough pieces on good squares to prevent any thoughts of checkmate with rook and two
kings, but he cannot bring the king into complete safety either. A perpetual soon emerges.
39 ...'it>d8 40 .l:!.g7 'it>c8 41 .l:!.g8+ 'it>b7 42 .l:!.g7+ 'it>c8 43 .l:!.g8+ 'it>b7 Yz-Yz
This was in some ways the most difficult game to try to understand, not despite it being
"only a draw", but rather, precisely because it was a draw and not a win by either of the
players. After all, in the days of the 19th century, it was expected that the natural result of
a game of chess is a win for one of the players. lt often takes great skill and sophistication
for both players to end up with a draw when play has been sharp.
In the first and third games, it was not too difficult, even in live play, to sense that one of
the players was starting to go wrong, and it was no great surprise that the player eventually
lost. Sometimes, though, the position became complicated; neither player ever had any clear
chances of forcing a win, but because the position was complicated, there might well have
been various minor errors on either side, perhaps swinging from a slight edge for White,
through to a slight edge for Black, and perhaps eventually ending up with a difficult draw.
But can the writer and analyst be certain that he has identified correctly what is going
on ? There is always the fear of ascribing a move as a mistake, when it is in fact a good
move; and there is the other fear of failing to identify a mistake. lt is much easier when a
player has lost. In that case, logically, one of the players must have made a mistake at some
stage, and so you give a question mark; at the worst, all that will happen is that you give
the question mark to the wrong move. Whereas with a draw, it is difficult to decide
whether, in a complicated position, there should perhaps be many "?!"s in the text; or
whether one should take the lazy approach, and decide that the players were both ex
tremely strong grandmasters, and why, therefore, should we question their moves?
lt took me more than a day to write up the notes to this game.
Game Six: Aronian-Kramnik

The play gets more interesting as the match goes on. We would have liked the match to
continue further, to game eight or beyond. The players have warmed up to the task, but
they are also well aware that if their play starts to get more fanciful, as in games one and
three, the opponent will find refutations.
Technically, what is most striking about this game is that, although the play gets
67
strange and complicated, and both players are fighting hard for tactical blows, it seems the
game did not stray far from equality, except for a brief phase just before the time control.
Most players would have got well beyond their depth, whether as White or as Black, and
therefore a loss is likely. Yet both Kramnik and Aronian kept their balance, not through
quiet "positional" chess, but fighting the position through, move by move.
28th April 2012

L.Aronian-V.Kra m nik
Ruy Lopez
1 e4 es 2 tLlf3 ttJc6 3 .tbs tLlf6 4 d 3
Aronian has failed t o break down Kramnik's favourite Berlin Defence i n the main line,
and looks for alternatives. lt may well be that in the years to come, 4 d3 will be seen as the
main line. White accepts that he cannot break through with d2-d4 in one go, but he still
has the advantage of the first move, and Black has to demonstrate that he can equalize.
4 -tcs s ttJbd2
••.
Heading further away from the main lines. This quite often happens in grandmaster
chess these days. There is not all that much point if both players play "book" chess, know
ing that the opponent is fully conversant in the main lines and will have analysed them. All
that will happen is a draw. If one of the players is interested in playing for a win, he will
need to try something slightly off-book. This usually means something slightly quieter, as
the "interesting " lines will have been analysed in great depth many years ago.
Here 5 c3 is the most popular continuation in this second-string attempt by White to
keep an edge in the Berlin. Then s ... d6?! is mistimed due to 6 d4.
s ... d6 6 c3 o-o
68
Kra m n ik - A ro n ian, Zu rich 2 0 1 2
7 0-0
Question: This looks natural, but hasn't White missed a good opportunity
for an edge? A knowledge of late 19th century chess would have helped.
As Steinitz recognized over a century ago, the manoeuvre with 7 tt:lf1 ! ?, followed by ei
ther tt:le3 or tt:lg3, can be highly effective. In such a knight manoeuvre, if White has already
castled he will have to use an extra tempo with .l:Ie1 to vacate the f1-square.
Also, again rem em bering some old Steinitz lines, if White is planning to set up an at
tack on the kingside, it might often be a good idea to delay castling kingside, and instead
to castle queenside, so that the king is not getting in the way of the attack.
When you have the opportunity, play through some of Steinitz' wins as White against
Chigorin in the 1892 World Championship match. Agreed, Chigorin did not defend with
great accuracy, and one can assume that Kramnik would have handled the defence far
more accurately. In other words, this is not going to be a re-run of Steinitz-Chigorin.
After 7 tt:lf1! ? �d7 or 7 ...tt:le7, a logical continuation for White might be 8 tt:lg3, with no
big attack, but challenging his opponent to try to equalize.
Shipov, admittedly having to write at speed (this was live commentary), gave 7 ...tt:lg4 as
a "refutation", although perhaps this was slightly garbled in the translation from Russian
to English. At the very least, 8 tt:le3 is equal. Moreover, White also has 8 d4 exd4 9 cxd4.
Then the "fried liver" sacrifice with 9 ... tt:Jxd4 10 tt:lxd4 tt:lxf2 (if 10 ...'iYf6, then simply 11 �e3)
11 'it>xf2 'iVf6+ 12 �e3 does not work, since Black has already sacrificed two pieces rather
than one. As a result, 12 ... �xd4+ 13 'iVxd4 'ifg S+ and 14.. .'it'xbs will still leave White a piece
ahead.
7 tt:le7
...
Black is probably equal now.

8 h3
69
8 d4 exd4 9 cxd4 jL,b6 does not give White any advantage.

8 ...lt:Jg6 9 .l:te1 c6 10 i..a 4 .l:te8 11 d4 jL,b6 12 jL,c2 h6
Black's position looks safe enough. A critical point is that the bishop is on the right side
of the d6-pawn. If the bishop were on e7, White would perhaps still have a slight edge.
13 a4
A slightly strange move, one might think, in that if White is going to play a2-a4 in the
Ruy Lopez, it is generally in the context of prodding the bS-pawn. Still, it gains some space
on the queen side, which might be useful later. White is waiting to see what Black does
next, and only then will he decide what to do with his minor pieces. Most particularly, what
is Black going to do with his bishops? If he tries 13 . . i.c7, for example, he releases the pres
.
sure on the d4-pawn.

13 ...jL,e6
Kramnik tries the other bishop move, preventing lt:Jc4, but there is still another knight
70
manoeuvre to reach for better squares.

14 tt::lf1
Either to e3 or to g3.
14...exd4
Neither player will want to open up the pawn centre, unless necessary, or unless it
starts to take the initiative. However, Kramnik was alert to the possibility that the ex
change here could be well timed.
Previously, 14...'i¥C7 was seen in E.Alekseev-R.Jumabayev, Moscow 2012. Then after 15
tt::lg 3 l:lad8 16 �e3, Black could still keep the centre closed with reasonable equalizing
chances. Instead, he tried 16 ... d5 ?, allowing 17 tt::l x e5 tt::lx e5 18 �f4, and White was substan
tially better.
Kramnik showed a much clearer understanding of timing, and it was soon Aronian who
was under pressure. If Black wants to try to open the centre, often the correct choice is to
do it promptly or not at all. A half-way approach, of playing a couple of moves quietly and
then opening the centre, can often be a mistake.
15 tt::lxd4
Question: In most such central structures, White would usually

recapture with the pawn, keeping the d4- and e4-pawns working
together. Why does he avoid playing 15 cxd4 here?
Because 15 cxd4? d5 ! leaves White's centre under attack. If he tries to win a pawn with
16 e5 tt::le4 17 �xe4 dxe4 18 l:lxe4, Black has 18 ... �d5 and (once the rook has moved)
... �xf3, regaining the pawn on d4, while still attacking the e5-pawn. Black is better in all
lines.
1S ...�d7
A loss of tempo, maybe, but Kramnik is happy that he has forced White to recapture on
71
d4 with the knight, rather than with the pawn. Black can now try to create pressure on the
e4-pawn.
16 f4
This looks ugly, with the knight being pinned, and both his e- and f-pawns being under
some pressure. The point is that he wants to allow his pawn to advance to es, if Black were
to try ... d6-dS. This is not really an attacking plan for White. lt is m ore a case of setting up
counterplay, if Black attacks the centre. White cannot be too wild, as his king can easily be
exposed.
The more straightforwardly defensive option is 16 tt::lg 3, and if 16 ... ds then 17 exds
J::rx e1+ 18 't�Vxe1 ltJxds, about equal. Instead, Black could maintain the tension, maybe
16 ...'i¥c7 17 .i.e3 J::rad8, again about equal.
16 ds...
Again, there are plenty of choices for Black, and plenty of ways of keeping tense equal-
72
Kra m n ik -A ro n ia n, Zu rich 2 0 1 2
ity. When the game i s close to equal, there are often several acceptable or good moves. If
the position is balanced, it will remain close to that balance, as long as no mistakes are be
ing made. Quite often, there is a much narrower range of possibilities when the position is
not equal. Indeed, if a position is worse, the defender may have to play extremely accu
rately. Any slight mistake, and the position can easily deteriorate to a quick loss. From the
other side, if a player is better, sometimes it is easy to maintain that advantage, whereas at
other times a player with a slight edge can easily drift to equality, without the most precise
play.
So it is extremely difficult to try to annotate an equal position, with equality kept over
several moves. At various stages of the World Championship a month later, the grandmas
ter commentary team had to admit they found it difficult to predict what was going on. If
there is genuinely no clearly best move, all the player can do is to pick one, decide for vari
ous reasons that he is happier with such-and-such a move, and play it. Don't be daft and
spend half an hour thin king about a level position.
Now for a piece of statistics. The computer engine I have been using indicates that, be
fore Black's 16th, there are well over twenty different moves that leave the position essen
tially equal or, at most, very slightly better for Black. This is a large number, especially given
that we are in a tense early middlegame, with no pieces and only one pawn each having
been exchanged. The move that Kramnik played was given as the only chance of gaining a
slight edge, but even then the game seems more likely to remain as tense and equal.
17 es lt:Je4
18 ..ixe4
Natural and direct. Even here, there appear to be three possibilities for White, with
(subject to later analysis) approximate equality.
The simplest approach, developing the queen and bringing it active, is 18 � 5 . Black
can then offer a queen exchange with 18 ...'Yih4 19 "Y!Vxh4 lt:Jxh4 20 ..lte3 - level, unless the
reader can suggest otherwise.
73
18 tt:ld2 is another option, apparently giving Black slightly the better of things, but it
may well be that this, after analysis, could end up with clear equality.
Unfortunately, I do not know how much time Aronian spent on this position, and what
lines he analysed. So...
Question: An analysis position.
You are welcome to use the computer to help analyse this position. Indeed, given the
context of the earlier argument, this is positively recommended. The challenge is whether
you can find a clear equalizing line after both 18 .. .'YlNh4 and 18 ...Ji.fs.
a) 18 ...'i'ih4 leads to much chopping in the centre after 19 tt:lxe4 dxe4 20 .l:!.xe4 .i.fs
(making use of the pin on the knight} 2 1 l:i.e2 tt:Jxf4 2 2 Ji.xf4 '¥Vxf4 2 3 Ji.xfs '¥Vxfs. Plenty of
moves to calculate, but no sacrifices. The analysis engine I am using gives Black an edge,
presumably in view of the isolated e-pawn. Play on a few moves, though, and after 24 as
(forcing Black to exchange the bishop} 24....1i.cs (or 24... .i.xd4+ 2S i¥xd4) 2S b4 Ji.xd4+ 26
'¥Vxd4, any hope of an edge has reduced almost to vanishing point. Black might continue
with 26 ... l:ted8 27 �c4 '¥Vd3 28 �xd3 l:txd3, but White can hold this after 29 e6 fxe6 30
.l:Ixe6 .l:rxc3 3 1 l:te7, even if temporarily a pawn down.
b) 18 ... .1i.fs 19 Ji.xe4 Ji.xe4 20 tt:lxe4 dxe4 leaves Black's advanced e-pawn in danger, but
there are potentially two pins on the white knight, so 2 1 11xe4?? at once loses to 2 1...cS.
Instead, 2 1 �h2 is possible, most likely equal. Alternatively, White could play more tac
tically, with 21 as ! ? ..txas 22 .l:.xe4 cs 23 �a4 cxd4 24 iYxas, which is lively but most likely
heading for a draw again.
Playing through these lines makes one appreciate that Aronian's 13 a4 was a wise idea,
and that the later a4-as becomes unexpectedly useful many moves later. The pawn push
was certainly not a waste of tempo.
Back to the main line.
74
Kra m n ik-A ro n ia n, Zu rich 2 0 1 2
18 dxe4
...
We have just seen something like this, in the line with 18 ctJd2 i..fs and then a double
exchange on e4. The difference here is that White's knight on f1 and Black's bishop on d7
are still on the board. This could well make a radical difference to the outcome of the next
few moves - or not. Possibly the strategies could be roughly the same.
19 a s!?
In my blog, I noted that this was "an extremely difficult position to assess. Translation: I
do not know what is going on, and I don't know whether Aronian's move was brilliant,
okay, or not all that good. At the very least, it would take me several attempts to make an
initial assessment at home, and maybe across several days, and even then I would not feel
any degree of certainty about all this." And much text to follow, I conceded that his posi
tion may well have been fully playable, but I could not understand why he chose it.
So let us try to understand why Aronian might have rejected 19 tLlg3 cs 20 ctJdfS.
75
Here White has kept the a-pawn, but has granted his opponent the long diagonal from
a? to g1. Perhaps in the end Aronian simply wanted to avoid any checks along that diago
nal, such as 20 ... c4+ 21 'it>h2 i.f2, threatening the rook, with a substantial gain of tempo if
White moves it away. Then 22 lt:Jd6 (not 22 l:te2?? ..ixfs ! 23 'iWxd8 .i.xg3+, winning a piece)
22 ... ..ixe1 23 'ifxe1 is playable for White despite the lost exchange, given his strong and
active d6-pawn and the chance of chopping a few black pawns, but it does not seem to be
advantageous. Most lines seem to end up in equality; e.g. 23 ...l:te7 24 �xe4 'iWc7 25 �xc4
�xc4 26 tt:Jxc4.
Maybe it is as simple as that. Given two complicated lines, with many variations and
sup-variations, he would rather give away a pawn than the exchange.
19 .txas 20 tt:Jg3
...
20 l:txe4?! recovers the pawn immediately, but after 20 ... cs Black will be able to make
good use of his bishop pair, against White's rather exposed king.
Question: Time to ask the readers ... What should Black do next?
20 i.b6
...
A difficult decision, but it seems that Kramnik played the best and most accurate move.
lt is tempting to prod the knight first with 20 ... cs 21 tt:Jdfs (21 lt:Jb3 ? favours Black, since
White's knights lack coordination) 21....i.xfs (2 1.. ...ic7 22 'iWg4 puts pressure on Black's
kingside) 22 tt:Jxfs, but White ends up better in the queen less late middlegame after
22 .. .'�Vxd1 2 3 .l:i.xd1, as 2 3 ... l:ted8 24 l:txd8+ .l:i.xd8 25 l:txas .l:i.dl+ 26 Wf2 .l:!.xc1 allows White to
coordinate his kingside with 27 g3!, so this is not a safe line for Black. Instead, 23 ... .i.b6 24
lt:Jd6 .l:!.e7 2 5 lt:Je4 f6 is a possibility, though Black will have to work hard for a draw.
The essential difference between this variation and the game is that, in the latter,
Kramnik has added ... .i.b6, and White has added 'it>h2. In view of the fact that the bishop on
as was under pressure, perhaps purely in positional terms 20 ... .i.b6 was the best.
76
Kra m n ik - A r o n ian, Zu rich 2 0 1 2
Iwas wondering whether Kramnik's move was a mistake, but it seems in fact that he
played accurately and well, forcing Aronian now to find good moves to try to hold the posi
tion. This is the wonderful thing about this game. So often it seems to swerve from advan
tage to Aronian, then back to Kramnik, then back to Aronian again - but l ooking it more
closely, the game was never ( apart from a blip in time trouble} anything other than equal.
21 'it>h2 cs 22 tt::ldfs
The simple question is: what has Aronian gained for his a-pawn ? Possibly not quite
enough. Or maybe he was okay.
22 .....txfs
White is obviously better if he can win the e-pawn without trouble, so Black's exchange
is necessary.
23 tt:lxfs 'iVxdl 24 ltxd1 ltad8
Black has three choices here - and remember that Aronian has already sacrificed a
77
pawn, so he cannot afford to play lazy moves.
Question: The choices are to complete development with 25 .Jte3;

to block the d-file and create a strong knight with 25 ttJd6; or to try
to recover the pawn quickly with 25 I:Ie1. Which seems best?
Complicated - not in the sense that there are any great tactics, but rather that there are
some obvious positional strengths and weaknesses for both sides, while it is far from clear
that either player can force a clear edge.
Black has an extra pawn, and if he keeps the pawn, he has excellent chances of winning.
He also has control, though not total control, of the d-file. Another point in Black's favour is
that White's bishop has not yet had a chance to enter the game.
The first impression might be that Black should be winning. And yet, if we examine the
four minor pieces as a whole, rather than focusing on White's cl-bishop, it becomes clear
that of the four pieces, only one, White's knight on f5, is a dangerous attacker. He can try to
win a pawn, either with lL'lg3 and lbxe4, or by forking two pawns with lL'ld6. If the e4-pawn
goes, the presumed advantage would swing to White. The final picture would be that
White would have an excellent kingside pawn structure, while Black would have problems
covering the a7- and c5-pawns.
Aronian played:
25 .Jte3
This is a combative move, ignoring Black's initiatives on the d-file, and quite simply de
veloping his own pieces.
White could plug the d-file with 25 l2'ld6, but the knight is then pinned. A likely result
would be a draw after, for example, 2 5 .. .f6 26 .Jte3 .l:le7 2 7 lL'lxe4 I:Ixd1+ 28 I:Ixd1 l2Jxe5 29
.Jtxc5 .Jtxc5 30 lbxc5 l2Jc4 31 b3 with a dull endgame. 2 5 .l:le1 I:Id3 26 lL'ld6 .l:le7 also seems
to lead to equality, with a possible repetition after 27 tt::lc 8 .l:le8 28 tt::l d 6. Clearly this is a
summary of several different possibilities, but the overall reading of these lines should end
up in equality.
What happens with the bishop move? lt is difficult to assess. We can only really answer
this a few moves later.
2 S l:Id3
...
Black can try to ram some pawns through by 2S ...c4 26 .Jtxb6 axb6 27 'it>g3 l:.d3+ 28
I:Ixd3, but after either pawn capture, White can hold the balance with 29 tt::l d 6.
26 I:Ie1
78
Kra m n ik - A r o n ia n, Z u rich 2 0 1 2
Question: What happens next? Or alternatively, what should be

happening next? (I am aware that I have not yet fully answered
the previous question. Maybe it is time to answer both.)
26...f6!
Kramnik did not want to give the e-pawn away without a struggle after, for example,
26 ... l:i.ed8 27 tt:Jd6 .l:i.d7 28 tt:Jxe4. Instead, his attack on the e5-pawn opens up lines for him.
Amazingly, the position is still about level, although Black probably has the slightly easier
play.
That being so, it seems that 25 .l:i.e1, 25 tt:ld6 or 2 5 ..te3 are all good and reasonable
moves. Perhaps that is not all that surprising. Chess writers, myself included, tend to like
finding the best move, or even the winning move, after an opponent has made a serious
mistake. But what happens if there is no serious mistake? lt is quite possible that several
moves might be equally valid.
27 exf6
27 tt:Jd6 .l:i.f8 28 tt:Jxe4 fxe5 29 fxe5 .l:i.d5 favours Black.
27 ... gxf6 28 tt:Jxh6+
79
Question: Which of the four legal king moves is best? Are they all
about equal? Or does one king move give Black an edge?
28 ...�f8?!
lt is difficult to decide which of the three reasonable king moves is best. (28 .. .<it>g7? 29
tt:JfS+ is an obvious loss of time.) But the move chosen by Kramnik is doubtful, and Aronian
momentarily has the initiative. The main problem is not so much of disengaging the joust
between king and knight, but rather that the king is now on the same diagonal leading to
cs (i.e . ..txcs would now be check), so Black's bishop on b6 is unable to free itself.
28 .. .'�h 7 29 tt:Jg4 �g7 is perhaps the most natural plan, but not necessarily the best. 30
tt:Jf2 .l:!.dS 31 g4 is not quite equal for Black, and he would need to play carefully.
However, there is no need to try to dislodge the knight. White would want to move it
anyway. Therefore, 28 ...�h8! 29 .l:!.a4 (by analogy to the main line) 29 ... ..te7 30 g3 as, and
Black is comfortably equal, as 31 ..ixcs is met by 31 ...tt:Jxf4!, still equal.
3o tt:Jf2 .l:!.d s 31 g4
"Comfortably equal" would tend to imply that the comfortable player will find it easier
to avoid making mistakes, as opposed to the uncomfortable player, who might need to
play highly accurate chess, to avoid making mistakes. This can be of considerable impor
tance as we start to reach the time control.
29 .l:i.a4
Mainly aimed at preventing Black from playing ... tt:Je7 and ... tt:Jds.
29 J1ds
..
To make it more difficult for the white knight to return.
80
Kra m n ik - A ro n ia n, Zu rich 2 0 1 2
Question: This seems to be the one golden opportunity for either

of the players to create a serious edge; i.e. after Black's inaccurate
king move. What should White do here?
30 c4?!
The problem with this as a winning attempt is that, with the pawn on c4, White's rook
on a4 no longer puts pressure on Black's pawn on e4. Instead:
a) 30 g4, with a general king side pawn advance, looks tempting, but after 30 .. .t2Jh4!
(aiming for f3) 31 �g3 �g7 32 �xh4 'it>xh6, play is close to equal.
b) 30 fs ! gives White an advantage, if not yet a decisive one.
b1} 30.. .<�:Je7 or 30...tt:Jes simply allows White to take a pawn with 31 l:txe4.
b2} 30 ... tt:Jh4 will get the knight trapped after 31 c4 .l:!.d7 3 2 'it>g3 . Black can save the
piece with 31 ... .i.C7+ 32 'it>g1 .l:!.e7, but then White wins the a-pawn with 33 l:txa7.
b3) 30 ...tt:Jh8 is not as awful as it looks, and there appear to be no immediate win. Nev
ertheless, Black has undoubtedly retreated, and there are various good lines for White,
starting with 31 ..ltf4 (the rooks attack the e4-pawn) or 31 g4.
b4} Finally, Black can insert 30 ... ..tc7+ 31 'it>g1. The problem in that case is that he has no
time to extricate the knight, while also saving the a7-pawn. If 31 ... ..ltg3, then 3 2 ftf1, and
the problem remains. Or if 31 ... tt:Jes, although Black is not yet ready to resign, White has a
clear advantage, probably decisive, after capturing on either a7 or e4.
A near-miss for Kramnik.
30 ....l:!.d3
Keeping control of the d-file, and therefore highly natural. lt now starts to look promis
ing for Black, but maybe it is still about equal.
81
31 b4?
With breathtaking ability to create wild and dangerous positions, Aronian makes his fi
nal attempt to win the match.
Or so it seemed at the time. As the reports emerged, it became clear that Aronian had
completely missed the exchange sacrifice, and had to improvise wildly to keep the position
alive.
Perhaps that figures. This game has been so tense that it would not be difficult to an
ticipate the players running short of time. That being so, mistakes creep in. Kramnik's
28 .. .<it>f8 and Aronian's 30 c4 were clear misjudgments, which could easily have altered the
result of the game. Aronian's 31 b4 was more serious, as it could easily have lost by force
against best play. Having said that, the winning line for Black is extremely difficult to en
visage, especially in time trouble.
31 tt:Jfs would have been a safe and steady move.
31 ....l:txe3 ! 32 .l:txe3 cxb4
82
Kra m n i k - A ro n ia n, Zu rich 2 0 1 2
Black now has three passed pawns, one of them looking highly dangerous and well pro
tected. White's pieces, at the moment, are badly coordinated.
3 3 llg3
The only sensible move. He has to attack the king, before Black can queen a pawn.
Question: What next for Black?
33 ... e3?
This is the sort of move that any reasonably strong player would be able to play and cal
culate extremely quickly. There is a simple enough variation to see, and if in the end Black
cannot find anything when he is about to queen, and if as a result it will end up as a draw,
then so be it. You cannot win a game out of nothing.
lt would take a stronger player to wind his way through the best line - except that there
is no player who could dare to suggest that they would do better than either Kramnik or
Aronian. Maybe there will at some stage be a player of 2900+ strength, not helped by rat
ing inflation (after the 2012 Olympiad, there are now fifty players over 2700). But it would
be a difficult ask. ..
Here 3 3 ...tt::l e 7!! is an exceptionally difficult move to find, not least because Black's three
king side pawns are apparently thoroughly immobile. The only thing that favours Black is
that his two passed pawns, helped by the bishop, are highly dangerous, whereas White's
two rooks and the knight are not well-placed for defence.
Perhaps the most difficult point in terms of thought processes is that, when trying to
attack, it takes a leap of faith (as Shipov noted} to make a quiet retreat, rather than strong
forward move.
The natural reply is 34 llxb4, but a second retreat, 34 ... ..ie7, creates the lethal threat of
... ..ixf4, pins and skewers working together. I have to admit that I signed off in my analysis
when I noted 3 5 .Mg4. Shipov's prod soon showed how Black can still win. He plays 3S .. .fs.
83
Then 3 6 .l:th4 protects the f4-pawn but is otherwise hopelessly out of position. Then
36 ... e3 does win for Black after, for example 37 l;lbl e2 38 !1e1 ll:lg6 or 37 �xb7 e2 38 f[bl
ll:lg6.
36 :xb7 is more lively, and indeed the computer temporarily gives White as better. Play
the line through though, and Black is winning after 36 .. .fxg4 37 �xc7 e3. The main tactical
point is that after 38 'it>g l e2 39 'iotf2 el�+ 40 Wxel li::l d S+, Black wins the rook.
White could try 38 .l::!.xa7 e2 39 §.al, but after 39 ... ll:lc6 40 :te1 ll:lb4, Black will win the
rook for the passed pawn. Rook and knight versus knight and four passed pawns is not as
exciting as it might sound. White's pawns will gradually drop off.
Going back, White could have tried 34 li::l g4, but now it is a quick positional grind for
Black, with the help of the two advanced pawns: 34...ll:lfs 3 S li::lxf6 ll:lxg 3 3 6 Wxg 3 1Ie7 3 7
li::ld s l::tg 7+ 38 'it>h2 e3, and s o o n . The b-pawn soon makes its presence felt; e.g. 39 li::lx b6 b 3 !
40 <t>g 1 b2 4 1 Il.b4 e 2 4 2 'it>f2 l::tx g2+! 4 3 'i£?e1 .l:!gl+ 44 Wxe2 axb6 and wins.
All, of course, computer-inspired analysis. Which leads to the question of how human
players over the board would be able to find such lines. Difficult.
34 lbg6 e2 35 l;la1
Just in time.
Question: Now what is the best move for Black?
3S i.f2
...
I hope everyone avoided 3 S ... el'ii'? ? 3 6 l!g8+ <3i/e7 3 7 �xel+ and the queen disappears.
Black needs to bring in the bishop first.
36 l::tg 8+ We7 37 �g7+ 'it>d6
And 37 .. .'�f8?? 38 .l::!.f7 mate would be just as bad.
38 l::tx b7 el� 39 �xe1 .txe1 40 li::lf s+ <t>cs
84
The last difficult position - and just after the time control, which i n theory means that
both players have time to think, although what quite often happens is that both players
continue to rush on with their earlier analysis without re-checking it.
Question: White soon forced a draw with 41 .l:IbS+.

Can he try for more with 41 .l:Ixa7 - ?
41 �b5+
Given the tension in the position and the dangers of a second pawn promotion, it is
understandable that, by this stage, Aronian will be happy with a draw.
Instead, the computer suggests 41 l!xa7, just grabbing a pawn while seemingly ignor
ing the other passed pawn. lt is doubtful whether a human player would even consider it.
Only the most ambitious and fearless would do so.
Play on a few moves and we soon see why humans are cautious. lt is a question of sur
vival.
41 ... b3 42 l:tb7 ..tb4 43 ctJd6 might be good enough as a winning try for White - if, and
only if, the opponent is not strong enough to find the correct replies. Black plays 43 ... .Ue3 ! .
85
Then 44 .l:.b5+? 'it>xd6 45 �xb4 'it>cs is badly mistimed as the b-pawn is still defended.
White has to wait for ... b3-b2 first.
Therefore White decides to advance his own pawns: 44 g4, when 44...b 2 ? 45 I!.bS+ \t>xd6
46 la.xb4 wins the black pawn on the seventh for nothing, while after 44 ... 'it>xd6? 45 .!::txb4
<t>cs 48 .l:.b7, with g4-g 5 to follow, White is a tempo up in the pawn race.
Instead, 44 ... �a3 ! puts White under pressure. Then 4S ll:lbs .tb4 46 ll:Jd6 i.a3 47 ll:lbs
..tb4 would be an unexpected repetition of position, but White is in trouble after 45 ....tc1!
46 �C7+ 'ot>b4 47 .l::!.b 7 b2 (avoiding another repetition) 48 ll:Jd6+ \t'a4.
Pawn snatches so often do not work, and Aronian is correct in aiming for a safe draw.
41 'ot>c6 42 lt:ld4+ 'it>c7 43 �CS+ Yz-Yz
...
This was the most difficult game of the match to analyse, and therefore the most inter
esting. Shall we add that this was also the best game?
lt was not quite perfect chess. There were a couple of identifiable accuracies, and in
deed a complete tactical oversight by Aronian, which gave Kramnik an extremely difficult
winning opportunity - but it was the sort of difficult position that a player would need to
think about for half an hour, rather than quickly playing a move in the time scramble.
lt was lively chess, as in the whole match. The next match, between Anand and Gelfand,
was inevitably going to be a much tenser encounter, this being for the World Champion
ship itself.
86
C h a pt e r Two
Anand-G e l fand,
Worl d Championship,
Moscow 2012
Many chess players had greatly looked forward to the friendly match between Kramnik and
Aronian, and were almost disappointed at the thought of a match between An and and Gel
fand. This was by no means a reflection of the players being in any sense "weak". lt has to be
remembered that the quality of chess has improved vastly in the last fifty years, and also in
the last ten years, or even the last five years. Gelfand in 2012 was a stronger player, if not
necessarily overwhelmingly so, than the combatants Spassky and Petrosian in the 1960s.
What is happening is that, although Anand was able to beat Kramnik and Topalov in of
ficial world championship matches, and has therefore been a fully worthy world champion,
younger players have, if anything, reached an even higher level. Currently Carlsen and Aro
nian have reached #1 and #2, and it looks likely that this will remain so, at least until even
younger players, or at least players of the same sort of age as Carlsen, are able to improve
and try to take over the top two slot. lt is with this background that the match between
Aronian, the second highest rated player in the world, and Kramnik, not too long ago a
World Champion, was of particular interest. A match between two players, both in their
mid forties, is, unfairly or not, of lesser interest than a match between two players in their
late twenties or mid thirties.
Having said that, it was an immense achievement by Gelfand to qualify to play in the
World Championships. He had to go through a marathon knock-out of 128 players, the
Chess World Cup, in Khanty-Mansiysk at the end of 2011. After some strong preliminaries,
he had to beat Jakovenko, Karjakin and Ponomariov, all redoubtable strong grandmasters
whose ratings at the time were, respectively, 2736, 2723 and 2739. (In the July 2012 list,
Karjakin is up to 2784.)
87
Tough enough, but these were still only the preliminaries. He still had to beat Ma
medyarov, Kamsky and Grischuk, in May 2011. This was clearly no soft opposition. Gelfand
fully deserved his shot at the title. If anyone wants to argue that Gelfand was a soft option
for the world championship, well, try it yourself!
Having said that, Gelfand was slightly fortunate in his timing. At the start of the World
Cup, he was the highest rated player in the tournament, at 2758, and thus the "favourite",
but only by the smallest of margins. There are currently, in July 2012, ten players higher
rated than this, not including Gelfand, now at 2738. Two of the top ten players (Anand and
lvanchuk) are around the same sort of age as Gelfand, one is less than ten years younger
(Kramnik), while the rest are significantly younger, and therefore have time to improve.
lt was therefore a remarkable achievement for Gelfand to qualify for a World Champi
onship match, but realistically, unless he were actually to beat Anand, this was likely to be
his last shot at the title.
Game One: Anand-Gelfand

A relatively quick draw, in 24 moves, but this was not quite as solid as it might appear.
Wasn't Gelfand better? One soon gained the impression that Gelfand was aiming solely at
holding the draw, especially with the black pieces, and not quite believing that he was ca
pable of beating Anand. The impression strengthened as the match continued.
Here it is useful to remember past encounters by the two players, both recent, and go
ing back over twenty years. Out of 3 5 full-length games, White has won ten times, but
Black has won only once. Both players have found it extremely difficult to win as Black, but
every so often they have a reasonable number of wins by White. Anand as White has scored
five wins, ten draws and no l osses. Gelfand as White lost once against Anand, but for the
remainder has achieved five wins and fourteen draws.
The players are therefore about level in playing strength, just one extra win for An and
over 3 5 games. Any lopsidedness is in the White-Black question. What we were later going
to see was that An and was just going to concentrate on trying to draw every game as
Black, aiming for quick draws if at all possible. He was reasonably confident that, over a
large number of games, he could achieve a plus score with White, and therefore an excel
lent chance to keep the World Championship.
Gelfand, as Black, played a far more open strategy. He did not concentrate on ultra
symmetrical pawn structures, but instead played "normal" chess, as in game one, albeit
while still trying to concentrate on solidity.
There was an odd resemblance between this game and the first game of the Kramnik
Aronian match. Kramnik as White wanted to show he could dominate the match, and
made full use of a very recent theoretical novelty, in which he hoped that his opponent
could not find the best moves. This, however, was only tested in one game, and the new
line, under top-level examination, proved not to be good. So, too, in the Anand-Gelfand
match. The main difference was that Aronian kept his nerve and held out for the win (al
though Kramnik played several inaccuracies later on), whereas Gelfand was only looking
for the draw.
88
A n a n d - G elfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
11th May 2012

V .An a nd-B.Gelfa nd
World Championship, Moscow 2012 (Game 1)
Grunfeld Defence
1 d4 'Llf6 2 c4 g6 3 'Llc3 ds 4 'Llf3 .i.g7 5 cxds 'Llxds 6 e4 'Llxc3 7 bxc3 cs
All standard so far for the Grunfeld Defence. Play in the centre usually opens up quickly,
sometimes with gambit attacks, as here. There is major tension on the d4-pawn.
8 .i.bS+
Known, if usually regarded as of no great importance. Anand wants to innovate, but
unless his analysis is watertight, some unpleasant surprises can happen.
Instead, 8 .i.e3 �aS 9 'Lld2 has had some good results recently. The idea is 9 ...cxd4 10
'Llc4, and if 10 .. .'�Vxc3? (10 ... �C7 is safer}, then 11 .i.d2 and the queen is trapped.
Just to show that occasionally it can be Aronian's opponents who sacrifice the queen, a
game continued 10... dxe3?! 11 'Llxas .i.xc3+ 12 'it>e2 .i.xas 13 'it>xe3 0-0, L.Aronian
E.Sutovsky, World Team Championship, Ningbo 2011. 1t just didn't work. Aronian later
won.
8 '2lc6
...
No need to worry about the pin.

9 ds?!
He tries it anyway.
89
Question: Why do you think that strong players do not trust this move?
White is not going to win the knight for n othing. Black will break the pin with 9 ... a6,
and if 10 .ta4, then 10 ... bs. To take the knight, White would have to give away the light
squared bishop, which is a big concession. In open pawn structures, with the chance of
roaming on open diagonals, the bishop pair is useful, more so than bishop and knight.
Players quite simply would not trust this as White. Black's pawn structure is not suffi
ciently wrecked to have any sort of decisive disadvantage. After all, Black's king side pawns
are strong and secure. 9 o-o is a more common response.
I noted at the time that: "lt is difficult to imagine that this pawn push by An and would
have been intended as his main line of attack in this match. ( ... ) Anand is presumably trying
to test his opponent in an unfamiliar line, with tactics that Gelfand will need to think
about." lt seems risky, unless his analysis is solid.
9 :�a 5
..
The players needed also to consider 9 ... a6 10 .ia4 bs 11 dxc6 bxa4, with probable equal
ity, but with a difficult pawn structure for either side. Maybe Gelfand was suspicious about
this, or maybe he quite simply felt that his own move was better.
The computer suggests that, after either move, White is not worse but even has a slight
edge. A small edge, plus a theoretical innovation, sounds good at the start of a match.
Perhaps we should briefly add that 9 ... .ixc3+? 10 .td2 .txa1 11 "ifxa1 ends up winning
material for White.
10 .l::!. b 1 a6 11 .ixc6+ bxc6
90
A n a n d - G e lfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
Question: What should White do next? The main choice i s 1 2 dxc6 or 1 2 0-0.
12 o-o?!
Haven't we all, with a sense of relief, castled quickly when it seems the right time to do
so, in a wild open central position ? Often it is correct and appropriate, but if there is a bet
ter move, he should play it. If necessary, White can castle later.
The computer suggests that 12 dxc6 �xa2 13 �3 �xb3 14 l'1xb3 is best, and the only
equalizing try for White - there is no chance of an initiative (and where did that supposed
edge go to?} - but even here, one would instinctively prefer Black, with his outside passed
pawn and his bishop pair.
12 ...�xa2
And indeed, why not? What is happening here, as in the first Kramnik-Aronian game, is
91
that the player with White has gone all-out with a wild idea, hoping to be 1-0 up and home
and dry, but has gone too wild, and without having analysed in depth in advance.
Black is now a pawn up, with an outside passed pawn and no obvious way for White to
defend against it. The black king is still in the centre, but there is no obvious way for White
to attack or aim for checkmate either. Black also has a good bishop pair, while White's
bishop and knight look ineffective. Ideally, White would have liked to have had a light
squared bishop, to hassle the queen, but of course Anand had exchanged that bishop many
moves ago.
The two main points of interest for the rest of the game was how An and was appar
ently able to hold the draw with great comfort, and, conversely, the question of whether
Gelfand should or could have tried for more.
13 �b2 'ii'a s !?
A highly practical response, keeping the queen out of trouble.
Some twitter feed suggested that Anand might have forgotten his analysis here, and
that he could have tried something more effective. More likely, he soon recognizes that he
has already gone wrong, and that his optimistic d4-dS push was not as good as he thought.
He now has an uphill struggle.
If Gelfand were to try the much sharper 13 ...'iVc4, trying to win all the pawns available,
Anand would presumably have found interesting ideas in this line.
14 d6
White is not aiming at a mating attack. All he is doing is try to defend actively, making
it difficult for Black to castle safely.
He might have considered 1 4 dxc6 instead, but after 14...0-0 Black would b e clearly bet-
ter.
Question: Black is a pawn up, with excellent chances of aiming for

more than a draw. Can you find a good line for Black?
92
14 J:!:a7?!
..
The first sign in this match of Gelfand playing far too defensively. The problem is that,
in his numerous earlier games against An and, Gelfand achieved a serious minus score as
Black, so it is unlikely that he would feel totally confident in playing open chess. His match
strategy must therefore have been to cut down any unnecessary losses, be happy with a
string of draws, and wait for the chance of an occasional win in the next dozen games.
Understandable enough - but in playing so cautiously here, he seems to have missed a
great opportunity of applying serious pressure as Black, as early as in game one. If your
opponent has played inaccurately, you must punish him, even if he is a 2800+ World
Champion.
The simple pawn snatch, 14...�xc3!, seems to give Black a big advantage:
a) 15 l::tc 2 �e6 and Black is happy.
b) 15 �d2 �xd2 16 tZ:lxd2 �e6 17 dxe7 'fie? 18 �a4 r:Ji;xe7 19 'fla3 l::th b8 {there are other
good moves) and Black has the initiative, as well as the extra pawn, in a simplified position.
lt is difficult to see how Anand could survive here or on previous moves. Definitely a
missed opportunity for Gelfand. He got too tied up in trying to ensure the draw, perhaps
forgetting that it was possible to try to win.
c) Instead, Shipov suggests 15 dxe7!, regarding this as, in effect, winning for White.
93
Then 1S ...i.xb2? 16 i.xb2 gives Black far too many weaknesses on the dark squares,
while 1S .. .<it>xe7? 16 i.g S+ f6 17 e s ! ! ends up as a win after 17 .. .fxg s 18 v&'d6+.
Sergey Shipov must be among the best of instant commentators, without the luxury of
writing notes some months later. Nevertheless, it seems that he is overpressing when try
ing to suggest that An and has been able to keep a good position. The problem of setting up
a pawn exchange on e7 is that Black no longer covers the f7-square. Therefore, 15 ...\\IVC] !
(not given by Shipov) seems very good for Black. He does not have to try ...'it>xe7; his queen
and bishops can snuggle up with ... .i.e6 and ... i.f6 (or alternatively 16 i.g S f6); and he has
covered the weakness on d8 for the time being. He can later expand his defences and coun
terplay by finding a way to take the e7-pawn eventually.
15 i.gs exd6
Again aiming for simplicity. The computer suggests 1S .. .f6, with a slight edge for Black,
but most of the earlier advantage has gone.
16 \i'xd6 l:td7 17 �xc6 \i'c7
An excellent way of achieving a draw against Anand, except that here it is a draw l ost,
rather than a draw gained.
18 \i'xc7 l:txc7
When I saw this, after a stroll in the garden, I had fully expected that there would be a
long slow grind, with a serious attempt by Gelfand to win as Black. I certainly had not ex
pected a draw, just half a dozen moves later.
94
A n a n d - Ge lfa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
Anand still needed to find good defensive ideas, but h e did what was needed.
19 ..if4 l:i.b7 20 l:i.c2 o-o 21 ..id6 l:i.e8
Question: What is the best way for White of handling the defence?
Remember that Black has the bishop pair and an outside passed pawn,
so White still needs to be careful.
22 tt:'ld2!
A good defensive move. The one way in which the usually less impressive knight can
outweigh the bishop (even as a half of the bishop pair) is that the knight can change the
colour of its effective squares (and also attack squares of the opponent's colour). The
knight is now covering c4 and e4, and if it can reach either of these squares, there is always
the possibility of a knight fork on d6, perhaps aiming to eliminate the bishop pair. The
95
more I look at this, the more I appreciate that Anand is just about holding the line - a good
result from an uninspiring position.
22 �xcs?! is more natural, but after 22 ...1::!.xe4 23 kd4 f6 (to keep the bishop pair),
White would need to work extremely hard to hold the position. Black's outside passed a
pawn, with the help of the bishop pair, is far more dangerous than the inside passed c
pawn.
22 ...fs
At the time, this looked a little like oversimplification of the pawn structure, and one
felt that Black should still be able to create an edge. lt seems not. Certainly Gelfand was
unable to find anything significant.
White is able to hold after, for example, 22 ... c4 23 es �fs 24 l:i.cc1 �d3 2 S l:i.fe1 .l:!.b6 26
ct:Jf3. Note how the knight defends both the pawn on es, and also the rook on e1 (so that
there is no longer an effective threat of ... 1::!.x d6).
22 ...kf8 is a possibility, but after 23 �xf8 �xf8 24 l:i.a1, the black a-pawn is under con
trol, and White can consolidate with f2-f3 and �f2. Any edge for Black, as suggested by the
computer, is purely minimal and is likely to fade to nothing after a few good moves by
White. There are other possibilities, but nothing special.
23 f3 fxe4 24 ct:Jxe4 �fs Yz-Yz
The knight is now on an excellent central square, an ideal defender. True, Black can ex
change it off for his bishop, but once the bishop pair has gone, the outside passed a-pawn
has little support, and cannot realistically aim for power.
Therefore, a draw was agreed. 25 .l::!.a 2 kxe4 26 fxe4 �xc3 27 .i.xcs is quite simply level.
Game Two: Gelfand-Anand

Not quite as solid a game as it looks. Gelfand again had the opportunity to put Anand un
der pressure, but he allowed unnecessary simplification, both in the middlegame and in
the early part of the endgame.
96
A n a n d- G e lfa n d, World C h a mp i o n s h ip, M osco w 2 0 1 2
12th May 2012

B.Gelfa n d-V.Anand
Semi-5/av with s a6 ...
1 d4 ds 2 c4 c6 3 lbf3 tt:Jf6 4 e3 e6 5 tt:Jc3 a6
Question: Why the early ... a7-a6 in the Slav? This seems to happen so many times
these days, especially at the top level, but shouldn't Black be developing instead?
With over forty players reaching 2700+, there is little m argin for error at the highest
levels, and players with Black do not want, if possible, to give White even the slightest edge.
Here Black is concentrating on setting up a pawn barricade, waiting to see what White is
going to do next.
If 6 .td3, the simplest line is to exchange with 6 ... dxc4 7 .txc4 bs, then the bishop re
treats, and Black can choose between either 8 ...tt:Jbd7, transposing to one of the lines of the
Meran (s ... tt:Jbd7), or playing more flexibly with 8 . . cs, a Queen's Gambit Accepted, with the
.
choice of moving the knight either to c6 or d7.

As in m any lines of the Semi-Slav, if White is not going to move his bishop, it is not so
clear what he is going to do next. He can enter the Exchange Slav, with 7 cxds cxds, but
... a7-a6 is useful in many lines. lt is by no means a wasted m ove.
The one practical problem for Black is that it is often very difficult to play for a win
against an opponent of slightly lower strength. This does not apply so much against an
opponent of equivalent strength, and it is unlikely that An and as Black would be too wor
ried about setting up a quick drawing line against Gelfand. Against a mere I M, the problem
would need some thought.
97
6 b3
The most popular choice at top grandmaster level. Whether White is better, is open to
question.
6 . �b4
. .
So that White cannot continue the fianchetto, 7 �b2 losing a pawn after 7 ... tt:le4 8 'Yic2
'�!Vas 9 .l:tc1 �xa2.
7 �d2 tt:lbd7 8 �d3 0-0 9 0-0 �d6
The bishop, having prompted its counterpart to a slightly uncomfortable square on d2,
is content to centralize, with an insignificant loss of tempo. Black is close to complete
equality.
10 l:f.c1
In recent years, White has generally preferred 10 .l:te1. lt is not totally clear what Anand
would have played in reply. He could simply have tried 10 ... es as in the game; while there
have been a few lines starting with 10 ... h6 11 e4 dxc4 12 bxc4 es, and if 13 dS tt:lcs, which is
perhaps more interesting than the game, but this would probably have been not what he
would have wanted.
In my blog notes, just after the game, I noted that "10 �c2 looks a possibility, if one is
thinking only about ...e6-es by Black; for example 10...es 11 cxds cxds 12 e4 dxe4 13 tt:lxe4
tt:lxe4 14 �xe4 h6, and White is slightly quicker in developing his pieces. There is, however,
a possible change of direction with 10 ... b s ! ?, and if the c-file is opened up, the queen is not
so comfortable on c2."
Gelfand tried the queen move in game four.
10...es
Indeed, 10...bs is to be considered here too.
11 cxds cxds 12 e4
12 dxes tt:lxes does not give White much. lt would be a different story if his bishop were
on b2. As it is, on d2 the bishop gets in the way of other pieces.
98
A n a n d- G e lfa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
12 ...dxe4 1 3 ttJxe4 ttJxe4 14 ..i.xe4
Question: Is Black genuinely equal ?
One suspects not. The pawn structure is close to symmetrical, and would have been
even more so before the exchanges on es. Clearly neither player has the better pawns. In
terms of pieces, though, White is a long way ahead in development. Black has not even had
the opportunity of moving his bishop on c8. lt would require him to play a couple of extra
good moves (a bishop move, then ...l:tc8) to eliminate the backlog in development. White, in
the meantime, can think about going on to the next stage of activating his pieces: a good
middlegame.
lt seems strange that An and would be interested in playing such a line. Maybe it is a
question of chess psychology. Over the years, he might well have noticed that Gelfand was
slightly uncomfortable in trying for an edge in open and symmetrical pawn structures, and
that he has allowed the opponents to equalize without creating pressure. One would ex
pect this position to lead to a SO-move grind, with Black having to work hard to keep the
position alive, rather than a quick 25-m ove draw.
14...ttJf6
If 14 ... exd4, then 15 ..i. g s is irritating; for example, 1S ...ttJf6 16 .l:!.c4, followed by l:txd4,
and it is going to be a struggle for Black to develop all his pieces.
99
Question: The choice here is between an endgame (with 15 dxe5) or a

middlegame (with 15 �g5). Which gives White the better chances of an edge?
Either is possible. The problem was that G elfand decided upon a quiet and modest edge
in the endgame - and then, a few moves later, went for an even quieter and modest edge.
He soon found that the position was only equal.
15 dxes
Gelfand thought for a long time here. If he were to be given a second chance, and An
and does not modify his opening, Gelfand would presumably be more than happy to try 15
�g 5 ! exd4 16 .tr.c4 - if, and this is always a big if, he could find a clear edge. (Note that this
position could also have arisen via 14... exd4 15 .ig s above.)
The natural continuation is 16 ... �e7 17 .tr.xd4 'iib 6 18 j.b1, but then Black is at least a
1 00
tempo down, and White can use this tempo to set up dangerous threats with �d3, and a
hit on h7, as later occurred in G.Sargissian-E.Hossain, Istanbul Olympiad 2012.
Instead, an earlier encounter continued (via 14... exd4 15 ..tg s etc) with 16 ...h6 17 l:txd4
hxgs 18 l:txd6 �e7 19 ..tb1 ..tg4 20 I:te1 ..txf3 2 1 l:i.xe7 ..txd1 22 l:i.xd1 .l:tfd8 23 l:tde1 bS,
A.Goganov-S.Rublevsky, Taganrog 2011.
lt looks like White has a reasonably clear, albeit small, edge - although the game ended
up with a straightforward draw, when White soon allowed a minor piece exchange (after
24 h3 g6 2 5 ..te4, when the half points were inevitable . .
So perhaps Anand has not quite found a tight defensive structure, despite earlier im
pressions, and Gelfand can play with confidence to aim for a plus score in this opening
(These comments were, of course, made just after the game). No clear win for either side,
naturally, but Black has to work to hold.
1S ...Cbxe4 16 exd6 'iVxd6 17 ..te3
101
Even here, Anand is not yet fully out of trouble, and presumably Gelfand had decided
that this was a good prospect for an edge.
11....1H5 18 'ifxd6 lt:Jxd6 19 ctJd4 .:fe8
Anand is not too concerned about allowing his bishop to be exchanged for a knight. A
more significant point is working out what to do with his rooks, and 19 .....ig6! allows Black
to see what White's pieces do, once the bishop is safe. Shipov gives 20 .:e7 .:ac8 21 .:fc1,
and now Black can defend in comfort with 21...f6!, followed by an exchange of rooks and
... .l:!.f7.
Question: Should White squeeze a bishop versus knight edge after 20 lt:Jxfs lt:Jxfs,
or should he keep his pieces in play and try for gradual improvements?
20 lt:Jxfs?!
Again, excessive simplification by Gelfand - one gets the feeling that he is willing to
take as m any draws as he can against Anand, which is not perhaps the best way to try to
win a m atch. (Again, comments made just after the game)
Away from the psychology of chess, White's knight is the best minor piece on the board,
as long as it is actively guarded by the bishop. Black's minor pieces are still slightly inse
cure, as a result of a slight slowness in earlier development. There is absolutely no need for
White to exchange either of his pieces. lt is up to Black to try to work out how to offer an
exchange himself, without a loss of tempo and perhaps the danger of the loss of a pawn.
White can improve with 20 Il.c7! ..ig6 2 1 .:fc1 .:ac8 (compare this with 19 .....ig6! above)
22 h 3 h6 23 ..if4 l:!.xc7 24 .l:i.xc? l:!.d8 2 5 ..ie3.
102
A n a n d- Ge lfa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
One can imagine that Carlsen or a younger Kramnik would have great zeal i n trying to
grind such a position out for a win.
20 tt:lxf5 21 .itcs h s 22 .l:!.fd1 .l:!.ac8 2 3 'lt>fl f6 24 .itb4 'lt>h7 2 5 .l:!.cs Yz-Yz
.••
Draw agreed. A tame finish.
Question: Could White still have tried for more?

After all, the bishop should count for something.
Anand has secured his defences well. After 2S ....l:!.xcs 26 .ixcs .l:!.e6, followed by ...b7-b6
and ...'lt>g6, White will be unable to find anything to attack.
First impressions after the first two matches? Anand is clearly not yet playing at his
best, and he also had a bad weekend in the recent Bundesliga. The chances are that he will
1 03
improve. If so, this could be slightly worrying from Gelfand's point of view. lf he cannot
create trouble when he has had the better positions - in both games - what will happen if
he starts to be under pressure?
Game Three: Anand-Gelfand

There is no great need to write in detail about the opening, or to spend days analysing this
line of the Grunfeld. lt is enough to say that Anand was familiar with the 3 f3 line, and that
it is not too difficult to find a recent game where A nand experimented with this (see the
note to Black's eighth move below). If I had noticed this, then Gelfand will have noticed it
too, much earlier than me. He had prepared this line, innovating at move 16.
One suspects that there were various possible alternatives between moves 8 and 15,
and that quite possibly there could have been some sort of improvement for White. Again,
this is beyond the scope of this book. The point is that Anand had played what seemed at
the time to be the best line for White, and that Gelfand had found a big improvement for
Black. Certainly it was not a winning improvement, but he forced Anand out of the comfort
zone.
14th May 2012

V.Anand- B.Gelfa nd
Grilnfe/d Defence
1 d4 tiJf6 2 C4 g6 3 f3
The start of a new thread of analysis, avoiding the main line Grunfeld with 3 tiJc3 ds -
and for good reason, as it could easily have ended up in disaster for Anand in game one.
This move is likely to end up in either a Grunfeld, albeit without ever reaching any of
1 04
A n a n d - G e lfa n d, World C h a m p i o n s h ip, M osco w 2 0 1 2
the main lines, or a King's Indian Samisch, although even here, as Anand was able to show
later on (see game eight), we do not necessarily reach the Samisch either.
A lot is relatively unexplored, but sharp - an ideal line for White when trying to plug
away for a plus score in a m atch.
3 ...ds 4 cxds tt:Jxds s e4 4Jb6
The point is that Black never had the opportunity of exchanging on c3.
6 tt:Jc3 i..g7 7 i..e 3 0-0 8 "iWd2
Question: For those who want to catch up on their opening theory, what
does Black usually play here? If you do not already know, try to think about it.
s es
...
This is, of course, preparation by Gelfand.

In a game played several months earlier, White won quickly after 8 ... 4Jc6 9 0-0-0 "iid6 10
'it>b1 �d8 11 tt:Jbs �d7 12 dS tt:Jes 13 't!Vc2 c6?? 14 tt:Jxa7, already with a winning position,
V.Anand-S.Mamedyarov, Ajaccio (rapid) 2011. Not a game to be looked at in depth, except
to note that Anand had experimented with this variation. Naturally, Mamedyarov would
not have played so badly at a normal time limit.
A more representative continuation would have been 12 ... a6 13 dxc6 'i¥xd2 14 l:.xd2
�xd2 15 i..x d2 axbs 16 cxb7 .txb7 17 .txbs fs 18 exfs, as in N.Vitiugov-A.Giri, Reggio
Emilia 2011112. The computer likes this for White, but Black has active play for his pawn,
and later m ade a draw.
There are also some lines in which White has played an early f3-f4. No deep theory here.
Try it out yourself, if you are interested.
9 d 5 c6
Presumably 9 .. .fs is reasonable enough, but Gelfand continues with his prepared line.
10 h4 cxds 11 exds tt:J8d7
105
This looks strange for Black, until one appreciates that a later . tt:Jf6 will create great
. .
pressure on White's d-pawn. The pawn structure has become thoroughly asymmetrical,
with White trying to keep the advanced pawn, and also trying to open up the h-file with
h4-h 5, while Black will try to do something with his potentially mobile e- and f-pawns ...
12 h S tt:Jf6 13 hxg6 fxg6
... or, if the f-pawn gets exchanged, he will try to do something with his half-open f-file.
14 0-0-0 .i.d7
lt looks like wild attacks against kings on opposite wings.

15 'it>b1
Except that An and is carefully bringing his king into safety, with this move and the
next. He is aiming for a longer-term initiative, making use of the advanced passed pawn,
rather than aiming for quick mate.
106
A n a n d - Ge /fa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
Question: How would Black h andle the position if

White were to attack with 15 i..h 6 - ?
lt is not as if Black is going to get choked on the g7-square, so he has some freedom, and
some counterplay, with 1S ... i.xh6 16 'ifxh6 .l:tc8.
If White tries to escape with the king now - a couple of moves too late - Black is very
comfortable after 17 'Et>b1?! �fS+ 18 �a1 l::!.f7, and if 19 g4?!, then simply 19 ...�xg4 20 fxg4
lLlxg4, followed by ...lLlf2, gives him the advantage. 17 d6?! lLla4 18 lLlge2 �6 also allows
Black to take over the attack.
Possibly 17 1M3, when 17 ... i.a4 18 l:td2 (18 i.xg6 'i¥d7 allows Black more initi ative than
the pawn is worth for White) 18 ... .l:tc7 appears equal, but if the computer is allowed to run
for a few seconds, White is given an edge after 19 �b1!, followed shortly by lLlge2. A differ
ent setting, but again 'it>b1 is the critical idea.
1S ...l:.c8 16 �a 1
Question: Now what? Isn't Black just significantly worse?
16 ...e41
This was a prepared variation. The big advantage of a prepared line is that the oppo
nent, if he has not also seen it, will find it far more difficult to find good moves with speed.
That, at least, is the theory. Innovations can be good, bad or indifferent, and here Gelfand
has to show that his move is at least playable.
Black has sacrificed the pawn to open up the long diagonal for the bishop on g7. With
out the pawn sacrifice, the bishop is merely bad. Without the pawn, Black can attack the
squares around White's king.
Even so, this must be regarded as a shock move, an attempt to confuse his opponent.
107
The move itself was probably the best, but can we guarantee that Black's previous moves
were position ally the most accurate? lt would seem, for example, that 15 ..ih6 would have
given White a slight edge. If the opening were to be repeated in another game, there would
be an opportunity for White to look for an improvement.
Question: What should White do next? The computer suggests half a dozen
promising moves, and if even one of these is good, Black could face problems.
17 ..id4!?
A good and safe move, with a slight edge, but by no means a winning line. lt is good
enough, and that is all that is important. Anand challenges the long diagonal.
Were there any other possibilities? What about other lines, maybe more ambitious?
a) 17 fxe4? would have been bad, as after 17 .. J::tx c3 18 �xc3 tt:lxe4 Black is at least more
than equal. After 19 �e1, there would be some pleasing knight forks: 19 ...tt:la4 20 i.d4
108
A n a n d - Ge lfa n d, World C h a m p i on s h ip, M osco w 2 0 1 2
20 ...tt:if2 !, intending 2 1 ..ltxf2? ..ltxb2+ 2 2 'it>b1 J::rxf2 2 3 'ifxf2 tt:ic3+ 24 �xb2 tt:ixd1+ fol
lowed by 2 S ...tt:ixf2.
21 .i.c4 is more solid, but even here there are some enjoyable tactics (and a player will
never become good at the game, if they cannot take pleasure out of tactics}: 21 ...tt:ixh1 22 d6+
Wh8 23 ..ltxg7+ 'it>xg7 24 tt:if3 tt:ixb2 25 'it>xb2 'ii'f6+ 26 'ifes tt:ig3, and the two queens are mu
tually pinned against the kings, which gives Black the chance to escape with the knight.
Of course, Anand stayed clear of such lines.
b) 17 .i.h6 is by now a defensive move, rather than an attack against the king, but per
haps White still keeps a slight edge after, for example, 17 ... ..1txh6 18 'ii'xh6 l::rf7 19 tt:ige2 (19
d6 tt:la4 is about equal} 19 ...'ife8 20 �d4 e3 21 g3.
c) 17 ..ltgs, pinning the knight, also looks good for a slight edge after 17 ... exf3 18 gxf3.
If there are three good bishop moves by White, and if, furthermore, White could have
deviated a couple of moves earlier with a probable slight advantage, the overall conclusion
must be that Gelfand was playing this variation for shock value, rather than trying to sug
gest that he has found a serious way to equalize.
17 tt:la4
...
Question: Is this move good?
In my day-after blog, I noted that "I suspect that, had this been a player other than a
candidate for the World Championship, I would have regarded this as something of a wild
hack. Black is doing nothing in the centre, and his attack against the queen is almost non
existent. 17 ... exf3 would have been quieter, with a slight edge for White after either 18
tt:ixf3 kg4 or 18 gxf3 .i.fs."
A bit harsh? In fairness, I noted the practical point that, strangely enough, "before long,
Gelfand is on top, and Anand is under pressure." Maybe the psychology is good, but the
move itself is not good.
1 09
Question: Inevitable question: if Black's last move should be

branded as an error, what is the best reply for White?
18 liJge2
Another difficult choice, one of many for both sides. Here 18 tLlxe4!? is promising, and if
18 ...i.fs, there are presumably several quiet moves that keep a slight edge for White.
The computer recommendation, 19 'ifb4, is not quite so clear-cut after 19 ...liJxe4 20
fxe4 �c7 21 liJe2 'ili'c2 22 'ii'd 2 i.xe4 23 'iVxc2 �xc2 with equality. Is there anything more?
Try to think back to the mid-19th century, and the fascination by Morphy with the great
1834 Labourdonnais-McDonnell matches. Morphy recognized that, in the heat of tactical
battle, his two predecessors sometimes got too interested in tactics, and often forgot about
development. Apply the Morphy idea to the current position, and we can try 19 �d3 !?
'ii'x ds 20 liJe2.
110
Suddenly White's pieces are fully developed and centralized, attacking on the kingside
(where Black's king is open) and developing on the queenside. Black's f6-knight is pinned, a
serious problem, since if 20 .. .'�Jxe4? 21 fxe4 ..txe4 2 2 ..txe4 ..Wxe4 then 2 3 ..txg7 'it>xg7 24
11Vh6+ with a quick win. Alternatively, 20... ..te6 21 b3 lbcs 2 2 lbxcs .l:txcs 2 3 �2 keeps
White in control, making full use of the long diagonal, and not allowing Black to take ad
vantage. Anand would clearly have been much better, with best play.
lB ..Wa s
...
Consistent with his sideways plan. 18 ...exf3 19 gxf3 lbcs 20 lbf4 is more natural, but
White keeps a slight edge, offering perhaps a bishop exchange with ..th3 .
19 lbxe4
White is happy to simplify.
Question: Is there anything wrong with 19 fxe4 - ?
I have to admit that I missed the idea in my blog scan, but both players would have cal
culated it. After 19 fxe4 lbxc3 20 lbxc3 lbxe4 21 lbxe4 �xd2 2 2 lbxd2 ..txd4, Black is slightly
better, as we approach the endgame with bishops of opposite colour.
19 .. .'i¥xd2 20 lbxf6+ �xf6 21 .l:txd2
111
Question: There has been some simplification over the

last few moves, but what should Black do next?
21 .l::[fs?
...
Too provocative. With complete accuracy now on both sides, it would seem that White
should be able to win, therefore a full question mark, "?", rather than a half question m ark,
"?!". Even so, the rest of the game is difficult and complicated for both sides.
Here 21 ....l:[d6 blockades the d-pawn and Black can soon win it back, helped by White's
l ack of kingside development: 22 ..ixg7 'iitx g7 2 3 l:th4 t2Jb6 24 .i::[h d4 (24 t2Jc3 is only a draw,
as after 24 ....l:[xd5, the knight is pinned) 24 .....ib5 25 g3 l:tcd8, and Black finally recovers the
pawn with a level endgame.
Shipov suggests instead 23 b3 t2Jb6 24 tLlf4 .l:[cl+ 25 'it>b2 l:Ie1. Equal? Unclear? Frac
tional edge to White? The most important point is that it is not a clear win for White,
which is useful if Black is trying to defend a difficult position.
22 ..ixg7 'it>xg7
112
A n a n d- G e lfa n d, World Ch a mp i o n s h ip, M osco w 2 0 1 2
Question: "Passed pawns must be pushed". So what should White do here?
23 d6?!
Too hasty. White opens up space, allowing Black's rooks to double on the c-file, which
allows him to create counterplay. Quite probably White should be winning eventually, but
there is extremely difficult play on both sides and, in time trouble, Anand was eventually
unable to find the win .
White should have waited before pushing. 2 3 l:i.h4! improves; e.g. 2 3 ...tt:lb6 24 l:i.hd4
�bS 2 5 b3, and it is not so clear what Black can do. He is, after all, a passed pawn down.
23 J:i:fcs 24 .l:!.d1
••
An extremely difficult position to analyse, whether over the board or at home. The ques
tion and answer can be given in two parts.
113
Question: First, is 24 .. .l::t c 2 satisfactory? If so, then find the main lines.
If not, what else can Black try, to make life more difficult for his opponent?
24 a s
...
Very puzzling, and I have to admit that, in my analysis, I got on completely the wrong
track, thinking that in one apparently critical line after 24 ... l:rc2, Black can escape with a
draw with a few tactics. There was, however, a second critical line in which White has a
win.
If 24.. Jk2, quite clearly the most natural m ove, does not work, what else can Black try?
Gelfand uses an old technique, that if you cannot do anything constructive with the pieces,
all that remains is to make progress with your pawns. Here ... a7-a5 is a desirable move to
flick in. Therefore 24... a5, and the pawn push makes its presence felt in later play.
However, this would probably have been Gelfand's second choice, if there were prob
lems with the immediate 24 ... .l:.c2. Let's see.
a) 25 .l:!.h4 looks promising and is certainly something worth analysing. Then Black could
transpose to the main line in the game with 25 ... a5, but he could also try captures on b2.
al) 2 5 .. .<�:Jxb2? 26 .l::tb 1 l2Ja4 is unsatisfactory, as the tactics do not work after 27 l:txb7
.l':!.xe2 28 .i.xe2 l:rc1+ 29 l:tb1 .l:.xb1+ 30 �xb1 lZ:lc3+ 31 �c2 l2Jxe2 32 .l:.e4, since if 32 ... .i.f5,
then 33 �d2 lZ:lg3 (33 ... .i.xe4? 34 d7 promotes) 34 .l:i.e7+ �f8 35 .l:!.xh7 �e8 36 l:rxa7 and
White h as far too many pawns to give Black any chance of a realistic defence with bishop
and knight versus rook. Sharp tactics, but White is still keeping the crucial passed pawn.
a2) Black can improve with 25 ...�xb 2 ! ? 26 l:rxa4 .i.xa4 27 �xb2 .i.xd1 28 lZ:lc3.
The critical position. The bishop is trapped(!), but Black can pick up the passed pawn by
28 ... .l:.c6 29 lZ:lxd1 .l:.xd6 with a probable draw, even if White keeps a slight pull. lt is difficult to
see how White can create another effective passed pawn, so long as Black keeps his structure
tight, not advancing to the centre ofthe board. This, however, is just one line of thought.
114
A n a n d- G elfa n d, Wo rld C h a m p i o n s h ip, M os c o w 2 0 1 2
b ) The more prosaic 2 5 b 3 gives White a comfortable edge - indeed a decisive edge - af
ter 25 .. .':'i:Jc3 26 tt:Jxc3 .l:i.8xc3 27 .i.d3 ! l:.xg2 28 l:.he1.
Suddenly White's undeveloped kingside pieces jump into activity, the bishop covering
the c2-square against the black rooks doubling in the second rank, while his own rook is
able to eliminate any defences against the passed d-pawn; e.g. 28 .. .<i£tf6 29 .l:Ie7 .ifs 30
l:.xh7 .i.xd3 3 1 d7, queening.
Still, this is not yet the whole story. Gelfand must surely have considered the strange
but resourceful 25 ... tt:Jb2. This can confidently be said, since Gelfand soon made the same
m ove himself. Play then continues with 26 l:.b1 tt:Jd3 27 tt:Jd4 .l:td2 28 .i.xd3 l:.xd3.
This is much as in the game, except there White had added l:.h4 and Black ... a7-aS.
Which insertions or omissions would be the more significant?
Gelfand presumably decided that he preferred his opponent's rook to be off the back
rank. After all, in the diagram position above, White has 29 l:.bd1 l:.xd1+ 30 .l:txd1, and if
115
30...'it>f6 then 31 f4 .l:!.d8 3 2 l:td2 (to avoid tempo hits with ....i.g4).
Now how would you enjoy handling the king and pawn ending? Remember we are al
ready almost ten moves deep into the analysis, and Gelfand would have had to consider
several other variations, including the ones already given. lt was no wonder that Gelfand
looked tense during the match.
Here it seems that 32 ....i.c6! 33 lLlxc6 bxc6 34 Wb2 We6 35 'it>c3 .l:!.xd6 36 l:txd6+ 'it>xd6 3 7
'it>b4 holds the balance after 37 ...h s 3 8 'it>as 'it>c7. Indeed, the only danger is of White over
pressing, bringing the king to a6 and pushing the b-pawn, since if then b4-b5 ??, Black sud
denly wins by ... c6-cs.
There are other lines which could be unearthed, but I would prefer the book to be ready
for publication in 2012 rather than 2013. Provisionally though, it would appear that
24 ... .l:!.c2 ! was more accurate, and certainly at least in practical terms. lt is difficult to find a
quick refutation to this move.
2 S l:th4
White can't do anything with the knight or bishop, so it has to be a rook that is needed
to activate his pieces.
25 ....l:!.c2 26 b3
116
A n a n d - Ge lfa n d, World C h a mpio n s h ip, M os c o w 2 0 1 2
26...ll:lb2
With great astonishment to the spectators - although Gelfand must have thought
about this while considering his 24th. The problem is not so much that the knight is in the
lion's teeth, but rather that Gelfand has slightly mistimed it.
Black could try to burrow down with 26 ...ll:lcs 27 'i!tb1 ll:le6, but would he be happy to do
so, a pawn down?
Question: Which is better: to attack the knight with 2 7 �b1 and

later to force simplification ? Or to try to put the knight into a cage
with 27 .l:tdd4 - ? Other moves may also be considered.
27 �bl!
A good and straightforward move, and indeed the best move, aiming for favourable
117
simplification and an excellent chance of forcing a win. Unfortunately, albeit under

standably, Anand was getting short of time.
27 .l:!.dd4 is also good, but is more complicated after, for example, 27 ...a4 28 bxa4 'Llc4
29 .l:!.he4 'it>f6 30 .l:!.f4+ 'it>g7 3 1 .l:!.de4 .txa4 32 I!e7+ 'it>h6 3 3 11c7
and now a final swing with 33 ... .i.b3 34 l:tfxc4 l:txc4 3 5 axb3 114xc7 36 dxc7 11xc7 3 7
'it>b2, and this time the pawns are equal i n number, s o the bishop and knight should even
tually outplay the rook. Perhaps not something to be tried during a time scramble though.
Best to play simpler lines.
The computer suggests that 27 11d5 might also give an edge, but it is not clear that this
is as effective as 27 .l:!.dd4.
27 ... 'Lld3 28 'Lld4 .l:!.d2 29 .txd3
The bishop does not budge for the first 28 moves, but then gets exchanged for the far
advanced knight, which has moved six times. This must surely suggest some gain oftempi
by White.
29 ....l:!.xd 3
1 18
30 l:i.e1
And now it is White, not Black, who is seriously attacking the king. The exchange of
tempi - looking back at 24 ... as 25 .l:!.h4 - has allowed White to provide extra defence on d4,
and so activate his other rook.
30 l:i.d2
...
Question: Assess the position. Who is better, and by how much ? What is White's
best? Do this exercise quickly. Remember, both the players were short of time.
I, too, did not have time to assess the position in "live chess". You concentrate on follow
ing the main part of the game, then there is the time scramble; you do not look at the time
scramble carefully before the annotations of the main part of the game get started.
I wrote some total nonsense, noting that "play is starting to fizzle out to a draw. White
is a strong passed pawn ahead, but his weakness is that he permanently has to be careful
119
about checkmate, on the back rank, or doubled rooks on the seventh. White has no chance
to coordinate his pieces for any sort of attack."
Well, perhaps 60% total nonsense, and a few reasonable comments.
In fact, An and has shown excellent composure when Gelfand's attempted attack was
raging. His next move should also have been winning a move further on: 31 1Ie7+ �f6 3 2
Wb1. Of course he must not allow checkmate with ...�cl.
3 1 'it>b1! i.fs+ 3 2 lt:Jxfs+ gxfs 3 3 .l:.e7+
lt is understandable that, with two pairs of rooks trying to give checks, and nothing else
attacking the king, the easy interpretation would be this was going to end up in some sort
of perpetual. Quite probably both players would have had similar thoughts in their mind.
3 3 ... 'itig6
Question: Find the win !
34 1:tc7?
This only ends up as a draw.
Opportunity in chess often happens only fleetingly. A couple of hours of intense con
centration and one player has a winning position but is short of time, having perhaps less
than a minute to make a move - concentration lapses, and suddenly the win has blown
into sand. Any serious chess player will know the feeling.
This time, passed pawns should definitely have been pushed. Here 34 d7! is enough to
tip the game in White's favour. lt is not so much that the new queen itself is winning. lt is
more that the mere threat of promotion forces Black to allow the disruption of his own
forces. This would be seen most clearly in 34 ... 1Id8 3 5 �hxh7 f4 36 Wc1 .l:i.d6 37 a4, and
White will have plenty of leisure to brink his king forward, with an eventual win.
What about immediate counterplay by Black? The immediate impression is that
34 ... 1:tcc2 forces a perpetual.
120
A n a n d - G e lfa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
Then White has the tricky 3 5 ILc4! ! and there i s n o perpetual, since after 3 5 ...l::tb 2+ 36
�cl, the rook covers the checking square on c2 - very easy to miss - while 3 5 ....l:!.xc4 36
bxc4 is an eventual win for White. He is a pawn up and has plenty of opportunity to ex
change his passed pawn for one of his opponent's pawns, maybe with a well-timed d8'i!V
and l::tx b7. White can advance his king, with �c2, 'it>c3, etc, before setting up the winning
exchange.
Another attempt for a perpetual is 34....l:!.d1+ 35 �b2 Iid2+, but this time the king runs
forward with 36 �a3 .l:!.cc2 37 �a4 and escapes into active play.
So everything fits, but how many players could be able to analyse this in ten seconds?
34...�e8
Black is now instantly equal. The main defensive threat is 3 5 ... .l:!.e1+ 36 Iic1 l::te e2.
3 S l1h1 �ee2 36 d7 l::t b 2+ 3 7 'Ot>c1 l::tx a2 Yz-Yz
Very tough, but to win World Championships you are expected to overcome extreme
121
difficulties. Anand should really have won this game, but Gelfand certainly had his win
ning chances in game one, and to a lesser extent in game two, so the "luckometer" was
therefore about level. None of the three draws was faultless, when examined closely and
with the help of the computer, but very few players would have gone unscathed, whether
as White or as Black, in the third game.
We are still a long way off from the ideal of perfect chess, and perhaps this is no bad
thing. Chess players have something to aspire to.
Game Fou r: Gelfand-Anand

As in game two, Anand gradually equalized with a symmetrical pawn structure and made
a steady draw (even if one feels that Gelfand might again have placed his opponent under
greater pressure). This sort of thing happens, and indeed it ought to happen a lot of the
time, at anything beyond intermediate level. Most of the time, though, it doesn't happen.
Players quite simply make mistakes, or try to overpress their position, or start to crumble
under pressure, against accurate and/or inspired play by the opponent.
lt is easy to skip through "dull and boring" games, but in terms of technique, if both
players have played well, a reader who wants to improve will learn lessons. Below grand
master level, there will almost always be mistakes made, and making mistakes is frustrat
ing. What do grandmasters do to avoid making mistakes? Or is it that they make mistakes
anyway, but their mistakes are smaller and less significant than those of the rest of us?
15th May 2012

B.Gelfa nd-V.Anand
World Championship, Moscow 2012 {Game 4)
Semi-Siav with 5 a6 ...
1 d4 ds 2 c4 c6 3 lZ'lc3 lZ'lf6 4 e3 e6 5 lZ'lf3 a6 6 b3 i.b4 7 i.d2 l2Jbd7 8 i.d3 o-o 9 o-o i.d6 10 �c2
122
A n a n d- G e lfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
10...es
I had suggested, when blogging on game two, that Black could have tried 10... b5. How
ever, Anand stuck to his main idea. lt is not so clear that Black was fully able to equalize
here, but Gelfand seemed to trust his opponent's judgment.
11 cxds cxds 12 e4 exd4 13 tt:lxds tt:lxd s 14 exds tt:lf6
Question: Isn't White clearly better?
Yes! White is not yet winning, but Black has to work hard to try to equalize. White is
ahead in development, if only slightly. He has brought his bishop from cl to d2, and his
queen from dl to c2, while Black's corresponding pieces remain unmoved. Black has no
chances of playing for a win, barring a clear error by White, and therefore such an opening
would be inappropriate if Black is, for example, aiming for a win in a weekender, or trying
to gain a few rating points.
This is matchplay, however, and in theory the aim is to give absolutely nothing away
with the black pieces, and try for a slight edge by White. To this end +1 =11 -0 would be a
good score, a World Championship score; whereas +5 =2 -5 would not be so good.
Anand wants his six safe draws as Black, and somehow to find a win for White. Mean
while, Gelfand as White has to pin his hopes on seizing on microscopic edges, and - very
difficult - taking full advantage of them.
In games two and four, Anand drew both times, but not quite convincingly. If Gelfand
were to play with extreme accuracy, he would have had the chance to press hard for a win
in either of these games.
15 h3
To eliminate any possible activity with ...i.g4. The naive 12 li:lxd4?! �e5 13 i.c3 tt:lxd5
gives Black a slight edge.
1S ...i.d7 16 J:iad1
1 23
With the rook off the long diagonal, lbxd4 is now a threat. Black has no time to try
... lbxds himself.
16 J:res
..
Anand concedes that he is losing a pawn - not really what he would have been hoping
for in playing ultra-solid chess. His main hopes are that he will be able to keep his pieces
solid and that, eventually, he will regain White's own isolated d-pawn. Nevertheless, it will
take a lot of time before Black can take the pawn safely.
Question: Can Black think about regaining the pawn

immediately with 16 .. J:tc8 17 ll¥h1 h6 18 lbxd4 lbxds - ?
Black is too far behind in development to allow this to work, even if the opposing pieces
and pawns are close to symmetrical. Sometimes the "advantage of the first move" counts
for nothing, but at other times the extra m ove can be vicious. Here 19 i.. e4 gets the attack
in first, and if 19 ... i.. e s 20 i.. x ds i.. xd4, White has a decisive advantage after either 21 i.. x b7
or 21 i..x h6.
Black would have try to defend with 19 ...i.. c 6 or 19 ...i.. e 6, but in each case his pawn
structure would be seriously disrupted after White takes it with the knight.
17 lbxd4 �c8 18 �b1 h6
124
A n a n d- Ge /fa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
Question: Impressive symmetry, but what is White going to do next?

The likelihood is that White's extra pawn will drop at some stage, so it
is important that Black is forced to make concessions before recovering it.
19 tt:lfs
White intends to gain the advantage of the bishop pair before Black can regain the
pawn. Plausible, but it does not give White a workable edge.
Instead, 19 lirfe1! keeps play more complicated and seems to be a genuine improve
ment. Either Black can exchange the rooks himself, when White has the chance to coordi
nate his pieces, or he can allow White to exchange rooks himself, after which Black's own
pieces will be slightly disorganized.
If, for example, 19 ...lt:lxd5 20 l:f.xe8+ i.xe8 21 �h7+ �h8 22 �e4, Black is under great
pressure on the h1-a8 long diagonal and on the d-file. White's gain of tempo converts to a
clear edge. lt is nothing yet decisive, but after 22 ... i.c6 23 lt:lxc6 bxc6 24 'ii'c 2, White has
weakened Black's queenside pawns and gained the bishop pair.
Alternatively, Black can exchange with 19 ... l:f.xe1+ 20 i.xel, when White has opened up
some space for the rook on the d-file. For instance, 20...lt:ld5 2 1 lt:lf5, and White is ahead of
the game.
Instead, 20....Uc5 attempts to reach something close to the line in the game, but then 2 1
�c4 is promising.
125
If 21 ... bs 22 .i.b4! bxc4 23 .i.xcs .i.xcs 24 bxc4, and White has rook and two pawns for
the two bishops, a slight material edge for White. Furthermore, the two pawns are strong,
connected and passed, while Black's two bishops do not coordinate well. White has a clear
edge here.
We must remember, however, that the defender is also allowed to play well. 19 ....i.cs
seems the toughest defence, one point being that after 20 .l:l.xe8+ .i.xe8 21 tt:Jfs, Black can
try to equalize with 21 .. .'iWxds.
All of a sudden, tactics emerge from even the quietest and most symmetrical openings.
lt is clear that White is not worse, since his position is still better developed. The only ques
tion is whether he can keep the edge.
Of course, there is a third possibility, that he can overpress and end up worse, maybe
through bad tactics. The computer at first recommends 22 .i.xh6? gxh6 23 tt:Jxh6 'it>f8 24 .i.fs,
but then notices that Black has far more effective play with 24 ... i.xf2+! 2 5 'it>xf2 'ii"c s+.
126
This i s not the sort of position White would be aiming for i n a quiet positional game.
Indeed, his king is so badly exposed that he is actually losing.
Rewind, and 22 .i.e3 ! is best, with a good positional edge.
This time the tactics would favour White, Black's queen standing awkwardly on the
open d-file. For instance, after 22 ... .i.c6 23 tt:Jxh6+ gxh6 24 .i.h7+ tt:Jxh7 25 .l:!.xds .i.xds 26
'ir'c2 as 27 a3 .l:!.c6 28 .i.xcs b6 29 �dl .l::tx cs 30 b4 axb4 31 axb4 .l::tb s 3 2 �g4+, foll owed by
11Yd7, White is winning material.
127
Players could not be expected to see all this as far back as move 19. What is needed is
for the player to have good tactical instincts, a sense that if one side has the more mobile
pieces in an open position, something is quite likely to work in his favour after the jousting
of attack and defence. lt is a matter of positional play, as well as of tactics. If a player has
the more active pieces, the tactics will often emerge, even if he has not seen everything in
advance.
And we have not yet found a convincing way for Black to equalize. A missed opportunity
for Gelfand, it seems.
19 ....i.xfs 20 .i.xfs .l:tcs 21 .l:tfel .l:txd s
The pawn has now been safely recovered. All that White has in his favour, in terms of
aiming for an advantage, is the bishop pair. lt is not enough, if there are no opposing
weaknesses to attack.
22 kc3
128
Question: How should Black equalize, with the greatest speed and comfort?
22 ....l:rxe1+?!
22 ... .ie5 is, quite simply, level. He will have neutralized White's dark-squared bishop, so
there is now nothing to attack with.
23 .l:rxel .ics
Question: White to play. What should he do now?
24 'iWc2?
Very strange. We can attribute this to the extreme tension of playing in a m atch for the
World Championship.
This quiet move is essentially a non-move, whereas the simple 24 .ixf6! gxf6 {not
24 ... 'iWxf6?? 25 l:te8+ .if8 36 .ih7+ and mates) 25 �e4 would give White a clear edge again,
since the black kingside is now wrecked. lt is unlikely that Black is as yet losing, but he
would have to defend extremely accurately for perhaps forty more moves to prove that he
is holding, and so will be under great pressure.
24 ... .id4 25 .ixd4 .l:rxd4
Now we have only a microscopic edge for White, and Anand is soon able to halve out.
26 'iWcs g6 27 .ig4 hs 28 'iWxdB+ .l:rxd8 29 .if3 b6 30 .l:rcl .l:rd6 31 �fl as 32 'it>e2 tt:Jds 33 g3
tt:Je7 34 .ie4 'it>g7 Yz-Yz (see diagram overleaf}
Even here, the position is not yet dead. There are still opportunities for White to play on
with 35 f4, and perhaps a l ater push with g3-g4. Probably in the end it would not amount
to anything more than a draw, but Gelfand could continue to nag his opponent.
Note carefully that it was An and, as Black, who offered the draw, and Gelfand that ac
cepted. This adds to the impression that, in the games so far, Gelfand was not confident in
129
playing for wins against Anand, and that he was playing for safety, making draws as often
as he could. Anand would clearly have sensed the vibes, and was happy to offer an early
draw on any reasonable occasion.
Ciame Five: Anand-Cielfand

Both players seem to have settled down, no longer wanting to play extravagant chess (An
and in game one, Gelfand in game three), and attempting to make sure that they make no
unnecessary mistakes.
After this game, there have been five draws in a row and, as the match goes on, one can
expect that play will become even more cautious. Eventually, it is likely to be a battle of
attrition, maybe with one of the players getting tired, losing control a little, and losing a
game. Or perhaps it will end up in a rapidplay finish after twelve draws.
If we are indeed looking for a match of drawish chess, who is more likely to be favourite,
if the balance were to tip? If anything, Gelfand, regarded as the underdog, might be re
garded as slight favourite.
(Written in the day-after blog, with a few typos corrected. Of course, there were several
more draws tofollow, but unexpected results also crept in.)
17th May 2012

V.Anand-B.Ge lfa nd
Sicilian Defence
1 e4
A change of opening.
1...cs 2 ti:Jf3 tt:Jc6 3 d4
130
A n a n d - Ge lfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
Anand wanted to try out a m ain line Open Sicilian. If h e i s unable to find an edge, then
there is still time to opt for something new.
3 ...cxd4 4 tt:Jxd4 tLlf6 5 tt:Jc3 es 6 tLldbS d6 7 ..ltgs a6 8 tLla3 bS 9 tt:Jds i.e7 10 i.xf6 i.xf6 11
C4 b4 12 tt:Jc2 0-0 13 g3 a S 14 iJ.g2 iJ.gS 15 0-0 iJ.e6 16 'iiVd 3
All mainstream. The only real comment to be m ade so far is that the fianchetto system
with g2-g3 and i.g2, in addition to c2-c4 rather than c2-c3, m akes it difficult for Black to
aim for sharp kingside pawn play - in that, if he were to play .. .f7-f5, White would have an
instant open long diagonal after e4xfs.
16 i.xds
•..
Gelfand, too, is looking for steady chess and, with the black pieces, the prospect of a
draw. There are reasonable ways to keep the tension, a recent game continuing 16 ...'1i'b8
17 f4 i..d8, J.Borisek-A.Moiseenko, European Team Championship, Porto Carras 2011. Black
eventually won, but the mistakes came much later on.
131
Question: How should White recapture? I leave this as an

open-ended question, in that all three recaptures have been
played in high-level chess. Don't rely on the databases. Just think
about this position yourself, and what you would play next in a game.
Each of the three recaptures on ds would allow White to claim that he can still play for
an edge. There is no obvious answer, but players of different temperament or, more impor
tantly here, the same player in a different tournament or match situation, would handle
the position differently. If, for example, a player is absolutely desperate for a win, and a
draw is simply not good enough, he would keep the position asymmetrical, probably with
17 exds. The second choice perhaps would be 17 �xds. If White wants to make absolutely
certain that he has no danger of losing, then a highly symmetrical position, allowing no
obvious weaknesses for either side, is good enough.
11 cxds
And here, as in the second and fourth games (albeit as Black), An and aims for symme
try. I am not sure whether it is better, or worse, than the other recaptures, but 17 exdS
might well create more danger for both sides.
Question: Where should the knight go?
17 'bb8?!
...
Here 17 ...'be7 looks genuinely equal, whereas Black's attempt to play more sophisti
cated chess ends up in unexpectedly creating a weakness. Black's idea is to manoeuvre his
knight to cs, a good square, and just about impossible if the knight is on e7. The trouble is
that, in a couple of lines, the knight on b8 turns out to be in the way of the major pieces,
and White could have kept a slight edge with alert play. See the next comment.
132
18 a 3
Question: Should Black defend with 18 lt:Ja6 or exchange with 18 ...bxa3 ?

... -
18 lt:Ja6
...
Black has the awkward problem of deciding which of two reasonable-looking lines to
choose, in each of which the opponent has a slight pressure. Perhaps Black's position is not
particularly bad, but it takes some care to find a way to equalize after, arguably, a slightly
inaccurate previous move.
What it boils down to is whether Black, after the exchange of the queen side pawn, is go
ing to be safer with the remaining pawn on as or on b4. Normally a pawn on as would be
safer than on b4, because if Black's b-file is fully open, he can create counterplay against
White's b2-pawn; whereas if Black has a b4-pawn instead, the problem is that his pawn
can then be more easily pressured than the white one on b2.
Black does not fully equalize with the line he chose, and White could later have tried to
improve. So the question is whether Black can equalize with 18 ...bxa3, or at least has less of
a disadvantage. This is always difficult, when trying to evaluate two positions a few moves
later, in which there are no obvious tactics emerging. lt will be a matter of what is going to
happen, not two or three m oves along the line, but in later manoeuvring. And yes, this is
more about what is likely to happen during the endgame, rather than in the middlegame.
This is not surprising, since it is the sort of middlegame which forces players to think about
future endgames.
As far as moves are concerned, if Black tries 18 . .bxa3, the most threatening recapture is
.
19 'iWxa3, keeping tabs on both the as- and d6-pawns, and so preventing Black from rede
veloping with ... lt:Ja6 or ... lt:Jd7. If 19 'iWb6, then 20 ..ih3 keeps the knight under partial con
...
trol.
133
lt is tempting to say that, visually, White has an edge, but perhaps it will fizzle to equal
ity after accurate play; e.g. 2o.Jt:Ja6 21 'ifxas 'i:Vxb2 22 'ifa4 Ct:Jc7 2 3 'i:Vc6 .l:.xal 24 .l:.xal tt:Jbs
25 .l:.a8 iie7 26 .l:.c8 g6.
Black is equal, as the pawn weakness on d6 is now no more significant than White's
pawn weakness on e4.
lt would take good defensive nerves for Black to try such a line, but it is an improve
ment from the one he actually played, in which White could have kept an edge.
19 axb4 tt:Jxb4 20 tt:Jxb4 axb4 21 h4 iiJ.. h 6
134
A n a n d- Ge lfa n d, World C h a m p i o n s h ip, M osco w 2 0 1 2
Question: What should White play here, if h e wants to keep a slight edge?
22 iL.h3?!
A missed opportunity. Anand finds a way to manoeuvre his bishop to c6, entrenched on
the opponent's side of the board, but so what? The bishop is attacking only thin air, unable
to put pressure on any pawn, and Black may later put counter-pressure on the e4-pawn.
Play soon tends to equality.
Instead, 22 'ifc4! is uncomfortable for Black, the pawn on b4 being genuinely weak. For
instance, 22 ...'ifb6 23 'ifc6 attacks the pawn that the bishop can no longer defend, or
2 2 .. ..l::tb 8 2 3 l1a7, in either case with pressure. Black's bishop ought, in theory, to be able to
give solid cover to the b4-pawn, but it cannot reach the queenside.
lt is possible that Black would find it prudent, in some lines, to abandon the pawn, hop
ing that White cannot find a win with only an isolated extra pawn, and with bishops of
opposite colour. However, Black would still need to work extremely hard, and there would
be reasonable chances for White to have tried for a win.
22 'ifb6 23 iL.d7
...
135
Question: Which is the best square for Black's b-pawn? On b4 or on b3?
23 ... b3
When I blogged, I was critical of this pawn move, as it seems much easier for White to
attack a pawn on b3, rather than on b4. Black could easily have kept the position safe with,
for example, 23 ....l:Ia7.
I now acknowledge that 23 ...b3 does not give anything away to White. Unless Black is
careless, any win of the pawn on b3 will soon result in Black winning the pawn back, once
the b-file has been opened.
Gelfand was thinking about the tactics, and what happens if White tries to win the b
pawn. The alternative view was one of purely positional thinking: that if the pawn is safe,
Black's position cannot be broken down.
24 .i.c6
24 .i.a4 will eventually enable White to swallow the b3-pawn, but Black seems safe
enough after, for example, 24 ...l:!.fc8 2 5 .i.xb3 l:tab8 26 .i.a4 'i!Vxb2, while 26 �a3 .i.c1 is no
help for White.
24....l:ta2
Consistent. Gelfand's aim is to open the b-file for his major pieces.
25 l:!.xa2 bxa2 26 'i!Va3 l:!.b8 27 'i!Vxa2 Yz-Yz
136
A n a n d- G e lfa n d, World C h a m p io n s h ip, M osco w 2 0 1 2
Question: Draw agreed, but could An and as White have tried a little harder?
lt is often difficult for the spectator to assess whether a draw is legitimate or prema
ture. One reason for this is that the players are vastly more knowledgeable, and if they
agree a draw, this generally means that they have analysed a few moves deep, seen what is
likely to happen if play continues, and quite simply offered and accepted the draw.
Sometimes there can be premature draw agreements, true, but what about here?
Anand could have carried on, but would it have led to much ? After 27 ...'i!Vxb2 28 'i\Yxb2
.l:!.xb2 29 .l:i.al g6 30 .l:!.a7, White has pressure on the f7-pawn.
lt looks difficult until one recognizes that, at long last, Black has time to get his own
bishop active: 30 ... �d2 ! comfortably holds the balance.
If White tries the natural 31 �e8?!, Black takes the initiative with 31 ... ..iel, as 32 .l:!.xf7?
13 7
.l:lb8 3 3 11e7 �f8 even win s a piece. But 3 2 ..ixf7+ -.t>f8 3 3 .Ud7 ..ixf2+ 34 'it>g2 ..ics+ 3 5 �h3
.'l:tf2 36 ..ie6 h S ends up in a draw.
Another try for White would have been 30 .l::!.a 8+ �g7 (30 ... ..if8 is slightly passive) 3 1
l:ta7, a s i n the previous diagram position, but with the king being o n g 7 , rather than o n g8.
However, 3 1.. ...id2 still holds, since 32 ..ie8 �f8 33 ..ixf7 ..ie1 just transposes to the earlier
line.
The draw is therefore justifiable, even if the spectators would have learned a little more
had it been played out.
Game Six: Gelfand-Anand

You've guessed it, another draw, another endgame, again not reaching the second time
control. What is disappointing for the viewers is not so much the succession of draws, but
rather that neither player at any stage seemed prepared to grind out for a win. There ap
pear to have been several minor inaccuracies by both sides in previous rounds, but neither
player seemed prepared to take on the initiative. We would have preferred to see long
endgame struggles, to move 60 and beyond, but the players were h appy enough to level
out, just wanting to reach the next round with as little stress as possible.
Anything new in game six?
18th May 2012

B.Gelfa nd-V.Ana n d
World Championship, Moscow 2012 (Game 6}
Semi-Slav with s a6 ...
1 d 4 ds 2 c 4 c 6 3 tt:::lc 3 'Llf6 4 e3 e 6 5 'Llf3 a6 6 'iVc2
Something new, at least for this match. lt is not so clear that Black is genuinely fully
138
equal after 6 b3, as i n games two and four, but Gelfand was unable to achieve more than a
draw over the board. Time to check out new ideas.
6 cs
...
Question: Why this second move with the c-pawn?

Isn't it simply just a waste of a tempo?
Such a move must be made carefully, for a less experienced player. Unless the timing is
correct, it can easily be a bad waste of time. Here White h as weakened his d4-pawn slightly
by moving his queen off the d-file, so Black wants to counter-attack with ... c6-cS and ... tt::lc 6,
as soon as possible. There may also be occasional hits on the queen with ...tt::lb4.
Then there is the battle with the pawn centre. If White were now to play 7 .i.d3 or 7
�e2, Black instantly regains the tempo by 7 ... dxc4 8 �xc4 with quick equality. Or he can
try, more ambitiously, 7 ... tt::l c 6. White has not gained anything.
If White were to exchange with 7 dxcs instead, Black again regains his tempo with
7 ... .i.xcs, since he has not spent time on playing ... Ji.e7 or ... Ji.d6 before the recapture.
7 Ji.d2 is too much of a quiet developing move to be dangerous. If we remember the
second and fourth games, Black positively encouraged .i.d2 by playing ... JJ.b4, even though
the black bishop would soon retreat.
This second push with the c-pawn is the most direct way for Black to equalize. After
6 ...tt::lb d7 or 6 ... Ji.d6 he would have to work harder.
7 cxds
So White makes his other central pawn exchange.
7 exds
...
Question: Why does Black allow himself to get stuck with an isolated d-pawn?
Surely he would want to keep his pawn structure fluid, with 7 ...tbxds or 7 ... cxd4 - ?
139
The question is one of tempo. Black has already used up an extra pawn move, with ... C7-
c6 and then ... c6-cS, and while ...a7-a6 is often a good and useful move, it does not help
develop his pieces. At the moment, White has three pieces in play, Black only one. lt would
be unwise to open up the centre too quickly, and therefore he wants to keep a pawn on dS.
a) 7 ....!t:Jxds 8 tt:Jxds 'ii'x ds leaves White several possibilities to keep an edge. One idea is
9 b4 c4 10 i.xc4 i.xb4+ 11 i.d2 i.xd2+ 12 'iYxd2, and Black will take several moves to finish
his development. Not 12 .. .'tlixc4? 13 .l:!.c1, skewering the queen.
b) 7 ... cxd4 8 'iVa4+ i.d7 9 'iYxd4 is also troublesome.
When playing as Black, it is sometimes necessary to make slight concessions with the
pawn structure, to give breathing space for the pieces.
8 i.e2
lt is too early to initiate the exchange on cs. After all, Black can do nothing constructive
with his c-pawn. 8 ... c4? 9 e4 swallows another tempo, and Black's pawns become weaker,
rather than stronger.
8 ... i.e6
The sort of breathing space that is needed after Black has played ... e6xds.
9 0-0 tt:Jc6 10 .l:!.d1 cxd4 11 tt:Jxd4 tt:Jxd4 12 l:txd4
Question: Any particular reason why 12 ... .\tcs or

12 ... i.d6 would be preferable to the other?
12 i.cs
...
Anand would have needed to calculate carefully over the next few moves, and indeed
he was content to sacrifice the d-pawn later on, just to ensure that all his pieces became
active, and more so than his opponent's.
A completely positional defensive response would be something like 12 ...i.d6 13 i.f3
0-0 14 h 3 'iYd7, followed by bringing the rooks into the centre. The only problem is that if
140
Black decides that, at some stage, h e needs to push the rook away from d4, with ...JL.cs, or
... ..tes, he would probably have lost a tempo. The counter-argument might be that the rook
will have to move away anyway for reasons of safety. For instance, if White were eventually
to play ..td2 while it is still on d4, the rook would be short of escape squares; while if White
were to try, probably unwisely, b2-b3, there would be uncomfortable attacks and pins with
... JL.es, and along the c-file.
My own personal preference would be 12 .....td6, but I am not suggesting that Anand's
move here was wrong. In most interesting and tense positions, there will be more than one
reasonable choice.
Question: And the question I did not ask before.

What are the tactics like after 13 l2'lxds � ?
13 .l:td1
Gelfand ignores the invitation. The tactics end up only level, and Gelfand still hopes for
a win by positional methods.
The sharp line is 13 l2'lxds ..txd4 14 l2'lc7+ 'it>d7 1S lbxa8 (15 t2Jxe6 fxe6 16 exd4 gives
White reasonable compensation for the exchange, but no m ore than that) 1S ... ..ta7.
141
The knight is trapped, and there is no clear way of bringing it back into play. This is
dangerous enough to reject the whole line. In fact, and surprisingly, White can force a
repetition with 16 �a4+ �c8 17 �f4! �d7 (or 17 ... �b8 18 'iib 4! �d6 19 1i'h6 �b8 20
't!Vxd8+ l:txd8 2 1 ltJa4 and the knight escapes with equality) 18 't!Va4+, a good line, but not
what Gelfand would have wanted.
Instead, he develops, waiting to see what Anand does next. There is indeed a choice
coming up.
Question: A queen move now, but which is better, 13 ...'t!Ve7 or 13 ... 'ii'd6 ?
-
13 JWe7
..
An extremely difficult position to try to evaluate. Anand is a vastly experienced grand

m aster, and World Champion, with great tenacity in defence. He decides that this move,
142
protecting the cs-bishop but abandoning control of the dS-pawn, i s sufficient to hold the
game. He loses a pawn, but his pieces are so much better developed, he would argue that
White has no realistic chances of playing for a win . We will try to assess this point of view
later.
Here 13 ... 'i¥d6 looks slightly uncomfortable, so maybe instincts should prevail. After 14
'i¥a4+ i.d7 15 'i¥h3 i.c6 16 i.f3 lld8 14 �d2, White is starting to create the standard pres
sure against the isolated d-pawn.
Question: White is about to win the d-pawn, but where does he want his queen?
On c2 or b 3 ? The choice is either 14 'i¥a4+ .\td7 15 'i¥h3 .ltc6 16 .ltf3 o-o, or, as in
the game, 14 .\tf3 0-0, in either case followed by captures on dS.
14 .if3
Keeping a minimal edge for White.
At first, it might appear that the suggested alternative, 14 'i¥a4+ �d7 15 'i¥h3 i.c6 16
.if3 o-o 17 tt::lx ds it:Jxds 18 i.xds, is a considerable improvement for White, as Black no
longer has the same pressure on c2 {after .. J::ta c8} as in the game.
However, Black can improve with 14...'it>f8 ! ?. The king is slightly awkwardly placed, but
not weak, and Black can perhaps get his kingside moving with ...h 7-h 5. Meanwhile, much
of the pressure on ds has been lost. After 15 i.f3, White has a slight edge, but probably no
more (see the comments in the game after 14 i.f3 .l::i.d8 15 'i¥a4+ 'it>f8}.
A more ambitious line, perhaps, would be 15 .id2! ?, completing his development, and
ignoring the isolated d-pawn for the time being. Black's main problem is how to complete
his own development, while avoiding the bishop being in the way of the rook pair. If
1S ... g6, there is an unexpected plan of attack with 16 'Llb1 ! ? (a computer suggestion)
16 ...tt::le4 (making it more difficult for White to control the long diagonal with .ic3) 17 �e1
'it>g7 18 i.f3 bS 19 'i¥c2 b4 (again keeping White off the long diagonal} 20 a3 as 21 .ixe4
143
dxe4 22 'ii'xe4, and White has finally won the pawn, even if his queenside pieces have been
temporarily undeveloped.
14 0-0
...
This was the last chance to abandon the pawn sacrifice. 14 ... l:td8 15 'ii'a4+ 'it>f8, as in the
previous note, could now be met by 16 b3, and if 16 ... .lii.b4, then 17 tt:lb1! with an edge. The
tactical point is that 17 ... ..id7 is answered by 18 ifxb4! �xb4 19 i.a3, with favourable sim
plification for White.
15 ..ixds ..ixds 16 tt:lxds tt:Jxds 17 l:!.xds l:!.ac8
Question: What should White do next? Can he keep the extra pawn successfully?
18 �d2
White has a microscopic edge. This move is only good if White is unable to keep an edge
144
by trying to hol d the pawn. Gelfand could not see anything for him by passively defending.
That being the case, this was a good attempt to keep a slight edge. Anand had played the
defence impressively. Few would have had the confidence to defend such a position a
pawn down.
The main problem for White is in bringing the bishop into play, and then the rook. Gel
fand achieves this immediately, at the cost of giving back his extra pawn.
Instead, 18 �e2 is natural, but after 18 ...�e4, Black's pieces are strong. For instance, 19
�d3 �xd3 20 .l:!.xd3 .l:!.fd8 21 �xd8+ .l:!.xd8 22 �fl .l:!.d1+ and it as clear that White is not
winning. He has still not developed his rook and bishop.
White wants to keep the e4-square covered, so possibly 18 �fs g6 19 �f3 .l:i.fd8 20
.l:i.xd8+ .l:i.xd8 21 g3, which is still difficult for Black to play. The best response seems to be
21 ....i.d6 22 b3 lies 23 �b1 �d7 24 �e4 and now 24 ...�c6! 25 �xc6 �d1+ 26 �g2 bxc6,
when a familiar positional draw results.
Naturally, it is easy enough to generate other lines, but the likelihood is that they will all
end up in a draw, with accurate play.
18 ... .i.xe3 19 .i.c3
19 �d3 is a simple draw offer. White still wants to try for more.
19 ... .i.b6 20 �fs
And indeed it is easy to see why Gelfand wants to carry on. He has chances, maybe small
but not negligible, to play for a win, whereas Black has no chance of more than a draw.
Question: What's the best way for Black to steady the boat?
20...�e6?!
A slightly strange move, pushing White's queen to a safer square. A nand seems to be
concentrating a little too much on defence, and less on finding the best squares for the
pieces.
145
Here 20 .. Jks is equal. If rooks start to be exchanged, Black's defence becomes much
easier. Indeed, 2 1 l:f.xcs 'i!Vxcs would be a good time to agree a draw; while 21 'i!Vg4 fS 2 2
'i!Vd4 l:f.xds 2 3 'ii'x dS+ 'i!Vf7 is also safe for Black. After 24 l:i.d1 'i!Vxds 2 S l:i.xds l:i.d8, the bishop
ending would be drawn.
21 'i!Vf3 f6
Question: White has just been handed a slight positional edge.

What is the best way of keeping up the pressure?
22 h4
A strange move, a defensive move, pretending to be the start of an attack. What is im
portant is not a pawn push to hs or even h6, but rather trying to put pieces on dominating
squares. To keep any sort of edge, White needs to try to control the centre, not make a ran
dom pawn push on the king side. The centre is open and he has good pieces there.
Having said that, it is not so obvious that White is going to be better anyway. Possibly
22 .l:!.el 'i!Vf7 22 'i!Vfs, to try to irritate Black again. He should not create extra weaknesses for
his pawns.
22 ...'i!Vc6
22 ....l:!.fe8 looks simple and equal.
23 hs
Perhaps the problem is the second pawn m ove, rather than the first. 2 3 .l:i.adl at least
avoids creating a pawn weakness on h S , but play should still end up equal after 23 ...I:i.cd8.
23 ...l:!.fd8 24 l':!.xd8+ l:!.xd8 25 'i!Vxc6 bxc6
146
A n a n d- G e/fa n d, World C h a m p i on s h ip, M osco w 2 0 1 2
If White still had a pawn on h3, rather than on hS, h e might probe for Black's queen side
pawn weaknesses. As it is, White has to be careful to protect his own king side pawns.
26 .l:te1 c;i;f7 27 g4 SL.d4 28 .l:tc1 SL.xc3 29 .l:txc3 .l:td4 Yz-Yz
After 30 .l:!.xc6 .l:!.xg4+ 3 1 c;i;fl as 3 2 .l:!.cs a4 3 3 .l:!.bs, White can cover his pawns. So draw
agreed.
This was the most interesting of the "quiet draws" so far. Anand's pawn sacrifice in the
late opening seemed to hold, both in play and in analysis, though the position was tense.
Readers will notice a pattern in the first six draws. Play was sharp early on, in games
one and three, when both players were happy to take the occasional calculated risk; but
later on, they both wanted to play quiet and technical chess.
Of the "quiet draws", several were perhaps not quite as accurate as they looked. Games
two and four should have kept a slight edge for White (Gelfand}, his pieces being slightly
but significantly better developed; whereas in game five, Anand (as White) could have cre
ated more pressure on one of the pawns. Small edges, and not necessarily enough to make
a significant advantage, but even finding just one chance of a genuine advantage would
alter the balance of the match.
People were starting to speculate as to what would happen if all twelve games were
drawn, and the match end up in a rapidplay finish, and who was going to be the more suc
cessful in that aspect of chess. Going through the statistics, Anand had a reasonably sig
nificant edge at the faster time limits, but of course one slip could easily change the bal
ance. And so, in the end, it happened. However, there were some massive surprises before
the players could think about the rapidplay.
Game Seven: Gelfand-Anand

After six draws, Gelfand, the supposed underdog, takes the first win, and convincingly so.
There were no sharp tactics, and no delicate endgame squeeze, but instead a middlegame
14 7
positional grind in which Anand, for once, did not handle his knights effectively, while Gel
fand's knights, with the help of queen and rook dominating the c-file, broke through with
great force.
lt is technically relatively easy to write up a game in which one player won and the
other lost. You need to find the losing move, the move which put the other player on the
slide. Therefore, after six draws, it is almost a relief not having to try to analyse the many
ups and downs of a drawn game, in the full understanding that most such possible inaccu
racies did not make the difference to the result of the game. There is often quite a large
margin of safety in a drawn game, whereas in a decisive one there is a far more distinct
boundary between being safe and either losing or being on the way down.
Organizational point: Yes, Gelfand is playing White two games in a row. This is easily
explained. The routine in the match was game-game-rest-game-game-rest, and so on.
Halfway though the match, the players flipped, so that now Gelfand has the first White in
the next sequence of two game pairs.
20th May 2012

B.Gelfa nd-V.An a nd
Semi-Slav with 5 a6 ...
1 d4 dS 2 c4 c6 3 t2Jc3 t2Jf6 4 e3 e6 5 lL'lf3 a6 6 cs

At least it is asymmetrical, perhaps no bad thing psychologically if Anand, as Black, is
fully geared towards highly symmetrical pawn structures, as part of his defensive plans.
Here the asymmetry lies only on the a- b- and c-files. The rest, from files d to h, is symmet
rical.
6 ...t2Jbd7 7 �c2
148
7 ... b6
Question: Is this a good move?
The computer suggests that the position is equal after the 7 ... b6 pawn exchange, and
yet it is difficult to be convinced that Black is fully equal. His pieces are very slightly mis
placed, and in the play which follows, Black is unhappy having the knight on b6, rather
than on c6 (after he has played ... c6-c5). Black remained under pressure throughout the
game.
There were, of course, good alternatives for Black, mainly based on the idea of ... e6-e5
and then setting up a wide broad pawn front, as opposed to the narrow front on the a- to
c-files.
For instance, 7 ... e5 is a far more direct response to White's flank pawn push. Black does
not even have to bother preparing with 7 .. ."�c7, as 7 ... e5 8 dxes tt'lg4 is lively, but probably
about equal.
In the context of the match, An and quite simply did not want to play anything lively as
Black. He just wanted to cut down anything unpredictable. This, it seems, was his basic
error in thinking.
8 cxb6 tt'lxb6
After 8 .. ."�xb6 9 tt'la4, followed by ..id2, White is probing the black queenside.
9 ..id2 cs 10 �cl cxd4 11 exd4
White is ahead in development, and it is still unclear as to what Black should do with
his b6-knight. lt is already an opening success for Gelfand. Note that there is nothing
wrong with White's isolated d-pawn : it helps create knight outposts on c5 or es.
u .....id6 12 .td3 o-o 13 ..igs h6 14 ..ih4 .tb7 15 o-o
149
1S .'i¥b8
..
An and, too, knows about any unusual thought processes by his opponent. His f6-knight
is pinned and, if given the chance by White, most players would feel confident about dou
bling the f-pawns, unless there is a definite improvement in playing differently. We have
already seen, in game four, how Gelfand rejected doubling up Anand's f-pawns on move
24, and how An and was able to equalize comfortably, rather than being slightly inferior.
Here Anand could have tried 1S ... tt:Jbd7, and only later breaking the pin with a queen
move. lt is possible that White has nothing better to do than to play 16 .ig3 anyway, in
which case Anand's queen shuffle would have been unnecessary.
Instead, Anand allowed his opponent to create the doubled pawns.
16 Ji.g3
Question: How much would White achieve by doubling the black pawns here?
150
The quiet bishop retreat was greeted with some surprise by the commentators. Gelfand
has certainly found a consistent plan: forcing the exchange of the dark-squared bishops so
as to take full advantage of the es- and cs-squares for his pieces. This keeps a slight edge
for a while, but with careful play Black can reach close to equality. If Black can defend his
sore squares, it is not so clear what White can do after that.
Structural pawn weaknesses, such as broken pawns, can create long-term problems,
right up to the endgame even if the remaining pieces are relatively active. With pieces, the
more active player always has to be aware that if the defender can gradually activate his
forces, and if his pawns are as good as his opponents', then he should gradually equalize
the game. There are plenty of examples in this book, notably in some of the earlier games
by Black by Anand.
This leaves the question of how seriously damaged the pawns are after 16 �xf6 gxf6 17
lt:le2. If you have not tried it out on the computer, then do so. Just go through a few sample
moves, suggested by the computer, and see what h appens a few moves down the line.
Ready?
The problem you might well find for Black is not so much the weakness of the doubled
f-pawns, but rather the isolated pawns, on a6 and h6, on both sides of the board. For ex
ample, after 17 ....l:!.c8 18 �d2, the white queen is excellently placed, with ideas of �as, as
well as keeping the h6-pawn under threat. Black m ay try to simplify with 18 ... .l:!.xc1 19 .l:!.xc1
�f8 20 lt:lg 3, but then he has pawn weaknesses on both sides of the board, and his pieces
are equally stretched to cover both sides. If Black were to try to keep his pieces active with
20...I!.c8 21 .l:!.xc8 �xc8 22 �as, the pawns are in danger of collapsing.
lt is difficult to see any obvious way for Black to achieve equality. White just seems sub
stantially better. The line that Gelfand actually played does not seem so impressive, and it
would seem that Anand needed to make further slight errors before he was worse, then
losing.
16 .l:!.c8 17 �e2 �xg3 18 hxg3
...
151
The main point of Gelfand's plan of attack would appear to be that, by exchanging the
dark-squared bishops, he n ow has the opportunity to take control of the es-square for the
knight, and later, the cs-square for the other knight. The only problem is that, in his at
tempt to dominate on the dark squares, his d4-pawn, a dark-squared pawn, is slightly
weak. Black should be able to hold the position.
18 'ir'd6 19 nc2
...
19 'Llbd7
...
The computer suggests that there are more than a dozen moves that give Black clear
equality. There are no tactics on view. In other words, this is a position a! battle.
Question: Is Anand's move bad, in that he is allowing White to play 'Lla4 - ?

Or alternatively, is he okay and just equal anyway?
At first glance, Anand's move looks a little too passive to be good. "Passivity" implies
quietly trying to defend a slightly worse position - but if his position is level, then he has
defended successfully, rather than defended passively. The critical position is not now but a
couple of moves further on. Black was holding the balance, but he made a mistake later.
A more combative approach might be 19 ... 'Llc4 (starting to put pressure on White's
pawns) 20 .l:f.fcl �4. Again, the computer suggests that there are about a dozen moves
which are equal here. Clearly, no annotator could be expected to go through them all and
summarize conclusions. There is so much more to try to work out in run-of-the-mill posi
tional battles than in wild tactics. In positional chess, the difficult point is sieving through
many ordinary-looking moves, trying to see if just one of these might give a slight edge (or
equality, or good defensive chances, or whatever). Whereas with tactics, there are fewer
moves which need to be considered - a check or a capture, say - but each choice of move,
whether by the player or the defender, will have radically different outcomes.
One possible drawn outcome would be 21 'Lld1 (to put pressure on the c4-knight)
152
A n a n d- G e /fa n d, World C h a m p i o n s h ip, M osco w 2 0 1 2
2 1...Wib6, when there i s an immediate repetition with 22 lLlc3. Black has also positioned his
pieces in such a way that White cannot profitably win a pawn by a capture on c4. If White
were to take there, then Black can open up the long diagonal, with ... dsxc4, or possibly first
... .l::txd4, after which Black is ready, among other things, to take on f3 and win the d-pawn.
And what about the other dozen drawish moves for Black? I leave this to you to try to
see if you can find something. My own positional instincts are that Black cannot possibly
be better, after reasonable play by both sides. There is no sense in attempting to win the
position by force.
The quickest draw, at first illogical but in fact reasonable enough, would be 19 ...lbg4 20
lLlg s lLlf6 21 lbf3, again with s repetition. The knight on g4 is useful, providing a different
way of covering any thoughts of lLles.
Perhaps White could try 20 llfc1 instead, but then 20 ...as, for example, is still about
equal.
20 .l::tfc1
According to the computer, there are now two choices for White which keep him fully
equal: the text move and 20 a3.
20.. J�ab8
Whereas here there are three fully reasonable choices for Black that aim for equality.
These are the text move, 20 .. �C7, and 20... lbg4. Other moves, the computer seems to sug
.
gest, might give White the very slightest of edges.

21 lba4
And now half a dozen reasonable moves by White, with complete equality. The move
that Gelfand chose was, in fact, well out of the top dozen, but once the move gets played,
the computer sees outcomes, and agrees that it is equal.
Question: What should Black play here? The main choice,

according to the computer, is between 21 ... lbe4 and 21 ... .U.xc2.
1 53
Actually, we can add to a third reasonable choice for Black: 2 1 ... as!?. If White responds
with 22 ctJc3, attempting to exploit the weakness on bS, then perhaps 22 ...'i'h4 or 22 ... 'i'b6
and Black is fully in play.
Instead, Anand tried:
21 ... 'Lle4?
This seems too quick a jump from defence (on the queen side) to attack (in the centre),
especially as Black is not really attacking anything. He needs to watch what is happening
with White's queen side pressure.
2 1...l:i.xc2 ! is not a surrender of the c-file, but rather is the start of counterplay along
that file. For example:
a) 22 .l:!.xc2 �c6 2 3 'Llcs? (23 'Llc3 is safer, with a possible offer of repetition with
23 ... �b7 24 ctJa4) 23 ... i..b s!, and now his bishop is no longer "bad", but is able to force an
exchange with the previously good bishop. White will have to watch out for his isolated d
pawn.
b) 22 �xc2 .l:!.c8 2 3 ctJcS looks impressive on the queenside, but White's centre is going
to be badly eroded after 23 ... e s ! . On 24 dxes, Black can even try for an advantage with
24...'i'b6! ?.
With best play, 2 1 ...l:i.xc2 would have been equal. So the game was not the totally one
sided win by Gelfand, as it might at first appear.
22 �xc8+ i..xc8
Black has give up control of the file, as 2 2 ... .l:!.xc8? 23 .l:!.xc8+ i..x c8 24 ..ixe4 dxe4 25 �xe4
wins a pawn.
Question: What should White do next?
23 �c2?!
Both players are becoming extremely nervous mid-match, and it shows in their choice
1 54
A n a n d- G e lfa n d, World Ch a m p i o n s h ip, M os c o w 2 0 1 2
of moves. Gelfand i s aiming for total security on the c-file, neglecting other avenues of at
tack. lt is time to concentrate on the dark-squared diagonals, rather than the files.
a) 2 3 a3?! ti:Jdf6 24 b4 ..td7 is too slow, using up too many pawn moves. After 25 ti:Jc5
..tb5 ! (pinning White's bishop}, Black is more than equal.
b) 2 3 "iWe3 "iWb4 might perhaps give White a slight edge, but he can also cover the b4-
square with his queen ...
c) 2 3 'ife1! is the best way to create pressure. White keeps an edge, with ideas perhaps
of iVa5. After 23 ... ti:Jdf6 26 ti:Je5, White has more pressure on the dark squares; or 23 ... i.b7
24 b4 ..tc6 25 tt:Jc5 ..tb5, and this time the d3-bishop is no longer pinned so White can play
26 tt:Jxe4 dxe4 27 ..txe4.
Whatever happens, Black is under pressure. He lost his positional balance on m ove 21.
Question: What should Black do next?
23 gs?
...
Certainly not this. Again, this an extremely nervous response, attacking with a pawn,
hoping he can achieve something, while ignoring the main threats along the c-file.
Black does not even have all his pieces in play, so 23 .....tb7! has priority and just about
holds the balance. The immediate defensive tactic is that 24 'Wie7?? runs into 24 ....l:!.c8, and
White loses material . Instead, 24 ti:Jc5 .l:!.c8 25 b4 is still tense, but one possibility for Black is
25 ...'¥Wb6 26 a3 ti:Jdf6, when he is starting to show some good defensive coordination.
24 "W/c7
Of course. Now White is significantly better.
24 ."'i\Vxc7 25 .l:!.xc7
.•
Position ally winning. Just take a look on the rook on the seventh, and Black's bad
bishop, which in the end gets l ost. lt took a dozen or so moves to mop up, Gelfand finding
ways to manoeuvre his knights to excellent outpost squares.
155
2 S ...f6
This may well be regarded as another weak move, but in reality the position cannot be
held anyway.
26 �xe4 dxe4 27 l2Jd2 fs 28 l2Jc4 l2Jf6 29 l2Jcs l2Jd s 30 .l:!.a7 l2Jb4 31 l2Jes l2Jc2 3 2 l2Jc6
There is probably not much that needs to be said about the last few moves. Just note
the way in which Gelfand takes over squares, move by move, with his opponent unable to
muster any real opposition.
32 ....l:!.xb2 33 '!J..c 7 '!J..b l+ 34 Wh2 e3 35 .l:!.xc8+ Wh7 36 .l:!.c7+ Wh8 3 7 l2Jes e2 38 l2Jxe6 1-o
This was not quite as smooth as it looked, but Gelfand's win was thoroughly deserved.
He had slightly the better of the play in many of the earlier drawn games, but he was held
back by An and's accuracy in slightly inferior positions - and also, one suspects, through
Gelfand's own lack of confidence in trying to grind out fractionally superior positions.
156
This time, Anand's defences failed. I n all the seven games up to now, An and was better
in only one.
Anand needed to up his game, even when he was still drawing. Sooner or later, Gelfand
would be able to take advantage of the various minor errors. Remember, Anand lost, not
because he made a single mistake, but rather because he went wrong twice. The first time,
Gelfand threw away his advantage by concentrating only on the c-file, rather than taking
note of the whole queenside.
After this first win, I felt at the time that we were seeing a new World Champion, in the
name of Boris Gelfand, and that he would be expected to halve out to five more draws. I
noted that "realistically, it is unlikely that Anand will be able to bounce back. Even so, it
takes only one piece of inspired play, and, it must not be forgotten, a mistake by the oppo
nent to allow things to level. At the time of writing, it is just under an hour before the next
game will start. Has An and composed himself to play for a win?"
Game Eight: Anand-Gelfand

Chess never ceases to surprise. Just as Boris Gelfand was, it seemed, about to become
World Champion, he managed to self-destruct in only 17 moves. lt was not a case of great
brilliance on An and's part. He merely did what needed to be done.
The win itself was relatively easy. The difficult part was the chess-thinking early on, the
way in which An and, presumably with the help of seconds before the match, was able to
find ways to induce his opponent to make mistakes.
Anand's opening was certainly very well thought out, and it would have been difficult
for Gelfand to handle Black's position, even in normal circumstances. The likelihood was
that he was so focused on concentrating on five more solid games, draw-draw-draw-draw
draw, that he found it difficult to handle a sharp position as Black, in full knowledge that
his opponent would have prepared the opening thoroughly. He did not want to bear the
tension of being Black in a long and difficult game, where he was unable to equalize
quickly, and so he improvised, disastrously.
There were some apparent similarities to the earlier match between Kramnik and Aro
nian - to the extent that Kramnik lost a game badly, but two games later, Aronian's play
fell apart, in what was basically an unsound queen sacrifice. Look closely, though, and Aro
nian and Gelfand collapsed in opposite directions. The match between Aronian and Kram
nik was a friendly match, without the extreme tension of a World Championship, and Aro
nian was prepared to show to his opponent, and to the world, that he could play chess styl
ishly, a win followed by another brilliant win. His hopes crashed. Kramnik did not have to
do anything particularly remarkable out of the opening, but defend coolly and accurately
after Aronian's queen sacrifice.
In contrast, it was Anand's original and strong handling of the opening which plunged
Gelfand into a tail-spin. Gelfand was not really aiming for creative ideas of his own. He was
merely improvising, and without a script it did not work for him.
157
21st May 2012

V.Anand-B.Gelfa n d
Benoni Defence
1 d4 l2Jf6 2 C4 g6
Would it have been a good or a bad idea for Gelfand to revert to the Slav and Semi -Slav
set-ups? He played such lines a year ago, but suffered a couple of reverses against Carlsen,
which Anand would have been aware of, and studied.
3 f3 cs
I suspect that this was only the third line of defence in Gelfand's opening preparation,
behind the Semi-Grunfeld (3 ... ds), and the King's Indian (3 ...Ji.g7 or 3 ... d6). Risky, given that
he only needed three more draws as Black to win the title.
4 dS d6 5 e4 i.g7 6 ttJe2 o-o
Question: Can you find a new way for White to make life difficult for Black?
Compare and contrast with the Samisch King's Indian : 1 d4 l2Jf6 2 c4 g6
3 l2Jc3 .ltg7 4 e4 d6 5 f3, and in particular the line s . .o-o 6 l2Jge2 cs 7 ds.
.
7 l2Jec3!
A full exclamation mark, since it was this move which ultimately won the game.
White's main problem after an early f2-f3, in analogous lines of the Samisch or the Benoni,
is working out what to do with his knight on g1, for which there is no easy route to a good
square. Sometimes in the Benoni, White is able to develop with l2Jh 3 and l2Jf2, once Black
has already played ... l2Jbd7. If, however, White has already moved the knight to e2, and the
other knight is on c3, he has to spend time manoeuvring behind the lines with, for exam-
1 58
ple, ..ie3, followed by tt:lcl, and perhaps tt:ld3, to enable the minor pieces to begin working
together.
A nand shows a simpler and more effective way to develop his knights, playing tt:lge2
and then tt:lec3, after which the other knight can decide where to go, probably to d2, per
haps occasionally to a3. lt all seems m arvellously effective.
There are many lines against the Benoni in which White develops his knights with tt:lc3,
tt:lf3 and tt:lfd2. Anand shows a way in which he achieve do this, and with the addition of an
early f2-f3 pawn push, to secure his pawns on the light squares.
The simple answer is, "I don't know" - and of course this is a perfectly reasonable an
swer, for the writer as well as for the reader.
Picture, though, how difficult the position for Gelfand would be at the board, after only
seven moves. The natural continuations here would perhaps be 7 ... e6 or 7 . tt:la6, but he
. .
must surely have sensed that Black is reaching a slightly inferior position in one of the
m ain lines. He would have been happier to go through all this while preparing at home,
rather than having to sort it all out over the board. In addition, he would have found it dif
ficult, having been geared up to play ultra-solid chess, to find a sharp innovation.
Should he accept a slight disadvantage? Or should he improvise? He improvised - un
successfully.
Should he have avoided any lines involving King's Indian or Benoni set-ups? Perhaps
that needs to be left as an open question.
7 tt:lhs?!
...
lt just cannot be good, but perhaps it is not quite as bad as it looks. "Knights on the
edges get stuck in the hedges" and all that.
1 59
Question: What should White do next? A kingside

pawn rush, or develop his pieces?
This position is already extremely tense, with both players by now being out of the
book.
My own personal reaction in such a position would be to charge the pawns forward,
gaining tempi after 8 g4 tt:Jf6 9 h4 h s 10 .l:.g l, with wild and complicated play to follow.
White still has the likelihood of a slight edge. If anything starts to go wrong, however, Black
would presumably find ways to create dangerous counterplay. One technical problem for
White is that, because of his earlier knight manoeuvring, he cannot create a quick attack
on Black's king side pawns with etJf4 or tbg3 . In the end, this is probably how both Gelfand
and An and would have assessed the position. There is not much point in trying to gain a
couple of tempi with a pawn rush when, a few moves later, he will probably have to use up
a couple of tempi with the knight to cover his own weaknesses in front of the king.
Anand is in a difficult match position in that, despite being a point down, he cannot go
all-out for an attack, since a second loss would have ended his chances of remaining World
Champion. He therefore plays it cautiously, aiming for quiet development. Who knows - if
his opponent makes one slightly strange move to break the positional balance, he might
well play another strange move, this time overpressing.
s .i.gs
8 ..ie3 is also good, when 8 .fs transposes to 8 .i.gs f6 9 .i.e3 fs below.
..
160
A n a n d- G e lfa n d, World Ch a m p i o n s h ip, M osco w 2 0 1 2
s .tf6?
...
Sometimes a top grandmaster can, through over-sophistication, look like a lower divi
sion club player. lt is not so clear that Black has done anything seriously wrong so far, but
this move tips the balance too far.
Instead, 8 .. .f6 is a natural way of starting counterplay. For example, 9 .ie3 fs 10 exfs
.ixfs 11 g4 .ixb1 12 "ifxb1 tt:lf4 13 'ii'e4 .tes is about equal. 9 .lth4 it slightly more difficult
for Black to achieve ...f7-f5 (there is pressure on the e?-pawn), but 9 . tt:ld7 looks compli
. .
cated and, again, about equal.

9 .ixf6
Gelfand was tacitly making a draw offer with 9 .ih6 .ltg7 10 ii.gs. Anand would never
had accepted it.
9 exf6
...
This is part of Gelfand's "cunning plan". He wants to exchange his front f-pawn for one
of the white pawns, while keeping the second f-pawn in place. Unfortunately, he is seri
ously behind in development. For example, Black would need at least four extra moves to
bring the rook from a8 to e8, and even then it is not clear where the bishop, and even more
so the knight on b8, will find good squares. White, too, has only one knight developed, but
his other pieces are more balanced, and can get into play quickly as required.
White is better - but it is hard to imagine that the game is already more than half way
through.
161
10 'iVd2
Covering the c1-h6 diagonal is a good consistent plan. He also makes an escape square
(d1) for the king, in case Black plays a later ... �4+. Anand was thinking quite a few moves
ahead.
10 g4 l2Jg7 11 h4 l2Jd7 is no doubt also playable, with an edge for White, but Anand
wants to keep his pieces, rather than his pawns, moving.
1o fs 11 exfs
...
Gelfand now had a long think, recognizing that queen, rook and knight are not enough
to attack the enemy king, if White has active defensive pieces. Black's other pieces are well
out of play.
The m ost obvious line is 11...�4+ 12 'it>d1 t2Jg3 13 'iVf2 lLlxfs 14 'iVxh4 l2Jxh4.
162
Pawns are level, neither king i s under threat, and both sets of pieces are equally devel
oped, in terms of getting off the back rank. There is, however, a horse kick at the end with
15 tt:Jbs, winning a valuable central pawn.
Instead, 11 ... .l:!.e8+ 12 �d1 ..ixfs 13 g4 transposes into the main line, which we shall
consider later. lt is just about playable, in the sense that it is not immediately losing, but
still unattractive.
Probably the best attempt to hold the position together is 11 .. ."�e7+ 12 �d1 ..ixfs. Then
White could try to avoid simplification with 13 tt::la 3 !?, or he could continue the thematic
line with 13 g4 ..ixb1 14 l:i.xb1 tt::lg 7.
11.....ixfs 12 g4 .l:!.e8+ 13 �d1 ..ixb1 14 .l:!.xb1
Question: Does Black have to retreat his knight with highly

uncomfortable play, or he can consider attacking with 14...'l!Vf6 - ?
What happens next? Clear analysis is required.
1 63
14 .. .'iVf6?
He loses in three moves!
Black could, of course, retreat his knight, but then his position, though not losing im
mediately, is very uncomfortable. lt seems that on a bad day, overexcited by his previous
success, Gelfand quite simply missed that Anand's 17th move was going to trap his queen.
15 gxhs!
15 'it>c2 l2Jf4 16 l2Je4 leaves White better, but it is not so clear-cut. Black can, for instance,
offer the exchange with 16 .. Jhe4.
1S .'�'xf3+ 16 'it>c2 "i!Vxh1 17 "i!Vf2! 1-0
..
The queen has been well and truly tagged. Gelfand presumably decided that it was
hardly worth bothering trying to play on, with 17 ... l2Jc6 18 dxc6 (but not 18 .i.h 3?? l2Jd4+
and the queen escapes to f3) 18 ... "i\Vxc6, as White can carry on with the presumed winning
attack with 19 .i.g 2 "i!Vd7 20 l2Jd5 "i!Va4+ 21 b3 "i!Vxa2+ 22 .l::!.b 2 "i!Va1. The queen is however on
a different corner, and this time it can be fully reactivated, if given the time, with ..."i!Ve1.
White must still be winning, but it would still need care. Perhaps the most convincing line
would be 2 3 lLlf6+ 'it>g7 24 h6+ 'it>xh6 25 .i.d5 !, aiming for a big check with the queen on f7
if Black's king is forced to retreat. The m ost direct reply would be 2 5 ..."i!Ve1, but then 26
l2Jg8+ 'it>h 5 27 .i.f3+ 'it>g5 28 "i!Vg2+ 'it>f4 29 �b1. White's checks have given himself the op
portunity to bring the rook into play, and in view of mating threats with .l:tf1, he would be
winning.
So yes, Gelfand could have played on, but it was well within An and's ability to find the
winning lines. lt would presumably have been too much to endure by Gelfand, with both
the queen being stuck in the corners, and the king being hunted in the open. He resigned
quickly.
1 64
Game Nine: Gelfand-Anand

After six draws, the two wins - one for either side - might be expected to liven things up.
Both players would be fully aware that they are each fallible. They have each lost a game,
so they would be concerned not to lose another one. Equally, their opponent would each
have dented armour, so it is possible for either to play for a win. Add to this the fact that
there were now only four more games of long-play for one of them to find a win some
where, and play started to get more interesting. Sooner or later, one of the players will
want to avoid halving out every game, and the onus would seem to be on Gelfand to try to
gain a result, before the rapidplay finishes. At long-play, the two combatants are very
evenly matched, but with quick time limits Anand has achieved a big plus score.
lt was n ot to be expected that Gelfand's play would collapse after his disastrous l oss. If
anything, I felt at the start of the day's game that Gelfand was still the favourite, if only
slightly. He seemed, overall, to have marginally the better of the six draws.
Of course, with only four games left, any result, even a draw, will affect the odds each
time. And in any critical game, individual moves can affect the whole result of the match.
An and was greatly under pressure in game nine, eventually scraping a draw. All credit to
him for seeing, well in advance, that this endgame was tenable. Gelfand tried hard to set
up the endgame, with reasonable chances of success, but he could not find a way to grind
down Anand's defences. An excellent defensive result by Anand, it seems.
Perhaps the most critical position of the whole match was when Gelfand had to consider
what to play at move 19. The problem was whether to go for a tactical solution, winning a
queen against various other pieces, or whether he should have played it positionally. Win
ning the queen looked good at the time, but Gelfand would have had a clear edge with the
positional solution. This does not necessarily mean that Anand was in any sense losing. lt is
more a case that he would have to defend rigorously against everything Gelfand could throw
against him. In the game itself, Anand needed only one basic defensive idea - a fortress posi
tion - and he defended it excellently and drew. Would he have defended it so confidently
against other options, with no simplification of the position?
Then, when we go back to an earlier stage, wasn't Anand playing slightly too extrava
gantly, giving away two bishops for two knights? Couldn't he have equalized instead, or at
least reached a manageable inferiority?
This was the most interesting game so far - not because it was in any sense perfect
chess, but rather because it was a tense and interesting battle, with several critical turning
points to discuss. lt was also the first serious endgame battle.
23rd May 2012

B.Gelfa nd-V.Anand
World Championship, Moscow 2 0 1 2 (Game 9)
Nimzo-lndian Defence
1 d4 tLlf6 2 c4 e6
165
Sooner or later, it was inevitable that Anand would drop the Semi-Slav with ... a7-a6. lt
seemed okay to try it out in games two and four, with comfortable draws in the end, but
even in these Gelfand was slightly better in the opening. The longer the match continued,
the more likely it was that, were Anand to continue to play ultra-symmetrically, Gelfand
would find some sort of weakness. So it proved in game seven.
The problem was not so much the opening itself, but rather the mentality for Black that
"my position is ultra-solid and symmetrical, therefore I cannot be worse". Tempo does mat
ter.
3 tt:Jc3 .i.b4 4 e3 o-o 5 .i.d3 dS 6 lZJf3 cs 7 o-o dxc4 8 .i.xc4 cxd4 9 exd4 b6 10 .i.gs .i.b7 11
'¥ie2
Nothing new. This was a favourite of Karpov's as Black. White has more space in the
centre, but still has the isolated d-pawn.
11...tt:Jbd7
A recent top grandmaster game went 11 ....i.xc3 12 bxc3 tt:Jbd7 13 .i.d3 '¥ic7 14 c4 h6 15
.i.d2 tt:Jg4 16 .i.e4 .i.xe4 17 'ifxe4 tt:Jgf6 18 'ife2 'ifc6 19 a4 llac8 20 llfc1 'ife4 2 1 .i.e3 with a
slight edge to White, later a draw, in H.Nakamura-A.Giri. Dortmund 2011. We will discuss
this below, after various transpositional possibilities.
12 llac1 llc8 13 .i.d3
166
A n a n d- Ge lfa n d, World C h a m p io n s h ip, M osco w 2 0 1 2
Question: How should Black handle the position?
13 ....ixc3
If you are prepared to play the Nimzo as Black, you need to be ready to give up the
bishop pair, as here.
The pin on the h4-d8 diagonal for Black is uncomfortable, but if he were to try to defend
with 13 ... .ie7, he would be a tempo down when compared with the Queen's Gambit De
clined, where Black breaks the pin in one go.
Black could insert 13 ...h6 14 .ih4, but then if 14....ixc3 15 bxc3 "Wic?, White has a better
square for the bishop with .ig3. lt is better to allow it to stay on g S .
l t is difficult t o find any other genuinely constructive m oves here. A more positive point
of view is that, by exchanging off White's knight on c3, Black has less serious opposition on
the long diagonal, and so his bishop remains strong. lt would, for example, now be more
difficult for White to aim for d4-dS.
14 bxc3 "W/c7
And now there is a genuinely constructive reason why Black wants to force the white
pawn to the c-file. The queen is safe here, and can keep some pressure on White's central
pawns.
15 C4
167
15 i.xf3 ?!
...
A strange move - and indeed Anand admitted thi s as so at his press conference. Allow
ing two knights against two bishops usually gives an advantage to the player with the
bishops, all other things being equal. The only genuine disadvantages is where are blocked
pawn structures (not here), or if the player with the knights is ahead in development (not
here), or if the player with the knights has the opportunity of exchanging one of the bish
ops, leading to single bishop versus knight. To justify the bishop exchange, Anand needs to
make a serious prod to the d4-pawn by ... e6-eS, with ideas, once the pawn structure has
been clarified, of playing ...ttJcs. Anand only half-achieves this at best, and Gelfand remains
with the advantage.
So what improvements could have been made?
1S .. JUe8 ! ? seems a far more natural way of achieving the same idea of an early ... e6-eS,
perhaps after an exchange on f3. If 16 i.xf6 (intending 16 ...tt:Jxf6 17 tt:Jes), Black exchanges
the bishop with 16 ...i.xf3 ! 17 'ifxf3 tt:Jxf6 and is comfortable. One bishop versus one knight
tends to be less threatening than two bishops versus two knights.
There are other possibilities, of course, with tactics following logical play on both sides
after 16 .l:tfd1 e s ! ? (he could also try 16 ... h6) 17 i..xf6 exd4 18 .i.xh7+ 'it>h8.
1 68
The tactics soon fade to approximate equality after 19 1i.e7 tbf6 2 0 l:i.e1 1i.xf3 2 1 �xf3
l:txe7 22 .l:f.xe7 'ifxe7 2 3 i.d3 .l:f.e8.
Also, since we have referred to the Nakamura-Giri game, 1S ...h6 16 1i.d2 (16 1i.h4 �f4
favours Black slightly) 16 ...tbg4 17 1i.e4 i.xe4 18 'ifxe4 tbgf6 gives White only a modest
edge. Note that Nakamura-Giri omitted l:tac1 and ...l:tac8. Would these extra rook m oves
make any difference? This is something the readers can explore.
So 1S ...l:tfe8!? looks the best, giving away as little as possible to the opponent, provided
he can make his way through the initial tactical battle.
16 'ifxf3 l:i.fe8
As we've just noted, this should probably have been played a move earlier; and 16 ... es? 17
ds does not work. If given the chance, the two bishops will dominate the knights after Ji.fs.
17 l:i.fd 1 h6
17 ...eS?! 18 1i.fs again gives White the more active position.
18 1i.h4
1 69
Question: Black is worse and, in order to survive, he has to

admit this. What is the best way to try to hold the balance?
18 J:\l'd6!
..
A brave decision. All he can do is to place one of his pieces on a good active square, and
hope that White has no tactical refutation. 19 cs is dangerous, but if Anand has calculated
that he is only slightly worse after the tactics, then he must try this.
Passive play will end up with a grinding disadvantage, and should be avoided. The like
lihood is that his opponent will eventually find a win. White is happy after, for example,
18 ... .l:rcd8 19 i..g 3 '&'c8, or again 18 ... es 19 i..f s.
Trying to simplify with 18 ...'&'c6 19 '&'xc6 .l::t xc6 does not work either. White has the
bishop pair against the two knights, and also has the more active pawns. Add this up, and
20 cs is strong. 20 ... bxcs is met by 21 i.. xf6, and White is almost certainly winning the end
game after 21...gxf6 22 i..b s l:tc7 23 dxcs.
lt is easy to generate some implausible alternatives - the computer suggests that
18 ... 'ith8 might be one of these. All one can say is that if 18 ... 'ith8 is the best move, White is
clearly better. Whereas if Black can find a sensible idea, with chances of setting up some
counterplay (pressure on d4, say), he should try it.
Time now for Gelfand to make a difficult decision.
Question: White has the quick pawn break 19 cs bxcs 20 dxcs, with the threat of
i..h 7+. The question is whether this is the best line. If it can be shown that Black
can hold afterwards, could White find something better here?
19 cs?!
Tempting, but it forces play far too much for such a minimal advantage. Given the op-
1 70
portunity, White should squeeze hard with the bishop pair, and get the opponent to make
concessions before any sort of tactics or simplification.
My first impression, in my day-after blog, was that 19 ..ie2 ! ? looked good, setting up the
cross-eyed bishops with ..ia4. However, Black can play 19 ...�4!, when he is still in the
game, as after 20 a3, he could sacrifice his queen with 20...Wixc4! 21 ..ih7+ tt:Jxh7 22 l:txc4
l:i.xc4 23 �7 tt:Jdf6.
Whether or not White exchanges on f6, Black is still going to defend actively with ... tt:Jds,
and although White has a slight material plus (queen versus rook and knight), it is difficult
to see how he can ever create a passed pawn. Defensive queen sacrifices are very much in
the air in this type of position. 20 ..ig 3 can also be met, with reasonable safety, by
20 ...'iYxc4.
Time, perhaps, to take a different viewpoint. White has his bishop pair. He should aim
to keep his pawns well defended, when he would have chances of a clear edge if Black can
not do anything immediately active. Therefore, 19 ..ie2! - quiet and unglamorous, but
Black is going to find it hard to equalize. If 19 ... es, for example, only then 20 cs, with a large
advantage to White.
19 ... bxcs 20 dxcs l:i.xcs
Better to sacrifice the queen for assorted pieces, than to allow his opponent a big
passed pawn.
21 ..ih7+ �xh7 22 �xd6 �xc1+ 23 l:i.d1 �ec8 24 h 3
1 71
The general view at the time was that White must surely have close to a winning advan
tage. After some further simplification (a rook exchange}, it is difficult to see how to defend
the a7-pawn, and then how to stop White from pushing his own a-pawn. No doubt Gelfand
would have been confident about his prospects here, but Anand is a superb defender in diffi
cult positions. He starts his defensive manoeuvres by a direct attack on the queen, gaining a
tempo with the knight - but isn't he merely weakening his pieces and pawns?
24...ll:Jes 2 5 1i'e2
2 S ll:Jg61
•.•
Smart defence under pressure. Anand is not too worried about having his pawns dou
bled, so long as his king remains close by. lt is much more important that, if one of the
1 72
A n a n d - G e lfa n d, World Ch a m p i o n s h ip, M osco w 2 0 1 2
knights i s exchanged, the other knight i s safe. His main defensive plan i s that if h e can
move his other knight to ds, the rook can return to the second rank and, with the knight
being on such a good square, he will not be zugzwanged. The rook can alternate between
c7 and e7.
Anand made a difficult endgame defence look relatively simple. Remember, Gelfand
probably felt that he had excellent winning chances, otherwise he would not have entered
the tactical simplification at move 19.
25 .. .'!t:Jd5 might seem more active, but there are too many pieces getting in the way of
each other, and with not enough safe outposts, supported by pawns. For instance, after 26
'it>h 2 M.xd1 27 'iYxd1, it is not clear how Black can keep everything covered.
Even so, Black might still have chances to hold with 27 . ..lDc4!?, and if 28 'iYa4 as. Or
maybe not. With so many of Black's pieces and pawns being required for mutual protec
tion, there is always the possibility of a mixture of zugzwang and White gradually improv
ing his pieces.
26 .i.xf6
The only sensible move. After 26 .i.g 3? M.8c2 27 'iYfl M.xd1 28 'iYxd1 M.xa2, Black is ahead.
26 ... gxf6 27 M.xc1 M.xc1+ 28 'it>h2
28 ...M.c7
A natural defensive square for the rook, but Anand later finds it difficult to complete his
plans of ...tZ:le7-dS and .. .f7-fS. He is fractionally short of making everything safe, which of
course makes the endgame all the more fascinating.
Alternatively, if 28 .. .fs 29 'iYb2 M.c7, then 30 'i!Vb8 makes things awkward, since 30...l:f.e7
(or 30 ...l:f.d7 31 'iYc8 M.e7) blocks the knight's desired path to dS. With accurate play by
White, it can be difficult for Black to coordinate his pieces.
29 'iYb2
A good square for the queen, eyeing the f6-pawn, while also covering the b-file and
helping White to push the a-pawn without any significant interruption.
1 73
29 'it>g7 30 a4
•••
Question: A thought experiment. Imagine that for the rest of the game, both
players play the very best moves. Is White going to win ? Or is Black going to draw?
No hedging bets allowed, like saying that "White is slightly better". For this experiment,
you have to decide whether White ought to be winning, or whether the final result, a draw,
would h ave been a correct answer.
lt seems finely balanced. When playing through the game live on the computer, I, like
many others, believed that White should win. I was mightily impressed that Anand was
able to hold out for a draw. Six weeks later, going through the game again, I started to be
lieve that perhaps, with extremely accurate pawn pushes on the queenside, White could
yet break through . but after trying to analyse several such pawn pushes, I have been un
able to find a convincing win, so maybe the position was only a draw.
lt's difficult. If you think Black has no problems and that it is only a draw, then either
you have the very clear instincts of An and, or maybe you just do not see what the problem
is. There are, of course, intermediate points of view.
Sorry to give such an indeterminate answer. The point of the question is to ask the
readers to think about what A nand and Gelfand would have been thinking about them
selves.
30 tt::le 7
•••
Bringing the knight to ds as planned.

31 as tt::ld s 32 a6 'it>h7
1 74
A n a n d - G e lfa n d, World Ch a m p i o n s h ip, M os c o w 2 0 1 2
Question: How should White handle the next part of the game?
33 �d4?!
In some ways, this is the easiest part of the game, but in other ways it is the most diffi
cult. The point is that the players were getting close to the time control, so neither player
can be expected to have to think long and hard about the next few moves. Better just to
reach move 40. Therefore, if you have a stable position, it can be simple enough to play
some nondescript moves quickly. This is well understood.
There is no obvious reason why the queen should be better on d4 than on b2, and cer
tainly no obvious justification to use up a tempo with this move. White could instead have
done something else. There is a natural reluctance to make a critical pawn move just be
fore the time control, since pawns cannot retreat, and squares nearby get weakened. Here,
though, it turns out to be important for Black that he is able to play ...f6-fS, and therefore,
it might be quite a useful idea for White to prevent that advance, by playing 33 g4! ? him
self.
A flick through the blogs and websites suggest that most commentators ignored this
point. Sergei Shipov, among the most informative of the instant commentators, noted that
33 g4 was "the harshest" test of An and's defensive plan. others merely mentioned that
Anand seemed to look more relaxed after playing .. .f6-fS.
1 75
White's basic plan, possibly after a few intermediate moves with the king and queen,
would be to force something through with f2-f4-f5, Black now being unable to play .. .f6-fS
himself. lt looks good, and even if White cannot find a quick pawn push, Black still has to
be careful about attacks on the h6-pawn with "i¥d2, and, in some lines, 'it>g3-h4-h 5.
This leaves Black playing natural quiet moves, while White has the chance to coordinate
his queen, his pawns, and, not to be forgotten, the king. Play might continue 33 ... Wg6 34
�8 Wg7 3 5 Wg3 .l:!.c3+ (Black can only wait) 36 f3 .l:!.c7 37 f4. Since Black has to protect the
a7-pawn on a7, and the rook needs to be guarded by the knight, he has no real mobility, so
White has time to centralize his king with, for example, 37 ....l:!.c3+ 38 Wf2 .l:!.c2+ 39 We1 .l:!.C7
40 �d2, etc, and then perhaps finding the right time to advance with f4-fS.
lt is progress, and genuine progress, but it has yet to be proven that White can force a
win against the very best play by Black. A challenge for the reader?
3 3 fs
...
1 76
Anand has finally achieved his defensive set-up, although naturally h e still has to play
accurately.
34 f4
This leaves the pawn structure slightly rigid. Gelfand, still close to the time control,
might well have felt that 34 g4 makes it too easy for Black to exchange off his front dou
bled pawn, and decided it was not going to be his best option.
Possibly 34 f3! ?, a waiting move, deferring his options until after the time control. He
still has the choice between playing g 2-g4 and f3-f4. He can also move his king around, just
to see what gives.
Pressure, but no obvious winning attempt.
34...�d7 35 'it>g3
He still has no desire to exchange pawns with 3 5 g4.
3S ...'it>g6 36 'ii'h B
Progress? Or a positional trap? Black was ready to play ...tt:Jf6 anyway, followed by ...CDe4.
So he is close to holding the draw.
36 ...tt:Jf6 37 'ii'b B
And now the queen runs to the queen side.
37 ... hs
To create an extra escape square for the king {on h6).
3 8 'it>h4 'it>h6 39 'iVb2 'it>g6 40 'i!Vc3 tt:Je4
1 77
Question: The time control has just been reached.

But isn't White just about to win the pawn on h S ?
41 �c8
Gelfand avoids a nasty little trap: 4 1 'il'f3?! .l::td 2 !, when all of a sudden Black's pieces are
aggressive, rather than defensive:
After 42 �xh s+? Wg7, White has "won" one battle, in that he has gained the pawn, but
his king and queen are absurdly placed, and he will have to lose material. Black will play
....l::td 8 and ....l::th 8(+).
The self-burial by 42 g 3 ? .l::th 2 43 'i¥xh 5+ 'l£tg7 44 g4 .l::tf2 is even stranger.

If zugzwang does not save the game, then there is a close relative, stalemate, which will
give the draw: 42 �3 (worse is 42 �e3 ? ! .l:!.f2) 42 .. J:txg2 (42 ...llf2? 43 �8 leaves White
1 78
A n a n d - Ge /fa n d, World C h a m p io n s h ip, M os c o w 2 0 1 2
better) 43 �xe6+ fxe6 V2-V2.
How many of these lines did you visualize?

41 ltJf6 42 'irb8 .l::i.e 7
•••
The entertainment has gone, and we are now l ooking at a draw.

43 g4 hxg4 44 hxg4 fxg4 45 �es
Gelfand's last attempt to enliven the position, but now An and shows a new fortress.
4S tLlg8 46 �gS+ �h7 47 �xg4 f6 48 �g2 �h8 49 'iVe4 Wg7 Yz-Yz
•••
Not a faultless draw, but Anand defended well in an awkward endgame.
1 79
Game Ten: Anand-Gelfand

Even though this ended up in another draw, not reaching the time limit, one gains the im
pression of a sense of urgency, of forcing the opponent (for both sides) to have to think
about every move.
24th May 2012

V.Anand-B.Gelfa n d
Sicilian Defence
1 e4 cs 2 tLlf3 tbc6 3 .ibs

Evidently, Anand has decided that Gelfand was sufficiently well booked up not to try
another Open Sicilian .
3 ... e6
Or 3 ... d6. Or, the most common, 3 ... g6. it's a matter of choice.
3 ... a6 ?!, on the other hand, would be a sign of inexperience. As we soon see in the game,
White is more than happy to exchange bishop for knight and double the black pawns. The
forward c-pawn is then a weakness, rather than a strength, and gets in the way of the
dark-squared bishop.
4 ..ltxc6
And indeed this is the standard response for White here nowadays, not bothering to
wait for Black to play ... a7-a6. If White were to delay this exchange, Black has the chance to
play ...tbge7, avoiding the pawn doubling, after which it is not so clear what White is going
to do with his bishop on bs.
4...bxc6 5 b3
Playable, but hardly mainstream. As the game soon shows, Gelfand had anticipated the
possibility of this opening by Anand.
1 80
A n a n d - G elfa n d, World Ch a mp i o n s h ip, M os c o w 2 0 1 2
lt i s rather like a Nimzo-lndian with colours reversed. The difference i s that if Black were
to play ... d7-d5, White would in effect be a couple of tempi up, in that he already has the
advantage of the first move, and in that, when compared with White's e2-e4 in one move,
Black has to play ... e7-e6 and ...e6-es to put pressure on the e-file.
Therefore Black does not usually play ... d7-d5, so perhaps it is really a "Nimzo-English"
with colours reversed.
s ...es?!
With a sense of paradox - one would assume that s ... d6 6 .i.b2 es is a simpler way of
proceeding. There are, of course, other choices for Black.
At the Tal Memorial, later in the month, the players were naturally interested in the
openings of the World Championship, and Radjabov faced this system three times as Black.
Each time, he played the older, m ore established set-up with s ... d6, and White tried 6 es.
He beat McShane, and drew with Caruana and Grischuk.
Innovations are not necessarily improvements. They are often slightly offbeat ideas, just
to try something new, to get the opponent thinking. The problem with Gelfand's move is
that he clearly loses a tempo. Then again, is this so much of a problem, if he has prevented
White from playing e4-e5 - ?
6 tt:lxes
He takes the bait, justifiably so.
6 .. ."i&'e7
Question: What next for White?
7 �b2!?
Anand later changed his mind about this position and tried something else: 7 d4 d6 8
tt:lxc6 'i&'xe4+ 9 'ii'e 2 'fkxe2+ 10 'it'xe2 �b7 11 tt:las ii.xg2 12 �g1 .i.h 3 13 tiJC3, Anand
Gelfand, World Championship, second tie-break game 2012. Perhaps White was still
1 81
slightly better anyway. The queens have been exchanged, and White is ahead in develop
ment. Both players have slightly broken pawn structures {for White, pawns on f2 and h2;
for Black, pawns on a7 and cs}, but Black's c-pawn is further advanced and therefore more
open to attack, and also cuts down the m obility of his dark-squared bishop. Black's bishop
pair was therefore unable to threaten White. Anand later won, after 77 moves of a rapid
play, though the traffic was not totally one-sided.
The computer also gives 7 ti:ld3 'ii'xe4+ 8 Vi'e2 Vi'xe2+ 9 \t>xe2 with, it is suggested, a
slight edge to White. Possibly so. lt is not clear how Black can fully equalize. White's bishop
on b2 is going to be more effective than either of Black's bishops.
Since White has three reasonable choices for a slight edge, Black's innovation seems to
be ultimately unconvincing.
7 ...d6 8 tt::lc4
Not 8 tt::lx c6 WJ/c7 and the knight is trapped. White can win a third pawn in return with 9
tt::l x a7, but this hardly seems good.
s ds
...
8. . .'ii'xe4 9 tt::le 3 leaves White well ahead in development. He can castle quickly, and put
pressure on the e-file.
9 tt::le 3 d4 10 tt::lc4 'ikxe4+ 11 Vi'e2 'ii'xe2+ 12 \t>xe2

My own assessment, at home, was that White was doing very well, with a standard
Nimzo blockade against the doubled pawns, and also keeping ahead in development.
Unlike the previous game, the bishop pair does nothing. I was surprised on seeing a few
tweets that Black was supposed to be doing well. Was I looking at another game?
12 .....ie6
1 82
Question: What should White play here?
13 d3?!
A pawn move is not in itself a developing move, and any unnecessary pawn push cre
ates weaknesses.
White can develop far more efficiently with 13 tt::lba3 !, followed perhaps by l!hel, Wfl
and, if given the chance, tt:les, allowing the other knight to move to the centre with tt:lac4.
White has a clear edge.
lt is also possible that he can develop with the other rook instead, with .l:Iael, and �dl.
The important point is to develop quickly, rather quicker than his opponent can do, and
then concentrate on manoeuvring against Black's pawns. lt is going to be difficult for Black
to get all his minor pieces working well together.
13 tt:lf6 14 tt:lbd2
...
The problem now is that each of White's pieces is blocking the others. Play is only level.
14 0-0-0 15 .l:the1 i..e 7 16 �f1 11he8 17 i.a3
...
1 83
Not really the greatest of diagonals, but what else is there to be done?
11 ttJds
...
A reminder for White that he no longer has any protection for the c3-square.
18 t2Je4 l2Jb4
Indeed, just for the moment, it is White, not Black, who faces difficulties on the queen
side. Anand will be fully aware of this, and he quickly neutralizes the difficulties.
19 l:te2 kxc4
Gelfand wants to remove the strong c4-outpost for his less than impressive bishop.
20 bxc4 f5
21 .ixb4
Question: But should White have exchanged Black's knight just here?
1 84
A n a n d- Ge lfa n d, World C h a m p i o n s h ip, M osco w 2 0 1 2
Black's knight i s no longer threatening, now that h e cannot move it to dS. Unless White
allows Black to take one of the pawns, the only way back to active play would be via a6. For
instance, 21 tt:lg3 g6 22 .i.c1 �d7 !? (to make good use of the b-file for the rook) 23 a3 tt:la6 is
tense, but equal.
Both lines look reasonable, but Anand's move cuts down any complicated lines. lt seems
he was already thinking about a draw.
21 ... cxb4 22 tt:ld2
Amazingly, it is White who now has the doubled c-pawns, and this makes it difficult to
take advantage of Black's slightly ragged pawn structure.
22 .i.d6
...
The computer suggests 22 ....i.f6 as with a slight edge for Black, though it is difficult to
believe this. 23 a3 bxa3 24 l:txa3 looks fine for White, whereas Black's bishop is not on a
good post to nurture his queen side dark squares.
23 .l:txe8 .l:!.xe8 24 tt:lb3 cs 25 a3 Yz-Yz
1 85
Question: A draw was agreed here. Could one of the

players have continued to try for an edge?
The draw agreement seems slightly premature. Black definitely has a slight edge after
2 5 ...bxa3 26 .l:!.xa3 'it>b7, in view of his outside passed pawn. White cannot, despite initial
appearances, force a perpetual or set up a simple rook exchange; while Black, for as long as
he can keep the a-pawn safe and slightly threatening, can force his opponent to have to
defend accurately. White has an extra pawn elsewhere on the board, in compensation for
Black's passed pawn, but clearly the backward doubled c-pawn is in no sense dangerous.
27 l:!.as .l:!.c8?! 28 l:!.b5+ looks like a stock equalizing plan for White, and no doubt both
players would have noted this. But the computer notes that 27 ... a6! ! is good for Black.
The point is that after 28 lt:lxc5+ 'it>b6 29 lt:lb3 .ib4 30 l:!.al (remember, Black is threaten
ing mate on el) 30 ... .i.c3, now that Black has opened up lines for his diagonal, his a-pawn is
rather dangerous. For instance, 31 l:!.d1 as 32 lt:lc1 a4 33 lt:la2 (not 3 3 lt:le2?? l:!.xe2, and
Black promotes on al) 33 ... .ias is unpleasant for White.
He could instead retreat with 27 l:!.a1 a6 (still a useful move) 28 .i:!.b1 'it>c7 and then, for
example, 29 l:ta1 l:!.a8 30 'it>e2 'it>b6 31 l:tbl 'ito>c6 3 2 lt:la5+ 'it>d7 (see following diagram), but
now Black has untangled his pieces, and White will have to work out how to defend
against the a-pawn for the next thirty m oves.
186
A n a n d- G e lfa n d, Wo rld C h a m p i o n s h ip, M osco w 2 0 1 2
As i n some of the earlier games, Gelfand should really have tried for more i n this end
ing. Perhaps he did not quite believe that he could beat Anand in technical endgames.
Anyway, another draw, and we are getting ever closer to the maelstrom of rapidplay
and blitz to decide the World Championship.
Game Eleven: Gelfand-Anand
26th May 2012

B.Gelfand-V.Anand
World Championsh ip, Moscow 2012 (Game 11)
1 d4 tLlf6 2 c4 e6 3 tt:Jc3 ..ib4

Again avoiding the Slav.
4 e3 0-0 5 ..id3 dS 6 lLlf3 cs 7 0-0 dxc4 8 ..ixc4 ..id7
Not new, but almost forgotten. Gelfand thought for more than half an hour in response
to Anand's modest developing move, apparently introduced by Bronstein. The main devel
opment plan is ... i.c6 followed by ... tt:Jbd7.
187
Question: Can you find a clear edge for White? Or is Black's opening fully playable?
9 a3
This was the one chance for Gelfand to take the initiative. If An and's move was genu
inely doubtful, and if there were a few ways to gain an edge, then Gelfand would have
found one of them. If the position after Anand's move was genuinely equal, then it would
have been hardly surprising that Gelfand could not find anything.
a) 9 a3, as played in the game, is the computer's main suggestion for a slight edge.
b) 9 'iVb3 is the computer's second selection, but then 9 . ..lbc6 forces White to scurry
away to avoid . ..lt:Jas. White cannot realistically be better here.
c) 9 tt::l e s is potentially interesting, but it needs to be remembered that the "threat" of
gaining the bishop pair is not of great significance. White uses up three moves with the
knight to capture on d7, while Black has made only one move with the bishop. The tempo
count suggests that Black should be equal after either 9 . tt::l c6 or 9 cxd4.
. . ...
lt seems that Gelfand is still on the right tracks.
1 88
A n a n d - Ge lfa n d, World C h a m p i o n s h ip, M osco w 2 0 1 2
Question: Three reasonable choices here, the third being 9 . cxd4.

. .
Which is best? Or which has the m ost surprise value in a level position?
Or are they all about equal ? Find a move, and hope that it is not a mistake.
g i.as
...
Apparently, this was a surprise to many of the top grandmasters commentating in

Moscow. Later moves were not predicted by the grandmasters either. This does not neces
sarily mean that the moves were particularly odd or unusual. lt just reflects the point that
if a position is close to equal, then quite often there are several ways to retain near
equality in several different lines. We can be grateful for that. If there was only one reason
able move in any position, chess would have been played out to its death by the time of
Capablanca.
The basic point about Anand's bishop retreat is that he has decided - in advance of
course - only to exchange his bishop for knight if White has already captured with d4xcs.
An and would prefer, if possible, not to allow his opponent to have a solid pawn structure
of c3/d4/e3, etc. lt is much less solid if the pawns are on c3, cS, e3, etc.
Instead:
a) 9 .. �xc3 10 bxc3 is a minor concession, though the main play has yet to come. One
.
problem in this line is that, once the b-pawn has switched files to c3, White can add some
pressure with l1b1. Black no longer has the simple equalizing option {as in two rounds ear
lier) of exchanging the cs-pawn on d4, followed by ...b7-b6 and ... �b7, so his position is
potentially slightly uncomfortable. Anand was looking for more direct ways to equalize.
b) 9 ..cxd4 10 exd4 is a standard IQP {isolated queen's pawn) variation, in which it is not
.
so clear that he really wants the bishop on d7. Again, he would have been happier with
... b7-b6, ... �b7, ...tt:Jbd7, etc. Black still has to prove that he can equalize.
Possibly Anand's move is indeed the best among equals.
1 89
10 'ii'e 2
10 dxcs �xc3 11 bxc3 'it'as is level.
10 ..tc6 11 l:td1
••.
Question: Black to play. What next?
11 ..txc3!?
••.
A puzzling move, indeed the most puzzling move of a short but difficult game. Black has
already retreated with ...i.as, declining to capture on c3, but now, a couple of moves later,
he plays ... �xc3 with loss of tempo. Neither player has altered the pawn structure in the
meantime, whereas White has played a couple of developing moves which, one would
think, would mean that Black does not have the luxury of giving away a tempo.
So why did Anand try this? Or, more specifically, why now? The point here is that while
White has made a couple of developing moves, Black has done the same. He has manoeu
vred his bishop to c6, and is ready to play ... tZ'lbd7. Once he is ready to achieve this, it is a
good idea to eliminate White's c3-knight, to avoid, for example, a d4-d5 prod by a white
pawn. Also, as we shall see later, Black can offer a favourable bishop exchange with ... �ds,
once the knight on c3 has gone.
lt works, in that Anand soon equalized. There is the question, though, of whether Gel
fand could have reacted more alertly, with the possibility of a slight edge.
Anand's play is arguably the most difficult to appreciate, at World Championship level,
since the days of Petrosian in the 1960s. lt is not so difficult to appreciate the games of
Fischer or Kasparov, even if admirers know that they would be unable to play as well. Kar
pov, too, gives nothing away, and takes whatever is given. If the opponent makes a slight
error, he will lose. Playing through the games in this match, Anand's most obvious similari
ties are with Petrosian and the deep and intricate manoeuvring, which looks puzzling, but
in which everything works out well in the end. We have already seen, several times in this
190
A n a n d- G e lfa n d, Wo rld C h a m p io n s h ip, M os c o w 2 0 1 2
match, how Anand has given away bishop for knight, when most players would prefer the
bishop. This was a Petrosian speciality as well.
The difference between Petrosian and Anand, and why Anand became a much stronger
player, was that, behind his defensive technique, An and was far more willing to attack
when given the chance - and, above all, he was able to analyse his openings in great depth
at home with, of course, the help of the computer, whereas Petrosian tended to be lazy in
this aspect of the game. Occasionally, Anand, like Petrosian, can go spectacularly wrong in
misjudging the pace of play in a position. But not here.
Returning to the game, there is still the puzzle as to why An and did not try 11 ... tt::lb d7!?.
After all, 12 dxc5 i.xc3 13 bxc3 could have arisen in the main line, supposing White had
played d4xc5 a couple of moves later. Alternatively, 12 d5 is n atural, when 12 ...exd5 (not
12 ... i.xc3?? 13 dxc6, winning a piece) 13 tt::lx d5 tt::lb 6 leads to sharp play, although after any
tactics, Black remains equal.
Well, at least that is what I said several weeks ago. On checking through various refer
ences, I notice that Shipov gives this variation as really strong for White after 14 tt::lx b6
(Shipov describes as "weak" the continuation of R.Knaak-A.Yusupov, Hamburg 1991: 14
tt::lf4 �c7 15 i.a2 l:Iae8 16 i.d2 i.xd2 17 l:!.xd2 tt::le4 18 .l:!.c2 �e7, which is equal) 14...�xb6
15 b4 cxb4 16 tt::l e 5 "and so on". However, it is hard to see why White's position could be
regarded as in any sense advantageous. Black has his extra pawn, and potentially two out
side passed pawns; he is better developed, and his king is safe enough. All that can be said
for White is that he has a pair of bishops after, say, 16 ...�c7 17 tt::lx c6 �xc6. The best I can
reasonably suggest is that White should be able to equalize with 18 i.d2. So I stick to my
earlier assessment, while noting that White has a possible improvement at move 12 (see
below).
Instant commentaries are often extremely useful, especially because the commentator,
like Shipov, is not merely churning forth the output of computer chess engines. The trouble
is that, as in chess over the board, mistakes and misjudgments slip in.
191
We return to a previous comment of mine, in which I suggest that White has a different
way to try to show an edge. One suspects that Anand would have been more worried about
12 ct:Jes ! ?. If Black takes the knight, then after 12 ...ct:Jxes (there are various alternatives,
probably each with a slight edge to White) 13 dxes lDd7 14 .ibs, White is starting to create
pressure on the d-file.
12 bxc3 ct:Jbd7
Question: What is White's last chance to squeeze out a slight edge?
13 ..id3
This m akes it much easier for Black to exchange bishops, a few moves later, with .....ie4,
neutralizing the bishop pair. After the move played, everything soon ends up in equality.
Instead, 13 ..ib2 Wile? 14 ..ia2, or perhaps 13 .i.a2 first, was the last chance before the
rapidplay for Gelfand to try for a plus score with the white pieces. After that, it is Anand
who has the advantage of White in game twelve. Black still has to prove he can equalize.
13 ...'�a5 14 c4
14 .i.b2 cxd4 15 cxd4 i.e4 is equal.
14 ... cxd4 15 exd4 'i!Vhs
192
16 ..if4
An unexpectedly provocative move, inviting Anand to break open White's pawn struc
ture with 16 . ...ixf3 17 gxf3.
.
Question: Should Anand have tried this?
Probably not. Anand was happy to halve out with his next move, safely.
Yes, he could give his opponent doubled and isolated f-pawns, but White then has two
bishops versus two knights, and while it is easy enough to see ways in which White can
improve his position, it is difficult to see anything constructive for Black.
Besides, Anand had just encountered a gruelling defence with two knights versus two
bishops, as Black in game nine. He would not want to go through all that again. Quite pos
sibly Gelfand would have reasoned that Anand was unlikely to speculate, at this stage of
the match.
16 .:.ac8 17 'Lles
...
He cannot keep the knight on f3 indefinitely.

1 1 ...'ifxe2 18 �xe2 'Llxes 19 .i.xes .l:tfd8
193
Question: Does White still have chances of playing

for a slight edge with his bishop pair?
The problem is that White's queenside pawns have been weakened, making it ex
tremely difficult for him to open the position in his favour. Meanwhile Black can try to ma
noeuvre his pieces to attack the hanging pawns. White turned out to be happy with a
draw.
20 a4 lL:le4 21 Il.d3 f6 22 i.f4 ..tes
Indeed, momentarily it looks as if Black is already better, but Gelfand has his defences
sorted out, and there was soon a handshake.
23 .l::!. b 3 .l:!xd4 24 .te3 .Ud7 Yz-Yz
194
Draw agreed, perhaps slightly prematurely from the point of view of the spectators, as
there is still play on the board - but if both players play accurately, the result would even
tually be a draw. White of course plays 2 S iJ..x a7, when a possible continuation might be
2S ... .i.g6 26 1J..e 3 lZ:ld6 27 cs lZ:lfS 28 �f3 lZ:lxe3 29 fxe3 J:txcs 30 .i.xb7 llas, and the draw
comes even closer.
Game Twelve: Anand-Gelfand

Was this going to be a tame draw in the final round, a time to relax before the stress of the
rapidplay? There seems little sense in going all out for a win, taking risks, on the last day of
the slow-play. Better to wait for the opponent to make mistakes in a quicker time limit.
lt turned out that the game was indeed a draw, and in 22 moves, but there was no lack
of creative spirit. There was enough interest in the first 21 moves to fascinate anyone.
After much of the tentative chess of the first half of the match, and the serious mistakes
in games seven and eight, both players seemed to settle down to play exceptionally high
level, creative chess. One senses that, had the match had been played over 16 games, or
even the traditional 24 games, there would have been far more outstanding chess to fol
low.
28th May 2012

V .Ana nd-B.Gelfand
World Championsh i p, Moscow 2012 {Game 12)
Sicilian Defence
1 e4 cs 2 lZ:lf3 lZ:lc6 3 .i.bs e6 4 1J..xc6 bxc6
195
5 d3
Question: What do you think about this move? Remember that

Anand had played 5 b3 a couple of rounds earlier.
The d2-d3 push seems rather tame here, especially given that Anand soon played b2-b3
and Ji.b2 anyway. The suspicion is that he simply did not want to allow Gelfand the same
variation as in game ten {5 b3 e5 6 t'Llxe5 'ike7, etc}, even though White could still have
kept, it seems, a slight edge at move 13. And what were the seconds there to do? Three days
before the next game was surely enough time to examine the opening and test it out, with
a few grandmasters and a computer. The move I suggested in the notes {13 t'Llba3) seems
logical enough, avoiding the extra pawn move {d2-d3}, while still completing his develop
ment. The computer suggests that it is good.
Quite possibly Anand, having been unable to demonstrate an edge over the board,
wanted to disregard the whole line with 5 b3 for this game. If he can find something prom
ising with 5 d3, then play it! Objectively, though, it is unlikely that 5 d3 is quite as accurate
as 5 b3, and in the second rapidplay game, Anand returned to 5 b3, making a different at
tempt at an improvement.
5 0-0 is also to be considered, a straightforward developing move, waiting to see
whether Black aim s for ... d7-d6 or, less likely, ... d7-d5.
s t'Lle7 6 b3
...
Again, if this is still Anand's plan, there was no great reason to insert 5 d3. Was this a
sign that, these days, Anand has tended to play overcautiously, as some commentators
have suggested?
6...d6
Thus, no 6 ... d5.
196
A n a n d - Ce lfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
Question: What should White play next?
1 es
Really, there is not much choice to be made here. If Black is allowed to play ... e6-eS, he
would have a solid and compact pawn centre, and a useful bishop pair, for when it starts
to open up.
Play starts to become intricate and complicated.
7 tZ:lg6
•••
Question: Should Black have taken the pawn immediately?
There is no obvious reason why the pawn cannot be snatched with ? .. dxes. White could
.
197
Fig h ting Ch ess: M o ve by M o ve
not try 8 ct:Jxes?, as after 8 .. .'iVd4 9 ctJc4 �xa1 10 ii.b2 �xa2 11 ctJc3 'ifa6 12 ct:Jd6+ 'it>d7 13
ct:Jxf7 l:.g8, he has started to run out of ideas for his rook sacrifice.
This leaves something like 8 ii.b2 f6 9 ct:Jbd2, making it a positional gambit.
lt is not absolutely certain that White has complete compensation for his pawn, but it is
not clear either that Black is definitely better. Indeed, White has his own positional pluses.
Probably the position is equal, or close to equal, though it can be difficult to assess what is
going on when there is a material imbalance.
The danger is, for Black, that he might feel obliged to show that he can do something
positive with his extra pawn. He has three main weaknesses in this rather strange position.
Obviously the c-pawns are doubled and isolated, and the squares in front of the pawns
create good outposts for White. The other two weaknesses are the two bishops. The one on
f8 has no real destination, blocked by the doubled pawns on cs and on es. The c8-bishop is
equally obstructed, by the other doubled pawns on e6 and c6. Black needs to think about
how to bring his bishops into play.
lt is possible that this might have been a slightly better way of playing this for Black, in
view of later comments. Such assessments can at best be tentative in such a difficult posi
tion. The move Gelfand played was not bad either.
198
Question: White to play. What next?
8 h4?!
By now Gelfand and Anand will have been well-tuned into each other's thought proc
esses, if not so much in the first eight games, then certainly in the later rounds. One senses
that, on the last move, Gelfand had instinctively decided that Anand would want to con
tinue the pawn sacrifice for as long as possible, and try to squeeze some sort of initiative
out of it.
Anand's push of the h -pawn was slightly strange, the only realistic explanation being to
try to force Black into taking the gambit. The advanced h-pawn did not add to the ideas
and strategies used later on. The possibility of a further h4-h6-h6, for example, was
unlikely ever to be played.
There is a much simpler way of handling the position for White, avoiding having to
gambit a pawn, and almost certainly giving him a slight edge. All he has to do is to play 8
exd6. 1f then 8 ... ..txd6 9 tt:lbd2, White's two knights, plus the bishop, are far better coordi
nated than the opposing minor pieces.
Black can try 8 ...�f6, gaining a tempo by the attack on the rook. Then 9 c3 �xd6 is
about equal, but 9 d7+! ..ixd7 10 c3 gives White a slight edge, and often this is all that is
really needed out of the opening.
199
Black's bishop on d7 is on a bad square, and this makes all the difference between a
slight advantage for White and mere equality. lt will take time for Black to put pressure on
d3 with the rook, and to attack that square with the bishop, as ... �c8 and ... �a6 would cost
two tempi.
10 ...tZ:lf4 11 �xf4 �xf4 12 �d2 .i.d6 13 �xf4 �xf4 14 tt:Jbd2 gives the sort of position
where two knights can occasionally outplay the two bishops. White will soon be able to
play tZ:lc4, a good square which cannot be attacked by pawns, or perhaps tt:Je4. The two
knights can work together.
If instead 10 ... \\!Vfs, the simplest option would be 11 Wie2, followed by offering the ex
change of queens with �e4. Again, Black's queenside pawns are damaged, and he does not
even have an extra pawn to compensate.
Black can also try opening up for the light-squared bishop with 10 ... es, but after 11
tt:Jbd2, it is not so clear that any of the bishop moves are that effective. Meanwhile, the
other bishop is caged in again by the pawns on cs and es. White keeps a slight edge.
200
A n a n d- G e lfa n d, World C h a m p io n s h ip, M osco w 2 0 1 2
Question: Black could again take on es, and this is playable. But i s there a quick
and simple way to simplify the position, to ensure that Black cannot lose?
s . . . tt:'lxes
The most obvious reply, but there are still, as the computer suggests, several different
possibilities of keeping the position equal, while lively and interesting.
Assume, though, that Black does not want to keep the tension alive. He is not interested
in provoking his opponent into overpressing. All he wants is to take the half-point away as
quickly and as conveniently as possible. A draw with Black against strong opposition is
usually a good result.
The simplest way would appear to be 8 . .. c4! ?, a return pawn offer, which immediately
mobilizes his dark-squared bishop. If 9 bxc4 then 9 . . ..�:Jxes and Black is comfortable, since
10 tt:'lxes?! is answered by 10 . . .'iVaS+ 11 c3 (not 11 �d2?? 'Ylixes+, winning the rook)
11 ... 'YlixeS+ with the advantage. Black also has at least equality after 9 dxc4 tt:'Jxes.
If instead 9 exd6 cxd3 10 'Ylixd3 'Ylixd6, Black has some slight difficulties with his queen
side pawn structure, but there is adequate compensation with his more active minor
pieces, and his bishop pair.
lt is difficult to imagine that White could be enthusiastic about trying to win in such
positions. (Again, what has he achieved with his h2-h4 push ?) Job done for Black.
9 tt:'Jxes dxes 10 tt:'ld2
Question: Effectively the same question as before. What is the best way to
handle the position here, to try to eliminate any danger of a loss?
Warning: This was perhaps the most difficult move so far in the whole of the match,
and Gelfand thought for an hour in making his decision. lt is a baffling position, and if a
201
player of Gelfand's strength cannot fin d a good position quickly, then few others could
have done better. Indeed, one would suspect, 99% of chess players would not have consid
ered the move that Gelfand played, and of the 1%, most players would have rejected it
quite quickly.
10 c4!?
...
An astonishing move, which got the audience buzzing. Black gives away the extra
pawn, just to activate his dark-squared bishop, and then, before long, gives away a second
pawn too.
Gelfand's move is not perhaps all that great a surprise if we take into account the pos
sibility of ... c5-c4 two moves earlier. Quite possibly he had thought about that, but decided
it was unnecessary. A little further on, there is an extra degree of urgency. Black has to de
cide quickly how to get his bishops into active play. Two moves ago, this would have been a
straightforward equalizing idea, whereas now it makes the whole game noteworthy.
The problem is that quieter lines, such as 10 ...i.. d6 11 i..b 2 f6 12 � 5+ g6 13 'i!Vf3 i..b 7
14 �g4, or 10.. .'ii'd 5 11 tt:Je4 !ii.. e 7 (now if 11...c4, 12 bxc4 gives White a slight edge) 12 'il¥g4
g6 13 'il¥g 3, still make life uncomfortable for Black, and therefore there is still a danger of a
loss in the longer run.
Suppose, on the other hand, that Black tries to keep the extra pawn, being prepared to
combat any tactics - for instance, if it was essential for him to play for a win, rather than to
secure a draw - there is still a possible choice. He could try 10 ...'i!Vd5 11 tt:Je4 f5 !? 12 'il¥h 5+
g6 13 tt:Jf6+ �f7 14 tt:Jxd5 gxh 5 15 tZ:le3 .
Black now h a s three sets o f doubled pawns! l t is unlikely that White will b e able t o re
gain a pawn quickly, but Black will have to work hard to keep them all together. Maybe
15 ... i.. a6 16 ..ib2 !ii.. g 7 17 0-0-0 .l:!.hg8 18 g3, tense but perhaps level. Not quite enough, if
the first objective is to make sure he cannot be worse, in the heat of the match battle.
11 tt:Jxc4
202
11 ..lta6
...
Question: Could Black have protected the front e-pawn with

11 .. .f6 - ? He can be content with having opened up lines
for his dark-squared bishop, and material is level.
This would not be advisable. 11...f6 12 'iVh S + g6 13 'iff3 leaves too many pawn weak
nesses, especially on c6 and f6.
12 �f3
Of course not 12 tt:lxes?? �aS+.
12 :iVd s
..
The next stage of Gelfand's defensive idea.

13 �xd s
Question: This queen exchange is obvious, natural, and wins a pawn. An and played
it very quickly. But were there any other interesting moves he could have tried?
Shipov suggested that 13 �g3 ! was the last attempt by An and to try for a win. Pawns
are level, but Gelfand still has structural problems, which are not yet fully resolved. Black
also has trouble with his dark squares, and it is difficult to bring his dark-squared bishop
into play.
The computer suggests 13 .. .f6 14 o-o as level, but White can be happy with this. Black's
sacrifice of the cs-pawn gives a bit more freedom to his pieces, but White's game still looks
better.
Another option is for Black to exchange his front e-pawn with 13 ... e4, but this opens up
the long diagonal, and after 14 o-o exd3 15 cxd3 l:!.d8 16 .l:!.d1, followed soon by ..ltb2, Black
still has to work out how to develop his kingside.
203
This would have been the last serious attempt by Anand to try to win the slow-play
match, before the tensions and dubious excitements of the rapidplay.
13 ...cxds 14 lt:Jxes f6
Gelfand has sacrificed his second pawn, ensuring that the five remaining pawns are
healthy and secure, with prospects of activity, revitalized by his free-ranging bishops.
1s lt:lf3 es 16 o-o �f7
Question: Superb though Gelfand's defensive plans might be, it is always

best in chess to remain a little sceptical. Is Black really totally level here?
17 C4
This simplifies the structure a little too soon if White is aiming to gain anything out of
the extra pawn. lt was not long before the players agreed a draw. Quite possibly, with it
being the final game of the main match, both players would have been extremely tense,
2 04
and each would have felt a sense of relief at not losing. Six all would be a highly respect
able score for either player, equal first in the slow-play; whereas s%-6V2 would have been a
real downer for the loser. They then had to start thinking about the rapidplay.
Objectively, Anand is still slightly better here. The important point is not to allow his
pawn structure to be weakened, if at all possible. 17 �b2 would at least have forced Black
to show how to defend a slightly worse position:
a) 17 ... e4 18 lt:ld4 exd3 19 cxd3 .1xd3 20 .l::i.fd1, and while Black now has level material,
he will still be under pressure with his isolated d-pawn. lt is impossible for two bishops to
focus on the same square, whereas bishop and knight can sometimes work well together.
b) 17 ... d4 allows White to open up a good square for the knight with 18 lt:ld2, and lt:lc4
before too long.
c) 17 ...Sl..b4, to block out the knight from reaching d2, would perhaps be one good
method of defence, and then after 18 c3, a quick retreat with 18 ...Si..d6. White's pawn struc
ture will again have been slightly weakened, and this gives the opportunity for Black to
defend.
The draw is not wholly trivial for Black, although Gelfand would surely have found a
good way of holding the game. The chances for White to try for an edge with 17 �b2 seem
less than those after 13 'ifg 3 earlier.
17 ...i.e7 18 �e3 i.b7 19 cxds
If 19 !Ifc1, all Black needs to do to defend is to centralize his pieces, and not worry about
pushing his pawns - therefore 19 ...�hd8.
19 ...�xds 20 �fc1 a s
In order to try and break down the outside 2-1 queenside white pawns.
21 i.cs .llh d8 22 JJ...x e7 Yz-Yz
Question: Many commentators suggested that the agreed

draw seemed a little premature. What do you think?
205
The draw offer by White was perfectly reasonable. White has the extra pawn, but Black
has the more active pieces, and White's pawns have various weaknesses, so it is difficult to
see how he could set up a viable passed pawn.
After 22 ...rJ;;x e7 23 .l:i.c7+, both 23 ....l:i.d7 and 23 .. .<it>f8 seem acceptable. The simpler line
would be 2 3 ...�d7 24 .l:i.xd7+ rJ;; x d7 25 tt:ld2 a4, when there is not much to be done for
White. Indeed, if he tries too hard, he has to be careful not to fall for any traps. For exam
ple, 26 tt:lc4 rJ;;c 6 27 b4 'it>b5 28 a3 i.xc4 29 dxc4+ c;t>xc4 30 �dl e4 3 1 .l:i.el f5 32 f3 f4! ?.
lt is now White who has to play for the draw: 33 fxe4! (not 3 3 .l:f.xe4+? 'it>b3 and White is
in trouble) 3 3 ...'it>b3 34 b5 rJ;;x a3 3 5 .l:f.bl with a repetition after 3 5...c;t>a2 36 l:tb4 rJ;; a 3 37
�bl.
But who needs all this extra excitement at this stage of the match ? Easier just to take
the safe and quick draw.
In my blog, I suggested this was "the best chess of the match so far". Now I am not so
sure. Technically, Gelfand should have played ... c5-c4 a couple of moves earlier, for exam
ple, while Anand had possible improvements at several points in the game. lt was un
doubtedly difficult and interesting, though, with much chess paradox squeezed into
twenty moves.
The Seq uel: Tie-Break Games

Anand won the rapidplay playoff, retaining the World Championship. He scored one win
and three draws, out of the four extra games. I have played through the games, but I do
not intend to annotate them fully. There was, of course, much entertainment, as is bound
to happen with two world-class players, each aiming for a win. Equally, the players cannot
be expected to play completely accurate chess, when the clock is so dominant. So there are
mistakes, so what?
lt is sad that we have resorted to rapid games to decide the World Championship. What
206
happens in the end i s that strength and stamina will decide, rather than chess understand
ing. The game that Gelfand lost was from a drawn endgame, with only one pawn on the
board and an equal number of pieces. Gelfand did not have any second time control and,
having to play ten seconds a move indefinitely, he slipped in an error or two. Not really a
good way for a world championship to finish.
For completeness, we give fragments of each of the games.
30th May 2012

B.Gelfa nd-V.Anand
World Championsh ip, Moscow 2012 (Tie-break game 1)
Semi-5/av Defence
This looked impressive when played through quickly. Sharp and open tactics ending up,
after sacrifices on either side, in a drawn rook and pawn endgame. Sometimes, though, the
draw may be the result of mistakes by both sides, and all the more so in a rapidplay game.
1 d4 d5 2 c4 c6 3 lLlc3 lLlf6 4 e3 e6 5 lLlf3 lLlbd7 6 'iVc2 .i.d6 7 i..d 3 o-o 8 o-o e5
Anand believes in his plan. 8 ... dxc4 9 i.xc4 bs is a more common reaction.
9 cxd5 cxd5 10 e4 exd4 11 ttJxd5 l2Jxd5 12 exd5 h6 13 b3 lLle5 14 lLlxe5 .i.xe5 15 .l:!.e1 .l:!.e8 16
.i.b2 i..d 7 17 �d2 'i!Vf6 18 g3 .l:!.ac8
19 a4?!
A strange move, which merely, and unnecessarily, ends up weakening his pawn struc
ture. The simplest way for White to try for an edge might well be the ultra-symmetrical 19
.i.e4, arguing that his bishop on b2 is superior to Black's bishop on d7.
19.. .'�f3
Straight away, An and pounces on Gelfand's weaknesses at b 3 and f3. Black is now bet-
ter.
207
20 .i.e4 �xb3 21 .U.ebl
Exactly the sort of position where A nand would have benefited from having twenty
minutes to think.
21 ... �xg3?!
This looks good, and the crowds enjoyed it, but the game soon ended up in a draw.
Instead, the computer suggests 21 ....U.c4!, quietly consolidating, and it is difficult to see
how White can improve his pieces. If 22 �xd4??, then of course 22 . .J::rxd4.
.
lt is easier to think through tactics quickly than to try to assess a position.

22 .l:ta3 �b6 23 �xd4 i.xh2+
Of course Anand had seen this in advance.

24 'it>xh2 �d6+ 25 .l:!.g3 .U.xe4 26 �xg7 c.t>h7 27 .U.xb7 .U.g8
208
And we can assume that both players will have seen their way through the tactics.
28 'iWxh6+
The only problem, from Anand's point of view, is that Gelfand was able to find the
drawing line. lt is extremely difficult for a player who has calculated a variation, over the
board, very close to its end, and seeing that it is not winning, then to change tack at the
beginning and start again (i.e. 2 1 .. J!c4 instead of 2 1 ... .i.xg 3 ) . lt is close to impossible to do
this in a rapidplay.
28 ...'iWxh6+ 29 i.xh6 I:i.xg3 30 'it>xg3 i.c8 31 I:i.c7 'it>xh6 32 I:i.xc8 I:i.xa4 Yz-Yz
A sharp draw, although there were undoubtedly mistakes.
30th May 2012

V.Anand-B.Gelfand
World Championship, Moscow 2012 (Tie-break game 2)
Sicilian Defence
1 e4 c5 2 tt:lf3 tt:lc6 3 .i.b5 e6 4 .i.xc6 bxc6 5 b3 e5 6 tt:Jxe5 W!ie7 7 d4

Deviating from game ten, in which Anand played 7 .i.b2 ! ?.
7 d6 8 tt:Jxc6 'iWxe4+ 9 'iWe2 Wixe2+ 10 'it>xe2 .i.b7 11 tt:la5 .i.xg2 12 I:i.g1 .i.h3 13 dxc5 dxc5
••.
14 tt:lc3 0-0-0 15 .i.f4 .i.d6 16 .i.xd6 I:i.xd6
2 09
Here it seems that Anand could have achieved a clear advantage with 17 .l:f.xg7, leaving
Black with four isolated pawns.
The game move, 17 l:tg s, was also possible, but leaves White only a slight edge. Maybe
Anand was distracted by the thought of attacking the black king, but, if so, he was unable
to demonstrate checkmate.
Nevertheless, Gelfand was getting short of time - a move every ten seconds indefinitely
- and although he reached what presumably have been a straightforward draw, with rook
and bishop versus rook, knight and pawn, he managed to lose the game. lt is much easier
in defending, when you have plenty of time on the clock, than when you have to play al
most instantly, while still trying to calculate whether you have allowed a potentially good
move by the opponent, and see what your own move would be in reply. Just going through
the mechanics takes up time.
This is the position after Black's 64th.
210
And this i s the position after White's 71st. lt i s easy to see that White has made pro
gress, but it is not so clear that he should have any winning chances. Even so, Gelfand
somehow lost, after allowing Anand to exchange minor pieces into a winning rook ending.
So, a game down over the match.
The question for the reader is to try to find a reliable way of holding the position,
whether after move 64 (you are aiming for a rock-solid blockade}, or after move 71. Use the
computer to help you analyse.
30th May 2012

B.Gelfa nd-V.Anand
World Championsh ip, Moscow 2012 (Tie-break game 3)
Slav Defence
1 d4 d5 2 c4 c6 3 tLlf3 ttJf6 4 e3 ii.f5 5 tLlc3 e6 6 tLlh4 ii.g6 7 tt:Jxg6 hxg6 8 ii.d3 tt:Jbd7 9 o-o
.i.d6 10 h3 0-0 11 �c2 �e7 12 nd1 nac8 13 c5 .i.b8 14 f4 ttJe8 15 b4 g5 16 nb1 f5 17 b5
gxf4 18 exf4 tt:Jef6 19 bxc6 bxc6 20 .i.a6 nc7 21 ii.e3 tLle4 22 .l:!.b2 g5 23 .l:!.db1
Anand is in massive trouble here. All he can do, in practical terms, is force Gelfand to
have to calculate not just one line, but several different lines. With any luck, the opponent
will miss something.
211
Anand opened play up with ...

2 3 ...gxf4!? 24 i..xf4 es
Two good captures on the e-file, but which is the better? In fact it doesn't really matter,
but it is time spent on the clock for Gelfand to analyse.
25 i..x es
25 tt:Jxe4 also wins: 2S ... exf4 26 ti:Jd6 and Black is squashed on the queenside; or
2 S .. .fxe4 26 i..x es tt:Jxes 27 dxes and the bishop on b8 is lost.
2s ...tt:Jxes
Somewhere in all the hacking of pieces on both sides, Gelfand missed 26 tt:Jxe4. Then
26 .. .fxe4 27 dxes is the same as before, and again Black's bishop drops; while if he moves
the knight away, White has 27 ti:Jd6.
Gelfand instead played...
26 .l:txb8?
... which is strange, as the bishop is dead wood. Taking the live knight would have been
far more sensible, but of course mistakes occur in rapidplay. Anand was now able to play...
26 ...'2lg6
... when he had survived for the time being.
Then m any, many more moves were m ade, just seconds per move, far quicker than one
would expect to ask grandmasters to play. After many mistakes, Gelfand emerged with a
winning rook endgame, two pawns up, and sacrificed one to promote the other.
So, the final chance for Gelfand to win, and he blew it...
212
A n a n d- Ge lfa n d, World C h a m p i o n s h ip, M os c o w 2 0 1 2
White to play
Gelfand gave away a totally unnecessary tempo with 61 l:'th7? 'it>d7 62 'it>g3 'it>e6 63 'it>g4,
when he had to admit that he was no longer winning, and so offered the draw. 61 h7? 'it>b7
is also a well-known draw. All White can do is move the king, and all that Black needs to do
is to keep the rook safe on the h -file, or, if necessary, check on the g-file should the white
king reach g6 or beyond.
So can you find the win from the diagram ? Remember, Gelfand could still have levelled
the whole match.
30th May 2012

V.Ana nd-B.Gelfa nd
World Championsh ip, Moscow 2012 (Tie- break game 4)
The final game ended up in a level rook and opposite-coloured bishop ending - no big
mistakes, just a steady draw, and An and retains his title.
There is a temptation to analyse this game, as an illustration that it is still just about
possible for the top grandmasters to conduct a good rapidplay game, since most of us are
not capable of playing with such accuracy in rapid finishes. On the whole, though, it seems
better to concentrate on the Tal Memorial.
The many blunders in the other tie-break games are, of course, painful to watch, and it
is much more interesting to analyse slow play. All the same, it is a valuable exercise to see
how the strongest players can make ghastly mistakes when they are forced to play quickly.
lt is an exercise in chess psychology. Are the mistakes they make at speed of the same type
as lower-level players with more time on the clock?
213
Afterthought
A couple of days after these notes were written, the British Championship had just been
completed - except there was the little matter of sorting out first place, with Gawain Jones
and Stephen Gordon tied on 9/11. Gawain won the play-off - but only after managing, in
one of the games. to lose the queen for not enough pieces, and then Stephen, when run
ning very short of time, made mistakes and lost.
The point is not so much that one player beat the other. Chess is, after all, a competitive
game. Rather, the quality of play has undoubtedly deteriorated through having to negoti
ate very fast time limits. lt would surely be preferable, for the author and for the general
chess public, to see chess being played at the highest possible level.
214
C h a pt e r T h re e
Seventh Tal Memorial,

Moscow 2012
Tal Memoria l, Moscow 2012
1 2 3 4 5 6 7 8 9 0
1 Carlsen Yz 1 Yz Yz Yz Yz Yz 1 Yz 5Yz
2 Caruana Yz Yz 0 0 1 Yz Yz 1 1 5
3 Radjabov 0 Yz Yz Yz Yz Yz Yz 1 1 5
4 Morozevich Yz 1 Yz *
1 Yz 1 0 0 0 4Yz
5 Aronian Yz 1 Yz 0 Yz Yz 1 0 Yz 4Yz
6 Kramnik Yz 0 Yz Yz Yz 1 Yz 0 1 4Yz
7 Grischuk Yz Yz Yz 0 Yz 0 1 1 Yz 4Yz
8 Nakamura Yz Yz Yz 1 0 Yz 0 *
Yz Yz 4
9 McShane 0 0 0 1 1 1 0 Yz *
Yz 4
10 Tomashevsky Yz 0 0 1 Yz 0 Yz Yz Yz *
3Yz
Sometimes, for no obvious reason, a tournament can fizzle out to dullness. This was
certainly not the case with the Tal Memorial. The draws were well fought, with no formal
20-minute handshakes. No player was out of their depth, and no player was able to run
away from the opposition. Play was so tight towards the end that it was unclear who
would actually win the event.
Caruana, the youngest player, was leading with 5/8, but lost badly against Aronian in
the final round. The critical position was at move 19.
215
L.Aronian-F.Ca ruana (round ni ne)

Black to play
Caruana could have snatched a couple of pawns with 19 . ..i.xc3 20 Vlixc3 Vlixe4+ 21 .i.d3
.
�xf4, but then he has exchanged off his fianchetto bishop, and his king looks open to at
tack. Nevertheless, it looks well worth trying, and seems much better than the tame
19 .. .'i&'e8 20 'i&'h4 .i.f6 21 ttJg s with pressure for White.
Try to analyse the position yourself, and remember the tension of the tournament
situation. Would you have taken the two pawns? Would it all have been safe? Would it
have been safe against a renowned attacking player such as Aronian? Could Caruana have
won the tournament? Or was he still under pressure?
Carlsen never dominated this tournament, but his convincing win against McShane in
the final round was enough to come sole first, with two wins and seven draws. Conversely,
McShane, had he beaten Carlsen, could h ave ended up equal first, instead of second from
bottom. However, he was able to win against Aronian, Kramnik and Morozevich, which is
still extremely impressive.
The other three games in the final round were all drawn. I leave it to the reader to de
cide whether Kramnik could or should have achieved more against Morozevich, or whether
Grischuk could have done so against Tomashevsky. The chess engine suggests so, but often
this is merely the starting point of analysis, rather than the conclusion.
The latest strongest gra nd masters

Sometimes the headline stories are more thought-provoking than the actual games, at
least at one level of thinking. And sometimes the picture is as interesting as the chess itself,
again at one level of analysis.
A photograph of the prize-giving shows images of Radjabov, Carlsen and Caruana, all of
undergraduate age, dressed in suits. One recognizes that, at this extremely prestigious
216
S e v e n th Ta l M e m o ria l, M os c o w 2 0 1 2
event, the three youngest players are on top - and that these, perhaps with a few others of
the strongest young grandmasters (Karjakin was not playing, and Giri will have his chance
in the future) will soon start to dominate the chess world. lt is fully appropriate that the
tournament was the Tal Memorial, in recognition that Tal himself had an immense impact
at the start of the 1960s, when he beat Botvinnik to become world champion at the age of
23. Now it seems that there are at least three other players who could have a similar im
pact, compared with just one - Tal - or perhaps Fischer a few years later.
Chess is on the move, and this Tal Memorial in 2012 suggests that we are at the begin
ning of the most exciting period of chess since the original impact of Kasparov in the early
1980s, and his development and consolidation thereafter. For the first time since Kas
parov's retirement, we have the sense that not just one player, Magnus Carlsen, but a se
lect few others too will in time be as strong as Kasparov.
This must surely be well worth celebrating. In the third part of this book, I wanted to go
through some of the strengths, and possible slight weaknesses, of these three young super
super-grandmasters, with a win and a loss by each of them. Of course, Carlsen did not lose
a game at all, so instead we include a tense draw by him.
There is an element of double-counting, in that Carlsen h ad a win against Radjabov,
and therefore Radjabov had a l oss against Carlsen - so an extra game has been included, a
win by Nakamura, who is of a similar sort of strength. This win was against Morozevich,
who had himself looked, in the early stages, to be threatening to run away with the tour
nament - until his loss to Nakamura prefaced two more, against Tomashevsky and
McShane.
Naturally, we should not get carried away with the idea that the younger players were
necessarily the greatest geniuses. What was happening during the Tal Memorial was that
Kramnik, the oldest player in the event, and Morozevich, not necessarily in the most robust
of health, got tired, and made various mistakes. Youth in chess is of enormous significance.
Young players do not get tired so easily - though of course, without excellent chess under
standing, even the most athletic of chess players will not reach the top.
Where does this leave Aronian, the world's second highest rated chess player? He is
seven years younger than Kramnik, but is significantly older than the three leading players
in Moscow. A clue can be taken from the cross-table: Aronian lost twice at the Tal Memo
rial, and also lost a game against Kramnik at Zurich. Three losses in fifteen games, an aver
age of one defeat in five of his games. This compares with Carlsen's recent record of an av
erage of one defeat in over twenty games. lt is difficult, therefore, to see how Aronian will
ever be able to beat Carlsen in a world championship match, unless he can cut down
sharply on his losses.
Round One: Carlsen-Kramnik

Magnus Carlsen has a remarkable ability to avoid defeat, which has allowed him to be
come the top-ranking player in world chess, ahead of Aronian, and further ahead of all
other players, including Anand, the world champion. Carlsen must be regarded as the fa-
21 7
vourite in the next cycle of the world championship. In matches, avoiding defeats is more
important than in tournaments. Carlsen has the ability to win just one game in a match,
and presumably more than that, while not losing any.
Carlsen just does not lose games. He is therefore able to increase his rating, despite hav
ing already reached 2800, by the apparently simple (but not really so simple) method of
playing solidly and avoiding trouble, while still finding wins against the top players.
As an indication, look at his tournament results over the past year and a half, after he
had already reached 2815. Then:
Medias (June 2011): Carlsen (2815) +3 =7; wins against Nakamura, Nisipeanu and lvan
chuk.
Biel (July 2011): Carlsen (2821) +5 =4 -1; wins against Pelletier (twice), Shirov (twice) and
Caruana (once); two draws against Morozevich; and, most impressively, a win and a draw
by Vachier Lagrave.
Sao Pa ulo/Bilbao (September/October 2011): Carlsen (2823) +3 =6 -1; wins against lvan
chuk (twice); win and a loss against Vallejo Pons.
Moscow (November 2011): Carlsen (2826) +2 =7; wins against Gelfand and Nakamura.
London (December 2011): Carlsen (2826) +3 =5, wins against Howell, Nakamura and Ad
ams.
Wijk aan Zee (January 2012): Carlsen (2835) +4 =8 -1, wins against Gashimov, Aronian, Gel
fand and Topalov; loss to Karjakin.
Moscow (June 2012; our featured tournament): Carlsen (2835) +2 =7; wins against Rad
jabov and McShane.
Biel (July/August 2012): Carlsen (2837) +4 =6; wins against Wang Hao (twice) and Bologan
(twice).
And, most recently, Sao Paulo/Bilbao (September/October 2012): Carlsen (2843) +4 = 5 -1;
wins against Vallejo Pons (twice) and An and; win and a loss against Caruana.
Only four losses against ninety grandmaster opponents, all at world-class strength or very
close to it. There was a time when Capablanca was regarded as almost unbeatable, having
lost only 35 games in his career, at about 6%. This seems tiny in comparison. Many of Capab
lanca's opponents were relatively weak, whereas Carlsen's opponents were all strong mod
ern grandmasters, only rarely dropping below 2600, and mostly at 2700+ level. To lose only
one in ten games would be quite an achievement, as in the matches by Aronian and Kram
nik, or Anand and Gelfand. But to lose on average only once in over twenty games, that is
mind-blowing accuracy. In terms of world championship chess, who would be able to find a
single win, plus a string of draws, in a ten-game match against anyone else?
Predictions often have the ability of going vastly astray, but one can feel reasonably con
fident that Carlsen will beat Anand in the next world championship. lt would be unwise to
think too many years ahead. After all, even younger players, rapidly advancing through the
2700s, will themselves have the valuable opportunity of learning from Carlsen, and trying to
218
S e v e n th Ta l M e m orial, M osco w 2 0 1 2
improve on him. Caruana and Giri are the obvious possibilities. They both have the talent to
become world champions, but can they play with extreme and consistent accuracy, move by
move, week by week, month by month? And there will be even younger players advancing
through the ranks. lt will be interesting to see how it all works out.
But now it is time to focus on the early summer (in the northern hemisphere) of 2012.
Specifically, the Tal Memorial of June 2012, a ten-player all-play-all with no back-runners,
every player having the confidence and the chess ability to exploit any tiny mistake by the
opponent. Only Carlsen in this event managed to remain unbeaten. No player scored less
than 31/219. There were no "weakies" here.
Carlsen's chess was not quite as self-assured as it might appear. He was a little rusty
through not having played in the previous five months, and some of his opening play was
too leisurely - impressive if the opponent responds too quietly, but downright dangerous if
the opponent takes the initiative and sacrifices.
In the following game Carlsen got his king stuck in the centre. Many felt that he was
soon in extreme danger. others, including myself, thought that, while he was extremely
close to the edge, Carlsen just about kept things safe with gritty defensive play. There is
often, in really sharp positions, only the smallest of difference between a tense draw and
catastrophe.
8th June 2012

M.Carlsen-V. K ra m n i k
Tal Memorial, Moscow 2012
1 d4 ctJf6 2 c4 e6 3 t2Jc3 .i.b4 4 'i'c2 o-o 5 a3 i.xc3+ 6 �xc3 dS
All this seems very solid for Black, with not many wins for White at the higher levels.
219
White has the bishop pair, but he is in danger of running behind in development, after
having spent time on 'iVc2, a2-a3, 'i'xc3, and almost certainly another queen move later.
7 e3
Varying from the more popular 7 etJf3, followed perhaps 7 ...dxc4 8 �xc4 b6 9 .i.g s .i.a6
10 'iVa4 cs 11 dxcs bxcs. The computer suggests this is slightly better for White, but it usu
ally ends up in a draw - as, for example, in M.Carlsen-V.Kramnik, Dortmund 2009. Carlsen
and Kramnik are, of course, fully entitled to have their own points of view.
7 .i.g s is also played.
7 b6 8 cxds exds
...
Question: What is the most accurate move for White?
9 .i.d3?!
As mentioned above, Carlsen was rusty. His move allows Black to equalize with comfort.
White would do better to delay developing his bishop until it becomes clear what Black is
going to do with his own bishop. 9 etJf3 .i.fs also looks comfortable for Black.
Instead, 9 "irc2 ! ? places the queen on a more flexible diagonal, so that if 9 ... .i.a6 10
..ixa6 tt:Jxa6 11 b4, he can set up the standard queen side minority attack. Or if 9 ... �e7, then
maybe 10 b4 again. Black still has to show that he can equalize.
9 .i.a6 10 .i.xa6 tt:Jxa6 11 �d3 �c8 12 tt:Je2 cs
...
Black is ahead in development, in the sense that he has already castled and all his minor
pieces are off the back rank. He now has to be careful not to slow down. If, for example, he
becomes too concerned about the isolation of his knight on a6, and starts to think about
quiet manoeuvring such as ... c7-c6 and ... etJC7, he will lose the impetus in his position. Like
it or not, he has to be prepared to play ... C7-c5, continuing to open up lines for his pieces,
even if it allows his pawns to be weakened.
220
S e v e n t h Ta l M e m orial, M osco w 2 0 1 2
Question: What should White do in reply?
13 b3
Black is going to be very comfortable if he is given the chance of playing ... c5-c4. There
fore Carlsen prevents that move.
The main alternative would be to exchange pawns with 13 dxc5, which may be the
more accurate move - although it is unlikely, given White's slowness in development, that
he can claim any edge. Then 13 ... bxc5 looks tense, but equal, while 13 ...tt:lxc5 14 't!Vd1 is
similar the main line {after 15 �d1 in the game). The difference here is that White's b
pawn is still on b2. Life is much easier with the pawn on this square. If, for example,
14 ... 'ii'a6, White can simply play 15 o-o, as there is no threat of ... tt:lxb3.
13 cxd4
...
221
Question: Could, or should, he have created a passed pawn, with 13 ... c4 - ?
Any real weakness in White's play is through the slowness of his development, not his
pawn structure, which is strong. Black needs to concentrate on generating as much activity
for his pieces as possible.
The passed pawn after 13 ... c4 14 bxc4 is not particularly strong, either with the queen
exchange (14 .. .'�xc4 15 'ifxc4 dxc4) or without (14 ... dxc4). Remember that White, too, will
have a passed pawn, and a fully centralized and protected one.
Question: Which way should White recapture?
14 'ifxd4
He would like to create pressure on the d-file and, ideally, on the long a1-h8 diagonal.
The naive 14 tt:Jxd4?! would obstruct both lines, giving Black the advantage after 14 ...tt:Jc5,
especially since 15 Wie2 runs into 15 ...tt:Jxb3! 16 tt:Jxb3 'ifc3+, winning a pawn.
The queen recapture keeps play far more open for White. The knight would not have
been on a good square anyway. Better to leave it on e2 and reserve options.
The computer suggests 14 exd4 as offering White a slight edge, but it is difficult to see
why this should be so. lt seems equal after 14 ... l:Ie8 15 o-o tt:Jc?. If anything, with the sym
metrical pawn structure, but with a pawn blockage on d4, Black's knights are more flexible
than White's bishop and knight. The line is playable, but certainly not inspiring for White.
14 tt:Jcs 15 'iVd1 'ifa6
...
Suddenly one senses a buzz on the internet. Isn't Black doing extremely well? I sn't he
just winning? There were plenty who suggested that he was. Even when the game ended
as a draw, many still suggested that somehow Kramnik had missed a win. lt is of course
not quite so simple.
222
S e v e n t h Ta l M e m o ria l, M os c o w 2 0 1 2
Question: So what should Carlsen do next?
16 tbf4?
Over-sharp. By covering the d3-square Carlsen aims to give away nothing to his oppo
nent. If Black attacks, then by all means White will fight back. If Black does nothing (which
is unlikely) White will start to attack himself with b3-b4, aiming to capitalize on his supe
rior pawn structure once his pieces are in safety.
After having gone through the Anand-Gelfand match, one feels that Gelfand in particu
lar would quickly have given away a pawn, so that the rest of White's position remains ab
solutely secure, with not even the slightest chance of attack by the opponent. This would
occur after 16 0-0! tbxb3 17 �xb3 'it'xe2 18 ..ib2 ir'c4 19 ir'd1, when Black's extra pawn is
isolated and needs protection, while White's dark-squared bishop has easily the better mi
nor piece.
223
Question: What are the drawing lines after each of 16 ... gs, 16 ... d4, and
16 . ..li:Je6 - ? Or is it possible that one of these may end up with an advantage?
Sometimes when play is wild and tactical, if one of the players in a sharp position goes
wrong, he can go decisively wrong, without any obvious later error.
16 ...d4!
Kramnik's instincts are to go for the win, and rightly so.
The simplest solution, though certainly not a winning try, is 16 . ./i:Je6 17 tt:Je2 (17 tt:Jxe6
fxe6 or 17 �d3 'i¥xd3 18 tt:Jxd3 d4 offer White nothing either) 17 ...tt:Jcs {he does not want to
allow White to escape with easy castling) 18 tt'lf4 tt:Je6 with a draw by repetition.
A more complicated drawing line follows after 16 ... gs 17 .itb2 ! gxf4 (or 17 ...tt:Jfe4 18 h4!,
since 18 ... gxf4?? now loses to 19 'i¥g4+ and mates) 18 .itxf6 tt'ld3+ 19 'it>f1 (not 19 'it>d2??
tt:Jxf2 with decisive threats, including ...tt'le4+ and ...tt:Jxf6) 19 ...tt'lc1+ 20 'it>e1 (and not 20
'it>g1?? tt'le2+ 2 1 'it>f1 tt'lc3+, winning the queen) 20 ...tt'ld3+ 2 1 'it>f1. Many would regard this
as unthematic, as it is Black who is scraping for the draw, rather than forcing White to
chase the half-point. All the same, it is still a draw.
Instead, Kramnik went all-out for attack - the correct decision - but he missed some
thing on the next move.
17 b4
The Tal Memoria/ 2012 Quiz Question: Find the win for Black - possibly.
17 dxe3?!
...
Tal would have enjoyed this. Both Kramnik and Carlsen missed the tactic, but one can
feel reasonably sure that Tal would have seen the idea instantly.
If one piece is under attack, then do not despair, try sacrificing the other! The point is
224
that the opponent can only take one piece at a time. For instance, if the opponent takes the
attacker's first piece (here, the knight on cs), the second piece might achieve its objective
with gain of tempo. Or if the second piece is taken, a defender might be deflected so that it
can no longer cover a critical square.
This would happen after 17 .. .<�:JdS!.
Tal magic! One knight is attacked, but the other knight becomes the attacker.
Of course, even Tal's magic cannot force a win from a level position, and to win a game
would be possible only if the opponent makes at least one mistake. Also, controversially,
Tal was not all that concerned even as to whether his magic might leave himself in a worse
position; all that he was concerned with was to induce the opponent to make the occa
sional, and fatal, mistake.
The knight sacrifice here is sound, so Black will not end up in a worse position, whereas
even the tiniest mistake by White will end up with a speedy loss; but it seems, after much
analysis, and quite a few prods by my editor, that White can just about hol d the draw. The
game ended up with a draw anyway, but Carlsen was not forced to have to find the very
best moves. lt is a shame that Kramnik, a great World Champion, did not play like Tal here,
another great World Champion. What a great finish it would have been !
Analysing the moves themselves, 18 bxcs? is the less complicated line but is not good.
After 18 .. .<�:Jxf4, Black barges through with his other knight. White has to capture - there is
nothing else - but after 19 exf4 .l::!.fe8+ 20 'it>d2 bxcs, it is not difficult to see that his king is
in trouble, with his pieces undeveloped. If 2 1 .i.b2, then 2 1....l::!.ab8 22 .l::!.b 1 (or 22 'it>c1 .l:te2)
22 .. .'�c4 (threatening 23 ....l::!.xb2+ 24 .l:txb2 'iWc3 mate) 23 'ilff3 Vi'a2 24 'it>c1 c4 wins easily.
The other capture, 18 tt:Jxds, followed by 18 ... tt:ld3+ 19 'it>d2, is critical.
225
No player could hope to analyse in depth all the main lines over the board; even at
home, with the help of several evenings and the use of the computer, this is very difficult.
My own instincts were to try 19 ... .l:tad8 first, putting immense pressure on the d-file, but
quite possibly this is not quite enough, the likely result after accurate play being a draw.
There are several endgames which need to be considered, mainly with asymmetrical piece
play. 19 ..JUd8 is an attempted refinement, less natural, but it frees the f8-square for the
king should White try 0Je7+. Finally, on a different strand of thought, Black can consider
19 ... dxe3+ 20 fxe3 0Jxb4, which we can regard as unclear, for the moment.
Flesh and blood humans cannot process all lines simultaneously, and my first choice
was to see whether Black can claim a clear edge after 19 .l:tad8. Maybe we can look to see
...
whether it is possible to find a big advantage with the exchange on e3, but then, after this
quick scan, it is back to what seems to be the critical line.
Trying 19 .l:tad8 first:
...
226
S e v e n t h Ta l M e m o ri a l, M os c o w 2 0 1 2
This i s very sharp - but with White being underdeveloped and his king out in the open,
and Black having open lines against the king, it would hardly be surprising if White were in
trouble. There seems to be just one way to try to hold the position: 20 e4!.
Before examining 20 e4 more closely, there are other moves which need to be consid
ered:
a) 20 'iikb 3 tt::lxf2 21 'Lle7+ �h8 22 bS 'ilkh7 23 tt::lc 6 tt::lxh1 24 exd4
This is given as the first choice in the computer analysis. White's king is somewhat safer
and his pieces are going to good squares {after ..tb2); unfortunately he is now down in ma
terial. To get his pieces working, he has to let the rook drop. Black is close to winning.
b) 20 'Lle7+ �h8 is the next line up.
White h as saved his knight with tempo, but what is he going to do next? Black is about
to set up a decisive attack against the king. For example, 2 1 'iikf3 dxe3+ 2 2 'it>c3 {if 2 2 'it>xe3,
then, among others, 22 ...tt::lx c1) 22 .. Jk8+ 23 tt::lx c8 .l'1xc8+ 24 'it>d4 'iikc4+ 2 5 �e3 .l'1e8+ 26
22 7
'it>d2 l1d8 27 'it>e3 tZ:le1 28 l1xe1 l1d3+ 29 'it>e2 l1d4+ 30 'it>e3 'iVd3 mate is a good old
fashioned king hunt.
White has to play 21 e4! tZ:lxf2 22 'iVe2 'iVxe2+ 23 'it>xe2 tZ:lxh1 24 .id2, which transposes
to critical 20 e4 variation below. Instead, 22 �fl tZ:lxe4+ is hardly worth analysing in
greater depth. Black is obviously winning. White's pieces are undeveloped or completely
out of place, while Black has two extra pawns for the piece and everything helps set up a
attack.
c) 20 tZ:lf6+ gxf6 21 'ir'g4+ \t>h8 is good, if it works. If it does not work, he has merely re
turned the piece for nothing.
Unfortunately for White, it does not work. Black is threatening a killer check with
... d4xe3, so 22 e4 tZ:le5 23 'iVh3 (White must cover d3} 23 .. ."�c4 24 f4, but then the pawn
crashes through with 24 ... d3 2 5 fxe5 '¥Yc2+ 26 We3 d2 and Black wins.
We are starting to reach the minor variations. Briefly:
d) 20 '¥ff3 dxe3+ 21 fxe3 tZ:le5 22 'iVe4 '¥Vb5 recovers the knight, with a clear positional
win.
e) 20 '¥fe2 dxe3+ and White's position collapses.
f) 20 'iVf1 dxe3+ 21 tZ:lxe3 tZ:lxb4+ 22 'it>e1 (22 'it>c3 l:Id3+ soon wins for Black} 22 ...tZ:ld3+ 2 3
'it>d1 tZ:lxf2+ 24 'it>e1 tZ:ld3+ 2 5 'it>d2, and after the knight has eaten a couple of pawns, Black
quite simply adds another rook to the attack with 25 ...l1fe8.
Now for the main line:
g) 20 e4! tZ:lxf2
228
S e ve n th Ta l M e m orial, M osco w 2 0 1 2
Question: Where should the white queen go?
Very difficult - a quiz within a quiz. The basic choice is between 21 iif3 and 21 1'ke2. As a
practical player, whether with Black or with White, I would want to concentrate my attention
on the sharp tactical lines with 21 'ii'f3 , not least because White is attempting to refute
Black's play, and an attempted refutation forces the players to think much harder than an
attempt to hold the game. lt is, of course, quite possible that the quieter option is more effec
tive than the aggressive one. My editor suggests that the quieter attempt is in fact better.
g1) I give the analysis unchanged on the line with 21 �f3, except that this move no
longer gets an exclamation mark. After 2 1 ...lZ:lxh1 2 2 i.b2, White, by sacrificing the rook,
has given himself time to get his pieces developed and active, and his king seems safer. He
is currently material down, but Black's knight is in danger of dropping, after which White
will be material ahead.
229
The computer gives this as equal, but dig a little further and Black can set up a second
run of the attack with 22 .. .f5 ! 23 tt:Je7+ 'it>h8 24 .l:Ixhl fxe4 25 'ii'xe4.
lt is clear that White's king is no longer so safe, and he has problems getting the bishop
and knight working together.
Here the obvious 25 ... l:tf2?! 26 'it>e1 .l:i.xb2 ends up in a quick draw after 27 tt:Jg6+ hxg6 28
'iYh4+ Wg8 29 'ii'x d8+. But Black has an impressive queen zigzag with 2 5 ...'/Wc4! (now
White's bishop is out of play) 26 'iWh4 'ii'f7! 27 .l:!.e1 (after 27 .i.xd4 �a2+ White's king is far
too exposed) 27 ... �f2+ 28 �xf2 .l:!.xf2+ 29 .l:!.e2 .l:!.xe2+ 30 c.t>xe2 .l:Ie8 and White loses a piece
before he can properly reach the endgame.
So is Black genuinely winning? Let's have another try:
g2) 2 1 'ii'e 2 forces a queen exchange much earlier, with 2 1 ...�xe2+ 22 'it>xe2 tt:Jxhl 2 3
tt:Je7+ Wh8 2 4 .i.d2 .l:tde8 25 tt:Jfs .l:!.xe4+ 26 'it>f3 .l:i.e5
I spent much time analysing this position at home, concentrating on 27 tt:Jxd4 l:td5, and
230
S e v e n t h Tal M e m o ri a l, M os c o w 2 0 1 2
now either 2 8 i.e3 or 2 8 i.C3, trying to see what h appens if Black wins the two minor
pieces for the rook while White wins the h i-knight or, alternatively, if after 28 i.C3 he in
serts 28 .. Jk8.
This is an interesting exercise, but White can improve with 27 g4!.
The computer, or at least the one I am using, seems to think that Black has a straight
forward clear advantage, but in fact it is far from clear-cut. If 27 ... h s , White simply blocks
with 28 h 3, and any exchange of pawns on g4 favours White since the h-file has been
opened. If Black tries an early ... g7-g6 instead, the dark-squared long diagonal could be
come seriously weakened; for instance, 27 ... g6 28 lZ:lxd4 l:tds 29 l:txh1 .l:Ixd4?? loses to 30
i.c3. Quieter moves seem to give no inroads to White's defences.
So time to consider something else, and 19 .. JUd8! ?.
This is not something a player would think of spontaneously, since it seems far more
natural to bring the rooks to the d- and e-files. We have already seen, however, that if the
231
other rook moves from a8 to d8, White keeps the possibility of tDe7+, after which Black has
to move the king to h8. Black is trying to gain a tempo by playing 19 ...l:!.fd8.
lt is easy enough to demonstrate that 20 tDc7? is senseless after 20... dxe3+ 21 �xe3
�c4, when Black is winning. So the knight fork is nothing to be scared of, even if it would
be more "natural", in one's head, to eliminate such danger.
The problem here is not the elementary tactic, but rather that the kingside has been
weakened by moving this rook to the centre. In all but one replies, 19 ... l:!.fd8 is at least as
good as, or even better than, moving the a8-rook. lt is not worth repeating all the above
analysis we have just done with 19 ... l:!.ad8.
The best reply for White after 19 ...l:!.fd8 is 20 e4 tlJf2 21 �f3 ! (much more dangerous for
Black this time) 2 1 ...tt:Jxh 1 22 i..b 2.
Black no longer has the saving move 2 2 .. .fs, as his rook on f8 has already gone. He could
perhaps try 22 ...�C4, but after 23 l:i.cl 'ifa2 24 .l:i.c2 m1 25 i..xd4, Black's pieces are comi
cally decentralized, and he has no protection from the white queen on the king side, with
2S .. .f6 merely inviting 26 l:!.c7. Maybe Black would want to try sacrificing the exchange on
dS, but it is unlikely that his position will hold.
Therefore 19 ...l:[fd8? does not work - White has an almost certain win. 19 ...l:!.ad8! ? gives
Black at least a draw, but it seems likely that, if White plays with great accuracy, Black does
not have more than that. So, equal, so far.
The analysis is not yet complete though. There is a third option: 19 ... dxe3+!? 20 fxe3
iLJxb4!
232
S e v e n t h Ta l M e m orial, M os c o w 2 0 1 2
This i s a computer suggestion, and it almost instantly was advocated on the Internet.
My own instinct was that it looked slightly speculative, and I felt that I wanted to bring a
rook into play as quickly as possible, so I did not try to analyse this in depth. lt would, I am
sure, be a good practical try. After all, even a Tal would not be able to analyse it in complete
depth over the board, and with a time limit. Nevertheless, with the luxury of home analy
sis, there is the obligation to tie up the loose ends, and this is a big loose end. Could it be
that Kramnik had missed a win against Carlsen ? Or is it, after all, only a draw?
There is also another time limit question to be considered. Neither I, nor my publisher,
would want the analysis to drag on through 2013, so on occasion "probably" might be a
legitimate answer.
In response, 21 axb4 �xa1, sacrificing the exchange, is reasonable, but White is still
under considerable pressure. His king is still exposed, and his kingside pawn structure is
damaged, so that Black has good outlets with the queen, for example ...iVes and ...'ii'e4, or
on the seventh rank via ...�a2. Black can also create difficulties on the d-file, and he now
has the option of setting up a passed pawn with ... a7-aS. This is something that might be
considered for White, but only in desperation.
A better defensive try would be to give up the queen with 21 t2Jxb4 .l:i.fd8+ 22 �c3 .l:i.ac8+
23 �b2 .l:i.xd1 24 .l:i.xd1.
233
White is slightly ahead in material (R+B+N vs. Q+P), but his pieces are badly coordi
nated, and Black can start to prod away with his queen side pawns by ... a7-aS. The only real
istic chance for White to defend is to try to set up a passive barricade with the pieces and
avoid losing any king side pawns.
Black now tries 24.. .'iie 2+ 25 .l:!.d2 'Wbs. On 26 .l:!.dS, he could offer a repetition with
26 .. .'ii' e 2+?!, but he should not be happy with this. Instead, 26 ...'�f1 aim s to gain a free
move after 27 .l:td2. White could try 27 .i.d2, sacrificing the g-pawn to develop his pieces,
but after 27 .. .'�'xg2 Black is much better.
A more interesting attempt is 26 'it>a2! as 27 lt:Jc2 'iYc4+ 28 'iii>b l. At last, White's king has
found his way into the castle. What is unusual is that it is White's rooks and minor pieces
which guard the castle, rather than the pawns. We can complete the defensive line with
28 ...�3+ 29 ..ib2 bs 30 .Ua2.
Now it is up to you, reader, to find a win or a clear advantage for Black. White's pieces
234
S e ve n th Ta l M e m orial, M os c o w 2 0 1 2
are stuck - the only possible try, at the moment, is to wiggle his rook between a1 and a2,
and perhaps a pawn push with h2-h3, but these pawn pushes cannot last indefinitely. My
positional sense is that Black is clearly better, and that if he can find a good way to advance
his own pawns, he should be winning. Still, I could be wrong. One point to be noted is that
if Black pushes his pawns too hastily, with .. .f7-f5 and ... g7-g5, White can try to spring back
with tLld4 (aiming for tLlfs), and if ...'ii'x e3 then .l::i.e 2.
Question: Can you find a winning plan for Black? (open ended question)
This whole game is extremely complicated. lt is pure tactical chess, except that after all
the tactics, the players need to think about the positional stuff, once things have simpli
fied. After 19 ...1::i.ad8, for example, the players will need to think about possible rook and
pawn endings, or rook versus minor piece endings, perhaps with an extra rook on each
side; or possibly, after 19 ... dxe3+ 20 exd3 tLlxb4, some complicated zugzwanging attempts
in the late middlegame.
"Tactical chess" does not always mean someone making a sacrifice with a forced
checkmate to follow a few moves later. All that is mostly textbook stuff - even if, in a game
not long afterwards, both Carlsen and Aronian (who should have won) missed such a com
bination, involving a rook and queen sacrifice and leading to checkmate.
Tactical chess, at a more sophisticated level, is a battle between two players, with nei
ther player ever being sure of what is going on, but having to calculate move by move -
and keep finding the best moves, often the m ost imaginative moves, while trying to avoid
serious tactical or positional blunders.
The "Tal magic", as commemorated in the Tal Memorial, may or may not force the win
in the end, but Kramnik could have pressed hard, and it would have been extremely inter
esting to see how Carlsen, the strongest and most solid player of his time, would have han
dled this.
Tal himself was not all that concerned whether all his games were analytically sound.
He wanted, given the opportunity, to play scarily wild chess, to see how his opponent
would react. Often it was just a bonus if his position, when he initiated the tactics, was
position ally and/or analytically better. He just wanted to outplay his opponents.
In most aspects of chess at the highest level, play has improved considerably since Tal's
day lt is not so clear, though, that there has been much of an improvement in totally wild
tactical chess. In this particular game, both Kramnik and Carlsen appear to have fallen
short of what Tal at his best was available to achieve. lt is also far from clear whether the
top modern players could have reached the strength of Kasparov either, when play got
wild.
Back to the game itself, and a comparatively disappointing quick draw by repetition.
18 bxcs i:Vas+
235
19 �f1
Question: Could Carlsen have tried for an edge himself with 19 �e2 - ?
The computer suggests that White is winning, but how much do
you believe in computer assessments? Have a long hard think.
Remember, too, that this is only analysis. For Carlsen and Kramnik, this was "real life"
and neither player would want to jeopardize their game. Carlsen had good instincts here in
aiming for a clear draw. 19 �e2 .:tad8 20 Vi'b3 (to offer a queen exchange with �4) would
have been far more speculative.
Then any good tactician would immediately consider 20.. J�d2+ 21 �xe3, when White's
king is fully exposed to the attacks of queen, two rooks and a knight. The trouble is that,
given the chance, White can reach relative safety with 22 �f3. For instance, 21 ...£i.e8+ 22
236
S e ve n th Ta l M e m o ri a l, M os c o w 2 0 1 2
'.t>f3 'i!Vxc5 ! 2 3 .i.xd2 �c6+ 24 tt:Jd 5 ! (the only move) 24...tt:Jxd5 2 5 '.t>g3 (or 2 5 llae1 lZ:lc3+ etc)
25 .. .'ii' g 6+ 26 '>t>h 3 'iWh 5+ is only a draw.
To go for checkmate, Black would need to sacrifice further with 21...'�'xc5+ 2 2 Wxd2, but
then Black, a rook and a bishop down, finds that he is running out of attacking chances.
Neither 22 ...�xf2+ 23 'it>d1 l:td8+ 24 lZ:ld3, nor 22 ...lZ:le4+ 23 �d3 lZ:lxf2+ 24 �e2 lZ:lxh1 2 5
�e3 will save the game. A nice try, but it has t o be rejected.
Black needs to attack position ally, rather than by direct tactics. He needs to recognize, in
his analysis, that he has good compensation for his bishop, with more active pieces, a vul
nerable white king, and of course a pawn count in his favour. In such positions, the de
fender, with the extra piece, will often find it difficult to retain equality, and even more
difficult to find an advantage.
So, quite simply, 20 ... exf2! and see what White does next.
A quick scan through the computer might suggest a possible draw after 21 .i.e3 bxc5 22
l:thd1 .i:[xd1 2 3 llxd1 J:Ie8 24 h3 (to prevent ...lZ:lg4) 24...g 5 2 5 lLld5 lZ:lxd5 26 'i!Nxd5 llxe3+ 27
'it>xf2 lle8 28 �xg 5+ with a perpetual, but Black can improve on this with 22 ...llfe8! 2 3
llxd8 (or 2 3 h 3 c4! 24 'ii'xc4 �e5) 2 3 . . .'i¥xd8! 24 lld1 'fie? 2 5 h3 g 5 and wins.
There are, of course, many possible subvariations which might be considered, but none
that appear to be a clear improvement for White. In which case taking a draw for White by
Carlsen seems a fair enough result. He did not, it seems, have an advantage after 19 'it>e2,
but would quite possibly have lost.
19 J:Iad8
...
237
Question: Where should the queen go?
20 �c2!
After a lot of thought. 20 'ii'f3 ?? �c3 would be disastrous; while 20 'i¥e2?!, covering any
threats on the a6-f1 diagonal, is risky after 20 ...exf2, as Black is setting up threats on the e
file instead. The best that could be said for White is that if he plays extremely well, he
might have chances to hold the position, which is hardly enticing. For example, 21 �xf2
.l:Ue8 22 'i¥c2 g S ! (far more accurate than 22 ...tt:le4+?! 23 '>t>f3} 23 tt:lh 3 g4 or 23 ....l::td s and
Black is attacking.
20 .'�b5+
..
There is nothing better.

21 'it>e1
lt has to be the king move. 21 �e2?? .l::td 1 mate, or 2 1 tt:le2?? �c4 2 2 �xc4 .l::td 1 mate,
would have been catastrophic.
21 ...'ii'a s+ 22 'it'f1 'ii' b s+ 2 3 �e1
Again the only move. 23 '>t>g 1?? �xcs wins for Black.
23 .'il'a5+ Yz-Yz
..
A remarkable game, and it might have been even more remarkable if Kramnik had
found 17 ...tt:Jds !.
As an indication of the overall evenness of the leading grandmasters in this tourna

ment, had Kramnik beaten Carlsen in the first round, and all other games remained the
same, there would have been a tie for first between Carlsen, Caruana, Radjabov and Kram
nik. Or perhaps if Radjabov had not lost concentration against Carlsen, or if Caruana had
made more steadily against Morozevich, or if Morozevich had avoided three losses in a
TOW ...
238
S e v e n t h Ta l M e m o r i a l, M os c o w 2 0 1 2
There are so many ifs and buts in chess. Few players can ever fully achieve the ideal of
never having a losing position, and never giving away a clear advantage. Not even, it
seems, Carlsen. Or at least, not yet.
Round One: Radjabov-Tomashevsky

A second game from round one, and a strange and memorable finish. Tomashevsky gets
highly centralized with his pieces - his queen and two knights - but they unexpectedly be
come over-centralized, ending up zugzwanging themselves. The two knights look good, but
if they stand a knight's move apart, they are potentially vulnerable, each h aving to defend
the other. In this game, the forward knight was both under pressure and pinned against
the queen, which was also tied to its defence, rendering all three centralized pieces immo
bile. Remarkable stuff, but surely Tomashevsky could have seen all this much earlier?
In truth, thi s was not one of Radjabov's greatest wins technically, since he allowed Black
to equalize far too easily in the opening, and so had to rely on a big misjudgment by his
opponent late on. However, when Tomashevsky went wrong, Radjabov pounced for a win
with remarkable speed.
8t h J une 2012
T. Radja bov-E. Toma shevsky
Scotch Came
1 e 4 es 2 tt:lf3 tt:lc6 3 d 4 exd4 4 tt:lxd4

The Scotch Opening, which has had a revival recently. The point is not so much that
White is genuinely much better. lt is more a case that the main lines of the Ruy Lopez do
not appear these days to give all that much either.
4...i.cs 5 tt:lb3 .tb6 6 tt:lc3 d6
239
Fig h ti n g Chess: M o ve by M o ve
7 'i¥e2
Question: Why this particular m ove? You are not being asked whether this is
better, or otherwise, than a few alternatives. The question is rather, what
White is aiming for, and what chance he has of trying for an opening edge.
Over the last few moves, White's pieces have been increasingly directed to queen side
play, and the natural continuation of the plan, arguably, is to get all his queen side pieces
developed, and then to castle long. The problem is what to do with his dark-squared
bishop. The natural square is e3, but 7 �e3 ? �xe3 8 fxe3 leaves White's pawns unnecessar
ily doubled and isolated. Therefore White needs to find a good safe square for his queen, to
cover the e3-square, and the logical solution is on e2. The other bishop is blocked, but that
can be sorted out later. The queenside has priority.
lt has to be admitted that Radjabov has not achieved much from his opening, and that
other lines should be considered. 7 �bs, in Ruy style, is possible, but again it does not seem
to achieve much after 7 . .'i¥h4. Alternatively, 7 tt:lds is certainly playable, gaining the bishop
.
pair after tt:lxb6, but remember that White has invested a lot of moves with his knight, just
to take the bishop. Play is probably about equal.
7 ...tt:Jge7 8 �e3 o-o 9 o-o-o fs
Position ally, the simplest way to equalize. He removes the central white pawn with a
swipe on the edges, Black's pieces are as mobile as White's, and neither side has any sig
nificant pawn weaknesses.
10 exfs �xfs
lO ... tt:JxfS seems slightly more direct. Equal.
240
S e v e n th Ta l M e m o ri a l, M os c o w 2 0 1 2
Question: Any chance of a microscopic edge?
11 h3
This looks reasonable, with the idea of g2-g4 and iL.g2. After Tomashevsky's reply, Rad
jabov thinks again, and decides not to continue with this plan.
Instead, 11 g3, followed by .i.g2, might still give chances of an edge. Or on second
thoughts, probably not. Black is equal, and whatever White tries, he is not going to achieve
anything, if Black continues to play accurately.
11 .i.d7
•••
Prophylactic chess. White has used up a move trying to make g2-g4 more effective, so
Black can use one up in return, and preferably a m ore constructive move, to make White's
h 2-h 3 ineffective.
11 . .tbg6 is possible, when his position is still playable, but it only makes total sense if
.
White has played, or is fully committed to, g2-g4.

12 �d2
Radjabov decides that 12 g4 lt:Jg6 is not quite satisfactory, as White would have weak
ened his dark squares, on f4 and e s.
12 ....i.xe3
Tomashevsky is happy enough to gain an almost insignificant tempo. Possibly 12 ....i.e6,
now that White no longer has the queen on the e-file, might have been very slightly better.
Or possibly not.
lt is almost inevitable that there will be much shuffling of pieces, when there is no clear
weakness on either side, and both sets of pieces are in play, but are not dominant. Either
there will be a steady draw at some stage, or one of the players will slightly loosen their
grip, allowing the opponent an edge.
13 �xe3 Wh8 14 iL.d3 �e8
241
15 f4
Question: What do you think of this move?
White has to be careful here, in that he is slightly weakening his pawn structure. His f4-
pawn could end up under attack, and it no longer protects the e3-square, which could be
significant if Black were to play ... tt::lfs and put pressure on the e-file. Having said that, if he
takes reasonable precautions, his position will be playable, and he has gained some
squares in the centre while preventing Black from playing ... tt::le s. Also, he might get his
other kingside pawns moving.
If White wanted to play something boringly equal, he could try 15 Ii.he1 'iff7 16 f3 Ii.ae8
17 .i.e4 .i.e6 18 �b1, and it is difficult for either side to m ake progress. Radjabov wants to
try something, and one cannot dispute the fact that he was successful in winning the
game, though Black's mistakes came later.
1S ...'iff7 16 .l:!.hf1 .l:!.ae8 17 'ii'd 2 tt::l b4
Indeed, Black is starting to look slightly better.
18 .i.e4
242
S e ve n th Ta l M e m o ria l, M os c o w 2 0 1 2
Question: To play for an edge? Or to keep steady

equality? And which move should he try?
18 i.c6
...
This simplifies and still keeps the position steady. He might have played for a slight
edge, without simplification, by 18 .. .'�Jec6! ? (threatening ...4Ja2+), and if 19 'iii>b 1 then
19 .....te6 with comfort. White could throw in 19 kxc6 tt:Jxc6, but then Black's pieces are
again clearly the more coordinated. In hindsight, White's 14 f4 seems premature.
19 :del ..txe4 20 tt:Jxe4 'i'c4 21 a3 tt:Jbc6 22 'ti'c3
243
22 .. Ji'd S?!
lt is hard to believe that Black was soon going to be in zugzwang in an open position.
Tomashevsky quite simply lost his sense of danger. However, there is always a potential
risk if the queen is stuck in the centre without a clear means of escape.
The simplest equalizing option would be to exchange queens with 22 .. .'�xc3 2 3 l'Llxc3.
If he wants to keep queens on the board, with the aim of maintaining the initiative,
that is fine, as long as he keeps the queen safe, and with good escape routes. For instance,
22 . .'i!r'bs is comfortable for Black.
.
2 3 l'Llbd2
The position seems much too quiet to set up a trap!
23 ...ll:lfs 24 g4 l'Llfd4
Black seems to be thinking that he is taking the initiative.
25 'ii'd 3
Question: What now?
2s ... bs?!
Closer to zugzwang. Black's knight on d4 is pinned, while the other knight and the
queen are finding it difficult to escape. After this, Black's battle is for survival. lt is no
l onger realistic for him to try to for anything more.
Where is the queen going to go, first of all ? An initial try might be 25 .. .'�a2!?, but then
26 l'Llg 5 is troublesome. Black can try 26 .. .'�'g8, but it is not all that clear why Black should
want to lure the knight to gS. White's knight is, after all, still on a good attacking square.
The next train of thought is to try 2 5 ...�g8! at once. This is far from obvious, but it
works. The queen is away from attack, and the king is also safe from the attacks on h7.
Meanwhile, the rooks and knights can gradually get coordinated. Black is not yet fully safe,
but he has made progress.
244
26 '.t>b1
A step closer to zugzwang. The king blocks any escape square at a2. Also, do not forget,
the knight on d4 is pinned, making c2-c3 is a winning threat, as Black no longer has
... ctJb3+.
26 ... b4
He has prevented c2-c3, but there are still a few other problems.
2 7 a4
27 ... h6?
Even here, Black can still play 27 ...'ii'g 8. In comparison with the line a couple of moves
earlier, he has weakened his queen side with ...b7-b5-b4, but this is not necessarily fatal.
28 liJb3
Question: Black is by now already on the slide, but he has one

last chance. Can he find a tactic which avoids immediate collapse?
2 45
28 ....l::!.e 7?
28 ....l::!.e 6 offers more resistance, since if then 29 ltJed2, Black can hit back with 29 ....l::!.xf4.
However, 29 .l:!.e3 .l::!.fe8 30 .l::!.fe1 as 31 h4 is a straightforward positional win. Black's queen
cannot move, and neither can his knights. His rooks are stuck between three squares on
the e-file, and his king and pawns can do nothing active. All White has to do to win is to roll
the king side pawns carefully.
29 ltJed2 .l::!.xe1+ 30 .l::!.x e1
Zugzwang - where the zugzwanged pieces have, at first sight, taken over all the main
central squares. Perhaps someone would object that this is not really a true zugzwang, as
Black can still wriggle around with his king and rook, and White's win will be through a
direct attack with his pieces and king side pawns. Black does not therefore have to give
away material, and White still has to find a definite win (though that should not be too
difficult).
Maybe so, but what would be the correct technical term ? The basic point is that, for a
while, Black's queen and two knights have been immobile, as a result of the partial
zugzwang of three important pieces.
3o...gs 31 fs 1-o
White will win with .l:!.e4, when Black's knight will drop.
Round One: Morozevich-Caruana

Alexander Morozevich is commonly regarded as one of the most talented of the very top
grandmasters, and yet he has never been consistently able to maintain a rating over 2800.
Somehow he manages to fall just short. For instance, he was leading, a point clear, after
five rounds of the Tal Memorial, but then his play deteriorated dramatically and he only
managed a single draw in the remaining four games. Not long afterwards, his health suf
fered a collapse, hopefully minor, at the Biel tournament, and he had to retire early on, al
lowing another grandmaster (Bologan) to take part in the remaining rounds. Chess is,
246
S e v e n th Ta l M e m orial, M os c o w 2 0 1 2
among other things, a sport, and his physical strength seems not be at the required level
for him to reach the very top.
Caruana was, at the time of this game, still a teenager, albeit an exceptionally talented
one. In his game against Morozevich, he went badly wrong in the opening. A couple of
years of further experience would no doubt have papered up many of his gaps, when en
countering very old openings. Here Morozevich surprised his opponent. Caruana was still
able to play imaginatively, with a pawn sacrifice on the a-file, just to keep his position
alive, and allow his pieces to take part in the game. He managed to reach the time control
with reasonable chances, but then missed a sacrificial pawn break which could easily have
held the draw.
8th June 2012

A.Morozevich-F .Ca rua na
Ruy Lopez
1 e4 es 2 tt::lf3 tt::lc6 3 i.bs a6 4 ..ta4 tt::lf6 s o-o tt::lxe4 6 d4 bs 7 i.b3 ds 8 dxes i.e6 9 'i!Ve2
Morozevich plays an old-fashioned line and soon reaps the dividends.
9 ..te7 10 .Ud1
...
Question: What should Black play here? Don't bother about trying to look up
the theory. lt is more a case of using one's positional common sense.
1o tt::la s?!
•••
This is the wrong knight to remove the bishop, and Black now has little chance to try to
equalize.
24 7
10...ttJcs, again followed by ...tL'lxb3, is much safer. Then once the exchange has been
made, the remaining knight is on c6, rather than on e4. On the latter square the knight is
vulnerable, and can easily be pushed away with gain of tempo. The knight on c6 would at
least be safe.
10 ... 0-0 is also possible, but then White has 11 c4 bxc4 12 .i.xc4, with the superior pawn
structure.
As mentioned above, Caruana was still a teenager at the time of the tournament, and
inevitably there will be occasional gaps in his knowledge and understanding, which older
players will be able to take advantage of. As he progresses through his twenties, he should
become an even stronger player.
11 c3 tL'lxb3 12 axb3 o-o 13 tL'ld4
Problems are starting to emerge for Black. If he had a knight on c6, rather than on e4, he
could quickly have exchanged on d4; whereas with the knight on e4, he now has to deal
with attacks by f2-f3.
13 'ii'e 8
...
Here, for example, he would quickly get his knight trapped after 13 ... cs 14 tL'lc6 'file7 15
tL'lxe7+ 'it'xe7 16 f3 ttJg s 17 h4.
14 f3 ttJcs 15 b4 tL'lb7
248
Not really the square that Caruana would have wanted for the knight, but h e was soon
able to fight back.
Question: What is the best way for White to handle the position ?
16 ..ie3?!
One senses that, over the next few moves, neither player seemed totally confident in
handling the manoeuvring battle.
Morozevich was clearly itching to get his minor pieces away from their starting squares,
but in doing this so quickly, he is not bringing them to better squares. He starts off with a
long manoeuvring plan - ..ie3, lt:Ja3, tt:Jac2, f3-f4, ..if2, and tt:Je3 - and while all this looks
extremely steady and solid, it is quite slow and gives Black chances to equalize. The critical
point is that e3 is not really such a good square for the bishop. So if he cannot move the
bishop to a good post in one go, perhaps he should delay moving it just yet.
Instead, 16 'i¥e1, or possibly 16 �f2, looks more effective. The idea is 'iVg 3, when the
bishop can move either to h6 or to ..ig s. The exchange of dark-squared bishops would tend
to favour White, leaving him with two good knights versus knight and a rather ineffective
bishop. Meanwhile, there is no hurry to develop the b1-knight so quickly. A few moves
later, he can perhaps move it to d2. A likely steady edge for White.
16 ... i..d 7 17 lt:Ja3
I approved of this at the time, but again it all seems so slow. White is still structurally
slightly better, but only slightly so.
17 ...c6 18 tt:Jac2 as
Black is now close to equal. The key point is that if White is forced to exchange pawns
with b4xas, Black can make use of the knight, pawn, or maybe the bishop on the cs-square.
Caruana has by now recovered his disadvantage.
249
19 f4 'ifc8
The simplest way of achieving equality is 19 ... axb4. Caruana prefers to maintain the
tension, and this first queen move is fully acceptable - but his later queen moves, just to
keep it going, seem far too indulgent.
20 h3 .l:!.e8 21 ..tf2 ..tf8 22 'iVf3 .l:!.a6 23 tt:Je3
Question: The computer suggests 23 a4 as equal .

...
Is there anything wrong with that?
23 .. -'iWas
The problem is that White has so much power in the centre, Black cannot afford to take
any pressure off the pawns. He needs somehow to make his presence felt against the b4-
pawn, so that White is forced to exchange, and then, just possibly, he can create counter-
250
play with ...tt:'lcs and ...tt:'le4.

If he were to try 23 ... a4?!, it would be easy enough to find many small steps to advan
tage for White, and much more difficult for Black to improve his position. A line might con
tinue 24 .i.g3 .i.e7 25 :1f1 .i.d8 26 fs ..tg s 28 tt:'ldc2, followed by ::tael, and Black is slowly
being ground down.
Caruana has either to find something more imaginative, or perhaps something simpler.
For instance, 23 .. .'ife7 looks equal.
24 tt:'ldc2
Question: What should Black play here?
This is very much an open question. Quite often, if the play is level and there are no
immediate tactics on either side, what is important is to give nothing away to the oppo
nent. Eventually, either one player will make a concession, or perhaps the position will re
main deadlocked indefinitely, until perhaps the players start to get bored and agree a
draw.
The tension between White's pawn on b4 and Black's pawn on as is still critical.
24 ... axb4? would be a big mistake, in view of 25 .l:i.xa6 �xa6 26 .l:i.a1 1Wb6 27 tt:'lxds. How
ever, Black can try 24 ... a4 this time, now that the white knight is not on d4. White would
have conceded a couple of tempi if he were to return with 25 tt:'ld4, and Black could con
sider 2S ...tt:'ld8 to bring his own knight back into play. lt is only equal now.
Of quieter moves, it is probably better for Black to keep his bishops on their own
squares. Any advance will risk leaving the bishop on a more exposed square. Perhaps the
simplest, if he is happy to take a quick draw is 24...�c8, inviting his opponent to offer a
repetition with 2 5 tt:'ld4 �a8.
24...'i¥d8?
Instead, Caruana tries to squeeze a tiny tempo, preferring the queen on d8, rather than
251
c8. This appears to be an oversight. Black needs to keep the queen covering the rook on a6,
whether from a8 or c8.
2 5 bxas
White finds a good time to break the tension on the b4- and as-pawns. After all, Black
cannot play 2S ... ttJxas? because of 26 b4 and the knight is pinned.
2 s ...ttJcs!?
Caruana immediately recognizes that he has made a mistake, and settles for gambit
play, rather than fear losing in a "the rest is a matter of technique" grind.
The obvious 2S ... .l:!.xas 26 b4 .l::!.xa1 27 .l:!.xa1 is uncomfortable. What, for example, is he
going to do with the knight on b7? The implied answer would be something like 27 .. .'�c7
28 tLld4 tLld8. other players as Black might prefer to try handling the defence here.
26 b4 ttJe4
Whereas now Black has genuine counterplay. True, White is a passed pawn up - an
outside protected passed pawn, often extremely useful in the late middlegame or end
game - but it does not make much of an immediate impact in the earlier part of the mid
dlegame, if the defender has genuinely active pieces. Indeed, the attacker has suddenly
become the defender.
252
S e v e n t h Ta l M e m o r i a l, M osco w 2 0 1 2
Question: What should White play next?
27 I1d3?!
White is still better, but the rook move is ungainly. 27 lLlfl is much more convincing,
covering the c3-pawn, while opening up lines for both the queen (a more economical way
of protecting the pawn since, unlike the rook, the queen can quickly move again) or the
bishop. If 27 . .f6, for example, then 28 ..ltd4 and White should be able to keep a very clear
.
edge.
27 f6
...
Question: What should White do next? lt would appear that Morozevich

missed a tactic when calculating his next move. For advanced players,
what is the problem in one of the lines with 28 exf6 or 28 .th4 - ?
253
28 .lii. h 4?!
28 exf6 lt:ixfs 29 ctJg4 is simple enough, and keeps the advantage without unnecessary
tactics.
2s...gs!
29 fxgs
Quite possibly Morozevich only now started to appreciate that 29 lt:ixd5 cxdS 30 .UxdS
gxh4 31 'i!Vxe4 'i!Va8! is, at best, only equal. There are many sacrifices, pins and pin avoid
ance which need to be considered, and even an excellent tactician will occasionally allow
something to slip through the net.
29 ...fxgs 30 .i.g3
Similarly, 30 lt:ixds cxds 3 1 l1xd5 might look dangerous, but it is White who is in danger
after 31 ... gxh4 32 �xe4 �aS.
30 .lii. g7
•••
And so Caruana has taken over the initiative, although he is still a pawn down. There
will be a tense battle before the next time control.
31 i.h2 .l:!.f8
lt is no longer the case that Black has made a speculative pawn sacrifice under pressure.
Now it can be stated that he has genuine compensation for the pawn. Caruana has kept
the game alive, but he still needs to play accurately.
32 'i'e2
254
Question: What should Black play next?
32 l:!.f2?!
•••
Even as late as after move 30, it is a good idea to complete your development, rather
than trying to start an attack against the king with insufficient pieces. Black's rook did not
stay long on f2.
Here 32 ... 'i¥e7 33 l:!.f1 .l:!.aa8 allows Black to recover the pawn on es. White is still slightly
better, but Black is not losing. A defender cannot always achieve everything.
3 3 'i¥e1 g4 34 l:!.d4 l:!.f8
Whoops! White was threatening l:txe4, and 34 ... g3 is met simply by 3 5 i.xg3.
The players were clearly in time pressure over the next few moves, so it is to be ex
pected that minor errors might have occurred, but nothing disastrous. Caruana concen
trated on keeping his bishop pair, rather than trying to win back the es-pawn. No detailed
comments, until we reach the last move before the time control.
35 .l:!.adl 'i¥e8 36 tt:Jxg4 i.fs 37 tt:Jce3 i.g6 38 tt::lf1 hs 39 tt::lg e3 .l:!.a7
255
Fig h ti n g Chess: M o ve by M o ve
Question: One left minute on the clock, before the extra time
is allotted next move. What would you play here?
40 tbg3?!
White has a sudden urge to simplify, but unnecessarily gives away the exchange.
lt is difficult to formulate a good plan quickly from several possibilities, and I have to
admit that, going through the lines suggested by the computer, I used up far more than
the allowed minute. Black still has the dangerous possibility of ... �af7, so White needs to be
careful.
A different method of simplification, again with an exchange sacrifice, would be 40 c4! ?
bxc4 41 tbxc4 dxc4 42 �xe4 .i.xe4 43 'i!Vxe4, and with two extra pawns for the exchange,
the superior pawn structure, and a strong centralized queen, White is better. Whether that
advantage is enough to win against best play is uncertain. In practical terms, though,
White has given himself a comfortable working edge.
40....i.xes
lt is unlikely that Morozevich would have missed this, which means that the exchange
sacrifice would have been intentional. The problem is that, unlike the 40 c4 variation, he
has not broken up Black's queenside pawn structure.
41 tbxe4 .i.xd4 42 cxd4 'i¥xe4 43 .i.es
256
Question: After a flurry of tactics, it i s time to take one's breath. Who is better?
Or is it equal ? And what is the most sensible continuation for Black?
43 'it>h7?1
..•
This is just about playable, as long as he handles the next few moves with complete ac
curacy, and does not try for a win. He has wasted a tempo, though, since the king is no bet
ter on h7 than on g8.
Instead, 43 .. Jbf7 is far more natural and should give at least a draw. Whether there is
anything more than that is open to question.
White can try 44 l:!.c1, as in the game, when 44...<it>h7 just transposes. If that is Black's
best, it does not really matter which move order he plays it in.
44.. J:tf2 is more direct, but not necessarily better. White has 45 a6, which is a dangerous
distraction. For instance, 4S ....Ub2 46 a7 lta8 47 �c3 l:lxa7 48 'i'xb2 �xe3+ 49 'it>h2 leaves
queen, rook and opposite-coloured bishop on each side, but Black is going to be uncom
fortable, with an open king and an inferior pawn structure.
However, there is still a good move for Black, and respect if you saw it!
He can play 44... c S ! ! - a superb pawn break.
257
Then 45 Irxcs? loses immediately to 45 ...l:i.f1+ 46 �xfl �xe3+, while after 45 bxcs b4
White is also in trouble as, totally unexpectedly, Black now has the more dangerous queen
side passed pawn.
If White can keep his balance, he should be able to find 45 �g3 ! with a likely draw. The
most direct line would be 45 ... cxd4 46 l2Jxds l:i.f2 47 lLle7+ (47 l:i.c6 l:i.xg2+! 48 �xg2 �el+
leads to perpetual check) 47 ...'it>h7 48 l2Jxg6 �xg6 49 �xg6+ 'it>xg6 50 .i.xd4 and a draw is
on the way.
In chess, straightforward attacking play can end up in very difficult tactics, but there is
often more than one good possibility in a critical position. On move 44, positional play
would also h ave held the game.
Try 44 �g3.
Then there is an entertaining snakes-and-ladders draw by repetition after 44...'it>h7 (not

44 ... l:i.f3 ? 45 gxf3 l:i.xf3 because of 46 �g S ! �xe3+ 47 �xe3 l:i.xe3 48 l:i.a1 and White wins)
258
S e v e n t h Ta l M e m o r i a l, M osco w 2 0 1 2
45 �g s .l:!.f2 46 'iYe7+ l:t8f7 47 'ii'd 8 �f8 48 'iYe7+ etc; while if Black tries for more with
47 .. .'it>h 6 ! ?, then 48 l:te1 Itfl+ 49 .:txfl 'ir'xe3+ so �h2 l:txfl 5 1 �8+ 'it'gs 5 2 'ii'd 8+ is an
other draw.
A draw seems to be a fair result.
44 .l::tc 1
Question: An essential point of technique: How does the defender

halve out when he is slightly worse but wants to avoid losing?
In other words, what should Black do here?
44 gaf7?
...
This is the second attempt by Caruana to start an attack with ...�f2 (see move 32), and
again it fails.
He could have tried an even older idea (from m ove 22 and beyond) - blockading the
white a-pawn with 44... l:ta6. He cannot then win, but he should be safe from losing.
lt is, of course, frustrating not to be trying for a win when the exchange up, but the
passed a-pawn, plus the weakness of Black's various queenside pawns suggest that he
should be cautious. As the further play shows, White can defend against anything Black
attempts on the king side.
45 \t>h2 �f2
Caruana's plan, but it is not a good one. His attack on the kingside does not work, and
he has no defence on the queenside. The computer suggests the passive 4S ... .l:tc8, but now
White is better after 46 'ifg3, with the possible idea of 'ikgs. Tempi matter, and Black
should have set up his queenside defences a move earlier.
46 .l:!.xc6
Simple enough. Of the three rooks on the board, it is the sole white rook which is the
most dangerous.
259
46....l:!.Bf3 47 .l:!.c7+ i..f7 48 'it>h1

With a clear defensive win . The direct 48 a6 is also decisive.
Black is overcommitted.
48 ....l:!.f1+
Nothing else is any better. 48 .. .'i\Vxe3 49 .l:!.xf7+ wins a piece; while after 48 . ..l:!.xh3+ 49
.
'it>g1, both black rooks are en prise and one will fall.
49 't\Vxf1 .l:!.xf1+ 50 t2Jxf1
Black will be unable to attack with only his queen. Sooner or later, White's pieces will
work together and the a-pawn will finish the game off.
51 ... �g6 5 1 .l:!.c6+ 'it>h7 52 t2Jg3 't!Vb1+ 53 Wh2 h4 54 t2Je2 't!Vxb4 5 5 1:tc7 'lt>g8 56 a6 1-o
Few people would want to claim that this game was in any sense perfect, as there were
identifiable mistakes on both sides, but it was quite imaginative and entertaining. As for Ca-
260
S e v e n th Ta l M e m o ria l, M os c o w 2 0 1 2
ruana, h e was still a teenager at the time of the Tal Memorial, albeit with a rating of 2770. So
we can say with confidence that, if he can eliminate a few weaknesses in his game, he should
comfortably reach 2800 before too long, and stay at that level for many years.
Round Five: Radjabov-Carlsen

Radjabov has developed into an extremely solid player, if not with quite the same aura of
invincibility as Carlsen. He has lost seven games over the same recent period (including
three times to lvanchuk), rather than just four at the highest level. lt is enough to note that
Radjabov only lost once at the Tal Memorial, and that against Carlsen, while Carlsen went
unbeaten. In this extremely tough event, every other player lost at least twice.
If Radjabov can consolidate his defensive play just a little more, it would be a tense and
difficult match if both Radjabov and Carlsen were ever to play for the world championship.
Just one win by either player could easily be enough, with the loser quite possibly given no
possibility to recover.
So what happened in their game at the Tal Memorial? In previous recent encounters,
both players had ended up with steady draws, but somehow this time Carlsen was able to
find a very small edge as Black, and squeezed, and continued to squeeze, until Radjabov
finally lost his guard and the game. Had Radjabov drawn, it would probably have been he,
not Carlsen, who would have won the tournament.
Carlsen has gained the reputation of being the best positional grinder in history, ex
ceeding even the exploits of Kramnik in his younger years. Radjabov, on this evidence, is
not quite at the same level, but the gap is small.
13th June 2012

T.Radja bov-M.Carlsen
Tal Memo rial, Moscow 2012
Scotch Game
1 e4 es 2 tt:lf3 tt:lc6 3 d4
looking at Radjabov's two games in this book, the impression might easily be that he
opens with 1 e4 all the time, and then, if the opponent replies 1 ... es, the Scotch, which he
had also played against Tomashevsky in round one. Not so. He plays 1 d4 and 1 e4 with
equal regularity, and nowadays very often 1 tt:lf3 as well. lt i s noticeable that he does not
make quite as much of an impact as Carlsen with the white pieces, with fewer wins and
very many draws. This is perhaps something he will need to think about if he aim s for the
very top.
3 ...exd4 4 tt:lxd4 .ics 5 tt:Jxc6
Radjabov had clearly decided that 5 tt:lb3, with which he won against Tomashevsky, was
not that impressive after all, and so he tries a different line.
s ...1i'f6
261
If this were a beginners' book, I would add a diagram here, and ask what would happen
if White were to try to keep the knight with 6 tt:Ja s.
6 'iVf3
Question: The more interesting questions start here, a move later.

First, what happens if Black plays 6 ...'iVxf3 7 gxf3 dxc6, going
tit-for-tat with the doubled pawns? Second, if Black does not
exchange on f3, which is the best of the three recaptures on c6?
6 ... bxc6
According to the computer I am using (and of course there are more recent models),
Carlsen's move was, if anything, slightly the weakest of the four captures. Top grandmas
ters think otherwise. U sually it is best to be guided by the grandmasters, who are likely to
have far better assessments than computers of long-term positional consequences of mi
nor differences in pawn structure.
Carlsen has recaptured with the pawn towards the centre, with excellent defensive cov
erage of the central squares. White has the pawn on e4, but there is not much hope of him
trying to dominate in the centre. lt is far too early to say that Black can possibly be better,
but at least he is solid, and he can hope to improve on that if his opponent gives anything
away.
6 ... dxc6 is a familiar pawn structure from the Berlin Defence to the Ruy Lopez, popular
ized by Kramnik and others. The small but significant difference is that in that, in the
Scotch, White captures on c6 with the knight, rather than the bishop. Therefore, he gets to
keep his light-squared bishop. White can try 7 'iVg 3 0-0 8 ..td3, followed by quick castling,
with the reasonable chance of a slight edge.
6 ...'iVxc6 could be interesting, avoiding doubled pawns altogether. However, this leaves
the queen in a slightly strange position, and moving it again will be with loss of tempo.
2 62
S e ve n th Ta l M e m o r i a l, M os c o w 2 0 1 2
Black still has to prove that h e i s fully equal after 7 t2Jc3.

Finally, there is 6 ...�xf3 7 gxf3 dxc6 (with queens off the board, he can be happier with
the centre open). The problem here is that White can cause trouble on the g-file after 8
.l:!.g 1, with a slight but clear edge.
Carlsen's choice, 6 ...bxc6, seems the safest.
7 �g3
7 tZ:ld2 has also been tried. lt is nothing spectacular.
7 ..d6 8 t2Jc3 'ir'g6
.
Clearly Carlsen considers that he is happy enough with his pawn structure, especially if
the queens are off the board - he remains slightly nervous with them still on the board.
9 .i. d 3
Or 9 .i.d2 �xg3 10 hxg 3 tZ:le7 11 f3 tt:Jg6, about equal, and agreed drawn after 116
moves in Wang Hao-E.Tomashevsky, Russian Team Championship 2012. Be warned. In such
a line, players will try to grind each other down for many, many moves.
9 ...tt:Jf6 10 t2Ja4
Radjabov gains the bishop pair and has claims for a slight edge, but he would need to
play extremely accurately to try and prove it.
10 ... .i.d 4 11 C3 �b6
Question: How best to try to keep a slight edge?
12 o-o?!
Too routine - only castle if you cannot find anything else to do. Carlsen was able to
equalize immediately by exchanging queens, followed by ... t2Jg4. The problem is not so
much that the doubled pawns are themselves weak, but rather that the pawn on g 3 is ob
structing an retreat square for the bishop (after .i.f4) and that, in addition, White cannot
push away a knight on g4 with h2-h3. Such things can make a difference between a small
2 63
edge and mere equality.

Given this, 12 �xg6!? hxg6 13 ..ltf4 would have been a more appropriate way to com
plete his development, and then perhaps 13 ...l2:id7 14 ..ic2 l2:ies 15 b3, still with a slight
edge. 12 i.f4!? is also promising, again with a slight edge.
White has various chances to throw in an exchange on b6, gaining the bishop pair.
There is no hurry. The bishop will not run away. lt is usually preferable not to simplify
when you are slightly better. Keep the tension going instead.
12 ... �xg3 13 hxg3 l2ig4 14 i.f4 f6
Just solidly equal, and an excellent chance for Carlsen to demonstrate his outstanding
grinding skills.
The problem which makes it very difficult for Radjabov to improve his position is that
his kingside pawn structure is immobile. If he still had a pawn on h2, rather than on g3, he
could keep his pawns moving with h2-h3, retreat the bishop from f4, and then arrange
things so that he can eventually play f2-f4.
Instead, all that White can do is tack around, hoping his position is not going to deterio
rate.
15 .l:!.ad1 hs 16 i.e2 i.e6 17 l2ixb6 axb6 18 a3 �e7 19 f3 l2ies
2 64
S e v e n t h Ta l M e m o ria l, M os c o w 2 0 1 2
Black i s already slightly better, and White will need to defend carefully. Black's bishop
and knight work well together and slightly outperform White's bishop pair. If, for example,
White gets frustrated with the knight remaining on es and decides to take it off, Black
would remain with the slightly better bishop.
The pawn structures also need to be considered. Black has a queenside pawn majority
which is mobile. White's kingside pawn majority i s static. The only pawn move on that side
of the board is an f3-f4 push (after the bishop h as moved out of the way), but this would
create a weakness on g4.
White cannot hope for an advantage. Black is able to try.
20 'it>f2 bs 21 i.xes fxes 22 'ite3
22 ... h4
Question: Should Black really allow his opponent to eliminate his doubled g-pawns?
265
lt is a question of priorities. White is now ready to push his pawns forward with f3-f4
and possibly f4-f5, so Black decides to press against the g3-pawn immediately. Matters
have changed in the last few moves in that White has exchanged the black knight, so there
is no longer the possibility of parking the knight on g4.
Black has to be content with opening up the h-file for his rook.
23 gxh4 .l:!.xh4 24 .l:!.hl .l:!.ah8 25 .l:!.xh4 .l:!.xh4
Now there is a long stretch of manoeuvring by Black, taking play until at least the first
time control. He wants to keep his rook on the open h-file for as long as possible. White has
a 3-2 pawn majority on the kingside, but if he cannot find a way to make progress on that
side of the board, Black will start to take the initiative on the queen side. One problem for
White is that, if Black's rook can stay on the h-file, the desirable g2-g3 would simply allow
Black to keep control on the seventh rank with ....l:!.h2.
Nevertheless, one would expect that the result should be a draw. In the game, Carlsen
was able to bring his king to the queen side, to give extra protection for his pawns to ad
vance. Even so, it is unclear whether he can break through on that side of the board.
26 .l:!.c1 .l:!.h2 27 'it>f2 .l:.h8
266
Question: Find a good way for White to consolidate for a

safe draw. Black's edge is still not nearly enough to win, and perhaps
with a few sturdy defensive moves White will be close to equality.
28 �e3
White could certainly consider 28 'it>gl, followed by g2-g3 and 'it>g2, when he has pre
vented any incursions on the h-file. The end result should be a steady draw. Should we
criticize Radjabov for not heading for safety, in view of the fact that eventually he lost the
game? This must be very much an open question. Radjabov's mistake came later, when he
was only under mild pressure. lt is possible that he was thinking about whether Carlsen
himself might be under slight pressure. After all, there were a few small pawn weaknesses
in Black's position.
Radjabov might have been considering 28 c4! ? .i.xc4 29 .i.xc4 bxc4 30 l:txc4, when it is
Black who needs to think about how to secure the draw, since White now has an outside
passed a-pawn, and Black's pawn structure seems less effective than before. Radjabov was
already geared up to play c3-c4, with his l:tcl. Was he still aiming to squeeze out a slight
edge?
Perhaps this is the key to what was happening. lt seems that Radjabov did not want to
agree a draw just yet, but he needed to be careful not to overpress. lt is easy to create a
narrative which suggests that Carlsen was playing absolutely superbly, and that he was
able to create an advantage, and later a win, out of thin air. In a competitive game of chess,
there is always another story about what was going on. Radjabov, too, hoped for a slight
edge, and wanted to keep his opponent under slight pressure. Somewhere along the line,
he misplayed it.
2s gs
...
To prevent f3-f4.
267
29 .i.d3?!
One senses that Radjabov was starting to drift slightly. There are ten moves or so before
the time control at move 40, and the natural reaction in such a position is just to keep
things alive by shuffling the pieces around, and only then try to think about creating a
slight edge. The danger here is that if one player keeps shifting pieces aimlessly, while the
opponent is able to find something constructive to do, the player with the greater sense of
purpose will gradually take over the initiative.
29 c4 still leaves White at least equal.
29 �d7
...
Now Black is making some progress, in the sense that his king covers the c6-pawn, and
so White cannot gain a tempo with c3-c4 and a double exchange, since the pawn is no
longer threatened. Even so, the difference is tiny, and 30 c4 is still equal.
30 �a1 .i.b3 31 .l:i.c1
268
S e ve n th Ta l M e m o r i a l, M osco w 2 0 1 2
3 1...'it>c8
lt seems that Radjabov is not going to bother playing c3-c4 after all, in which case Carl
sen still has a glimmer of hope in bringing his king over to the queenside, specifically to b6,
enabling him to play ... c6-cS, after which a new phase of the endgame will commence.
32 'it>f2
Radjabov seems to believe that the simplest way of holding the position is to move
backwards and forwards, and of course sideways, with his pieces.
There is still time to play 32 c4. Not much has changed structurally in the last few
moves. Pawns and bishops get exchanged, and the kings are slightly differently placed, but
with no major improvements on either side. All that Radjabov needs to do is play c3-c4 be
fore it is too late.
32 ...'it>b7 33 'it>g3 J..e 6
34 l:f.a1?!
The procrastination has gone on too long, and for the first time it is possible to claim
that Carlsen has a clear edge, and that he is not just waiting for his opponent to make mis
takes.
Even now, 34 c4 looks fine, when one could argue that Radjabov's manoeuvre with
'it>e3-f2-g3 would have been the correct approach, keeping an eye on the gS-pawn. If
34 ... 'it>b6, then 3 5 b4 blocks off the as- and cs-squares, and the position remains level.
34 'it>b6 3 5 l:f.c1 cs
.•.
269
36 l:i.a1?
Even here, when Black is finally starting to make progress with his pawns, White con
tinues his rook shuffle. 36 c4, optional for a long time, is now essential. If 36 . b4, then 3 7
. .
.l:i.a1 defends.
36 C4
...
Even if this is not yet winning, Radjabov will need to rethink his play - not easy when he
is only four moves from the time control.
37 ..tc2 'it>cs 38 l:i.e1 c6 39 ..tb1
Question: What is Black's plan here?
First, it is important to recognize that Carlsen is indeed searching for a genuine plan, in
contrast to Radjabov's evident belief that once a position is solid, that position will remain
2 70
solid whatever happens (within reason). The position i s still blocked at the moment, but
Black can, with a few extra moves, unblock it in the centre.
39 'it>b6!
...
Having manoeuvring his king to the queenside to help advance the doubled c-pawns,
Carlsen then sets up a return journey, to e7 and then to f6, so that he can advance the
pawn to ds.
40 .tc2 'it>c7
The time control.
41 'it>f2 'it>d7
Question: What next?
42 a4?
Badly mistimed, and in effect the losing move. Radjabov wakes to the thought, just af
ter the time control, that he will need to make a pawn push at some stage, before Carlsen
can get his central pawns going, but he should have waited a move.
Instead, 42 Wg 3 ! still holds the balance. For instance, 42 ... g4 43 f4 and White is comfort
able. Or if 42 ... We7, continuing his march towards the kingside, then the c6-pawn is no
longer protected and 43 a4! would have been well timed. After 43 ...bxa4 44 .txa4 Ilb8
(44 ... cs 4S l:!.a1 .l:!.a8 46 Ita3 offers Black nothing) 45 .txc6 Itxb2 46 �hl, White, for the first
time, takes control of the h-file with his rook, and 46 ... Itc2 47 l:!.h7+ 'it>f6 48 Ith6+ 'it>f7 49
.tds .txds so exds leads to equality.
The inserted king moves, 'it>g3 and ... 'it>e7, make an enormous difference to the final
outcome. On e7, the black king no longer protects the pawn on c6; while with the white
king on g 3, there is no longer a check after ... Itxb2. King positioning is important in the
endgame.
Carlsen has shown greater positional stamina in this game. He has noted many minor
2 71
inaccuracies by his opponent, mostly in the delaying of c3-c4, and decided it was very
much worthwhile to continue playing in the hope of a more significant inaccuracy, on
which he will be able to pounce. Many players would have taken, or offered, a draw some
twenty moves earlier, but when Carlsen is given even the slightest chance of playing for an
edge, and possibly later a win, he will take it.
42 ... bxa4 43 l:!.a1
43 .i.xa4 .l:!.b8 44 .l:!.b1 WC7, followed by ... d6-d5, leaves Black with a clear advantage.
43 ....l:!.b8 44 1:!.a2 d5 45 exd5
He could perhaps try 45 .i.xa4 immediately, but Black is still considerably better after
45 ... d4. If White, at some stage, tries to cover the advanced d-pawn with his king, Black can
switch the attack to the h-file with his rook.
45 ... cxd5 46 .i.xa4+ 'it>d6
A transformation! Black's pawns are now extremely mobile, and he has excellent
squares for his pieces; whereas White's pieces are scattered and do nothing active, the only
role for the rook being to protect the b2-pawn.
47 .i.c2
On 47 �e3, Black has 47 .....tfs, threatening ....i.b1.
47 ...d4 48 .i.e4 l:!.b6 49 'tt>e 2 g4 50 fxg4 i..xg4+ 51 'it>d2
2 72
Question: Black has m ade the first part of the

breakthrough, but what is he going to do next?
5 1.....te6
White's bishop is the only piece, other than the king, to cover Black's passed d-pawn
and potential passed e-pawn. lt is therefore a good idea to try and exchange the bishops
with ... i.d5. White can resist, but it will still probably end up with the same rook and pawn
ending as in the game.
Not yet 5 1...d3?, as the passed pawns get quickly blockaded. After 52 'it>e3 'it>e6, White
can free his queenside pawns with 53 b4 - though Black would still have an edge, his posi
tion being so strong.
52 'it>c2 ..tds 53 ..txd5 d3+ 54 'it>d2 �xd5
2 73
Even now, one might easily gain the impression that, with rook and three versus rook
and three, the game might easily be drawn, with White's king blockading the passed
pawns. The key here is that Black can attack the opposing king, with the rook on the g-file
and good support from his own king and all the pawns. Once the white king is pushed off
the blockading squares e3 and d2, the rest is easy.
55 'it>e3 �g6 56 �aS+
56 'it>f3 e4+ 57 'it>f4 (or 57 'it>e3 1Ig3+ 58 'it>f2 e3+! 59 'it>xg3 d2 etc) 57 ...�f6+ 58 'it>e3 �fl
soon wins for Black.
S6 ...'it>e6 57 We4 �g4+ 58 Wf3 �f4+ 59 We3 �f1 o-1
White cannot do anything against the threat of ....l:Ie1+, followed by ....!:!.e2.
Round Six: Morozevich-Nakamura

What went right for Morozevich in this tournament and what went wrong? He started off
superbly with wins against Caruana, Grischuk and Aronian, and draws against Carlsen and
Radjabov. Then, suddenly, he had three consecutive losses against Nakamura,
Tomashevsky and McShane - and, finally, another draw against Kramnik, who was also
having a rough time. Morozevich's depth of imagination requires great respect, but there
is also a reputation that his play can fade quite sharply, through tiredness, or perhaps se
vere self-doubt, once he starts to lose games.
lt was a bad summer for Morozevich, with three losses in a row in Moscow, and then be
ing too exhausted to carry on playing at Biel. Sadly, it is unlikely that he will ever become
world champion. His play is often original and extremely good, but sometimes his play can
be "original" in the bad sense. This next game was the first of his three losses. Meanwhile,
Nakamura was having a quiet tournament, unable to keep up with his contemporaries,
Carlsen, Caruana and Radjabov. If Nakamura is able to achieve a couple of really good
tournaments, he will quickly edge up towards 2800 as well.
lt is easy to be slightly frivolous about their game below, noting that, between move six
(when he had to retreat his bishop) and move thirty-four (when he had shot his bolt and
was already losing), White did not make a single retreat - an unusual record. Lovers of
chess trivia will no doubt be interested in searching out even longer sequences without
any retreats. Quite probably this has already been done.
Behind the trivia, there is the implication that Morozevich is a player who concentrates
on attack, attack, and more attack, and that he is an extremely dangerous when attacking.
Before this game, he had just beaten Aronian, and he was also close to beating Carlsen,
appropriately with two rooks on the seventh, but he could not clinch the win.
Morozevich has never consistently reached 2800 - although, had he managed three
draws instead of three losses in rounds six to eight, he would have been up to 2795. Was
he starting to get tired at this tournament? Or were his opponents able to find systematic
weaknesses?
2 74
14t h J une 2012

A.Morozevich-H.Naka m u ra
Tal Memorial, Moscow 2 0 1 2
Queen 's Gambit Accepted
1 d4 dS 2 c4 dxc4 3 e4 tt::lf6 4 es tt:Jds 5 �xc4 tt::l b6
6 �d3
A necessary retreat. The position is now tense, but about level. White has more space in
the centre, but his d- and e-pawns are loose. Black has less space, but he is able to manoeu
vre effectively within that space.
6 tt::lc6 7 tt::le 2 �g4 8 f3 �e6 9 tt::l b c3
•••
9 .'ii'd7
..
2 75
Fig h t i n g Chess: M o ve by M o ve
Nakamura runs a tight ship. In an earlier game, A.Morozevich-Ar.Timofeev, Russian

Championship, Moscow 2011, Black tried a more open defence with 9 ... .tc4 10 i.xc4 tbxc4,
when Morozevich choked up Black's kingside with 11 e6 fxe6 12 0-0. For trivia followers,
the next retreat by White was at move 24, after 12 ... tbb6 13 tbe4 g6 14 tbcs �d6 15 tbxb7
'ii'd s 16 i.e3 �g7 17 tbf4 'ii'fs 18 l:!.c1 tbxd4 19 ..ixd4 i.xd4+ 20 'iVxd4 es 21 W¥b4 'ifxf4 22
'\Wb s+ c6 2 3 'ilixc6+ �f7, and now 24 tbcs. White was already close to winning, but the
knight cannot be expected to move forward from b7.
10 tbe4 i.ds 11 tbcs 'it'c8 12 a 3 e6 13 'ifc2
All roughly equal so far.
13 ...-txcs
Question: How should White recapture?
14 dxcs?!
This leaves both the es- and cs-pawns unnecessarily exposed. 14 'i!Vxcs is still about
equal.
14 tbd7 15 f4
...
2 76
S e v e n t h Ta l M e m orial, M os c o w 2 0 1 2
Question: How can Black take over the initiative? Look for a
tactical attempt to exploit the weak cs-pawn. The second, more
difficult, question is whether Black has a genuine advantage.
1S .i.xg2
...
This is certainly playable and should have been equal, but Morozevich later treated it as
a gambit and ended up losing.
The computer suggests that 1s .. :iiVd8!? is a considerable improvement, with a clear ad
vantage for Black, the main line being 16 0-0 .i.b3 (a sudden tactical resource) 17 'iVc3
(best; after 17 �xb3? tt:'lxcs and 18 ... 'iVxd3, Black wins a pawn safely) 17 ...tt::'Jx cs 18 .i.bs.
Although Black is again a pawn up, he has too many pieces under pressure to be com
pletely h appy (and yes, this is a human response, not a computer response). The position is
2 77
Fig h ti n g C h e s s : M o ve by M ov e
extremely complicated, Black's minor pieces are hanging, s o why should he try t o calculate
much further?
Play on a move or so and we reach 18 .. .'iVds 19 llJd4 �a4! 20 �xc6+ (or 20 �e3 �xbs 2 1
lDxbs lDa6) 20...bxc6, when Black seems t o keep an edge - but how many chess players
would be confident to calculate all this in advance?
Nakamura seems to have missed a very difficult line, though Morozevich's game was
not easy anyway.
16 .l:!g1 ..if3
Question: What next for White?
17 �e3?
Unnecessarily allowing Black an extra tempo, which should have won the game for him
relatively comfortably.
White should have preferred 17 .!:rxg7, and if 17 .. .'it'd8, then maybe 18 lDg1 ! ? ..ih s 19
�xh7 (19 ..ixh7 llJd4 is possible for White, but he will need to be careful with the queen)
19 ....l::txh7 20 ..ltxh7 lDd4 21 "ikc3 (again, he needs to handle the queen safely with the king
being exposed) 21...�4+ 22 'ii'g 3 "ii'x g3+ 23 hxg3. The position is tense, but about equal.
White could instead try to complete his development (finally!) with 18 �e3, but after
18 ... �4+ 19 lDg3 �6 20 .l:!g s 0-0-0, Black, with his own pieces now developed, looks to
have slightly the more flexible position.
In a complicated early middlegame, where various pieces have not yet been brought
into play, it is often a difficult judgment as to what priority should be given between de
velopment, or taking a pawn, or bringing the already developed pieces into more active
play. Here, taking the pawn should have been preferred. lt is not just the number of pawns
that needs to be considered; it is also that White, by taking on g7, would have damaged
Black's kingside structure.
2 78
S e ve n th Tal M e m o ria l, M os c o w 2 0 1 2
17 ...g6
Whereas now, White must struggle.
18 0-0-0 CiJe7 19 .i.e4 �xe4 20 'ii'xe4 b6
Or perhaps 20 ... CiJds.
21 CiJc3
Preventing the opposing knight from reaching ds.
21 .0-0
..
Question: it's starting to look desperate for White. He is a pawn

down, and Black is nearly ready to play ...CiJfs, after which there
is no hope of an attack against the king. What can White do next?
22 c6! CiJbS 23 fS!

Magnificent gambit play. White is already a pawn down, but he is ready to sacrifice two
more, not so much to provide clear compensation, just to open up lines to attack the king.
Win or lose, Morozevich showed resourcefulness. Without these pawn sacrifices, he would
quickly have gone under.
Despite the end result, affected by time trouble, Morozevich had given himself good
chances to hold the game.
23 ...CiJxfs 24 ii.gs
And the bishop is now in full play. If it reaches f6, Black will have to be worried about
mating attacks on g7, or along the h-pawn.
24 ... 'i¥e8
Black needs to bring his queenside pieces into play.
2 79
Question: What now? There is an obvious move, but how far can you calculate it?
25 't!Vf4?
Amazingly, Morozevich did not try it. Here 2 5 l::td 8 forces Black to give up the queen,
since 2S ... 'iix c6?? 26 l:i.xb8! 'ir'xe4 27 l:.xf8+, followed by ltJxe4, wins a piece.
Therefore he has to play 2S ...'Wxd8 26 i.xd8 l:f.xd8. Black is slightly ahead in material
(having rook, knight and two pawns for the queen), but his pieces are undeveloped. We
play on with 27 ttJbs (this has to be done quickly) 27 ... a6 28 tiJC7 l:!.a7
29 4Ja8! (the various computers suggest this move) and, after 29 ...4Jxc6, White makes
his first retreat since move six by 30 4Jxb6, with a likely draw after 30 . Jk7 31 Wbl. This
.
would have been a far tidier finish.

In the game itself, it is perhaps kindest not to try to analyse every move in depth, as
2 80
S e ve n th Ta l M e m orial, M osco w 2 0 1 2
both players were clearly short of time. The main creative part of the game has been and
gone. Now all White can do is try to find decent moves to reach the fortieth.
2S ...f6
lt is understandable that Nakamura did not want to risk 2 S .. .'ti'xc6 in time trouble.
White might, for example, try 26 �bl followed by tt:Je4; or there might be other possibili
ties. We'll leave it at that. Any slight mishap could easily end up with a quick defeat, not
something he would have liked.
26 �xf6 "ifxc6
Black has jettisoned his f-pawn in order to create more space for his pieces.
27 �b1 tt:Jd7 28 tt:Je4 tt:Jcs 29 tt:Jxcs bxcs 30 h4
Question: Quick assessment, as Black is in time trouble. What

would you do if you had to make a move in a minute?
30 ...l:!.f7?!
.
A loss of tempo. This rook move is unnecessary for his defence.

The computer suggests 30 ...c4, though after 3 1 .tl.c1, it is difficult to believe that Black
has any significant edge.
The correct continuation, it seems, is 30 ... .Uab8, and if 31 h S , only then 31 ...c4, and Black
is still genuinely better. For example, after 32 hxg6 h6 33 g7 .l:!.fc8 34 .l:!.c1 c3 35 .Uc2 .l:!.b3, he
will double up on the b-file. Black continues his attack until the very last moment, playing
a defensive move only when absolutely necessary. The general ratio is something like three
attacking moves to one defensive move, when there are attacks against kings on opposite
flanks. This is only an approximation, of course, but it is a reminder that it is best not to be
over-defensive in an advantageous but sharp position.
Maybe a quick-thinking reader would be able to envisage this in a minute. For the rest
of us, the important point is that Black, by pressing through with his attack immediately,
281
Fig h ti n g C h e s s : M o ve by M o ve
has gained crucial time.

31 h S If.d7 32 hxg6 h6 3 3 !txd7 �xd7
34 �f3
After all the strange vicissitudes, the game is now level again. Unfortunately, in time
trouble it did not remain equal.
34.. Jib8 3 5 ftdl �e8 36 Jad2 C4 37 'it>a2 a S
Question: What should White play here? Again, make your decision quickly.
3 8 �d1??
According to Morozevich, just about any reasonable move would have been fine, other
than the one actually played. He missed his opponent's pawn hack.
38 ...c3! 39 bxc3 lbe3
282
Suddenly, all his pieces are after the white king. I n contrast, White's advanced g-pawn
merely blocks any attack against the black king.
40 �e2 lt:lds 41 �d3
If 41 �c4, Black has an instant win with 41 ...�a4 42 �xa4 lt:lxc3+ 43 �a1 l:f.b1 mate.
41...a4
Followed, if allowed, by ... l:f.b3.
42 .l:!.b2 .l:!.xb2+ 43 �xb2 �b8+ 44 �cl �b3
White's bishop is useless.

45 �a6 �xc3+ 46 �d1 lt:le3+ 47 �e2 �c4+ 0-1
A strange game.
Round Eight: Caruana-Kramnik

In truth, this was hardly a great game by Kramnik. Once a player, even a great player, has
started to hit the mid thirties, tiredness will occasionally set in, and more frequent mis
takes will occur. Kramnik had, in the previous two rounds, played two very long and tiring
games, the latter being a loss (see below). Resilience can start to decrease, even if only
slowly, and there will be dangers of further losses.
Of course, only a really strong player would be able to exploit such minor weaknesses,
and Caruana is indeed such a strong player. He had, at the time of this game, slightly more
than a decade before needing to worry about similar signs of decline. His current problem,
as exemplified in his game against Morozevich, is that he can still make mistakes through
inexperience. His peak will come some time, perhaps, in his late twenties.
As for the long games by Kramnik: it took him 36 moves to win Tomashevsky's a-pawn,
to reach an endgame with rook, knight and four kingside pawns, versus rook, knight and
three king side pawns. This was going to be a long struggle, and Tomashevsky lasted out
until move 83. Then against McShane, just after the time control, the players reached an
intense queen and rook ending with three pawns on each side, two of them passed.
2 83
L.McShane-V.Kramnik (round seven)

Black to play
Kramnik was probably better in the earlier stages. After all, it is easy enough to imagine
how dangerous his central passed pawns could have become, if given the opportunity.
However, it is now White who h as the initiative. His outside passed h -pawn, almost of mi
nor significance during the middlegame, is now an enormous threat. All White has to do is
advance his h-pawn safely and it is difficult for Black to defend that part of the board.
Kramnik tried 44 ... e4? here, after which McShane's outside passed pawn proved decisive,
and Kramnik resigned on move 94.
The task for the students, making full use of their analysis engines, is to find a better at
tempt to defend the position after 44 ... :f.c1 4S l:txc1 ir'xcl+ 46 �g 2. lt is not easy, even
though the computer engine gives the position as stone-cold drawn in three different
lines. Good luck if you can find something secure.
Given a free hand, I feel sure that I would have given this game in detail in this book. My
intention, however, was to look only at Carlsen, Caruana and Radjabov. Only one game not
featuring one of the three target players has been annotated, that being Nakamura
Morozevich in round six. This leaves twenty more games, including twelve with a decisive
result, for readers to investigate on their own.
Meanwhile, returning to the current game, remember that Caruana beat Kramnik - a
truly impressive result.
17th June 2012

F.Ca ruana-V.Kra m n i k
Scotch Came
1 e4 es 2 lllf3 lllc6 3 d4 exd4 4 lllxd4
2 84
S e ve n th Tal M e m orial, M os c o w 2 0 1 2
Yet another Scotch. I n this book, we have already seen Scotch - and Scotch Four Knights
- games by Kramnik (successfully) and Radjabov (both successfully and unsuccessfully).
Now Caruana beats Kramnik. Could this be the start of a new fashion? Or is it all a one-off?
Probably this was just a brief flurry - and it is incidental that these games have been
picked up for detailed analysis in this selection. In fact, most of the 1 e4 es clashes at the
Tal Memorial were in the Ruy Lopez (which was seen eight times). lt is difficult to imagine
that there will be another sudden surge of interest in the Scotch as a result of these few
games.
4 4Jf6 5 4Jxc6 bxc6 6 eS �e7 7 �e2 ltJdS 8 C4 .i.a6
...
8 ... 4Jb6 is the other main line.

9 4Jd2 g6
Question: One for students of chess psychology. Would you be happy with a quick
draw by repetition here? Would it make a difference if you felt that your opponent
would try and avoid it? Finally, what is the move we are being asked to think about?
10 4:lf3
The line in question is 10 4:lf3 �4+ 11 tt:Jd2 "ilie7. Caruana would, I presume, have been
happy with a quick draw as Black against a recent World Champion. Naturally, I am unable
to get inside the head of a leading grandmaster, or his opponent, but there seems to be a
reasonable narrative in the next few moves - in that Caruana makes an implicit offer of a
draw, and Kramnik decides he has to try something else, which turns out to have been ei
ther not so good, or at least a more difficult route to equality. This seems to make the most
sense of the way Kramnik managed to find himself a pawn down while still in the opening.
Or it could just have been a bluff by Caruana. Maybe he was dangling the prospect of
an early draw by repetition, intending to play a new move next time round, rather than
repeat the position again. If nothing else, it would be a good opportunity to see what
285
Kramnik is thinking, and then decide what to do later.

There are other lines, notably 10 b3 ..ig7 11 ..ib2, with either equality or a minimal edge
for White.
10...�g7
As we have just seen, 10 ...iVh4+ 11 'Lld2 'ifle7 12 ctJf3 would repeat the position. Of
course this is not a forced line for either side, but the various alternatives - for example, 11
�d1 'iVa4+ 12 'ii'c 2 - should still end up equal.
11 �gs f6
11 ... 'iYh4+ 12 'ir'd2 would now give White a slight edge.
12 exf6 'ii'x e2+ 13 �xe2 'Llxf6 14 0-0-0
This, surprisingly, is the critical position in the whole game. Given that Kramnik soon
lost a pawn, there was clearly something wrong with his handling of the opening.
14 0-0-0
...
Black's two main problems are, first, what to do with his king, and second what to do
with a couple of small pawn weaknesses (a7 and d7). Kramnik castled quickly on the
queenside, but left his a7-pawn dangerously weak. Possibly he underestimated this. In
stead:
a) 14 ... 0-0? would lose the other pawn after 15 ..ixf6, followed by l:txd7. There is no rea
sonable compensation to be seen.
b) 14...l:tb8 merely delays the problem. To counter the possible threats to b2, the most
efficient solution seems to be 15 �e3, followed by ..id4. If Black were to try 15 ...'Llg4 16
..id4 �xd4 17 l:txd4 'Llxf2??, the knight soon gets captured after 18 .l:!.fl.
c) 14 ... 'Lle4 looks effective (maybe Kramnik had intended this?), but after 15 l:the1! 'Llxg5
15 'Llxg 5, Black's king is uncomfortably open. If 16 ...0-0 17 .l:!.xd7 .l:!.xf2, White has a hidden
286
S e v e n t h Ta l M e m o r i a l, M os c o w 2 0 1 2
tactic i n 18 l:!.xg7+ 'it>xg7 1 9 tt:le6+, followed by a fork with 20 tt:lxe7.

Despite initial impressions, this is not quite winning, since Black has 19 ...'it>g8 (avoiding
a later check on the f-file) 20 tt:Jxc7 .l:Ixe2 21 .l:Ixe2 .i.xc4 22 I1c2 .l:Ic8 23 .l:Ixc4 .l:Ixe7. How
should one judge this rook and pawn endgame? Black's pawn structure is slightly dam
aged, with isolated a- and c-pawns. This of itself, and with all other things being equal,
guarantees White a slight edge with accurate play. White has at least a draw, though it is
far too early to say that he has serious winning chances. On the other hand, it is uncom
fortable for Black and, given the chance, he would prefer to find complete equality instead.
d) There is one fully sound move: 14 ... d6!. Black's pawn structure has now been
straightened up and, if necessary, he can cut out any threats to the a7-pawn with ... c6-c5.
More significantly, he can castle safely on the kingside. Play is equal - if 15 ctJd4, then sim
ply 1 5 ...0-0, and if 16 tt:Jxc6?, then 16...ctJe4 with advantage to Black.
15 .i.e3
Now White has a solid positional edge.
1S ....l:Ide8
Question: Any improvements? In particular, could he

have tried to defend the pawn with 1S ... 'it>b8 - ?
The a7-pawn is not directly attacked after the m ove played, but each time Black leaves it
undefended, he remains under pressure. A quiet defensive plan with 1S ... 'it>b8 17 l:!.he1
.tb7 is certainly possible, but there is no clear and simple equality. The problem is that,
while Black would like to play both ... d7-d6 and ... c6-cS, how is he going to achieve this? If
he plays ... c6-cS first, his pawn will be taken. If he plays ... d7-d6 first, White will start an at
tack with ctJd4. lt is already difficult for him.
16 .i.d3 d6
287
Question: Kramnik has achieved his first objective, managing to play ... d7-d6 safely
while keeping the knight away from the centre. What is Caruana going to do next?
17 cs
Now 17 lLld4 achieves nothing for White after 17 ... cs, and if 18 lZJc6, then simply
18 ... Wb7.
17 .txa7?! cs is just about playable if White continues with 18 b4, but few would con
sider this as an attempt for an advantage.
White could try something quiet such as 17 h3 (to prevent ...lZJg4) 17 ... cs, but his edge is
going to be very slight.
Better to open up the queenside, so that all of Black's pawns are weakened. Or at least,
that is how the argument goes.
17 ....txd3 18 llxd3
288
Question: How should Black continue the defence?
Perhaps Black's position is not as bad as it looks, but he still has to defend imaginatively
and find the correct reply.
1s tt:Jds?!
...
Kramnik gives away his worst pawn in order to find good squares for his pieces. All the
same, a pawn is a pawn, and nowhere did Kramnik find a way to recover it. White's passed
a-pawn remained to the end.
Instead, 18 ... ds 19 tt::ld4 'Et>d7 20 f3 is just about playable for Black, though it looks un
comfortable. Once any immediate tactics have been resolved (e.g. after 20 ... lhe3 2 1 l:txe3
..ih6), Black still has to work out what to do with the a-pawn, and the two doubled c
pawns. There will be no quick win for White, but he will he able to grind for a long time.
The best reply seems to be 18 ... dxc s ! 19 ..ixcs l:te2. it goes against the grain to allow tri
pled isolated pawns, and an extra isolated pawn nearby, but for the first time Black has
good squares for his pieces and is in no great trouble. For example, 20 ..txa7 c s ! 21 ..txcs
(21 l:tb3 tt::ld 7 or 21 .l:!.hd1 .l:!.xf2 22 tt::le s l:If4! 2 3 l:Ib3 l!b4) 21...tt::l e4 22 b4 tt::l xb4 23 cxb4 .l:!.xa2
is equal.
In this line 20 l:te1! seems more testing, with Black's queen side pawns being trouble
some to defend. There is no obvious win for White, but equally there is no obvious way for
Black to find a clear draw. Perhaps something like 20 .....ih6+ 2 1 'iii>b 1 .l:!.he8, and it would
have been interesting to see what would happen next. This must still be better than being
a pawn down.
19 cxd6 cxd6 20 ..txa7
Question: Black to play. What next? How can he set up a new line of barricades?
2 89
20 ... Wc7?!
lt seems Kramnik is not quite as good as he used to be in defending difficult endgames.
This showed, for example, in his first match game against Aronian - though of course eve
rything is relative and he is still an immensely strong player, battling to stay at 2800 level.
Black certainly needed to work extremely hard on this position, whereas Kramnik
merely played passively, attacking White's passed pawn, and not doing all that much else.
Instead, 20 ....l:f.e2 leads to a quick hacking of material with 21 .td4 .txd4 22 tt:Jxd4 .l:f.xf2
2 3 tt:Jxc6 tt:'lf4 24 .l:i.xd6 cJ;c7 2 5 .l:f.hd1 .l:f.xg2, but after 26 h4 it is difficult to get back the extra
pawn quickly. If he can only achieve this slowly (the direct 26 ... .l:f.h 2 27 tt:'la7 .l:f.xh4 28 l::td 7+
sees Black's king suddenly in trouble: 28 ...'lt>b6 29 l::t 1 d6+ or 28 ...�b8 29 tt:Jc6+ cJ;c8 30 .l:f.a7),
White will have two dangerous, outside, connected passed pawns.
Black's best plan might be to complete his development with 20 ....l:f.hf8, making sure
that all his pieces are centralized and on good squares, without worrying about attacking a
pawn on the edge. One of the points is that he is creating pressure on the f-file, which
makes it difficult for White to improve his minor pieces. For example, 2 1 .l:f.e1 .l:f.xe1+ 22
tt:Jxe1 �b7 2 3 ..te3 tt:Jxe3 24 fxe3 .l:f.f2 and Black is reasonably safe. Perhaps 2 1 .te3 .l:Ie4,
when play continues, and White is only slightly better.
21 .te3 l::ta 8 22 a3 l::ta 4
lt is not so clear that his rook is any better on a4 than on e8, and of course he has spent
two tempi getting it there.
2 3 1::te 1 l::tf8
To provoke the knight move, and therefore not a waste of time. 2 3 ...tt:Jxe3 24 fxe3 .l::te 8
still sets up a tough defence, but White should probably win eventually.
24 tt:Jgs l::te 8 25 .tb6+ 'it>d7 26 l::txe8 �xe8 27 ..te3
The old story: the more pieces get exchanged, the more an extra pawn increases in
value. The bishop is on the same square as six moves ago, but White is happier now a pair
of rooks has gone.
290
S e ve n t h Ta l M e m o ri a l, M os c o w 2 0 1 2
27 .. /Jitd7
Again, 27 ... ttJxe3 28 nxe3+ is to be considered. At the moment, the knight looks stronger
than the bishop, but if the bishop is given the opportunity to move to a safe square, the
balance will change.
28 nb3 'iitc 8 29 kd2
The bishop is now safe. The general idea for White is to find better squares for his other
pieces and to cut down the opponent's options. Still, progress is not going to be quick.
29 ... hs 30 'Df7 �c7 31 ng3
White has prodded the king side pawns successfully and forced Black to retreat one of
his pieces.
31 ...lbe7 32 b4
Now that his pieces are safe, it is time to start pushing the pawns.
3 2 ...'iitd 7 3 3 nd3 d s 34 'it>c2
And the king is on hand to offer support, push away the opposing pieces, and eventu
ally enable his own pawns to advance. This is classic technique in the endgame.
34...ttJfs 35 kc3
291
3 s ... d4
3 S ....l:!.xa3 36 �xg7 .l:!.xd3 3 7 'it>xd3 lLlxg7 3 8 lLles+, followed by l2Jxg6, gives White
enough kingside pawns to ensure that he will win the knight endgame.
36 Wb3 .l:!.aB 37 lLlgs 'it>d6 38 f3 'iti>d s 39 l2Je4
39 l2Je3??
.•.
A time-trouble blunder. After 39 ....tf8 40 .td2 cs 41 l2Jxcs .txcs 42 bxcs 'it>xcs, it would
still require a lot of hard work for White to force victory. The next stage might be 43 .i.b4+
'it>ds 44 a4, advancing the pawn. In itself this is not enough to win; White will have to find
some way to make use of the extra pawn on the king side, with g2-g4 at some stage.
40 l:txe3 1-0
After 40.. .fxe3 41 .i.xg7 White has two pieces and a pawn for the rook, and the e-pawn
will soon drop off too, so Black resigned.
For a teenager, indeed for anyone, to beat Kramnik is quite an achievement.
292
Final Notes
There are, of course, no final notes in chess. lt is all part of a long historical process, which
carries on from the past and through the future. Older players will gradually see their
strengths decreasing, and younger players will improve, sometimes very quickly. Since
2005, it has to be acknowledged that there has been no improvement at the very top, in
that the number one spot has been vacant since Kasparov retired, with no player as yet
being of the same strength. There has been a vast increase of new 2700+ grandmasters,
who are all intensely strong, but not quite as strong as Kasparov.
What we have seen in the last few years, however, is that some of these players are
starting to go beyond An and and Kramnik, and have genuine aspirations to reach the level
of Kasparov, or even beyond. For instance, Carlsen, at 21, is just a couple of good wins from
reaching Kasparov's peak rating (2851 in July 1999, aged 36}.
There are statistical questions as to whether the FIDE rating system has an inflationary
tendency, and whether ratings in 2012 are fully equivalent to ratings in 1999. Carlsen, on
his own website <www.magnuscarlsenchess.com/>, suggested that Kasparov's rating of
2851 was actually equivalent to a 2886 rating in November 2010. Furthermore, according
to Carlsen's recalculations, Kasparov's rating of 2675 in July 1982, at the age of 19, would
have been 2811 in modern terms. There are many ifs and buts about all this, and no doubt
different statisticians would have different points of view, but Carlsen's argument makes
sense. He certainly does not claim that, in his early twenties, he is already the best chess
player ever. He knows he still needs to prove himself.
Carlsen also notes that he reached 2800 in November 2009, at the age of 18 years and
11 months, slightly faster than Kasparov. There is just one other player who is currently
achieving a similar speed of improvement. That is Anish Giri, aged 18 years and 3 months
at the time of writing, with a rating of 2730, and an increase on average of two rating
points per game. Even Caruana's development looks very slightly slower.
Time to have a closer look at the most recent ratings, up to the beginning of the 2012
Olympiad. According to the international ratings (updated daily at <www.2700chess.com >)
just before the I stanbul Olympiad, there are currently eight players rated over 2770. These
can be divided into three age groups (actual ages given in brackets}:
293
25 and under: Carlsen (21) 2843, Radjabov (25) 2785, Nakamura (24) 2783, Karjakin (22)
2778, Caruana (20) 2770
26 to 30: Aronian (29) 2816
31 and over: Kramnik (37) 2797, Anand (42) 2780
Then three more players, aged 25 or under, with ratings of 2730 or more (so within rea
sonable striking distance of 2750): Wang Hao (23) 2742, Giri (18) 2730, Tomashevsky (25)
2730.
There are other young players who might be capable of reaching this higher level, but at
the moment we have eight players (of age 2 5 or below) who can be regarded as excellent
candidates for the world championship (remember that Carlsen is not World Champion
yet). This is a definite generation gap. Only Aronian, at 29, straddles between the younger
age group and the old guard of Kramnik and Anand.
lt has been a splendid opportunity to be able to write about the major shift of the chess
generations in mid-2012. Before long, Carlsen will almost certainly become the strongest
ever player in chess, ahead of Kasparov. There will be others, too, who will overtake Kas
parov - and there are at least two obvious possibilities in the above list. This does not mean
that these players will become m ore of a "genius" than Kasparov, nor even that Kasparov is
necessarily a greater genius than Fischer. Every generation has the opportunity to learn
from its predecessors, and every player can also learn from their contemporaries. Even
more important perhaps, when considering the massive step-change in playing strength in
this short period of time, is the increased sophistication of computer technology.
Chess is becoming even more interesting than a few years ago.
294
Index of O penings
(figures refer to page numbers)
Benoni Defence 158

Four Knights Game 35
Gri.i.nfeld Defence 89, 104
Nimzo-lndian Defence 1 6 5 , 187, 219
Queen's Gambit Accepted 2 75
Ruy Lopez 28, 4 7. 68, 24 7
Scotch Game 239, 261, 284
Semi-Slav Defence 13, 56, 207
Semi-Slav with s a6
... 9 7, 122, 138, 148
Sicilian Defence 130, 189, 195, 209
Slav Defence 211
295
Index of Games
Anand.V-Gelfand.B, World C h a m p io n s h i p, Moscow 2012 (Ga m e 1) . ... . ......... ..... ..... ... .. 88 . . . . . . .
Anand.V-Gelfand.B, World C h a m p io n s h i p, Moscow 2012 (Ga m e 3) ....... . ......................... 104

Anand.V-Gelfand.B, World C h a m p io n s h i p, Moscow 2012 (Ga m e 5) ................................. 130
Anand.V-Gelfand.B, World C h a m pion s h i p, Moscow 2012 ( G a m e 8) . . 157 ... ..... . ..................... .
Anand.V-Gelfand.B, World C h a m pion s h i p, Moscow 2012 (Ga me 10) .. . . 180 ...................... ... .
Anand.V-Gelfand.B, World C h a m pion s h i p, Moscow 2012 (Ga me 12) . . . . . . . . . . ...... .. 195 . . . . . .. . . . . .
Anand.V-Gelfand.B, World C h a m p io n s h i p, Moscow 2012 (Tie-brea k ga m e 2) 209 ..............
Aronian.L-Kramnik.V, Z u rich 2012 ( G a m e 2) 27

..................................... . . . . . . . . . . . . . . . . . . ........................
Aronian.L-Kramnik.V, Z u rich 2012 ( G a m e 4) ... . ... .. .. . . . .. .

........ . . .. ....... 47. .. .. .. ... . ... ....... . .....................
Aronian.L-Kramnik.V, Z u rich 2012 (Ga m e 6 ) ... . . . . ... . ... ..... .... .. . ... . ..
. . . . ... . . .. 67 . .. . . .. . ............................
Carlsen.M-Kramnik.V, T a l Memori a l , Moscow 2012 ..... . . . . . . . .... . . . ..... .... .... .. .. . ... 219
. . . . . . . . . . . . . ... . . . . ..
Caruana.F-Kramnik.V, Ta l Mem o r i a l , Moscow 2012 . . . . . . . . . . . . . . . . . ... . .... . . . . . . .. 284

. . . . . . . . . .. . . ... . . . . . . . . . . . . .
Gelfand.B-Anand.V, World C h a m pio n s h i p, Moscow 2012 (G a me 2) . . 96 ................................ . .
Gelfand.B-Anand.V, World C h a m pio n s h i p, Moscow 2012 (Ga m e 4) ................................. 122

Gelfand.B-Anand.V, World C h a m pio n s h i p, Moscow 2012 (G a me 6) .. . . . 138 ........ . ........ ........ ...
Gelfand.B-Anand.V, W o r l d C h a m pions h i p, Moscow 2012 ( G a m e 7) ...... .. . . .. . .... . . 147 . . . . . .. ... . ...
Gelfand.B-Anand.V, W o r l d C h a m pions h i p, Moscow 2012 ( G a m e 9) . . 165 ............................ . ..
Gelfand.B-Anand.V, World C h a m p io n s h i p, Moscow 2012 (Game 11) . . . 187 ...... ... . . . . . .............
Gelfand.B-Anand.V, W o r l d C h a m pion s h i p, Moscow 2012 (Tie-brea k ga m e 1) 207 ...............
Gelfand.B-Anand.V, W o r l d C h a m p io n s h i p, Moscow 2012 (Tie-brea k ga m e 3) ............... 211

Kramnik.V-Aronian.L, Zu rich 2012 (Ga m e 1) . . . .
............................ 13 ...............................................
Kramnik.V-Aronian.L, Z u rich 2012 (Ga m e 3) . . . . . .. ..... ... ..... . . . .. ... ......... .. ... .
. . . . . . . . . . .34 . . . . . .. . . . ........... ... .
Kramnik.V-Aronian.L, Z u rich 2012 (Ga m e 5) . . . . . . . 56

........................................................... . ... . . . ..... .
Morozevich.A-Caruana.F, Ta l Memorial, Moscow 2012 . . . ..... . .... . ... . . . . ... 24 7

... ..... . . ... . . . .. . . ............
Morozevich.A-Nakamura.H, Ta l Memori a l , Moscow 2012 . . . . . . . . . 2 75 ......... . . . . . . . . . . . . . . . ..... . .... . ... . . . .
Radjabov.T-Carlsen.M, Tal Memoria l, Moscow 2012 .............. . ... .. 261 . . .. .....................................
Radjabov.T-Tomashevsky.E, Tal Memori a l , Moscow 2012 .. . . . . ... . ... . ...... ............ 239
......... . . . . . . . . . .
296
The fol lowi n g a re some sample pa g es
fro m a nothe r g reat move by move boo k :
Cyrus Lakdawala
Kra m n i k
move by move
r
'
I S B N : 9 7 8 1 8 5 7 44 9 9 1 4 - 4 1 6 pages
Kra m n ik: M o ve by M o ve
Came 28
V.Kra m ni k-V.Topalov
Nice (rapid} 2008
King 's Indian Defence
1 d4 tLlf6 2 c4 g6 3 tLlc3 .ig7 4 e4 d6 s tLlf3 o-o 6 .ie2 es 7 o-o tLla6
Question: Doesn't this decentralizing move violate principle?
Answer: Yes, but the move is an exception to the normal rule since the post is only tempo
rary. Black's knight has access to all sorts of good squares, such as cs, or even b4 and c7,
depending on how White sets up. The idea is to avoid the theoretical monolith stemming
from 7 . ..CtJc6, from which my database announces nearly 34,000 games. I remember being
surprised by 7 ...tLla6, by a master way back in the 1980's. I thought his move was outra
geous and expected to crush him. Unfortunately, I underestimated his opening and went
on to lose in confused fashion. So I learned then not to underrate the move.
8 .te3
White's best shot at an edge is to retain the central pawn tension for as long as possible.
s . . .tLlg4 9 .igs 'i&'e8
Question: An unnatural move?
Answer: This bizarre move is now quite normal in the position. Instead, 9 .. .f6 10 ..tc1 (10
.ih4 is also played here) 10 ...�h8 11 h3 tLlh6 12 dxes fxes 13 .te3 tLlf7 14 'i&'d2 tLlcs was
A.Karpov-G.Kasparov, World Championship (7th matchgame), New York 1990, when I pre
fer White slightly after 15 b4 tLle6 16 l:tfd1.
Riding t h e Dyn a m i c Ele m e n t
10 .Ue1 exd4 1 1 tt:Jd s

V.Anand-J.Polgar, Leon (human+computer rapid) 2000, saw 11 tt:Jxd4 'ir'es 12 lt:Jf3 'ir'cs
13 Si.h4 tt:Jes 14 tt:Jxes dxes 15 .l:!.c1 .Jie6 16 tt:Jds f6 17 a3 c6 18 b4 �d4 19 'ii'xd4 exd4 20
lt:Jf4 i.f7 21 ..ig3 1Ue8 22 ..ifl Jlac8 23 lt:Jd3 and again I prefer White, who managed to
blockade Black's passed d-pawn.
11... d3
lt would be a short game if Black played 11 ... 'iVxe4?? 12 Si.d3.
12 ..ixd3 c6
Question: Why can't Black take on b2?
Answer: He doesn't have the time. His dark squares grow grievously weak after 12 .....ixb2?
13 eS!.
13 lt:Je7+ 'it>h8 14 tt:Jxc8
White picks up the bishop pair.
14...l1xc8 15 i.f1 tt:Jcs
Question: Can he grab b2 now?
Answer: He can, but Kramnik would undoubtedly give up the exchange for a huge initiative
once again with 1S ... .Jixb2 16 �b1 Si.c3 17 .l:!.xb7! Si.xe1 18 tt:Jxe1!.
16 't!Vxd6! lt:Jxe4
Double attack: Black simultaneously attacks f2 and White's queen. The over-exuberant
e4-knight rushes out into the street and looks for someone - anyone - to high five. But he
may be celebrating prematurely, since he is pinned.
17 'i!Va3 fs
Kra m n ik: M o ve by M o ve
Question: Isn't f2 hanging?
Answer: lt is defended tactically. Black drops a piece after 17...lbgxf2? 18 h 3 ! (eliminating

the knight's exit route) 18 ...�d7 19 ..th4 lbxh 3+ 20 gxh 3 and Black doesn't get enough for
the piece.
18 h3 lbes 19 �f4!
Question: Whose position do you prefer?
Answer: White stands better for the following reasons:

1. Bishop pair in an open position.
2. The pinned e4-knight gums up the works, like the slow moving supermarket cashier
trainee who takes forever with each customer before you in line.
3. a7 is hanging.
4. Black's dream of king side or central counterplay is an arid desert where life of any
sort never takes root.
19 ...lbd7
Black loses the initiative after this unforced retreat, but good alternatives are hard to
find:
a) 19 ...lbxf3+?? drops a piece to 20 gxf3.
b) 19 ... bs 20 cxbs cxbs 21 'ii'xa7 lbc4 22 a4 bxa4 23 .l:txa4 lbxb2 24 , when White has the
initiative and Black's offside knight on b2 is in some danger as well.
c) 19 ... a6 20 �ad1 lbf7 (what else?) 21 'i'b4 bs 22 'i!Va3 and Black remains under strate
gic pressure.
20 "ifxa 7?!
20 l:radl! keeps Black twisting in the breeze.
Riding t h e Dyn a m ic Ele m e n t
20...i.xb2?!
He decides to ski the slopes, ignoring the prominent avalanche warning sign up ahead.
After this move Topa submits to endless defensive drudgery without complaint. He should
go for 20 ... l:ta8! 21 �xb7 (21 'iVe3 .i.xb2 also looks okay for Black) 21 .. /bdcs 22 'ii'h4 l:ta4! 2 3
� 6 .l:ta6 2 4 'fie7 tbe6 2 5 'ii'h 7 tb6cS! with a draw by repetition.
21 �ab1 i.g7 22 �xb7 tt::Jd cs 23 'iVb6 l:.f7 24 tt::Jg s! l:tb7
Farmer Topa plucks the withered turnip from his dying field and thinks: "will this drought
ever end?" We are at a crossroads: should White sac his queen here by taking on b7? We can
decline - reasoning that, just because a novelist writes a revolver into the hands of a charac
ter, it need never be fired). Or we can take a chance and sac - thinking that we carry regret
for actions not taken when we should have, if only we had the courage.
Exercise (critical decision}: Would you sac or not?
Answer: The queen sac is completely sound.

25 'ii'xb7!
GM Baburin writes: "One should not call this a sacrifice, as for a queen White will have a
rook, bishop and pawn." And a strong initiative, I might add.
2S ...tt::Jx b7 26 l:.xb7 'it>g8?
26 ... cs was called for.
Black's defenders are pilings on an old dock, worn away by the ocean's pounding and by
time. There lies a path where White's piece activity grows exponentially.
Kra m n ik : M o ve by M o ve
Exercise {planning}: Find one powerful move and you open

the spigot, as White's pieces pour out against Black's king.
Answer: Clearance - free the light-squared bishop of his constraints by clearing c4. Black's
position collapses quickly with the addition of the new attacker, as Kramnik's vacuum irre
trievably sucks all life energy out of Black's game.
27 cs! h6 28 ..tc4+ 'lt>h8
Exercise (combination alert}: Sometimes you simply get

an intuition that the disparate elements in one side's
position are subtly out of whack. White has a mind-bending
shot - a shot we all dream about. Do you see it?
Riding t h e Dyn a m ic Ele m e n t
Answer: Overload. Neither Black's queen nor bishop can touch White's intruder.
29 .tes!! hxgs 30 .txg7+ 'it>h7 31 .tf8+1 'it>h8 32 .te7! l:lb8
Exercise (combination alert}: The black king's corpse, icy to the touch,
is ready for the forensic examination to determine cause of death.
Answer: Removal of defender: White clears f6 for his bishop to deliver checkmate.
33 �xe4! 1-0
Game 29
A.Morozevich-V.Kra m n i k
Ta l M e m o r i a l , Moscow 2009
1 d4 lLJf6 2 c4 e6 3 tLlc3 .tb4

A bold choice, since Morozevich boasts a nearly 70% score with the Capablanca line of
the Nimzo-lndian.
4 �c2 0-0 5 a3 .txc3+ 6 '!Wxc3 d 5
A painter needs more than just one colour t o work the canvas. 6. . .d5 is a no-nonsense
move; Black stakes a claim in the centre. In the previous chapter we looked at 6 ...b6 (see
Game 16 ) .

Fighting Chess Move by Move - Colin Crouch 2012

Uploaded by

Copyright:

Available Formats

Fighting Chess Move by Move - Colin Crouch 2012

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fighting Chess Move by Move - Colin Crouch 2012

Uploaded by

Copyright:

Available Formats

Colin Crouch

Copyright© 2012 Colin Crouch

All rights reserved. No part of this publication may be reproduced, stored in a

British Library Cataloguing-in-Publication Data

ISBN: 978 1 85744 993 8

Everyman Chess Series

Typeset and edited by First Rank Publishing, Brighton.

Dr Colin Crouch is an International Master, a tremendously experienced tournament player

Also by the Author:

Rate Your Endgame

1 Vladimir Kramnik-Levon Aronian, Zurich, April 2012 13

2 World Championship, Anand-Gelfand, Moscow, May 2012 87

Final Notes 293

Game One: Aronian-Kramnik

21st April 2012

1 tt::lf3 ds 2 d4 tt::lf6 3 c4 c6 4 tt::lc3 e6

Aronian is happy to play in "normal" Queen's Gambit chess.

12 '1Wb3, and suddenly Black h as problems with his f7-square.

Question: What do you think should White play next?

lt has to be played sooner or later, and this seems to be a good time.

At last, there is a genuine pawn.

Question: What is White's idea with this move?

Question: Which is better? 16 tLld5 or 16 cxd7 - ?

Question: What should White do here?

So he tries it, but Aronian i s careful with his tactics.

Obviously, he cannot allow White's advanced pawn to survive.

Whereas White's knight is exposed to attack.

Or 41 tLlg2 .l:tal+, winning the knight.

This is what I wrote in my day-after blog:

Game Two: Aronian-Kramnik

22nd April 2012

popular, while 9 ...h6 or 9 ...i.d7 are also fully playable.

10 i.gs, when Black's pawn structure suddenly loses its flexibility.

Question: Isn't this a good time to offer a draw?

The ... b7-b6 push is needed, Kramnik decides.

Question: What should Black do next, to defend his position?

Everything i s covered, if only just. Black has defended it extremely economically.

Game Three: Kramnik-Aronian

24th April 2012

1 e4 es 2 lbf3 ltJc6 3 lbc3

rather than Marsh all.

The best equalizing opportunity is 11 .. .f6!.

Still, Black seems to equalize comfortably with 1 2 ...lt:Jcb4!.

Question: How should Kramnik handle Aronian's piece sacrifice?

Question: More tactics to consider, every move. The computer

Question: Should White try to speed things up,

words, h e i s winning ! That i s all that really matters.

Game Fou r: Aronian-Kramnik

2 5th April 2012

Another Berlin Defence and another draw in the main line.

For 13 .l:lfd1!? see the next note.

White's knight on f3 is now pushed away from any attack square.

Some suggested a n swers

Game Five: Kramnik-Aronian

27th April 2012

Question: Why this queen retreat? Is there nothing else?

Question: Which way should Black recapture?

thought for a long time here.

Question: Should White have retreated, rather than exchange queens?

simplification and an excellent chance of forcing a win. Unfortunately, albeit under

An extremely difficult position to try to evaluate. Anand is a vastly experienced grand