beating the grünfeld.pdf

Beating the Griinfeld

Beating the Griinfeld

Beating the Griinfeld ANATOLY KARPOV Translated by John Sugden

B.T. Batsford Ltd, London

First published 1992 © Anatoly Karpov 1992 ISBN 0 7134 6468 2 British Library Cataloguing-in-Publication Data. A catalogue record for this book is available from the British Library All rights reserved. No part of this book may be reproduced, by any means, without prior permission of the publisher Typeset by Lasertext Ltd, Manchester and printed in Great Britain by Dotesios Ltd, Trowbridge, Wilts for the publishers, B. T. Batsford Ltd, 4 Fitzhardinge Street, London WIH OAH

A BATSFORD CHESS BOOK Adviser: R. D. Keene GM, OBE Technical Editor: Andrew Kinsman

Contents Preface

7

1

Seville Variation

11

2

Exchange Variation with 7 �c4 - other systems

41

3

Modern Exchange Variation

52

4 Russian System

108

5

4 �f4 System

135

6

Fianchetto System

1 52

Index of Variations

1 87

Preface I have to admit that the title of this book has an air of sales-talk about it, as similar titles generally do. Anyone trying to think up a recipe for beating this or that opening will scarcely succeed, and the Griinfeld Defence is no exception. But seriously: to chessplayers whose repertoire includes the Griinfeld, the author imagines that the present work will be of considerable interest and use. Hence it is indeed quite possible that the book will help many readers to score wins - with Black if they are adherents of the Griinfeld, and with White if they are looking for a way to combat this defence. The book consists of a collection of games (or fragments of games) which reflect the contemporary state of Griinfeld theory. It is constructed around twenty-five paradigms of play by con­ noisseurs of this opening (like many authors I have a weakness for round figures), but the total number of examples is about ten times higher! The notes to each of the twenty-five principal games constitute a thorough discussion of the currently popular variations. As a result, all the most fashionable systems occurring in grandmaster practice in the last few years have found their way into the book. The selected games are not arranged chronologically but grouped according to themes. For this reason, the freshest examples - those from the recently concluded duel for the world chess crown, which incidentally are the most fully annotated - do not form the culmination of the book, but are placed in the middle of it . . . . This book is written in the same format as the four-volume work probably already familiar to the reader : The Open Game (Semi-Open Game I Closed Openings I Semi-Closed Openings) in Action. That is to say, the scores of all the principal games are given in full, and in analysing the opening the reader will mostly be able to study its relation to the middlegame or even the

8 Preface

endgame. In some games, the later stages are of independent interest and consequently receive fairly detailed notes. The informal layout of material has permitted the author to select contemporary games according to his own discretion (in contrast, say, to an opening monograph, where you also have to include systems that have not been used in practice for a long time). Whereas the four-volume work just mentioned covers the development of theory in the period 1984-7 (the last-but-one World Championship cycle), all the principal Griinfeld games in the present book are from the later period 1988-90, including the most recent World Championship match. In my encounters with Gary Kasparov, the Griinfeld Defence has figured prominently. In our last three matches the most varied systems were tried out, and in our preparations we utilised all the most important theoretical material. As a result of these matches, the theory was in turn substantially enriched. Many ideas employed for the first time by Kasparov and myself have seen an onrush of further developments. Suitable examples of this are included in the book, while practical sources prior to 1 988 are incorporated in the notes to the principal games. I should mention that I usually play the white side of the Griinfeld while Kasparov plays Black, and that both of us strive for victory (not only when playing each other). Such is the nature of this sharp and uncompromising opening, in which playing for the draw is inimical to both sides - although, to be sure, our fierce contests have often ended peacefully. For those who like statistics, let me give my overall score against the Griinfeld in each of these matches. London/Leningrad 1986 : +3 = 6. Seville 1 987: + 2 -1 =7. New York/Lyon 1990 : +1 = 3. As you can see, we have played almost an entire World Championship match of Griinfelds - with a definite plus score in my favour. If you like, you may conclude from this that I have unearthed the secret of Beating the Grunfeld . . . . Obviously it is not only the World Champion and ex-champion who repeatedly contest this popular opening. It belongs to the repertoire of many prominent players. Suffice it to mention such names as Vassily Ivanchuk, Boris Gelfand and Jan Timman. Some of the valuable discoveries by these super-grandmasters will be found in this book. In the course of working on the text, I have utilised a great

Preface 9 many sources (Informator, books and magazines), and this is reflected in the sheer quantity of references to games by masters and grandmasters. In the majority of cases where a variation or individual move is of major significance, its originator is mentioned. In conclusion, I must thank Soviet Master Evgeny Gik, my co­ author in many books, for his help in preparing the manuscript. Anatoly Karpov

1 Seville Variation Game No. 1 Kuzmin-Henkin

Moscow 1989 I 2 3 4 5 6 7 8 9 10

d4 c4 .!Llc3 cd e4 be �c4 .!Llel �e3 0-0

.!Llf6 g6 d5 .!Llxd5 .!Llxc3 �g7 c5 .!Llc6 0-0 �g4

The game has followed the main system of the Griinfeld. I would just remind the reader that in the event of 1 0 . . . cd 1 1 cd .!Lla5 1 2 �d3 �e6 1 3 d5, theory states that White obtains a dangerous attack for the exchange. ll

1 w

f3

.!Lla5 (1)

12 Seville Variation But now Black can answer 1 2 .td3 with 1 2 cd 1 3 cd .te6, and the exchange sacrifice (14 dS) is less dangerous to him, since in several lines he has an important queen check on b6 (the details can be found in any reference work on the Griinfeld). However, in this situation White's exchange sacrifice is by no means forced. We shall later acquaint ourselves with a recent game with 14 .l:l. c l instead (Game No. 6 , Yusupov-Kasparov). . . .

12

.txf7+

So International Master Kuzmin, one of my seconds, has decided to participate in the theoretical debate launched by Kasparov and myself in Seville. After that 1 987 match, Kasparov stated that the plan chosen by White with 12 .txf7 + was unpromising. But to judge from the way the play went in the five games in question, it cannot be said that Black easily solved his opening problems. And in the post-match duel to which we shall presently turn (Game No. 2), he was thoroughly routed. The pawn structure that now arises gives White every reason to count on the initiative, besides which he has an extra pawn. But then again, the position is highly dynamic and may very well suit the taste of the player of the black pieces. Anyway, interest in this variation has not died down in three years, and numerous games enriching its theory have been played. Most of them will be mentioned in this book, over the first five main games. Of course, to some extent it will also be necessary to refer back to the Seville games. 12 13 14

fg ..txfl

.l:l. xf7 .l:l.xfl +

Let me emphasise that the main feature of the position is not the extra pawn; the freedom of Black's game compensates for this minor deficit. White's basic plan is to block up the enemy bishop on g7 by means of the pawn chain c3/d4/e5/g5/h4. Black will rely on tactical devices to enable his bishop to escape onto the h6-c l diagonal. 14 "it'd6 (2) This queen sortie occurred in three games in the Seville match and two of our later encounters. It has also been played in many other games in recent years. Let us nonetheless mention some alternatives. In the ninth game in Seville, Kasparov preferred 14 ... cd 1 5 cd ,

Seville Variation 13 2 w

11V b6, and after 16 �g1 11Ve6 17 11Vd3! I returned the pawn while keeping all my positional trumps. The offshoots arising from 1 4 . . . c d 1 S c d will be examined i n d e tail i n Game No. S (apart from 1 S . . . "it"b6, the moves 1 S . . . "it"d7 and 1 S . . . eS have been seen). The Seville match was not yet over when the new move 14 "it"d7 occurred in a game Chernin-Gavrikov, Lvov 1 987. That game proceeded : 15 de J: f8 + ( 1 S . .. "it"xg4 16 �f4!) 16 � g 1 "it" xg4 1 7 �f4 ( 1 7 "it"d3 �c6 1 8 h3 is not bad either) 1 7 . . . "it" xd 1 + (but not 1 7 . . . J: xf4? 1 8 .txf4 "it" xf4, on account of 19 "it"d8 +) 1 8 : xd 1 .txc3 1 9 � d S (White gains nothing from 1 9 �e6 : c8 2 0 J: d7 �f7 2 1 �gS+ �e8 22 J: d3 .tb4, or in this line 2 1 .tgS? .tf6! 22 .txf6 �xe6) 1 9 . . . .tf6 20 .th6 (the advantage is with White, but Black manages to hold on) 20 . . . J: e8 (but not 20 . . . : f7 2 1 � xf6 + : xf6 22 e S :rs 2 3 g4, and White wins) 2 1 � xf6 + ef 22 J: d7 : xe4 23 J: g7 + �h8 24 : c7 �g8 2S J: g7+ (2S .td2 �c6 26 : xb7 J: e7) 2S . . . �h8 26 : c7 �g8 27 J: g7 + �h8 28 l: c7, draw. Gutman suggests 20 g4! : e8 21 J: b 1 a6 22 �f2, followed by �e2 and .td2, with the initiative. In answer to 14 . . . "it"d7, a more logical move seems to be 15 g5 (3), as played in Karpov-Gavrikov, European Speed Chess Championship, Spain 1 988. Despite the 'non-serious' nature of the contest, the game is of considerable interest : 15 "it"e6 16 eS! "it"c4 ! 17 �g1 J: d8 18 "it" e 1 �c6 1 9 .tf2 a 6 (wouldn't 1 9 . . . b S ! ? have been better?) 2 0 a4 �aS 2 1 h4 .tf8 22 de! �b3 23 .l:!.b l ! (more precise than 23 J: d 1 : xd 1 24 "it"xd 1 lbxcS) 2 3 . . . e6 (but now 23 . . . �xeS loses t o 24 : b4 "it"dS 2S J: d4) 24 c6! be 2S �d4 lbxd4 26 cd "it" xa4 27 "it"c3 (White has an obvious endgame advantage) 27 . . . J: c8 28 hS! gh 29 "it"h3 J: e8 30 "it"xhS J: e7 (30 . . 'lf c2 3 1 J: b7) 3 1 g6 11V c2 32 ooo

o· ·

.

14 Seville Variation 3 B

gh + .l:l. xh7 33 11fg4+ �h8 34 .l:l. f l 11ff5 (34 . . . 11rd3 35 -*.h4!) 35 11' xf5 ef 36 -*.e3 .l:l. h5 37 g4 ! .l:l. h3 38 .l:l. xf5 .l:l. xe3 39 .l:l. xf8 + �g7 40 .l:l. f4! (the rook ending is won for White) 40 . . . a5 4 1 �f2 .1:1. b3 42 .l:l. f6 .l:l. d3 (42 . . . .l:l. c3 43 �e2 a4 44 �d2 etc.) 43 .l:l. d6 a4 44 �e2 .l:l. g3 45 .l:l. xc6 .l:l. xg4 46 �d3 47 .l:l. a6 �f7 48 �c4 .l:l. g3 49 d5 1-0. From diagram 3, a game Gligoric-Popovic, Yugoslavia 1 988, continued differently : 15 .l:l.d8 16 �g1 e6 (an evident improve­ ment is 1 6 . . . cd 1 7 cd �h8 1 8 "ifd3 ll:lc6 1 9 .l:l. d 1 ll:lxd4 20 ll:lxd4 e5) 17 .1:1. b1 ll:lc4 ( 1 7 . . . cd 18 cd ll:lc6 19 d5 ed 20 ed 11fxd5 2 1 "iWxd5 .l:l. xd5 22 .l:l. xb7 is good for White) 1 8 j.f2 b5 1 9 11rd3 a6 (White can now obtain the better game with 20 h4 ll:le5 2 1 11fh3, but Gligoric prefers to move his other rook's pawn) 20 a4 ll:le5 2 1 11fc2 (now 2 1 11fh3 ll:lc6 2 2 ab a b would lead t o unclear play - but not 21 . . . ba 22 de a3 23 .i.e 1 ! etc.) 21 . . . ll:lg4! 22 e5 b � (Gligoric gives the variation 22 . . . ll:lxf2! 23 �xf2 b4! 24 de .i.xe5!, and Black has no problems) 23 .i.g3 11fc6 24 "iWa2 ll:le3 25 ll:lf4 (White has a slight edge in the endgame, but it isn't enough for victory) 25 . . . ll:ld5 26 ll:lxd5 11fxd5 27 "iWxa4 cd 28 cd "iWxd4+ 29 11fxd4 .l:l. xd4 30 .l:l. b6 �f7 3 1 .l:l. xa6 .l:l. d5 32 .l:l. a7 + �g8 33 h4 ..txe5 34 -*.xeS .l:l. xe5 35 .l:l.e7 .l:l. e4 36 g3 .l:l. e 1 + 37 �f2 .l:l. e5 38 �f3 .l:l. e 1 39 �f4 .l:l. e2 t - t . Perhaps a n even sounder answer t o 1 4 . . . 11r d7 i s 1 5 h3. This occurred in Yusupov-Popovic, Belgrade 1 989. There followed 1 5 . . . ll:lc4 1 6 .i.f2 cd 1 7 cd e5 1 8 de ll:ld2 + 1 9 � e 1 .txe5? (according to Yusupov, unclear play results from 19 . . . .i.h6 20 11fc2 .l:l. c8 2 1 11fb2 ll:lxe4 22 "iWb3 + ) 20 .l:l. c 1 .l:l. d8 21 "ifc2!, and in the endgame White made no mistake in exploiting his extra pawn. In addition to 14 . .. cd, 1 4 . . . 11rd6 and 14 . . . 11rd7, Black has . . .

Seville Variation 15

one other option : 1 4 1Wc8. This move was tried out in Portisch­ Korchnoi, Reykjavik 1 988. After 15 1Wa4?! Black obtained an active game with 1 5 . . . cd 16 cd lbc4 17 .tf4 a6 1 8 g5 b5 19 1Wb3 e5 20 de 1Wc6. At this point, instead of 21 lt d 1 , White should have preferred 21 1Wd3!? lbxe5 22 1Wd5+ 1Wxd5 23 ed b4, with about equal chances (but not 23 . . . lbd3 24 lt d 1 lbxf4 25 lbxf4 ltf8 26 g3 .te5 27 d6 .txf4 28 gf : xf4+ 29 "'e2, with initiative for White). Korchnoi answered 2 1 ltd1 with 2 1 . . . lt f8!, whereupon Portisch played 22 ltd5, overlooking the blow 22 . . . 1Wxd5!. After 23 ed lbd2+ 24 "'el lbxb3 25 ab lt d8, Black gained a decisive endgame advantage - though it took him fifty more moves to achieve the win! Evidently White should react to 14 . . . 1Wc8 in the same way as if the queen had gone to d7, with 15 h3 or 15 g5. Seirawan­ Kudrin, USA Ch. 1 989, went 14 . . . cd 1 5 cd 1Wc8 16 g5, when the black queen utilised its possession of the c-file with 1 6 . . . 1Wc4. However, after 17 "'g1 lt d8 18 1We1 lbc6 19 e5 e6 20 1Wf2 lbb4 21 lDf4 ltc8 22 ltfl , it became clear that the raid with the queen had achieved nothing and the white centre was invulnerable. In the endgame after 22 . . . lDd3 23 lbxd3 1Wxd3 24 1Wf7+ "'h8 25 .tf2, White had a material and positional plus, which he duly converted to a win. • • .

15

e5

In the 1 1 th match game in Seville, I chose 15 "'gl . For the current state of theory on that move, see the notes to Game No. 4, where the completely new 1 5 1Wa4!? will also be discussed. 15 1W d5 (4) 15 -. e6 has also been seen. After 16 g5 lbc4 ( 1 6 . . . 1Wc4 1 7 "'g1 transposes into Karpov-Gavrikov, where play resolved itself clearly in White's favour), White has to avoid the trap 1 7 1Wd3? 1Wf5 + ! which costs him a piece; but even with 17 .tf2 lt f8 1 8 "'g1 1Wf7 1 9 1W e 1 lba3 20 ltc l 1Wxa2, h e achieves nothing. A game Makarov-Hodko, USSR 1 988, went 16 h3 lbc4 17 1Wd3 ltd8 1 8 1We4! ( 1 8 lbf4 1Wa6!) 18 . . . 1Wc6 19 1Wxc6 lbxe3 + 20 "'f2 lbxg4+ 21 hg be 22 lt b 1 cd 23 cd c5 24 "'e3! cd+ (24 . . . lt d5 25 "'e4 e6 26 : b5 .tf8 27 : a5 is bad for Black) 25 lbxd4 : d5 26 lbc6. At this point, in Makarov's view, Black should have played 26 . . . lt c5; after 27 lt b8+ .tf8 28 lt c8 "'g7, White has no more than a slight edge. Quite a good reply to 1 5 . . . 1We6 seems to be 1 6 lDf4 1Wc4+ 17 •••

16 Seville Variation 'ife2! : f8 1 8 'ifxc4+ 'Oxc4 1 9 �e2, with the better ending for White. One other possibility, 16 � g1 was tried in Hansen­ Fereec, Aosta 1 989. Black restored the material balance, but after 1 6 . . . 'ifxg4 1 7 'iVd3 'ife6 1 8 -*.g5! : f8 1 9 h3 cd 20 cd -*.xeS 2 1 de 'ifxe5 22 -*.c l ! 'ifxa1 2 3 "ird5 + e6 24 'ifxa5 : c8 2 5 'iVd2 b5 (doubtless a more accurate line was 25 . . . 'iVb 1 26 �h2 'ifc2 27 1We3 "irxa2 28 'Oc3 "irb3 29 .i.d2, with a minimal plus for White) 26 a3, his position was fairly difficult. ,

4 w

16

.i.f2

In this case 16 'Of4 is weak, since the queen gets to e4. A game Lichak-Asrian, Leningrad 1 990, went 16 . . . : f8 1 7 �g1 "ire4 1 8 "irf3 "irxf3 1 9 gf .i.h6 2 0 10d5 -*.xe3 + 2 1 'Oxe3 cd 22 cd 'Oc6 23 : b 1 b6 24 �g2 'Oxd4, and Black had an endgame advantage. 16

:rs

So far, the play coincides with game 5 in Seville. In the 7th match game, the black rook preferred to go to d8, forcing the white queen off the central file. In later games, 16 . . . : d8 completely replaced the transfer of the rook to f8; we shall go into details in the notes to the next game in this book. 17 gS (5) A valuable novelty. The game in which this position first arose (number 5 in Seville) continued 17 �g1 .*.h6 1 8 h4 "irf7 19 -*.g3 .i.e3 + 20 �h2 "irc4! 21 : b 1 (2 1 de is met by 21 . . . "irxg4, with a good game) 21 . . . b6 22 : b2 (but here 22 de is sounder) 22 . . . "ird5! 23 "ird3 'Oc4 24 : b 1 b5, and the sharp duel should have ended in a draw (although a blunder by Kasparov eventually enabled me to win). However, 24 . . . g5 would have given Black good winning chances.

Seville Variation 17

With 17 g5, White achieves his principal aim of shutting off the opposing bishop. Admittedly his king is dangerously placed opposite the black rook, but this is just a temporary problem.

5 B

17

Wf7

Gutman gives the variation 1 7 . . . "ife4 (or 1 7 . . . tt:Jc4 1 8 �g1 "ife4 1 9 tt:Jg3 "iff4 20 "if e2 b5 2 1 tt:Je4, with the better game for White) 1 8 tt:Jg1 ! tt:Jc4 1 9 ltlf3 ltle3 + 20 .i.xe3 'if xe3 2 1 "iib 3 + �h8 22 J:te 1 !, and Black's position is not to be envied. 22 . . . J:txf3+ may look inviting, but after 23 gf "if xf3 + 24 �g1 "ifg4+ 25 �f2 "ifh4+ 26 �e2 "ifg4+ 27 �d2 "ifxg5+ 28 �c2, there is no perpetual. 18

"ife1

h6

In reply to 18 'if f5 Henkin gives these variations : (a) 19 h4, and now : (a1 ) 19 "ifg4 20 �g1 h6 (or 20 . . . ltlc4 2 1 tt:Jg3! "if xh4 22 tt:Je4, with a distinct plus) 2 1 gh .i.xh6 22 ltlg3! cd 23 cd "if xh4 24 tt:Jf5! "ifg5 25 tt:Jxh6+ "if xh6 26 "if xa5, and wins. (a2) 19· h6 20 gh .i. xh6 21 ltlg3, with a clear advantage. (a3) 19 "if e4 20 �g1 cd 2 1 cd ltlc4 22 J:t c l ! (but not 22 g3? .i.xe5 23 de tt:Jxe5) 22 . . . ltle3 23 ltlf4!, and if 23 . . . l:t xf4 24 .i.xe3 ,J xh4 25 J:tc8 + , White has a considerable initiative. (b) Another playable line is 19 ltlg3 "if xg5 20 tt:Je4 'iff4 21 �g1 cd 22 cd tt:Jc6 23 .1: b 1 ! with the advantage. In a game Kuzmin-Malishauskas, USSR 1 989, Black played a new move, 1 8 ltlc4 ( 1 8 . . . ltlc6 is also possible). There followed 19 tt:Jg3 tt:Jb2 20 tt:Je4 tt:Jd3 2 1 "ife3 (2 1 "ife2 tt:Jf4 !) 2 1 . . . tt:Jxf2 22 • • •

• • •

• • .

...

•• •

,

18 Seville Variation

lt:lxf2 (Kuzmin recommends 22 . . . "it'c4 + , without the exchange in the centre; after 23 �g1 .z:td8, Black has the initiative) 23 cd "it'c4 + 24 �g1 .z:td8 25 "it'b3 "it'xb3 26 ab llxd4 27 llxa7 J.xe5 1-t (Black would lose after 27 . . . llb4 28 lt:ld3 ll x b3 29 lla8 + �f7 30 e6+ !). A much stronger answer to 1 8 . . . lt:lc4 is 19 lt:lgl! followed by 20 lt:lf3. 19

gh

Stronger than 19 lt:lg3 hg 20 lt:le4 cd 21 cd "it'f5. 19 20

J.xh6

� g1 lt:lc4 An inadequate alternative is 20 . . . J.d2 2 1 "it'xd2 "it'xf2+ 22 � h 1 �g7 23 "it'd3 lt:lc6 24 lldl . 21

lt:l g3!

J.dl

Otherwise White is a clear pawn up. The bishop move to d2 would also have been a good reply to 21 ..th4. 22 "it'el ..txc3 23 llfi! 23 e6? "it'xf2 + 24 "it'xf2 llxf2 would lose for White. "it'e6

23

If 23 cd, then 24 e6! is immediately decisive. On 23 lt:ld2?, White has 24 .z:td 1 cd (24 . . . "it'xa2 25 lt:le4!, or 24 . . . "it'f4 25 de!) 25 llxd2 J.xd2 26 "it'xd2 d3 27 -*.xa7. • ••

• ••

24 25

de! lle1

..txeS "it'dS

25 . . . "it'f7 or 25 . . . llf4 would be safer. 26

lld1

If 26 "it'g4 (26 lt:le4 ..td4!), Black has 26 . . . llxf2! when Henkin's analysis goes : 27 "it'xg6+ �f8 28 �xf2 (28 "it'h6+ J.g7) 28 . . . J.d4 + ! (28 . . . "it'd4+ 29 �fl lt:le3 + 30 llxe3) 29 �fl lt:le3 + 30 : xe3 (but not 30 �e2 "it'xg2 + 3 1 �d3 "it'c2+) 30 . . . ..txe3 3 1 "it'f5 + "it'xf5 3 2 lt:lxf5 ..txc5, and Black seizes the initiative. 26

ll xf2

After 26 . . . "it'e6 or 26 . . . "it'f7, Black could still offer resistance, but now the game ends at once. 27 28 29

30

llxd5 lt:lxe2 c6! : xeS! 1 -0

llxel

� f7 b6

Seville Variation 19 Game No. 2 Karpov-Kasparov Belfort 1988 1 d4 ll:lf6 2 c4 g6 3 ll:lc3 d5 4 cd ll:lxd5 5 e4 ll:lxc3 6 be .tg7 7 .tc4 c5 8 ll:le2 ll:l c6 9 .te3 0-0 10 0-0 i.g4 1 1 f3 ll:la5 12 .txf7 + .:t xf7 13 fg .:t xfl+ 14 �xfl 'it'd6 15 eS 'it'dS

16 i.f2 .:t d 8 Mter the 7th game in Seville, in which Kasparov played this move for the first time, the rook manoeuvre to d8 became standard practice in this position. On d1 the queen was quite conveniently placed, but now the threat of 1 7 . f .txeS compels it to leave its post. White gains nothing from 17 ltlf4 'ifc4+ ( 1 7 . . . 'iff7 1 8 ll:lh3!) 18 11Fd3 'ifxd3+ 19 ll:lxd3 cd 20 cd ll:lc6 21 ltlcS ll:lxd4 22 ll:lxb7 .:tdS 23 .:t e l : xeS 24 : xeS .txeS 2S ll:ld8 a6, with equality. There are three ways for the queen to vacate the d-file : along the d l -a4 diagonal (to a4 or c2), or with 17 'ifel. We examine 17 'ifa4 in the present game, and 17 'ifcl in Game No. 3. But first, let us recall how the Seville game proceeded : 17 'ife1 'ife4 1 8 gS 'iffS! 19 h4 ltlc4 (he should have put pressure on the centre with 19 . . . ll:lc6 20 �gl 'ife4, leading to sharp play) 20 �g1 'ifg4 (20 . . . bS was more precise, as after White's next move, the black knight will feel uncomfortable) 21 a4 h6 22 : a2! hg 23 'it'bl ! (White thus succeeds in exploiting the weakness of the a2-g8 diagonal) 23 . . . gh 24 'it'b3 'ife6 2S ll:lf4 'iff7 26 ll:lxg6 'ifxg6 (I shall refrain from further comment until we reach one particular critical position; this book is not the place for exhaustive notes on the Seville games, they are no doubt well known to the reader already) 27 'ifxc4+ �h8 28 .:t b2! cd 29 cd 'ifg4 30 'iff7! .J:I. xd4 3 1 .txd4 'ifxd4+ 32 .:t f2 'ifxe5 33 : fs 'ife l + 34 :n 'ifeS 3S �hl? (although we are quite a long way out of the opening, it is appropriate to take stock of the situation. With 3S 'iff4!, White should come out on top : 35 . . . 'ifxf4 36 .:txf4 i.f6 37 : c4 �g7 38 .:t c7 b6 39 �fl . with a won ending) 3S . . . b6 36 'iff4 'ifhS (a withdrawal that was impossible a move earlier. If now 37 'it'b8 + �h7 38 'ifxa7, the white queen is far from the scene of action, and after 38 . . . 'ife2 Black has enough initiative to draw. But White has another way to exploit the awkward position of the black queen on the rook's file) 37 'iffS? (at this point, 37 .:t f3! i.f6 38 1Wb8 + �g7 39 'ifxa7 1Wc5 40 1Wa6 was decisive) 37 . . . 1We2 38 .:t e l ? (here was White's third successive opportunity to win this .

20 Seville Variation game. He should have checked first - 38 "it'c8 + �h7 - and only then played 39 :t e l , threatening 'iltc2 + . If 39 . . . 'iit h 5, then 40 'iltc2+ �h8 41 'iitd l ! etc.). The game lasted another 40 moves, but all my efforts to overcome my opponent were in vain. Although Black was on the brink of defeat more than once in this game, we have seen that the opening was not to blame. After 1 7 'ilt e l , Black had the means to obtain a perfectly reasonable position. 17

'ilta4

(6)

6 B

Now after the reply 1 7 . . . b6, my queen would withdraw to c2. You might ask what point there is in provoking a useful move of Black's b-pawn instead of placing the queen on c2 at once. The explanation is that in some variations (after Black's . . . "it'c4, for example) White can play 'ilte4, gaining a valuable tempo by attacking the knight which would by now have returned to c6. So 1 7 'ilta4 has its plus side. But I will not offer a categorical verdict as to where the queen feels more comfortable, although I have tried both possibilities in my games. 17

b6

In the event of 17 .. . lLlc6 1 8 "it'b3 c4 19 'iltxb7, White has the better game. After 17 lLlc4 18 lLlf4 'iltf7 19 g3 lLld2 + 20 �g2 lLle4 2 1 'iltc2 lLlxf2 22 'iltxf2 White would again have a plus, but at move 20 Black has the powerful 20 . . . g5 !, pointed out by Gutman. He suggested that White should play 18 g5! himself, with the continuation 18 . . . l:t f8 19 �g1 'iltf7 20 ..tg3 lLle3 21 ..tf4 lLld5 22 l:t fl . Black's initiative has been neutralised and his bishop imprisoned, which is to say that the opening contest is settled. .. .

Seville Variation 21

After 17 l:f8 1 8 g1 'llf7 1 9 .*.h4! ll:lc4 20 'ilb3, the bishop is again unable to break out and free itself: 20 . . . .*.h6 21 g5! .i.xg5 22 .*.xg5 'ilf2 + 23 '>Ph 1 'ilxe2 24 h3, with a substantial advantage to White. 18 'ilcl There is nothing more for the queen to do on a4. • • •

18

l:f8

After this, the black pieces come under pressure. Evidently 18 l: c8 (7) was more exact, trying to obtain counterplay along the c-file.

•••

7 w

From diagram 7, here are some important recent examples : Tisdall-Thorsteins, Reykjavik 1 989, went 19 de be ( 1 9 . . . l: f8? 20 cb ll:lc4 21 g 1 ) 20 l: d 1 'ilxe5 21 'ila4 l: f8 22 l: d3 c4 23 l: f3 'lidS 24 l: xf8 + xf8 25 .i.xa7 ll:lc6 26 h3 'ile4 27 'ilb5! and White had a considerable plus. However, 20 . . . 'llc4! is more accurate : 21 .i.g3 'ilxg4 22 h3, with chances for both sides. Lputian-Dzhandzhgava, Simferopol 1 988, went 19 'ildl l:d8 20 'ilcl (20 'ilc2 repeats moves) 20 . . . l: f8 (the position that has now arisen is the same as in my game with Kasparov, except that the white queen is on c l . The difference is not too important, but . . . ) 21 h3 (this is a serious inaccuracy. Correct moves are 2 1 g 1 and 2 1 g5) 2 1 . . . W"f7 2 2 W" e l .th6 2 3 ll:lg3 ll:lc4 24 e 6 W"g7 25 ll:le4 .te3 26 W"e2 b5 27 '>Pg 1 .txf2+ 28 ll:lxf2 cd (Black has won his pawn back and seized the initiative; the game doesn't last much longer) 29 cd W"xd4 30 :Z. c l g5 31 l: c2 ll:le3 32 l: d2 'ila l + 33 ll:ld l l:fl + 34 h2 W"e5+ 35 g3 'ile4 36 l: d8 + g7 37 'ilb2 + r;Ph6 0- 1 .

22 Seville Variation

In Douven-Ilincic, Alma-Ata 1 989, after 1 9 1W d 1 l:t d8, White played 20 ll'el , an improvement on the previous example. There followed 20 . . . .!tlc4 (a more precise method was 20 . . . cd 2 1 cd -th6 22 �g1, and only then 22 . . . .!t!c4) 2 1 g5 1We4 22 �g1 1Wg4 23 .ltlg3 cd 24 cd 1Wxg5 25 1Wb4 l:t c8 26 .ltle4 1Wf4 27 1Wxe7 �h8 28 We6 l:t f8 29 1W xc4 1Wxe4 30 : e 1 , with a clear advantage. 19 .ltlf4 is well answered by 19 . . . 11'f7 or 19 . . . 1Wc4+ 20 1Wd3 -th6. Another inadequate line is 19 1W d 2 : f8 20 1Wg5 1Wf7 2. 1 1Wh4 -th6! (but not 2 1 . . . .!tlc4? because of 22 g5!, as in Schneider­ Ljubojevic, Pernik 1 988) 22 �g1 .!tlc4!, and the initiative is with Black. At move 20, the unfortunate 20 �g1? led to a quick loss in Schulz-Filipovic, Korimofen 1 989 : 20 . . . -*.xeS 21 l:t d 1 -txh2 + ! It only remains t o add that 18 1W c4 would be met b y 1 9 1We4 with favourable consolidation, for instance : 1 9 . . . l:t f8 20 �g1 ! -th6 2 1 -th4!. Notice one peculiarity of the position resulting from 1 7 . . . b6: Black now constantly has to reckon with the exchange d4xc5. . ••

19

�g1

1Wc4

The bishop cannot now break out onto the open board: 1 9 . . . -th6 20 h4 1Wf7 2 1 .ltlg3, or 1 9 . . . .!t!c4 20 h4. 20

1Wd2!

White continues with his plan of restricting the bishop's mobility. After 20 1We4, he would have to reckon with 20 . . . -th6 and especially with 20 . . . .!tlc6!?, threatening to capture on e5; naturally, 2 1 1Wxc6 1W xe2 cannot arouse White's enthusiasm. 20

1We6

20 . . . 1Wf7 is not good, since after 21 .ltlg3 everything fits together for White : his knight is transferred to e4, and his queen to e2. Nor does 20 . . . -th6 21 1Wxh6 1W xe2 rid black of his worries, on account of 22 1We3 1Wxg4 23 de be (23 . . . .!tlc4 24 1Wd4) 24 1Wxc5. 21 22

h3 ll'gS! (8)

.!tlc4

A critical moment. In addition to .!tle2-f4, White will now be threatening -tf2-h4 in some variations. 22

b6

22 . . . -tf6 is no good, if only because of 23 ef (23 .ltlf4 is also strong) 23 . . . ef 24 .!tlf4. 11'£7 23 ll'cl To obtain counterplay Black had to opt for 23 . . . b5, intending

Seville Variation 23 8 B

(for example) 24 ll:lf4 'flf7 25 ll:ld3 b4 !? even though 25 .tg3 maintains a plus for White. 23 .. . 11d5 would not work in view of 24 'flc2!, while after 23 . . . h5 the queen would return to g5. -

24

.tg3

g5

This move was roundly condemned by the commentators. But I do not recall that any serious alternative was suggested to 11d5 (for example) has been give Black counterplay. 24 recommended, but then 25 ll:lf4 'fle4 26 ll:le6 would put Black in a very dangerous position. There are three variations - let us look at them : (a) 26 .l:l. c8 27 'ilb 1 ! 'fle3 + 28 .tf2 'flxc3 + 29 'flxg6 'flxa 1 + 30 �h2, with unavoidable mate. (b) 26 ll:le3 27 11d2 cd 28 cd .l:l. c8 29 .l:l. e 1 .l:l. c2 30 .l:l. xe3 'flc6 3 1 d5, and Black is lost. He could have tried driving the queen to e2, with 27 . . . ll:lc4 28 'fle 1 ll:le3 29 'fle2 cd. Then 30 cd .l:l. c8 would give Black strong counterplay, but 30 ll:lxf8 .txf8 (30 . . . d3 3 1 11f2) 3 1 'flf3! 'ild3 32 cd 'ilxd4 33 .l:l. e 1 clears up the situation completely - since 33 . . . ll:lc2+ 34 .tf2 ll:lxe 1 is refuted by the intermediate 35 'ilb3 + . (c) 2 6 cd 27 ll:lxf8 ltle3 2 8 11d2 d e 2 9 'fle2 .txf8 3 0 'flf3, with a big advantage. 25 'flc2 11d5 •• •

. . •

. . .

• • •

26 27

.tf2 ll:lg3

b5 .1:1. £7

Forced. If 27 . . . b4, then 28 ll:lf5 is unpleasant, while 28 . . . .l:l. f7 is met by 29 e6 'flxe6 .l:l. e 1 11d7 3 1 cb. 28 .l:l.el I f White had wanted, he could have put a stop to Black's queenside counterplay with 28 : c t .

24 Seville Variation 28 29

1Wg6

b4 �f8

29 . . . be loses immediately to 30 o!Llf5 f8 3 1 e6 .l:l. xf5 32 gf o!Lld6 33 de. 30

o!Lle4

30 o!Llf5 is also strong: 30 . . . e6 3 1 o!Llxh6 .l:l. f4 32 1W xg5 , or 32 1Wh7. .l:l. xfl

30

The exchange sacrifice is incapable of improving matters. All it demands from White is a modicum of accuracy. 31 32 33 34 35 36 37 39

�xf2 'tiffS+ 1Wc8+ 1W xc5 g1 o!Llg3 o!Llf5 .l:l. c 1 1-0

be �g8 h7 1Wf7+ c2 .i.f8 �g8

Black's pieces on the kingside never succeeded in breaking free. As this game was played in the World Cup in the French town of Belfort, someone jokingly called 1 7 1Wa4 the Belfort variation . . . . I might add that this game was declared the best game played in the first half of 1 988 (Informator, vol. 45). Also, it proved to be our last decisive encounter until the following (already our fifth) World Championship match. Game No. 3 Karpov-Timman

Rotterdam 1989 1 d4 o!Llf6 2 c4 g6 3 o!Llc3 d5 4 cd o!Llxd5 5 e4 o!Llxc3 6 be J.. g7 7 J.. c4 c5 8 o!Lle2 o!Llc6 9 J.. e3 0-0 10 o-o J.. g4 1 1 f3 o!Lla5 1 2 J.. xf7 + .l:l.xf7 13 fg l:Xfl+ 14 �xfl 1Wd6 1 5 e 5 1Wd5 16 J.. f2 .l:l. d8 In the fifth game in Seville, Kasparov chose 1 6 . . . .l:l. f8 (see Game No. 1 ). As we know, the rook move to d8 was first employed in the 7th game of the match. 17

1Wc2

In the first game with this line, I chose 17 'tlfe l (see Game No.

Seville Variation 25 2); in Belfort 1 988, I preferred 1 7 'ilt'a4 (Game No. 2). 17 'ilt'c4 In Griinberg-Ilincic, Prague 1 989, an equal game resulted from 1 7 . . . .l:t c8 1 8 illf4 'ilt'f7 1 9 'ilt'e4 J.h6 20 g3 .l:t f8 2 1 ¢>g2 J.xf4 22 gf 'ilt'xf4 23 'ilt'xf4 .l:t xf4 24 ¢>g3 .l:t f8. 18

'it' b2

But not 1 8 g5? on account of 1 8 . . . cd. If 1 8 'ilt'e4, then 1 8 . . . .l:t d5!, threatening 1 9 . . . .l:t xe5, is quite good for Black. J.h6 18 In Ljubojevic-Timman, Linares 1 989, Black moved his rook again : 18 .l:t f8. There followed 19 ¢>g1 J.h6 20 .l:t dl ! 'ilt'a4 2 1 .!:te l cd (better was 2 1 . . . illc4 2 2 'ilt'b3 'ilt'xb3 2 3 ab illb2) 2 2 illxd4 'ilt'c4 23 h3 b6 24 illf 3 .l:t d8 25 .td4 (25 g5! is also playable) 25 . . . J.f4 26 ¢>f2 1Fd5 27 'it' b l ill c4 28 'ilt'e4 ill b2 29 'ilt'c2 'it'b5 30 .l:t b l illd 3 + 3 1 ¢>fl 'ilt'c4 32 'ilt'e2 b5 33 ille l ! illx e5 34 .l:t xb5 'ilt'f7! 35 ¢>g1 illc 6 36 illf 3 illxd4 37 cd .l:t c8 (but not 37 . . . J.d6 38 illg 5 'ilt'f4 39 'ilt'e6+ ¢>g7 40 'ilt'xe7 + with advantage, or 39 . . . ¢>f8? 40 .l:t f5 + and wins) 38 .l:t c5 .l:t b8, and the players decided to draw by repetition. It remains to be noted that at move 28, White could have kept the initiative with 28 .l:t e2! After 18 . . . .l:t f8 19 ¢>gl J.h6 20 .l:t d 1 , an innovation, 20 'ilt'e6, was played in L. Hansen-Jasnikowski, Warsaw 1 990. (Instead, 20 . . . 'ilt'f7 2 1 .!Llg3! illc4 22 'ilt'e2 'ilt'xf2+ 23 'ilt'xf2 J.e3 24 ille4! .l:t f4 25 .l:t d3 is hardly good for Black.) However, after 2 1 h3 .!Llc4 22 'ilt'xb7 .te3 (22 . . . ille 3 23 .!:ta l 'ilt'f7 24 J.xe3 J.xe3 + 25 ¢>h l 'ilt'f2 26 'it'b5 cd 27 illx d4!) 23 J.xe3 illx e3 24 .l:t d3 'ilt'f7 (or 24 . . . .l:t fl + 25 ¢>h2 ill xg4+ 26 hg 'ilt'xg4 27 .l:t f3 'ilt'h5 + 28 .l:t h3 'ilt'xe2 29 'ilt'xe7 .l:t f7 30 'ilt'xc5 and wins - Hansen) 25 'ilt'f3! 'ilt'xf3 26 gf .l:t xf3 27 de! White had a clear endgame advantage. ...

• ••

19 9 B

h4 (9)

26 Seville Variation

This position first occurred in Karpov-Kasparov, Amsterdam 1 988, which proceeded as follows : 19 11f7 20 �gl .J:I. f8 21 .!Dg3 (2 1 .ig3 is met by 21 . . . .ie3 + 22 �h2 11e6 23 g5 11g4) 21 . . . .!Dc4 22 'ife2 11xf2 + ! (other continuations lead to advantage for White : 22 .id2 23 .!De4, or 22 . . . cd 23 cd 11xf2+ 24 11xf2 .ie3 25 11xe3 .!Dxe3 26 .J:I. c l ) 23 11xf2 .ie3 24 11xe3 (24 .J:I. fl ? .J:I. xf2 25 .J:I. xf2 cd) 24 . . . .!Dxe3 25 de .J:I. c8 (but not 25 . . . .!Dxg4 26 .J:I. b l , when White i s already better) 2 6 .J:I. b 1 .J:I.xc5 2 7 .J:I.xb7 .!Dxg4 28 .J:I. b4 (White may also play 28 : xe7 : xc3 29 .!Dfl : a3 30 e6 .J:I. xa2, which quickly draws) 28 . . . h5 29 .!De4 .J:I. xe5 30 g3 (30 �fl .!De3 + 3 1 �e2 .!Dd5 32 .J:I. b 8 + is simpler) 30 . . . �f7 3 1 �g2 �e6 32 �f3 (a more accurate choice is 32 .!Dg5+ �f6 33 .J:I.f4+ .J:I.f5 34 .J:I.e4 .J:I.f2 + 3 5 �g1 .J:I.xa2 36 .:t f4+ �e5 37 .J:I.e4+ �d6 38 .J:I. e6+ �d7 39 .J:I. xg6, and a draw can be agreed) 32 . . . �f5!. Black now has slightly the better endgame, not that there is much danger to White. After a further two dozen moves, we agreed a draw. The question arises whether the strange retreat with the queen to f7 can really solve all Black's problems. It turns out that this is not quite the case. In Vyzhmanavin-Ernst, Stockholm 1 990, Black was dealt an annihilating blow, radically altering the assessment of the variation. The game went 19 . . . 11f7 20 g5! (White isn't afraid of the pin on his bishop. The main thing is to shut the opposing bishop out of play) 20 . . . .!Dc4 2 1 e6! (this is just the intermediate move that constitutes the refutation of Black's defence. White deflects the queen, now allowing his opponent to double on the f-file, and only afterwards picks up the b-pawn) 2 1 . . . 11f5 2 2 .!Dg3 11xe6 2 3 11xb7 .J:I. f8 24 .J:I.e 1 ! (now Black can't avoid loss of material) 24 . . . .!De3 + 25 �g l .J:I.xf2 26 gh cd 27 cd .:t f8 28 11b2 lljc4 29 .J:I.xe6 1 -0. Perhaps Timman had a foreboding of such misfortunes, and therefore avoided the move that justified itself in Amsterdam ( 1 9 . . . 11f7), preferring t o occupy the f-file with his rook. • • .

19 20

.1:1.£8 g5!

The stock manoeuvre; here too it guarantees White a plus. 20 11d3 21

'ilbl!

Gaining a couple of tempi for bringing the queen to the defence.

Seville Variation 27 A mistake would be 2 1 gh? lbc4 22 1i' c l 1Wf5 23 it'e l lbe3 + 24 � g l lbc2. 21 22 23

it'el � gl

1i' e3 R.g7

But not 23 lbc l ? cd and everything is fine for Black, as after 24 1t'xe3de, he wins. 1t' e4 (10)

23 10 w

24

lLig3!

White returns the pawn but seizes all the key squares with his pieces. A familiar precept! 24 25

1t'xh4 lbe4

25 lLif5 is also strong : 25 .. . 1t'xg5 26 lLixg7 irg4 (26 . . . �xg7 27 R.h4 1i'g4 28 R.xe7, etc.) 27 d5 �xg7 28 R.xc5 1t'd7 29 1i'e4, and Black is in a bad way. l:txf2

25

An exchange sacrifice (again not entirely voluntary!) analogous to the one that occurred in Karpov-Kasparov at Belfort. 25 . . . 1t'g4 does not help either : 26 lLixc5 b6 27 1i'e4 1t'xg5 28 1i'd5+ l:t f7 29 lbe6, and White should win. 26 27

lLixf2 l:tdl !

cd d3

On 27 . . . lLic6, I had in mind 28 cd lbxd4 29 1i'e3 lbc6 30 1t'b3+ �f8 3 1 1i'xb7 1t'c4 32 1i'c8 + �f7 33 lbg4!. Nor is 27 . . . de 28 1i' xc3 lbc6 29 1t'b3 + any better for Black. lLic6 28 1We3! 28 . . . lLic4 is well answered by 29 1W d4.

28 §eville Variation 29 30 31 32 33

If 33

. .

tl:lxd3 'lrf3!

e6 tl:lf4 tl:ld5!

'lra4 'lraS tl:ld8 .i.eS 'IreS+

. tl:lxe6, then 34 l:t fl ! is decisive. 34

�h1 1-0

There is no defence against the threats of 35 tl:lf6+ and 35 : fl. Game No. 4 Naumkin-Neverov

Moscow 1989 1 d4 tl:lf6 2 c4 g6 3 tl:lc3 d5 4 cd tl:lxdS 5 e4 tl:lxc3 6 be .i.g7 7 .i.c4 c5 8 tl:le2 tl:lc6 9 .i.e3 0-0 10 0-0 .i.g4 11 f3 tl:laS 12 .t.xf7+ .:.x£7 13 fg J: xfl + 1 4 �xfl 'ir d6 15 �g1

In the 5th and 7th match games in Seville I continued 1 5 e5, and the current state of theory on that line is given in the notes to the first three games of this book. In the 1 1 th match game, however, I played the king to g1 here. White is prepared to return the pawn in order to obtain a sturdy centre. But before going on to a detailed discussion of the prospects for either side, it is worth mentioning one other possibility : 15 'lra4!? This occurred in Zakharov-Henkin, Voronezh 1 989, where Black replied 1 5 . . . 'lrxh2! (11) . ll w

A bold piece sacrifice. Instead, the timid 1 5 . . . b6 1 6 � g 1 cd 1 7 cd 'lre6 1 8 d5! 'lrxg4 1 9 :tel , followed by 20 h3, is in White's favour.

Seville Variation 29 The game proceeded : 1 6 'ifxa5 l:l. f8 + 1 7 �e1 (but not 1 7 J.. f2? on account of 1 7 . . . 'if h4! 1 8 g3 'ii' h 1 + 1 9 �g1 'ifh2, and wins) 1 7 . . . 'ii' h 1 + 1 8 �g 1 ! (on 1 8 .!.g1 ?!, Black has the very strong 1 8 . . . ..th6! 1 9 'ifxc5 'if xg2, threatening 20 . . . 'iffl mate) 1 8 . . . 'ifxg2 (but here, 1 8 . . . J.. h 6 comes too late : 1 9 �d2 l:l. f2 + 20 �d3 c4+ 21 .;;.xc4 J.. x e3 22 'ii' d 5+ .;;.g 7 23 l:l. e 1 , with the better endgame chances for White; if he prefers, he can give perpetual check by 23 �f3 'ifxa 1 24 'ife4+) 19 'ii' b5 ..th6 (after 19 . . . cd 20 cd 'if xe4 2 1 'ife2 and 22 l:l. d 1 , White consolidates; but now, a highly unusual distribution of forces comes into being) 20 'ife2 'ifg3+ 21 �d2 l:l. f2 22 J.. x h6 l:l. xe2 + 23 �xe2 'ifh3! 24 J.. e 3! 'ifxg4 25 l:l.fl ! h6! (a very strong move according to Henkin, who gives the following variations : 25 . . . cd 26 J.. h 6! de+ 27 �c1 'ifc8 28 �d4, threatening �e6 and l:l.f8 +; 25 . . . h5 26 �f4, followed by l:l. g 1 ; or 25 . . . �g7 26 �f4 g5 27 l:l. g 1 'ifh4 28 l:l. xg5 + ! . Now the game heads towards a draw) 26 J.. x h6 'ifxe4 27 l:l. f8 + �h7 28 .!.g5 'ife6 29 a4 cd 30 cd 'ifa2+ 3 1 .;;.el 'ifxa4 32 lH7 + .;;.g 8 33 .J:I. xe7 'ii' b 4+ 34 �f2 a5 35 J.. f6 'ii' b 6! 36 J.. e 5 a4 37 �f4 'ii' b 2+ (Black loses after 37 . . . �f8 38 J.. g 7+ !, or 37 . . . a3 38 �e6!) 38 �g1 ! 'ifc1 + 39 �h2 'ii' b 2+ 40 �g1 'ii' b 1 + 41 .;;.h 2 'ii' b2 + t-t. So the extravagant 1 5 'ifa4 is not dangerous to Black. It must be acknowledged that 1 5 e5, which we have already examined, is the most precise; White can move his king if the need arises. Nonetheless, 1 5 .;;.g 1 should also be studied. 15 'ife6 In Polajzer-Anka, Dortmund 1 988, Black played 15 l:l.d8. It isn't clear that the rook on d8 is useful to Black after 16 'ii' d 3 'ife6 1 7 g5!. However, White replied 16 'if a4 'ifa6 1 7 .J:I.e 1 'ii' d 3 1 8 J.. f 2 �c4 1 9 'ii' b5 l:l.f8 20 'ii' b l 'ii' d2 2 1 'ii' b3 b 5 2 2 'ii' x b5 �e3 23 h3, and now, in Gutman's view, 23 . . . J.. h 6! would have given Black a powerful initiative. 16 'ii' d3 'ifc4 So far, the play has followed the 1 1 th game in Seville, and the text move is better than the capture on g4, which occurred (with a transposition of moves and the insertion of the pawn exchange on d4) in the 9th match game. In Chernin-Malishauskas, Lvov 1987 (played while the Seville match was still in progress), Black didn't hurry to exchange queens, preferring 16 cd 17 cd l:l. d8. After 18 g5 �c4 1 9 J.. f2 b5 20 a4 (20 h3 was worth considering) . • •

• • •

30 Seville Variation

20 . . . ba 2 1 lllf4 'flf7 22 lll d 5 .l: f8 23 i.g3 lllb 6 (but not 23 . . . lllb 2 24 'flc3 a3 25 h3, followed by i.e5, with advantage) 24 h4 (24 lll x b6 'ilb3!) 24 . . . lllxd5 25 ed 'ilxd5, the chances were equal. Instead of 1 8 g5 White can play 18 h3, but after the queen exchange Black's chances, once again, are no worse : 1 8 . . . 'ilc4 19 'ilxc4 lllxc4 20 i.f2 e5 2 1 d5 i.h6 22 h4 (22 a4 llld 6!) 22 . . . i.d2 23 a4 llld 6!. Note that variations involving the pawn exchange in the centre ( . . . c5xd4, c3xd4) will be examined more closely in Game No. 5. 17 'ilxc4 It hardly pays White to avoid the queen exchange : 1 7 'ild2 cd 1 8 cd 'fla6! or 1 7 . . . 'fle6. lllxc4 17 18 i.f2 Vasser Seirawan, who has also incorporated this variation in his repertoire, twice played 18 i.gS (12) here. Both games are worth examining. 12 B

Seirawan-Lputian, St John 1 988, continued 18 h6 1 9 i.xe7 cd (Black can win a piece with 1 9 . . . .l: e 8 20 .i.xc5 b6, but after 2 1 i.xb6 ab 22 e5, White obtains more than enough pawns for it. Furthermore, once his king is centralised and the knight jumps to e4 via g3, the white pawn mass can become highly mobile) 20 cd .l: e8 21 .l: c 1 lll a 5 (2 1 . . . .l: xe7 22 ::txc4 .l: xe4 23 �f2 .l: xg4 24 J:t c8 + �h7 25 .l: c7 is no better for Black, but 2 1 . . . b5 was worth considering) 22 .l: c7 lllc 6 23 i.c5 J:t xe4 24 ci>f2 J.xd4+ (in the event of 24 . . . lllxd4, Lputian gives 25 .1: xg7 +! �xg7 26 i.xd4+ �f7 27 �f3 J:t e6 28 i.xa7 J:t a6 29 J.e3 J:t xa2 30 i.xh6 b5 3 1 lllc 1, with advantage) 25 lllxd4 lllx d4 26 .l: xb7 lllc 6 .• .

Seville Variation 31 27 h3 :la4 28 a3. White has an extra pawn, which he eventually exploited to win (though it took him 30 moves!). Seirawan-Hort, Lugano 1 988, went 18 cd 19 cd eS 20 :lc1 (20 d5 h6 2 1 .i.cl �d6 22 �g3 : c8 23 .i.e3 :lc3 24 f2 .i.f6! 2S a4 .i.gS led to equality in Dlugy-Nikoloff, Toronto 1 989; Black can also play the immediate 20 . . . tLld6 21 �g3 : c8 22 .i.e3, with adequate counterplay) 20 . . . bS 2 1 de .i.xeS 22 :ld 1 : c8 (22 . . . : e8 is more accurate and gives equality} 23 .i.f4 ! .i. g 7 2 4 :ldS a 6 2S �f2 :le8 2 6 f3 f7 27 h 4 � e S + 2 8 .i.xeS : xeS 29 : d3 b4 30 �f4. Here again, White has an extra pawn, but this time Black managed to hold out. Interestingly, his loss to Seirawan made such a strong impression on Lputian that he took the first opportunity to play this variation with White. In Lputian-Hansen, Dortmund 1 988, Black played (from diagram 1 2} the immediate 18 e5. There followed 19 d5 b5 ( 1 9 . . . h6 20 .i.cl is sounder for Black} 20 : b1 :b8 21 f2 aS 22 thc 1 h6 (22 . . . �a3 23 :lb3 b4 24 cb cb 25 d6) 23 .i.e3 �xe3 (in Lputian's view, 23 . . . tLld6 24 f3 :lf8 + 25 e2 c4 26 e1 �xe4 27 :lxbS �xc3 28 :lxaS :ld8 would have given equality; but not 23 . . . :lf8 + 24 e2 �xe3 2S �xe3 :lfl 26 :la 1 ! and 27 �b3!} 24 xe3 c4 2S �e2 .i.f8 26 tZ'lg1 .i.cS + 27 e2 .i.xg 1 28 :lxg 1 f7 29 a3 rj;e7 (29 . . . b4 would have retained some saving chances} 30 :lb 1 , and White won the rook ending. After the game, Lputian explained that at move 2 1 , the correct course was 21 a4! b4 22 cb cb (or 22 . . . :lxb4 23 tZ'lc3 �d6 24 :le 1 , preparing 2S .i.e7) 23 �c1 .i.f8 24 �b3 �d6 2S �d2 b3 26 .i.e3 a6 27 fl! :lb4 (28 . . . b2 29 e2) 28 .i.c5 :lxa4 29 .i.xd6 .i.xd6 30 :lxb3, and White has the better chances. • • .

• . •

cd

18

Black can play 1 8 . . . eS at once, without opening the c-file. After 1 9 dS ( 1 9 de :ld8 20 gS .i.f8} 1 9 . . . b6 20 gS :lf8, he has a perfectly secure fortress. 19

cd

e5

After 19 . . . b6 or 19 . . . bS, White plays 20 :lbl! with the better chances. 20

d5 (13)

20 :lcl �d6 2 1 de .i.xeS is not dangerous for Black. 20

�d6

A refinement on the 1 1 th game in Seville. But before going any further, let us recall how that game (which played an important

32 Seville Variation 13 B

part in the match) proceeded. It might, incidentally, have been worth including it (accompanied as it is by some new annotations) among the 'basic' games in this volume. But we will not break our rule: only games played in 1 988 or later are to be specially numbered. 20 . .. .t..h6 21 h4 .t..d2 22 'fJ.d1 .t..a5 If 22 . . . b5, White has 23 lt! c 1 !, heading for b3. Gutman quotes the following analysis by H. Wirthensohn : 23 . . . aS (23 . . . .t..x c1 24 'fJ.xc1 lt!d6 25 'fJ.e 1 a6 26 .t..g 3 'fJ.e8 27 fl !? cd 1 2 cd llJd7 (and here, the right continuation is 1 2 . . . llJ c6 1 3 d5 J.xf3 14 gf ltld4) 1 3 e5! l:l. c8 14 J.e3 llJb8 1 5 h5! 1rd5 1 6 hg hg 1 7 1rd2 J:Hd8 1 8 l:l. h4 ! .i.xf3 1 9 gf J.xe5? 20 l:l. b5! 1re6 2 1 l:l. xe5, and i t i s all over. . . .

• • •

10 11

J.b7

1rd3 (29)

29 B

11

J.a6

In earlier games, Black played 1 1 . . . cd 12 cd e6 13 .i.g5 1rd6, but as a rule failed to equalise. Here is one example : 14 'iWe3 l:l. c8 1 5 l:l. fd 1 llJd7 1 6 h4 l:l. c2 1 7 e5 1rc7 1 8 J.d3 l:l. c3 ( 1 8 . . . l:.Xa2 1 9

58 Modern Exchange Variation l:t bc 1 ) 19 'lre2 .*.d5 20 : be l h6 2 1 .*.e3 l:t dS 22 h5! with a big advantage; Lputian-Lalic, Sarajevo 1 9S5. 12

'lre3

1 2 'lrc2 promises White less. Chernin-Dorfman, Moscow 19S4, continued 1 2 . . . cd 1 3 cd 'Ir eS 14 'lr d l .*.xe2 1 5 'lrxe2 'lra6 1 6 'lr xa6 lbxa6 1 7 .*. a 3 l:t fdS 1 S d5 f5 19 .*.xe7 l:t d7 2 0 d6, and the players agreed a draw. 12

. • .

12 cd e6 is too timid; after 13 de .*.xe2 14 'lrxe2 be 1 5 .*.f4

'IreS 1 6 .*.d6 :t dS 1 7 e5 'lrc6 l S h4 lbd7 1 9 c4! a6 20 : fe l lbb6 21 'lre4, Black has no prospects whatsoever; Lerner-Gorelov, Ivano-Frankovsk 1 9S2. In the last few years, Black has tried moving his queen to cS or d7 without exchanging pawns in the centre. These experiments have been unsuccessful, for example : 12 'IreS 1 3 d5 .*.xe2 1 4 'lr xe2 .*.xc3 1 5 e5 'lrf5 1 6 : b 3 .*.a5 1 7 lbh4 'lr d 7 1 S .*.h6, with advantage; Halifman-Lau, Rotterdam 1 9SS. Or 12 'lrd7 1 3 de be ( 1 3 . . . .*.xe2 14 'lrxe2 be is safer) 14 l:t xbS!? l:t axbS 1 5 .*.xa6 'lr a4 16 'lre2, with a won position; Gelfand-I. Sokolov, Yugoslavia 1 9SS. 'lrd7 13 cd A recommendation of Grandmaster Gavrikov. A much earlier game Browne-Martz, USA 19S2, went 1 3 . . . .*.xe2 14 'lrxe2 llJc6 1 5 d5 lbd4 16 llJxd4 .*.xd4 1 7 l:t d l .*.g7 1S .*.a3 'lrd7 19 e5, with a clear advantage to White. • • •

. . .

14 .*.a3 14 .*.xa6 lbxa6 1 5 'lra3 lbc7! gives White nothing; Halifman­

Epishin, Leningrad 1 9SS, continued 1 6 :t d l :t fcS 1 7 .*.b2 lbb5 lS 'lrd3, and now Halifman recommends lS . . . e6! Black also obtains a good game after 14 l:t dl?! 'lra4 1 5 .*.d2 .*.xe2 16 'lrxe2 :t c8 17 e5 lba6 18 e6 f6 ( 1 8 . . . f5!) 19 d5 'ifc4 20 'lrxc4 : xc4; Lputian-Zilberstein, Blagoveshchensk 1 988. How­ ever, Salov's innovation 14 d5!? deserves attention. After 14 . . . .*.xe2 1 5 'lrxe2 'lra4 1 6 .*.g5 l:t e8 1 7 l:t fc l , White obtained a slight edge in Salov-1. Sokolov, Haifa 1 989. .*.xel

14

Another possibility is 14 l:t e8 IS d5 .*.xe2 16 'lrxe2 'lra4 1 7 'lre3 lbd7 l S l:t fc 1 : ac8 1 9 h 3 l:t xc l + 2 0 l:t xc l llJf6 2 1 lbd2 .*.h6!, as in Gelfand-Malishauskas, Vilnius 1 988. However, Pribyl gives IS : fd! as more accurate. • • •

Modern Exchange Variation 59 15

'ir xe2

.l:l. e8

Black is not in a hurry to bring his knight out. In Gelfand­ Groszpeter, Palma de Mallorca 1989, Black played 1 S . . . c!Llc6, and after 16 dS c!LleS 17 c!LlxeS -*.xeS 1 8 f4 �d6 19 .i. b2 e6 20 .1:1. bd 1 White gained the initiative and soon won. 16 17

.l:l. fcl .l:l. c3

e6 .l:l. c8

1 7 . . . c!Llc6? loses to 1 8 'irbS .l:l. ac8 1 9 18

.I:!. be l

a6 20 'ir xb6.

.l:l. xc8+

1 8 .l:l. bc 1 .l:l. xc3 1 9 .l:l. xc3 c!Llc6 20 "irbS .l:l. c8 2 1 .l:l. c4 'irb7 leads to equality. 18 'ir xc8 19 .l:l. c l 'irb7 An alternative is 19 . . . 'ira6 20 "ire3 c!Lld7 2 1 .l:l. c7 c!Llf6 (2 1 . . .

jfa4 22 h3) 22 c!LleS .l:l. c8! 23 .l:l. xf7 "irbS 24 g3 jfbl + 2S �g2 jfxe4+ 26 jf xe4 c!Llxe4 27 .l:l. xa7 �xeS 28 de .l:l. c2, with full equality; Dzhandzhgava-Malishauskas, Uzhgorod 1 988. A better line for White, according to Halifman, is 20 jf xa6 c!Llxa6 2 1 �fl. with some advantage. 20

'irc4

c!Lla6 (30)

30 w

This position had first occurred only a few days earlier, in the game S. Ivanov-Pribyl from the same international tournament (Leningrad 1989). After 2 1 dS ed 22 ed .l:l. d8 23 d6 c!LlcS! 24 �xeS be 2S 'ir xcS 'ira6 26 h4 .l:l. xd6 27 c!LlgS h6, the players agreed a draw. The main game we are examining was to end in victory for Black, but that was hardly the rightful outcome. For the moment Halifman doesn't want to settle for a draw, and adds fuel to the flames.

60 Modern Exchange Variation 21 22 23 24 25 26 27 28

h4 e5!? h5 .l:l. c3 'irc6 .l:l. xc6 .l:l. c7 ..txf8

29

d5!

.l:l. d8 h6 gh b5! 'irxc6 lt! b8 .U8 �xf8 (31)

31 w

White's last chance - he hopes for success by sacrificing another pawn. As Pribyl has pointed out, a forced draw results from 29 .l:l. xa7 lt!c6 30 .l:l. b7 lt!xd4 3 1 lt!xd4 .l:l. xd4 32 .l:l. xb5 .l:l. d l + 33 �h2 .l:l. d2. 29 30 31 32 33

lt!d4 f4 .l:l. c5 .l:l. xd5

ed .l:l. e8! .l:l. e7 lt!a6

In the event of 33 .l:l. c6 lt!c7 34 .l:l. xh6, Black can save himself with 34 . . . �g7! (but not 34 . . . lt!e6? 35 lt!f5 .l:l. c7 36 .l:l. h8 mate) 35 .l:l. xh5 lt!e6 36 lt!f5 + �g6 37 lt!xe7 + �xh5 38 lt!xd5 �g4. 33 34 35 36 37

.l:l. d6 lt! c6 f5 f6?

lt!c7 lt!e6 .l:l. c7 lt!g5

Now all White's previous work comes to nothing. After 37 e6! Black would still have to struggle for the draw.

Modern Exchange Variation 61 37 . . . �g8 38 �fl 'it>h7 39 lll b4 a5 40 lll d5 .:t e l + 41 'it>e2 .:t el + 42 'it>e3 .:t xg2 43 .:t d7 .:t g3 + 44 'it>e2 .:t g4 45 'it>d3 h4 46 'it>e3 .: a4 0-1

Game No. 1 0 Halifman-Henkin

Leningrad 1989 1 d4 lll f6 2 c4 g6 3 lll f3 J.g7 4 lll c3 d5 5 cd lll xd5 6 e4 lll xc3 7 be c5 8 .:t b1 0-0 9 J.e2 lll c6 10 d5

White is unsuccessful with 10 J.e3 cd 1 1 cd 'ifa5+ 1 2 J.d2 'ifxa2; Paunovic-Ristic, Vrnjacka Banja 1 983. In comparison with the variation 9 . . . cd 10 cd 'ifa5 + , which we examine in Games 1 1 - 1 3, Black has the useful extra move . . . lll c6. 10

lll e5

Safer than taking the pawn. Let us look at that possibility : 1 0 . . . J.xc3 + 1 1 J. d 2 J.xd2+ 1 2 'if xd2 lll a 5 (moving that knight to b8 or d4 is worse. An example of the latter is 12 . . . ll:ld4 1 3 lll x d4 cd 14 'ii' xd4 'ifa5 + 1 5 'ifd2 'ifxd2+ 1 6 'it> xd2 .:t d8 1 7 'it>e3 b6 18 .:t bc l e6 19 J.c4 e5 20 J.b3 J.d7 21 .:t c7 a5 22 d6, and Black has numerous problems; Kasparov-Natsis, Malta 01 1 980) 1 3 h4 (White may also castle, with e4-e5 to follow) 1 3 . . . J.g4 14 h5 J.xf3 1 5 gf! (not allowing the knight onto c4). Despite the simplification, White's attacking potential is very noticeable, for example : 1 5 . . . e6 1 6 'if h4 'iff6 1 7 hg 'if xg6 1 8 'ifh2 'it>h8 1 9 'it>d2! f6 20 .:t bg l 'iff7 21 de 'ife7 22 .:t g6, and the attack continues; Bjarnason-Griinberg, Biel 1 985. J.xe5 lll xe5 11Vd2 (32) White can also play 12 c4 at once, for example : 12 11 12

• • •

'if d6!?

( 1 2 . . . 'ifa5 + occurs more often) 12 'ifd2 J.d4 14 J.b2 J.xb2 1 5 : xb2 e 5 1 6 0-0 b6 1 7 : b3, with a complicated position; Gelfand­ Ghinda, Halle 1 987. Alternatively Black has an interesting counter­ sacrifice of a pawn : 12 f5 1 3 ef J.xf5 14 .:t xb7 'if d6 1 5 .:t b3. Perhaps he should now go in for immediate exchanges with 15 . . . J.e4 1 6 J.f3 J.xf3 1 7 : xf3 : xf3 1 8 'ifxf3 : f8, trying to exploit his advantage in development. Danner-Shvidler, Biel 1 982, saw instead 1 5 . . . J.d7 16 'ifd3 .i.f5 17 'ifd2 J.e4 1 8 J.g4 J.xg2 1 9 J.e6 + 'ifxe6 ( 1 9 . . . �h8 20 .:t g 1 J.e4 2 1 .:t h3 with an attack!) 20 de •• •

62 Modern Exchange Variation .i.xh 1 2 1 11fe2 l:tf5 22 f3 .J:I. af8 23 .i.h6 .J:1. 8f6 24 11fe4, and Black has worries. A move that looks fairly logical is 12 11fc7, covering the pawn on b7 and preventing White from castling. A correspondence game Danner-Nesis ( 1 983-5) continued 1 3 h3 f5 1 4 ef .i.xf5 1 5 .J:I. b3 ..td4 1 6 0-0 e5, and Black achieved a sound position. • • •

32 B

In the first half of the 1980s, Black almost invariably played 1 2 . . . e 6 here, and after 1 3 f4 h e retreated his bishop t o g 7 o r h8. I do not plan to give a thorough study of the positions which then arise, but it is worth reminding the reader of the difference that the choice of retreat square makes - or more exactly, of why it is essential to withdraw the bishop into the corner. Consider the variation 12 e6 13 f4 .i.g7 14 c4 .J:I. e8 1 5 e5 f6 1 4 d6 fe 1 7 .i.b2 ef 1 8 .i.xg7 e8 29 .te6, and 24 . . . fe 25 llJxe5 .te6 26 : h7 + ri>f6 27 f4 : bc8 28 ttJf3! .tg8 29 ttJg5! .txh7 30 : xh7 : xd3 + 3 1 'i>xd3 : f8 32 �e3 llJc6 33 : c7 llJe7 34 llJ h7 + ) 25 'i>xd3 .txf3 26 : h7 + �f8 (61) . 27 : 1 h3 (forced, since 27 ef llJg8! 28 : l h3 .txe4 + ! draws; nor does 28 : 1 h4 help, in view of 28 . . . g5!, and when the rook moves,

100 Modern Exchange Variation 61 w

Black again has 29 . . . -*.xe4 + !) 27 . . . -*.g4 (the attempt to weave a mating net for the white king, with 27 . . . -*.g2 28 J:l. g3 -*.fl + 29 �e3 fe 30 .IH3 + �g8 3 1 J:l. xe7 J:l. d8, fails to the simple 32 J:l. xe5) 28 J:l. g3 f5 29 �e3 J:l. e8! 30 J:l. h8 + ll'lg8, and Black is out of danger. It remains for us to consider 28 ef J:t d8 + ! 29 �e3 (29 �c3 J:l. c8 + ! 30 �b2 -*.xh3 3 1 fe + �e8 32 J:l. xh3 �xe7 33 J:t h7 + �e6 34 J:l. xa7 J:t f8, and Black has everything in order) 29 . . . ll'lg8 30 J:l. g3 ll'lxf6 3 1 J:t h8 + �e7 32 J:l. xd8 �xd8 35 e5 ll'ld5+ 34 �d4 -*.e6 35 J:l. xg6 �d7, and Black's defence cannot be breached. So the move suggested by the annotators, 24 J:t bh 1 , fails to justify their hopes. �xf6 24 25

ll'ld4

25 e 5 + �g7 (25 . . . �e6?! 26 ll'lg5 + ) is not dangerous for Black. .l:t b7?! (62) 25

62 w

Modern Exchange Variation 101 This move was probably based on miscalculation in the main variation. But even after the more precise 25 .l:l. h8 (25 . . . �g7 leads to an unattractive minor piece ending after 26 .l:l. bh l .l:l. h8 27 .l:l. xh8 .l:l. xh8 28 .l:l. xh8 c;;, x h8 29 c;;, f4) 26 l:t bh l .l:l. xh2 27 .l:l. xh2 c;;, g 7, White can play, for example, 28 f3 (28 �f4 is premature in view of 28 . . . l:t d8 29 c;;, e s g5! 30 lDe6+ ..txe6 3 1 c;;, xe6 .l:l. xd3 32 c;;, x e7 .l:l. d2!), and it is not easy for Black to save himself. Has Black no other suitable defence, then? Zaitsev suggests 25 l:t e8!?, with the idea of answering 26 .tb5 (26 c;;, f4 .1:1. bd8) with 26 . . . lDf5 + ! 27 lDxf5 (risky alternatives are 27 c;;, f4 g5 + ! 28 �xg4 .l:l. xe4+ , and 27 c;;, d 3 .l:l. ed8! 28 .l:l. b4 .l:l. xd4+ ! 29 .l:l. xd4 .te2 + !) 27 . . . .l:l. xe4+ 28 c;;, x e4 .txf5 + . A complete demon­ stration of the geometric co-ordination of the black pieces. • • •

. • •

26

f3

At this point, White could have put his opponent in a critical position by playing 26 .l:l. h4!. How is Black to defend? The obvious-looking 26 .l:l. bd7 (26 . . . .te6 loses to 27 e5 +) suffers a fiasco after 27 e 5 + ! �xeS (27 . . . �g5 28 .l:l. xg4+ ! �xg4 29 .l:l. g l + is crushing) 28 .l:l. b5 + lDd5 + 29 .l:l. xd5 + c;;, x d5 (Black comes out a piece down after 29 . . . .l:l. xd5 30 lDc6 + c;;, f6 3 1 lDxd8 �g5 32 .l:l. h8) 30 .l:l. xg4. It is notable that this line was indicated by Mephisto, the World Microcomputer Champion. True, Black can still struggle on : 30 . . . .l:l. e8 + !? 3 1 �f3 (after 3 1 ..te4+ �c4!, the black king slips across to the queenside, creating counterchances; while after 3 1 �d2 .l:l. f7 32 f3, Black is not devoid of hope although he clearly stands worse), and now 3 1 . . . .;;, c s 32 lDb3 + ! or 3 1 . . . �e5 32 lDc6+ allows White to re-group effectively, so Black must play 3 1 . . . .l:l. f7 + or 31 . . . .l:l. f8 + ; though in either case, after 32 �g3, the two pieces are much stronger than the rook and pawn. Another try is 26 lDfS + . If now 27 lDxfS gf 28 f4 (if 28 ef, then 28 . . . .l:l. xd3 + , while after 28 .l:l. g1 .l:l. bd7 29 .tb5 .l:l. e7 30 f3 c;;, g 5 3 1 .l:l. hh 1 Black is guaranteed a draw with 3 1 . . . .l:l. xe4+ ! 32 fe f4+ 33 �f2 .l:l. d2 + etc.) 28 . . . fe 29 .txe4 ..tf5!, there follows a lengthy combination with a happy end : 30 .l:l. h6 + ! c;;, g 7 3 1 ..txf5 �xh6 3 2 .l:l. h 1 + c;;, g 7 33 .l:l. h7 + �f6 3 4 .l:l. xb7 �xf5 3 5 .l:l. xa7 .l:l. e8 + ! 3 6 �f3 .l:l. e 1 3 7 : f7 + c;;, e6 3 8 .l:l. b7 c;;, fs 3 9 .l:l. xb6 .l:l. fl + 40 �e3 .l:l. f3 + ! 4 1 c;;, d4 (capturing the rook leads to a classic stalemate) 4 1 . . . .l:l. a3 42 .l:l. b2 c;;, x f4 43 c;;, c 5 : as 44 .l:l. b4+ �e5 45 a4 .l:l. c8 + , and the draw is elementary. • • .

• • •

102 Modern Exchange Variation Nonetheless there is a way for White to win. He should answer 26 . . . lDfS + with 27 ef l:l.e7+ 28 .i.e4! gf 29 l:l. xg4 etc. 26 l:l. bd7 If 26 . . . .i.d7 or 26 . . . .i.c8, then 27 f4 is strong. 27 l:l. b4 .t e6 The best defence. 27 . . . .txf3? is refuted by 28 l:l.f2, and 27 . . . lDc6 by 28 lDxc6 l:l. xd3+ 29 �f4 .thS 30 eS + !. If 27 . . . l:l. e8, then 28 l:l. a4. The matter is more complicated after 27 . .th5 28 f4! lDc6 29 e S + �g7 (29 . . . �f7 30 .i.c4+ is bad for Black) 30 lDe6 + �g8 3 1 lDxd8 lDxb4 32 .i.c4+ lDdS + 33 �e4 l:l. xd8 34 .txdS + xg2 fe 27 j_xe6+ oi>h7 28 : c7 liJf4 + 29 oi>fl : ae8 30 d5 liJxh3 3 1 Wg3 W xg3 3 2 : xe7 + : xe7 3 3 fg liJg5, although Black's chances would still have been slightly better. 24 . . . : xd1 + 25 j_xd1 liJf4 26 g3 1Wxh3 27 j_f3 liJd5 And Black went on to win. 13 14

liJe4

eS (90) liJd5 In Rizhkov-Epishin, Leningrad 1 986, Black played 1 4 . . . liJc5 at once, which led to equality after 15 j_c4 e6 16 b4 1i'a3 1 7 liJe7 + liJxe7 1 8 be j_e4.

90 w

In the 'main' game which we are examining, the white queen's bishop calmly retreats to h2 - the familiar theoretical continuation. In the 'sensational' encounter in London ( 1 9th game, 1 986), I played the unexpected 1 5 : xc6!? In the digression which now follows, we shall examine that game as well as the current theoretical verdict on White's exchange sacrifice. 15 : xc6 ef So Kasparov decides to decline the sacrifice. Many commentators thought that accepting it would virtually lose outright : 15 . . . be 16 liJ e7 + �h8 17 liJxe5 j_xe5 1 8 liJxc6, but now after 1 8 . . . "ild2! 19 j_xe5 + f6 it is White who suffers material loss. I was intending 1 7 liJxc6, and if 17 . . . "ilb6 ( 1 7 . . . Wc5 1 8 liJcxe5 "fle7 19 liJd4), then 18 liJcxeS. After 18 . . . j_e6, White has three choices : 1 9 j_xe6 1i'xe6 20 b3, with two pawns for the exchange; 19 liJc4

140 4 J.f4 System .i.xc4 20 .t.xc4 lDc5, with unclear play; or 19 11fc2 .t.xb3 20 11fxe4 .i.e6, as in Gavrikov-Kochiev, Tallinn 1 987. 16 l:t c7 .i.e6 If 1 6 . . . fe 1 7 11fxe3 .i.xb2, then 1 8 lDd4 is strong; or if 1 7 . . . lDd6, White has 1 8 l:t d 1 , with mounting pressure. 1 7 1i'e1! The natural-seeming 17 : xb7 .i.xd5 1 8 l:t b5 would give Black the better game after either 18 . . . 11t'a6 19 .i.xd5 l:t ae8 or 1 8 . . . lDc3 1 9 be .i.xf3 20 gf 11fxc3. In a later game Szihigyi-Schmidt, Hungary 1 986, Black played the even stronger 17 . . . lDd6! (depriving the rook of the b5-square) 18 lDe7 + ..th8 19 lDc6 'it'c5 20 .i.xe6 11f xc6, and White came out a piece down. The queen move to e 1 was planned in my pre-game analysis, but later it was established that 17 lDe7 + is stronger : 17 . . . ..th8 1 8 l:t fc 1 -*.xb3 1 9 ab, with the initiative ( 1 9 . . . fe 20 11fxe3 lDd6 21 'it'f4). 1 7 . . . 1i' b5! The endgame after 17 . . . 11fxe1 1 8 l:t xe 1 .i.xb2 19 lDe7 + �h8 20 .t.xe6 fe 21 ef l:t xf4 22 l:t xb7 .t.c3 23 l:t e2 is unattractive for Black. 18 lDe7+ �h8 19 -*.xe6 fe 20 11t'b1! The white queen has made an unusual tour, 1i' d 1 -e2-e 1 -b 1 , and now unexpectedly aims a t the opponent's kingside. 20 . . . lDg5! The only move. 20 . . . 11t'b6 is simply answered by 2 1 : re t , while if the knight goes anywhere else, 2 1 lDxg6+ is decisive. 21 lD h4! (91) 91 B

After 2 1 lDxg5 (2 1 lDd4 11fe5) 2 1 . . . 11f xg5 22 ef J: xf4 23 l::t xb7 l:t e8 24 lDc6 11fc5 25 lDxa7 .i.d4, Black is dangerously active.

4 j.,f4 System 141 21 . toxh3+ !? In this extremely sharp duel, each of us was hoping to outwit the other. No doubt for that reason, Kasparov refrains from 2 1 . . . fe, after which I could have forced a draw with 2 2 tohxg6+ hg 23 toxg6 + �g8 24 toe7 + . 22 � h2 Of course not 22 gh 'lfg5 + 23 tog2 f3. 22 . . . 'lfh5? A serious error. 22 . . . toxf2! 23 l:t xf2 fe 24 .l:l xf8 + .l:l xf8 25 �h3 e2 26 'lfe4 'lfh5! 27 .l:l c4 g5 28 'lfxe6 gh leads to a draw. 23 toexg6+ If 23 �xh3, Black replies 23 . . . g5. Many observers thought that capturing with the other knight was stronger : 23 tohxg6 + hg 24 'lfxg6. As we shall see, this variation is indeed unpleasant for Black, but it happens to be unimportant which knight takes on g6 first. 23 . . . hg 24 'lfxg6 An inaccuracy throwing away the win! 24 toxg6+ �g8 25 toe7 + �h8 26 'lf g6! (rejecting the perpetual check with 26 tog6+) would have given White a clear advantage; this is the same position that was arrived at in the previous note. I shall now quote the interesting variations given by Halifman : 26 'lf eS (26 . . . .l:l f5 is met by 27 gh fe 28 'lfxh5 + .l:l xh5 29 tog6 + �h7 30 fe!) 27 �xh3 fe (27 . . . .l:l f6 28 �g4!) 28 'lfg4! .l:l f6 29 'lfh4+ j_h6 30 f4! 'lfxb2 3 1 .l:l b 1 !, etc.; a more stubborn defence is 26 'lfh7, but here again, after 27 gh .l:l f6 (27 . . . j.e5 28 .l:l c5 ! ; or 27 . . . fe 28 'lf xe6 .l:l f6 29 'lf xe3 with two extra pawns) 28 'lfg4 (28 'lfxh7 + �xh7 29 .l:l g 1 is also good), Black is in a critical position since taking on e3 is no good: 28 . . . fe 29 fe! .l:l xfl 30 tog6 + �g8 3 1 'lf xe6+ .l:l f7 3 2 'lf xf7 mate. 24 'lfe5! An ingenious retort which gets Black out of danger. White cannot now play 25 �xh3, as the rook on c7 is en prise (whereas with the knight on e7, the capture 25 . . . 'lfxc7 would be unplayable : 26 'lfh5 + and mates). 25 .l:l/7 Not 25 .l:l xg7 fe+ 26 f4 'lf xg7 27 'lf h 5 + �g8 28 gh 'lfxb2 + 29 �h1 .l:l f7!, or 26 'lfg3 'lfxg7 27 tog6+ �g8 28 toxf8 tog5! 29 tod7 .l:l d8 30 toe5 e2 3 1 .l:l e l .l:l d l 32 tod3 'lfh7 + 33 �gl 'lfxd3 34 'lf xg5 + �f7 etc. On the other hand, 25 'lfc2 would have forced .

.

. . •

•. .

. . .

142 4 �f4 System a draw : 2S . . . fe+ 26 cot>xh3 'iPg8 27 f4 e2 28 fe ef('if 29 .l: xg7+ cot>xg7 30 'ifg6+ etc. 25 . . . .l: xf7! 26 'ifxj7 fiJg5! After the game I discovered that the computer had indicated 26 . . . 'if bS here, but there would follow 27 fiJg6 + .,Ph7 28 fiJe7 'ife8 29 11t'xe6 fOgS 30 11t'fS + cot>h6 31 .l: h 1 ! and in spite of his extra peice Black has no defence. 27 fiJg6+ cot>h7 28 fiJxe5 liJxf7 29 liJxf7 'it>g6 30 fiJd6 fe The position has become simplified, the tension has abated, and Black has even emerged with a slight edge, though its significance is purely symbolic. 31 fiJc4 ef 32 .l: xf2 b5 33 fiJe3 a5 34 .,Pg3 a4 35 .l: c2 .1:/8 36 'it>g4 �d4 37 .l: e2 �xe3 38 l:t xe3 .1:/2 39 b3 .l: xg2+ 40 .,Pf3 .l: xa2 41 ba J-J. 15 .i.h2 fOeS 15 fiJf6 is very strongly answered by 16 .l: xc6! fiJxdS 17 .l: d6 fiJe7 1 8 e4, with a clear advantage; Pinter-Rogers, Szirak 1 986. But a more popular continuation is 15 �e6 16 .l: fd1 (92) . Let us look at some interesting games with this line from recent years. • • •

• • •

92 B

Farago-Smejkal, Baden-Baden 1 98S, went 16 .,PhS 1 7 �c4 .l: ad8 1 8 b4 fiJxb4 1 9 fiJxb4 .l: xd 1 + 20 .l: xd 1 11t'xb4 2 1 �xeS, with a slight advantage to White. Gleizerov-Vakhidov, Uzhgorod 1 988, went 16 .l: fe8!? 1 7 11t'c4 fiJf6 1 8 e4 .l: ac8 1 9 11t'c3 11t' xc3 20 fiJxf6 + �xf6 2 1 .l: xc3, and a draw was agreed. Basin-Kozlov, Naberezhnie Chelny 1 988, continued 16 .l: fd8 1 7 11t'c4 (the exchanges with 1 7 11t'e 1 .l: d7 1 8 11t' xaS fiJxaS 1 9 fiJc7 fiJxb3 20 fiJxa8 fiJxc1 21 .l: xc 1 f6 are to Black's liking; Rajna• • •

• • •

• • .

4 J..f4 System 143 Nikoloff, St John 1 988) 17 . . . lll d4? (the correct line is 1 7 . . . lll f6 1 8 e4 .l:l. ac8 1 9 lll g 5 lll d4 20 lll e7 + �f8 2 1 tll xe6+ �xe7 22 lll x d8 .l:l. xc4 23 J.. xc4 J.. h6, and Black's chances are no worse; Huzman-Dorfman, Lvov 1 988) 1 8 ed J.. x d5 1 9 "it'e2 J.. h 6 20 l:t c2 J.. f4 2 1 J.. x d5 "it'xd5 22 lll xe5 J.. x e5 23 J.. x e5 lll g 5 24 "it'g4 lll e6 25 "ifh4 g5 26 "it'h6 "it'e4 27 "it'f6 �f8 28 .l:l. cd2 "it'g6 29 d5 lll c 5 30 d6 1 -0. 16

e4!?

An interesting novelty. After 1 6 J.. c4 .:. adS? 1 7 b4! lll xb4 1 8 lll e 7+ �h8 1 9 lll xf5 gf 20 lll x e5, White had a big advantage in Vainerman-Epishin, Norilsk 1 987, but Black has the much better 16 . . . e4!, with chances of seizing the initiative. 16 : adS! (93) 1 6 . . J.. xe4 loses to 1 7 J:t xc5 J.. x f3 1 8 "ife3 J.. h 6 1 9 "it'xh6. White also has the initiative after 16 . . . J.. e6 17 J.. c4, with a2-a3 to follow, or 16 lll xb3 1 7 ab J.. e6 1 8 .l:l. a 1 "it'c5 1 9 .l:l. fc 1 "it'd6 2 0 .l:l. d 1 "if b8 2 1 b4. But 16 . . tll xe4 !? leads t o sharp play. .

. . .

.

93 w

17

"it'e3

An inaccuracy. In Ftacnik's view, an improvement is 1 7 .l:l. fd 1 J.. xe4 1 8 .l:l. xc5 J.. x f3 1 9 "it'e3 J.. x d 1 2 0 .l:l. xa5 J.. x b3 2 1 a b lll x a5 22 lll e7 + �h8 23 J.. x e5 lll c6 24 J.. x g7 + �xg7 25 "it'c3 + f6 26 lll xc6 be 27 "it'xc6, with somewhat the better ending. 17 18 19 20

ab J:t c5 "it'xc5

lll xb3 J.. xe4 "it' xc5

20 lll f6+ J.. xf6 2 1 "ifxc5 .l:l. d5 22 "ife3 was more exact.

144 4 J.f4 System 20 21 22 23 24 25

'lfe4 gf J.g3 'lfb5 'lfxb7

: xd5 .i.xf3 : fd8 : dl l0d4 l0e2 +

Black could have retained the better chances with 25 . . . Jl xb2 26 'lfe7 : f8 27 'lfxa7 l0xf3 + 28 �g2 : xb3. 26 27 l8 29

�gl �xg3 'lfxa7

l0xg3 J.b6 : xb2

'lf b6 The position has levelled out, and the game gradually heads towards a draw. 29 30

: at

: d3 �g7!

Precisely played. Not 30 . . . e4? 3 1 'lf b8 + J.f8 32 : aS : xf3 + 33 �h4. Also 30 . . . J.f4+ 3 1 �g2 : dxb3 is dangerous in view of 32 : aS + �g7 33 'lfd8, with an attack (Ftacnik). 31 32 33 34

: a6 'lfe7 �gl

J.g5 J.f4+ : dl

'lf b6

l-t Game No. 2 1 Belyavsky-Gavrikov

Moscow 1988 1 d4 l0f6 2 e4 g6 3 l0e3 d5 4 J.f4 J.g7 5 e3 e5 6 de 'lfa5 7 : ct de

In the fifth game of the return World Championship Match, London 1 986, Kasparov played 7 . . . l0e4 here. The game was quite interesting, so let us take this opportunity to give it in full : 7 . . . l0e4 8 cd l0xc3 9 'lf d2 'lfxa2 10 be This variation came into use after the famous game Petrosian­ Fischer, Candidates final 1 97 1 , which continued : 10 'lfa5 1 1 J.e4 l0d7 12 l0e2 l0e5 1 3 J.a2 J.fS 1 4 J.xe5! J.xe5 1 5 l0d4 'lfxc5 16 lOxfS gf 1 7 0-0, with a dangerous attack. Afterwards, various alternatives were proposed for Black; the most effective of them belongs to Mikhalchishin : 12 l0xe5 1 3 0-0 0-0 14 f3 • . .

• • .

4 J.f4 System 145 e5! 1 5 J.g3 b5 1 6 J.a2 1Wb6! 1 7 �h1 a5, with adequate counterplay. Recently the investigations have switched to 12 10f3, which so far has brought variable success. In Agzamov-Gulko, Frunze 1985, the interesting continuation was 1 2 . . . /Oxc5 ( 1 2 . . . 0-0 1 3 0-0 /Oxc5 1 4 J.e5 J.xe5 1 5 /Oxe5 f6 1 6 .J:I. a 1 10e4 led t o sharp play in Razuvayev-Mikhalchishin, Minsk 1 985) 1 3 J.e5 J.xe5! (a game between the same opponents a few months earlier in Sochi had ended quickly with 1 3 . . . 0-0 1 4 0-0 f6 1 5 .J:I. a 1 1Wd8 16 J.c7! 1Wd7 17 d6 + e6 18 /Od4 1Wf7 19 .J:I. a5 b6 20 .J:I. xc5! be 21 /Ob3 1Wd7 22 1Wd3! .J:I. d8 23 1We4 1 -0) 1 4 /Oxe5 f6 15 10f3 0-0 1 6 /Od4 /Oe4 1 7 1Wb2 /Od6 1 8 J.a2 J.d7 1 9 0-0 .J:I. ac8 2 0 e4! 11'c5 21 .J:I. fe 1 , with the initiative. 10 . . . 'll xd2+ 11 �xd2 10d7 12 J.b5 After 12 c6 be 1 3 de ltlb6 (or 13 . . . ltlf6), White achieves nothing. 12 . . . 0-0 13 J.xd7 This time, 1 3 c6 would be met by 1 3 . . . ltlc5. Now Black obtains the two bishops, but it soon becomes clear that they will both be constricted. 13 . . . J.xd7 14 e4 f5 The alternative 1 4 . . . .J:I. ac8 is weaker : 1 5 J.e3 f5 1 6 f3. 15 e5 e6! A critical moment. The line that had previously been tested was 1 5 . . . .J:I. ac8 16 c6 ( 1 6 e6 J.a4 17 c4 .J:I. xc5 18 J.e3 .J:I. c7 19 /Of3 .J:r. fc8 20 �d3 b5! favours Black) 16 . . . be 17 d6 ed 18 ed .J:r. f6, with the advantage; Schmidt-Gross, Naleczow 1 984. Why does Kasparov reject it? The answer is that after 1 5 . . . .J:I. ac8 White has the much stronger 1 6 c4 ! .J:I. xc5 17 J.e3. In Seirawan-Adorjan, New York 1 987, there followed 17 . . . .J:I. c7 ( 1 7 . . . .J:I. a5 18 f4 e6 1 9 d6) 1 8 10 f 3 b 6 1 9 c 5 b e 2 0 .J:I. xc5 .J:I. xc5 2 1 J.xc5 .J:I. c8 2 2 J.xa7, . and White won. 16 c4 .J:r.fc8 After 1 6 . . . g5 1 7 J.xg5 .i.xe5 1 8 10 f3 J.g7 1 9 .J:I. b 1 and .J:I. he 1 , White would complete his development while maintaining powerful pressure. 1 7 c6! The extra pawn cannot be retained, but in returning it White extracts the maximum profit : he creates a passed pawn and limits the scope of Black's rooks and light-squared bishop.

146 4 j_f4 System 1 7 . . be 18 d6 c5 Now the light-squared bishop acquires a little freedom, but the dark-squared one begins to suffocate. Perhaps the fate of the bishops should have been decided 'the other way round', with 1 8 . . . g5 1 9 j.xg5 j.xe5 2 0 c 5 .l:l. cb8, and Black has hopes o f counterplay. 19 h4! h6 20 li:Jh3! (94) .

94 B

Kasparov must have been examining the more natural-seeming 20 li:Jf3, which he could answer with 20 . . . j_c6!, leading to complex play with mutual chances. But I arrived at a solution to the position on mathematical lines. The white knight is heading for its ideal square, d3, by the only suitable route! Once the knight reaches its destination, the black bishop on g7 will be interned for good inside the cage formed by the pawns on d6 and e5 and the bishop on f4. At the present moment, White forestalls the freeing advance . . . g6-g5. 20 . . . a5 21 f3 a4 22 .l:l. hel! Over-protecting the pawn on e5. The hasty 22 li:Jf2 could be met by 22 . . . g5! 23 hg hg 24 j_h2 f4 25 li:Jd3 .te8 26 j_g l , when the situation i s fairly obscure. 22 . . . a3 23 li:Jf2 a2 24 li:Jd3 .l:l. a3 25 .l:l. al g5 The alternative 25 . .1:1. b8 is inadequate on account of 26 : ee l g 5 2 7 h g h g 2 8 li:Jxc5 (but not 2 8 .txg5 .1:1. bb3 2 9 li:J xc5 .1:1. b2 +) 28 . . . j.a4 29 li:Jxa4 : xa4 30 j_xg5 j.xe5 31 d7. 26 hg hg 27 j_xg5 (95) Now after 27 . .1:1. b8, the contest could have ended in a study­ like draw : 28 j_f4 .l:l. bb3 29 li:Jxc5 .l:l. b2 + 30 �c1 l:l xg2 3 1 j.d2 j_h6! 32 j.xh6 .l:l. c3 + 33 �d1 .l:l. d3 + ! 34 li:Jxd3 j.a4+ 35 �c1 .l:l. c2 + , with perpetual check. But there is quite a simple .

.

.

.

4 J.f4 System 147 95 B

refutation of the 'study' in 28 �e2! l:t bb3 29 .lt.:lxc5 ll b2+ 30 �fl . and White wins. 27 . . . �f7 28 J..f4 l:l. b8 29 l:l. ecl J.. c6 30 l:l. c3 l:l. a5 31 l:l. c2 l:l. ba8 32 .lt:lcl ]-{) 8 J.xc4 0-0 9

.lt.:lf3

A couple of years ago, when Black was achieving good results with this Griinfeld variation, the rare move 9 .lt.:le2 was tried out in a game Barlov-Gulko, New York 1 988. There followed 9 1t' xc5 10 1t'b3 .lt.:lc6 1 1 .lObS 1t' h5 1 2 .lt.:lc7, and now after 1 2 . . . 1t'a5 + 1 3 1t'c3 1t'xc3 + 1 4 .lt.:lxc3 Black equalised. In lnformator 45, Gulko considers 1l l:t b8, and judges that White stands better on the basis of 1 3 J.xf7 + l:t xf7 14 l:t xc6 1t'a5 + 1 5 .lt.:lc3 .lt.:le4 1 6 .lt.:ld5! e6 1 7 l:t c7. But this assessment was later refuted by Krasenkov in his game against Toth, Mazatlan 1 988 : 16 . . . .lt.:lxc3 1 7 .lt.:lxc3 ( 1 7 l:t xc3 J.xc3 + 1 8 be J.e6 19 J.xb8 J.xd5 20 1t'b2 J.xg2 2 1 l:t g 1 was more tenacious, although even then Black would be better) 1 7 . . . be 1 8 J.xb8 ( 1 8 1t'xb8 J.xc3 + 1 9 be 1t'xc3 + 20 �e2 1t'c4+ 2 1 �f3 1t'e6! 22 �e2 1t'xa2 + , and White is in a bad way) 1 8 . . . e6!, and in view of the threatened 19 . . . l:t b7 and 20 . . . l:t xb2, White loses his bishop on b8. A game Zlochevsky-Krasenkov varied with 1 7 .lt.:lxe7 + �f8 1 8 l:t xc3 J.xc3 + 19 be l:t xf4! 20 ef �xe7 21 0-0 J.e6 22 l:t e 1 1t'b6, and White resigned after ten more moves. A little later, Krasenkov ascertained that all these variations are redundant, as the simple 1 5 . . . e5 (attacking the bishop and threatening 16 . . . be 1 7 1t'xb8 1t' xc7) settles matters at once. After 1 6 l:t xf6 J.xf6 1 7 .lt.:ld5 ef 1 8 .lt.:lxf6 + �g7, White's position is completely hopeless. • . •

• • •

148 4 i./4 System 1i'xcS i.b3 1 0 ll:\bS looks like an active move, for example : 10 11t' b4+ 1 1 ll:\d2 ll:\a6 ( 1 1 . . . ll:\e4 is more precise) 1 2 a3 11t' a5 1 3 b4 11t'd8 1 4 i.e5 with advantage; Huzman-Tseshkovsky, Tashkent 1 987. However, Black has the powerful retort 1 0 .i.e6! (96) . 9 10

. . .

• • •

96 w

1 1 ll:\c7 (after this, Black seizes the initiative. Unclear play results from 1 1 .t.xe6 11t'xb5 1 2 .i.c4 11t'xb2 1 3 0-0; if 1 2 .i.b3, then 1 2 . . . ll:\e4) 1 1 . . . .i.xc4 1 2 b3 (in Popchev-Lalic, Bosna 1 988, White opted for 1 2 ll:\d2, but after 1 2 . . . b5 1 3 b3 ll:\d5 14 ll:\xa8 ll:\c3 1 5 .l:l. xc3 i.xc3 1 6 be 11t'xc4 ! 1 7 11t'e2 .i.xd2 + 1 8 11t'xd2 ll:\c6 1 9 11t'e2 11t'b4+ his position was hopeless) 1 2 . . . 11t'a5 + 1 3 11t'd2 11t' xd2+ 1 4 ll:\xd2 i.d3 1 5 ll:\xa8 ll:\d5 1 6 ll:\c7 .l:l. c8 1 7 ll:\f3 ll:\xf4 1 8 ef i.b2 19 �d2 i.xcl + 20 .l:l. xc 1 i.e4 21 ll:\b5 .l:l. xc l 22 �xc1 i.xf3 23 gf ll:\c6, and Black realised his advantage in the ending; Inkiov­ Lputian, St John 1 988. 10 11

0-0

11t'aS ll:\a6!?

A novel idea. The knight . heads for c5, so as to eliminate the bishop on b3 in some variations. Instead, 1 1 . . . ll:\c6 1 2 h3 i.f5 produces a position which we have already examined in Game No. 20. Gavrikov had introduced 1 1 . . . ll:\a6 a few months earlier, but in a less prominent tournament than the USSR Championship. Hence we have chosen Belyavsky-Gavrikov as the 'main' game. 12

ll:\eS

White utilises the fact that the e5-point is undefended. Stationing his knight there, he takes aim at f7. An equal game results from 12 .i.c4 ( 1 2 a3 i.g4 1 3 h3 .l:l. ad8!) 12 . . . ll:\c5 1 3 a3 i.e6 1 4

4 i..f4 System 149 i.. x e6 ltJxe6 15 i.. e 5. In a later game Magerramov-Henkin, Podolsk 1 989, White played 1l 1t' d4, and there followed : 12 . . . J: d8 1 3 1t'e5 ltJd5 14 1t'g5 h6 1 5 1t'h4 ltJxf4 1 6 ef ltJc5 1 7 i.. c 2 i.. e6 (but not 1 7 . . . e6? 18 b4, and White wins) 1 8 f5 gf 19 b4 1t'c7? (a serious error; after 19 . . . 1t'a6! 20 be i.. x c3 21 i.. xf5 i.. g 7! 22 i.. x e6 1t'xe6 23 J: fe l , the game is about equal Magerramov. Now White obtains a strong attack) 20 be i.. x c3 21 J.xf5 i.. xf5 22 J: xc3 �g7 23 J: e l J: d7 24 J: ce3 J: ad8 25 ltJe5 J: d 1 26 1t'g3 + i.. g 6 27 h4 1t'a5 (more stubborn resistance was offered by 27 . . . J: xe 1 + 28 J: xe 1 h5 29 f4 1t'xc5 + 30 �h2 e6 3 1 ltJxg6 fg 32 J: xe6) 28 J: xd 1 J: xd 1 + 29 �h2 1t'xc5 30 J: f3 ! (30 h5? 1t' c l , and it is Black who wins) 3 0 . . . � h 8 3 1 lll x g6+ fg 32 1t'b8+ 1 -0. I2 ltJcS (97) 97 w

I3

i.. xf7 + !

From the material viewpoint this operation benefits Black, but great complications now arise. A quieter move was 13 i.. c4. I3 I4 IS

ltJxf7 b4!

J: xf7 �xf7

The consistent follow-up to White's idea. IS I6 I7

lll dS 1t' xdS+

1t'xb4 ltJxdS

ltJe6 Better than 17 . . . i.. e 6 18 1t'xc5 1t' xc5 19 J: xc5 i.. xa2 20 i.. e 5! b6 2 1 J: c2 i.. b 3 22 J: c3 i.. x e5 23 .l:t xb3 aS 24 f4 i.. g7 25 J: xb6 a4 26 J: f2 a3 27 J: a2 i.. b 2, with advantage to White (Gavrikov). IS J: c4 Wb2 (98)

150 4 j_f4 System 98 w

19

J:t fcl?

The black pieces are rather awkwardly placed, which compen­ sates for White's material deficit. The position may be assessed as double-edged with approximately equal chances, for example : 19 J:t c7! (threatening 20 'it'c4) 19 . . . �f8 20 J:t xc8 + J:t xc8 21 'it'xe6 J:t c l 22 g4, when the resulting situation is completely calm. 19 j_gJ was also possible. The move played allows Black to extricate himself at once. j_d7!

19

The point is that after 20 'it'xd7 J:t d8 21 'it'a4 b5!, Black wins. If White had played 19 ..tg3, he could have answered 19 . . . ..td7 with 20 'it'xd7 J:t d8 21 J:t f4 + . However, in that case Black has 1 9 . . . ..tf6, and i f 2 0 J:t fc l , only then 20 . . . ..td7! 20

h3

White loses not only after 20 'it'xd7, but also after 20 J:t 4c2 ..tc6! ll d8

20

20 . . . ..tc6 is even stronger; then 2 1 J:t xc6? be 22 J:t xc6 fails to 22 . . . 'it' b 1 + 23 �h2 'it'f5. 21 'it'aS Interestingly enough, this very position arose in the first game that featured 1 1 . . . �a6 : Lukacs-Gavrikov, Debrecen 1 988 (we come to it at last). In that game White chose a different queen move : 2 1 'it'f3, but after 2 1 . . . �e8 22 J:t 4c2 'it'b5 23 ..tg3 ..tc6 24 'it'g4 'it'f5 25 'it' b4 J:t d3 26 J:t c4 'it'd5 27 e4 'it'b5, he had to resign. Belyavsky presumably wasn't acquainted with that game; at any rate, the move he plays leads to the same result. a6 21 22

..tc7

J:t c8

4 i/4 System 151 23 24 25 26

.l:t 4c:Z 'irxb5 ib6 .l:t xc6

'ir b5 ixb5 .l:t c6 ixc6

Black's endgame advantage is obvious, and he exploits it with no particular trouble - although it takes a full thirty moves. 27 f3 32 �d:Z �c7 37 42 �at 47 .tc3

ll:lf8 28 �f:Z ll:ld7 29 ia7 ie5 30 e4 .td6 31 �e3 e5 �e6 33 .te3 .te7 34 �c:Z b6 35 �b:Z �d6 36 .l:t d l + h4 a5 38 h5 g5 39 h6 ll:lf8 40 .td:Z ll:le6 41 .tc3 .td6 ll:ld4 43 �bl ll:le:Z 44 .tb:Z b5 45 .l:t d:Z ll:lf4 46 a3 ll:le6 b4

Slowly, but surely, Black accomplishes his task. 48 ab ab 49 : xd6 �xd6 50 .txb4+ �d7 51 �c:Z ll:lf4 52 g4 ll:le:Z 53 .tc5 .tb5 54 �d:Z ll:lf4 55 �e3 .te:Z 56 .tb4 .tdl 0-1

6

Fianchetto System Game No. 22 Karpov-Timman

Candidates Final (2nd game) Kuala Lumpur 1990 The King's Fianchetto System (g2-g3, .ifl -g2) occurred twice in the London/Leningrad match and twice in Seville. The exchange in the centre (c4xd5 c6xd5) which took place in those games has recently become highly popular; it follows that this 'symmetrical variation' is worthy of close examination. In the present book I have included my three most recent games with it. Two of them were played in what may be considered the highest-calibre event in the interval between the last two contests for the world crown : my Candidates Final match with Jan Timman. In those two games I played White, yet in the third example I had the black pieces. With the large amount of supplementary material incorporated in the notes, what is offered here is the most up-to-date survey of the variation in question. 1 2 3 4

d4 c4 .!0(3 g3

.!0(6 g6 .ig7 c6

The advance . . . d7-d5, which actually defines the Griinfeld Defence, is postponed by one move; in the event of an exchange in the centre, Black is preparing to recapture on d5 with the pawn. The other popular line, 4 . . . d5 Sed .!Oxd5, will be considered in Game No. 25. 5 .i g2 d5 6

cd

On a previous occasion I decided to avoid exchanging in the centre. Here is what happened : 6 .!Oc3 0-0 7 1Wb3 e6 8 0-0 .!Obd7 9 .if4 .!Ob6?! (9 b6 is more appropriate) 1 0 c5 .!Oc4 1 1 1Wc2 . . .

Fianchetto System 153 ltlh5 12 b3 ltlxf4 1 3 gf ltla3 14 1Wd2 b5 (on 14 . . . b6, White has 1 5 1Wb2 be 16 1W xa3 cd 1 7 ltla4 d3 1 8 .l:l. ac 1 de 19 .l:l. fe 1 , or 1 5 ltla4 b5 1 6 ltlc3, with the better chances) 1 5 .Z:. fe 1 .l:l. b8 1 6 �h 1 a5 1 7 e3 f5? ( 1 7 . . . b4 1 8 ltla4 ltlb5, or 1 7 . . . 1Wc7 followed by . . . f6 and . . . e5, was safer) 1 8 .tfl .td7 (now 1 8 . . . b4 is less good : 1 9 ltla4 ltlb5 20 ltlb6 ltlc7 21 a3) 1 9 .te2 .tf6 20 .l:l. g 1 �h8 21 .l:l. g3 1We7 22 : ag1 .Z:. g8 23 1W c l ! b4 24 ltla4 .Z:. g7 25 1Wfl .l:l. bg8 26 1Wh3 .te8 27 ltlb6 1Wd8 28 1Wh6 .l:l. c7 29 ltle5 .l:l.cg7 30 .td3 ltlb5 31 .txb5 cb 32 f3 .th4 (he can't save himself with 32 . . . .te7 33 ltlbd7! J.xd7 34 .l:l. xg6 J.f6 35 .Z:. xg7 .l:l. xg7 36 .Z:. xg7 J.xg7 37 ltlf7 + �g8 38 1W xg7 + ri;xg7 39 ltlxd8) 33 .l:l. h3 J.f2 34 .Z:. xg6 J.xe3 35 .Z:. xg7 .Z:. xg7 36 .l:l. g3 1We7 37 .l:l. xg7 1W xg7 38 1Wxg7 + b7 ¢>xh3 40 ¢>xa7 ll:ld7 4 1 b4 g5 42 a4 g4 43 a5 g3 44 �b7 g2 45 ll:le2, and White wins) 39 ¢>b7 g3 40 ¢>xa7 ll:ld7 41 b4! �f3 42 a4 e2 43 ll:lxe2 ¢>xe2 44 aS ¢>d3 45 ¢>b7 ¢>c4 46 a6 �bS 47 a7 ll:lb6 48 h4! and wins. (b) 37 . . . gS! enables Black to hold the position : 38 ¢>c6 ¢>e3! 39 ¢>b7 ltld6+ 40 ¢>xa7 �d2 4 1 ltld5 e4 ! (not 4 1 . . . ltlb5 + 42 ¢>b6 ltlc3 43 ¢>c5! e4 44 ¢>d4! ltlxd5 45 ¢>xe4 ltlc3 + 46 ¢>f5 and White wins but 43 . . . ltlxa2! draws-G. Flear) 42 �b6! (less precise is 42 a4? ¢>c2 43 b4 �b3 44 a5 �c4 45 ll:le3 + ¢>xb4 46 �b6 ltlc4+ with a draw) 42 . . . e3 43 ltlxe3 ¢>xe3 44 ¢>c6 .!ticS 45 ¢>d7 ltlb6?! (45 . . . ltla7? is inferior: 46 a4 �d4 47 ¢>e6 �c3 48 ¢>f5! �xb3 49 �xgS ¢>xa4 50 h4 ltlc6 5 1 ¢>f6! and wins; or 47 . . . ¢>e4 48 b4! ltlc6 49 bS ltla5 50 b6 �f4 5 1 ¢>d5 ¢>g3 52 ¢>c5 ¢>xh3 53 �bS, again winning for White) 46 ¢>e6 �e4! 47 a4 ltld5 48 a5 ltlb4 49 ¢>d6 ¢>d4!, or 49 ¢>f6 ¢>f4, with a draw in either case. These variations show that although Black achieves a satisfac­ tory result, it is only with immense difficulty. • • •

28

f4?

This throws away White's advantage, since the exchange of a pair of kingside pawns increases Black's chances of a peaceful outcome. 28 29

fS!

ltla4

Having let my winning chances slip, I commit a further inaccuracy. 29 �f3 was simpler. 29 30

¢>f6 .!ticS

Preventing 30 . . . e5 on account of 3 1 ltld7+. 30 31 32

ltld3

ltlb6 ltld7

ltlb4 I should have brought the king forward : 32 ¢>f3 e5 33 fe + ltlxe5 + 34 ltlxe5 ¢>xe5 35 h4! with a draw. 32 33

ltlc6

34 35

ltld4

fe +

eS a6 ltl xe5

Fianchetto System 159 The pawn ending would be lost for White. 35 36

�e7

�n 36 �f2 is answered by 36 . . . etlg4 + . 36 37 38 39 40

�d6 �e2

etlcl

� dS �e4

a4 b4

.!Of3 etld4+

In the event of 40 . . . .!Oxh2!, White could maintain the balance with 41 b5 ab 42 ab �d5 43 .!Od4! .!Og4 ! 44 b6 �d6 45 .!Oe6! �c6 46 .!Of8 .!Of6 47 �f3 �xb6 48 �f4 �c6 49 �g5 .!Oe4+ (or 49 . . . tt!h5 50 �h4 �d5 51 .!Oxh7) 50 �f4 �d6 51 lt!xh7 �e7 52 g4; alternatively 42 . . . lt!g4 43 b6 tOeS 44 b7 lt!d7 45 ltlb4 lObS 46 .!Od3! and again White draws, as Black cannot win without the use of his knight. 41 42

lt!xd4

bS

�xd4

t- t

After 42 . . . ab 43 ab �c5 44 �e3 �xb5 45 �f4 �c4 46 �g5 �d3 47 �h6 �e3 48 h4! �f3 49 �xh7 �xg3 50 �xg6, both sides' resources are completely exhausted. Game No. 23 Karpov-Timman

Candidates Final (4th game) Kuala Lumpur 1990 1 d4 lt!f6 2 3 4 s 6 7 8 9

c4 lt!f3 g3 J.g2 cd .!Oc3

tOeS

g6

J.g7 c6

dS

cd o-o

e6

o-o

In our discussion of this game, we shall scrutinise the fine points of the opening play arising from White's 9th move. The middlegame will proceed at a slow pace, yet the ending (after the adjournment) will be exceptionally interesting, and although this does not strictly

160 Fianchetto System belong to the subject of the book, we shall give a large number of entertaining endgame variations. lt)fd7 9 The best move. After 9 . . . lt)c6 10 li)xc6 be 1 1 �a4 lt)d7 1 2 .tf4 White has a clear positional advantage; Akhmilovskaya­ Chiburdanidze, Women's World Championship match 1 986. 10 f4 (102) 102 B

Karpov-Kasparov, 3rd game, London 1 986, continued differ­ ently : 1 0 lt)f3 (the exchange 10 lt)xd7 .txd7 gives White nothing : 1 1 e3 lt)c6 1 2 b3 "fle7 1 3 .tb2 .C. fc8 t-t ; Portisch-Nunn, Budapest 1 987) 10 . . . �c6 1 1 .tf4 lt)f6 1 2 lt)e5 .td7 (after 1 2 . . . li)xe5 1 3 .txe5, White's chances are to be preferred) 1 3 "fld2 li)xe5 1 4 .txe5 .tc6 1 5 .l:lfd 1 ( 1 5 .C. ac 1 , with the prospect of f2-f3 and e2--e4, would have preserved the initiative) 1 5 . . . lt)d7 1 6 .txg7 q;xg7 1 7 .C. ac 1 �f6 1 8 "flf4 "fi bS 19 "flxb8 .C. axb8 20 f3 .C. fd8 2 1 q;f2 .C. bc8 22 e 3 lt)e8 23 .C. d2 lt)d6 24 .l:l dc2 �f8 2 5 .t n �e7 26 .td3 f5 27 h4 h6 28 b3 (in his notes to the game, Kasparov states that after 28 g4 Black would be in quite a dangerous position. But it seems to me that the continuation 28 . . . .C. f8 29 g5 �e4 + 30 .txe4 fe 3 1 f4 h5, or 30 fe fe + 31 q;e2 ed + 32 q;xd3 hg 33 hg .C. f5!, promises White nothing) 28 . . . g5 2� lt)e2 .td7 30 .C. c5 b6 3 1 .C. c7 .C. xc7 32 .C. xc7 .C. a8 33 lt)g1 �e8 34 .C. c 1 .C. c8 35 .C. xc8 t-t . 10 lt) c6 In game 1 3 of the 1 986 match, my opponent drove tbe knight back with 10 f6, but after 11 lt) fJ �c6 12 .te3 �b6 l 3 .tf2 f5 14 li)e5 .td7 1 5 "fld2 li)c8 16 "fle3 q;hs 1 7 .C. fd 1 ( 1 7 .C. fc l was . . .

Fianchetto System 161 stronger) White generated considerable pressure - although the game was eventually drawn after blunders by both sides. White can retain the initiative by retreating his knight to d3. The game Hulak-H. Olafsson, Wijk aan Zee 1987, is instructive : ( 1 0 . . f6) 1 1 o!Lld3 o!Llc6 1 2 e3! (more precise than 1 2 .i.e3 o!Llb6 1 3 b3 .i.d7 14 o!Llc5 J:l b8 1 5 "it'd2 f5 1 6 J:l fc 1 .!Llc8 1 7 .i.f2 o!Lld6, with equality; Nikolic-Nunn, Linares 1 988) 1 2 . . . f5 (Andersson-Nunn, Brussels 1 988, is worth mentioning : 12 . . . o!Llb6 1 3 b3 ..td7 1 4 ..t a 3 J:l e8 1 5 "it' d 2 o!Ll e 7 16 o!Ll c 5 J:l b 8 1 7 J:l fe 1 f5 1 8 .i.fl o!Llec8 19 J:l ac1 .i.c6 20 o!Lld3 o!Lld6 21 o!Lle5 ..tf8 22 "it'b2! J:l e7 23 J:l c2 J:l c7 24 J:l ec1 J:l bc8, and now instead of 25 o!Llb1, which led to approximate equality, White could have played 25 .i.c5!, when according to Nunn the threat of 26 "it' a3 gives him a noticeable plus) 1 3 .i.d2 (a good alternative is 1 3 o!Lle5 o!Lle7 14 b3 o!Llf6 1 5 ..ta3 ..td7 1 6 J:l c 1 J:l e8 1 7 J:l f2 o!Lle4 1 8 J:l fc2 o!Llc6 1 9 .i.fl a 6 20 o!Llxe4 fe 21 ..td6 .:l c8 22 h4 o!Llxe5 23 de "it'aS 24 hS; Dzhind­ zhikhashvili-Mestel, Reykjavik 1 990) 1 3 . . . o!Llf6 14 J:t c l ..td7 1 5 o!Lle5 J:l e8 1 6 h 3 o!Llxe5 1 7 de o!Lle4 1 8 o!Llxe4 de 1 9 "it' b3 .i.c6 20 ..tb4, with advantage to White. Shortly after the London/Leningrad match, Kasparov played this variation himself with White, against Nunn (Brussels 1 986). The exchange on e5 unexpectedly led to a crushing defeat for Black : 10 . . . o!Llxe5 1 1 fe o!Llc6 1 2 e4! de 1 3 ..te3 (103) .

103 8

With Black at the moment a pawn up and not having made any noticeable mistake, it is hard to imagine that he will resign after six ( ! ) more moves. 1 3 . . . f5 14 ef J:l xf6 (the play up to here was still on familiar lines. For example, this position had arisen by transposition in Sveshnikov-Mikhalchishin, Lvov 1983; after

162 Fianchetto System 1 S .J:I. xf6 .txf6 16 ll:le2 ll:lb4, White's advantage evaporated. Kasparov doesn't waste time exchanging rooks) 1 S ll:lxe4 .J:I. xfl + 1 6 1Wxf1 ll:lxd4? 1 7 .J:I. d 1 eS 1 8 lOgS, and Black stopped the clock. The following beautiful variation is possible : 1 8 . . . 1We7 1 9 .tdS+ .te6 20 l:l xd4 ed 2 1 .txe6+ *h8 22 lll f 7+ *g8 23 /Od8+ *h8 24 .tgS! 1Wb4 2S /Of7+ *g8 26 ll:leS + *h8 27 /Oxg6+ hg 28 1Wh3 + . Incidentally, all this actually happened in a corre­ spondence game Hjorth-M. Andersson ( 1 986), and at this point Black resigned. Instead of 1 6 . . . ll:lxd4, Black might play 1 6 . . . .txd4 1 7 .txd4 /Oxd4 1 8 l:l e l eS 1 9 1Wf6. Kasparov judged this position to be in White's favour, in view of 19 . . . 1W xf6 20 lll xf6 + *g7 2 1 /Oe8 + *f8 2 2 .J:I. xeS with the better ending, but Gutman has suggested 19 . . . 1!fb6 20 *hl .td7 21 1!fxe5 l:l f8 22 lOgS .tc6. However, after 23 .tdS + ! (G. Flear) Black can resign! 12 e4, then, seems quite strong but White can also play the solid 1 2 .te3!?. This position arose in Karpov-Timman, Amsterdam 1 986, though by a different move-order ( 1 0 . . . /Oc6 1 1 .te3 ll:lxeS 12 fe). Let us look at a few moves of that important encounter. 12 .te3 f6 13 ef .J:I. x/6 Better than 1 3 . . . .txf6 1 4 1Wd2 .td7 1 S * h 1 .J:I. f7 1 6 .tg1 .te8 1 7 l:l ad 1 .tg7 1 8 l:l xf7 .txf7 1 9 e4, with the advantage; Makarov-Glek, Moscow 1 986. 14 1Wd2 .td7 15 *h1 l:l xf1 + 16 l:l xfl 1We7 (104) 104 w

This is a safer post for the queen than aS ( 1 6 . . . 1WaS 1 7 a3 l:l f8 1 8 l:l xf8+ .txf8 1 9 .tg 1 ) or b6 ( 1 6 . . . 1Wb6 1 7 .tg1 /Oxd4? 1 8 l:l f4).

Fianchetto System 163 1 7 '4 d1 Over-protecting the d-pawn and preparing ..te3-g1 and e2-e4. But White can also withdraw his bishop to g1 at once. Ribli-Nunn, Dortmund 1 987, went 1 7 ..tg1 '4 d8 18 a3 �h8? ( 1 8 . . . ..tc8 1 9 ..te3 '4 f8 i s safer) 1 9 e4! (the fiasco against Kasparov has not put Nunn off this variation, but once again the advance of the centre pawn puts him in a difficult position) 19 . . . de 20 llJxe4 ..tc8 2 1 ..te3 (21 llJg5 ..tf6) 2 1 . . . .l:l£8 22 : xf8 + 1Fxf8 23 b4, with a clear advantage. 17 . . . '4 c8 In Karpov-Chiburdanidze, Bilbao 1 987, the Women's World Champion played 17 . . . �h8 18 a3, and only then 1 8 . . . : c8. After 1 9 .*.g5 1Ff8 20 :n 1Fg8 21 e3 h6 22 ..tf6 .*.xf6 23 : xf6 : f8 24 : f2 1Fg7 25 : xf8 + 1Fxf8 26 e4 de 27 lt:lxe4 b6 28 �g1 'ik g7 29 d5 ed 30 1Fxd5 1Fd4+ 31 1Fxd4 + llJxd4, the game had turned from an opening into an ending, with White retaining a minimal edge. But Maya defended precisely, and obtained a draw in spite of all my efforts. 18 a3! White's intention is to retreat his bishop to g1 and seize the centre. But in reply to the immediate 1 8 ..tg1 , Timman had prepared 18 . . . 1Fb4 ! . Then after 19 e4 de 20 .*.xe4 ..te8, Black can secure his position with . . . .*.e8-f7 (2 1 d5? ..txc3!). If instead 20 llJxe4 1Fxd2 2 1 : xd2, Black has the tactical device 2 1 . . . llJxd4 22 : xd4 (22 ..txd4 .*.xd4 23 : xd4 : c 1 + 24 ..tfl ..t b5) 22 . . . ..txd4 23 ..txd4 : c 1 + 24 ..tgl .*.c6, with counterplay in a complex endgame. 18 . . . ..t/6 White easily refutes 1 8 . . . llJa5 with 1 9 llJxd5 ed 20 .*.g5 llJc4 21 ..txe7 llJxd2 22 .*.xd 5 + �h8 25 : xd2. 19 .*.g1 ..tg5? A more precise move was 19 . . . 1Fg7, hindering e2-e4 for the moment, on account of 20 e4 de 21 llJxe4 ..txd4 22 .*.xd4 llJxd4 23 1Fxd4 1Fxd4 24 : xd4 : c 1 + 25 .*.fl ..tb5 ! . 2 0 1Ve1 llJd8 2 1 e4 de 22 1Vxe4 b 6 23 d5 Black could not prevent this breakthrough, and has to go over to wholly passive defence. The results of the opening can now be summed up as miserable for Black. The game, however, lasted another 35 moves and ended in a draw. 11 .*.e3

164 Fianchetto System At this point the exchange on e5 leads to variations we have looked at already ( 1 1 . . . �dxe5 12 fe f6 etc.). In the main game we are examining, Black drives back the centralised knight and follows with . . . �b6. I should mention that on two occasions in Seville, Kasparov played 1 1 . . �b6 at once, and equalised easily in both cases. We shall thoroughly discuss the immediate knight move in the notes to Game No. 24. .

11

f6

�b6 12 �d3 'fle7 13 b3 13 ..td7 would transpose into the Nikolic-Nunn game that we have already seen. So it may be said that we are only now breaking new ground. . . .

14 15 16 17 18

19

a4 .tel e3

..ta3 l:t c l ..txf8

..td7 l:t fd8 ..te8 'flf7 ..tf8 'fi xf8 (105)

105 w

As the result of some complex manoeuvres, a certain advantage for White has emerged on the queenside. 20 g4 'fie7 l:t ac8 21 'fld2 22 �e2 l:t c7 23 l:t c5 �c8 24

f5

gS

White is slightly better after 24 . . . �d6 25 �df4 (25 fe? �e4 26 ..txe4 de 27 �df4 'fi xc5, and Black wins) 25 . . . gf 26 gf �xf5 27 �xd5 ed 28 : xf5.

Fianchetto System 165 25

�g3

2S fe 1i'xe6 26 �c3 (26 �g3 �g6 27 �fS �8e7) 26 . . . � 8e7 27 e4 is less clear. Now the threat of bringing the knight to f5 is more dangerous : 2S . . . .i.f7 26 fe 11'xe6 (26 . . . �xe6 27 �h5 .l:l. f8 28 �b4!) 27 �f5. To avoid this, Black sacrifices a pawn. e5!? (106) 25 106 w

26 Wet An inaccuracy, committed after much thought. The sheer variety of possibilities put me in rather a quandary, and I decided to decline the pawn sacrifice although Black's position would almost have become critical if I had calmly accepted it. After 26 �xd5+ (26 .l:l. xd5 ed 27 ed .i.f7! 28 .l:l. xd8 + Wxd8 gives Black counterplay for the pawn) 26 . . . �h8, White has a wide choice : (a) 27 � xc6 J.xc6 28 1W c1 ed 29 e4 !, or (b) 27 .l:l. fcl ed 28 e4! . This strategic operation involving the return of the pawn and the inevitable break with e4-e5 secures White a considerable plus. (c) 27 de �xeS 28 .l:l. xc7 1Wxc7 29 .l:l. c 1 'fle7 30 �xeS 1W xeS 3 1 1W d4! is also good. (d) 27 1Wcl .l:l. xd5 28 .l:l. xdS b6 (28 . . . �b4 29 l:t cS!) 29 de �b4 30 �xb4 .l:l. xcl 3 1 .l:l. xc l , or 27 . . . b6 28 .l:l. xc6 J.xc6 29 J.xc6 ed (29 . . . .l:l. d6 30 dS .l:l. xdS 3 1 .i.xdS) 30 ed .l:l. xd4 (30 . . . .l:l. d6 31 dS .l:l. xdS 32 .l:l. e 1 !) 3 1 .l:l. e 1 1Wd6 (3 1 . . . .l:l. xd3 32 .l:l. xe7 �xe7 33 Wet �xc6 34 1We8 + �g7 3 S �hS+ �h6 36 1Wf8 + and mates) 32 .l:l. e8 + �g7 33 l:t e6 - with a very strong initiative for White in either case. (e) 27 .i.g2 (in some of the foregoing variations the white king's

166 Fianchetto System bishop departs from the board, but it is also possible to bring it home before anything else) 27 . . . ed 28 e4!, followed by e4-e5 (if necessary supported by a rook - : fe l ). If 28 . . . lt:le5, then 29 : xeS! fe 30 f6 and 3 1 'ill xg5 is crushing. The queen move to c1 allows the centre to be blocked; if l had been playing for that, I should have continued 26 : fe l l e4 27 lt:lb4! - possession of the c-file is not without significance. The trouble is that I had 'too many irons in the fire' (trying to control the c-file and f-file at once). Now Black nullifies the danger in the centre and obtains a fully viable game. The ensuing phase, as I have said already, is a trifle dull; the most interesting events occur only after the adjournment. 26 b6 27 28 29 30

: c2 lilf2 1i'd2 l:tfcl

e4 lt:ld6 .l: dc8 aS

Black also stands quite well after 30 . . . lt:la5 3 1 : xc7 : xc7 32 : c3 'ill d 8 33 'ill c2 : xc3 34 'ill xc3 'ill c8! 31 -*.fl If White hadn't made the superfluous queen moves, this bishop manoeuvre would now be extremely dangerous for Black. The attempt to build up with 3 1 'ill c 3 would be countered by 3 1 . . . -*.d7 and 32 . . . lt:le8. 31

lt:lb4

3 1 . . . lt:lxd4! looked inviting, but after 32 ed e3 33 -.d3 ef+ 34 �xf2 : xc2 + 35 : xc2 lt:le4+ 36 �f3, the draw is not far away (the exchange of the good knight on c6 for the bad one on f2 is not advantageous to Black). The match situation compelled Timman to maintain the tension in the hope of seizing the initiative. 32 33 34 35 36 37 38 39

: c3 lild1 : xc6 : xc6 lilc3 �f2

40

�el

�e1 �d1

-.d7 : c6 : xc6 -.xc6

�f8

�e7 � f8 -.cs � g7

Fianchetto System 167 Under pressure from the clock, we have both been marking time with our kings a little, and I should mention that Black missed quite a good opportunity : 40 . . . hS! 41 gh (41 h3 with equality) 4 1 . . . lll xfS 42 .J.. h 3 .J.. d 7! 41 42

lll a 2 11rxa2

lll x a2 11rc7

A possibility was 42 . . . •c3 + 43 �f2 bS 44 ab lll x bS (but not 44 . . . .i.xbS 45 lll h S + �f7 46 .i.xbS lll x bS 47 b4!). 43 44 45 46

�f2 1i' b2 .tel �el

�f8 �e7 �d8 �c8

�d2 •ct

�b7 11re7

46 . . bS!? is sharper. .

47 48

For drawing purposes, 48 . . . 11rxc1 + was simpler. .i.d7 49 �el 50 51

Ill eS �f2 11Vhl ! (107)

107 B

I didn't feel like settling for a draw, and decided to provoke mind-bending complications by abandoning my queenside to its fate while going ahead on the kingside. Obviously there could be no question of calculating the variations to the end. 51 52 53 54 55

h4! hg 11rxh7 11re7

11V b4 11V xb3 fg • xa4 11rc6

168 Fianchetto System 56 57 58 59

1hg5 'fll e7 'fll xd6 .t. dl

a4 'fll d6 ltlxd6 .tb5?!

It was hardly a good idea to block the path of his pawn. After 59 . . . b5, it is most likely that the game would soon have been drawn; at any rate Black would not be risking anything. Here are some sample variations : 59 . . . b5 60 f6 (60 ltle2? b4 61 ltlc1 �c7) 60 . . . .txg4 61 .txg4 ! a3 62 .te6 �c6 63 ltle2 (not 63 f7 ltlxf7 64 ltlxe4 ltlg5! 65 it:lxg5 a2, and Black wins) 63 . . . a2 (63 . . . b4 64 ltlcl) 64 .t.xd5 + �xd5 65 ltlc3 + �c4 66 ltlxa2 b4 (nor does he gain anything from 66 . . . �b3 67 ltlc1 + �b2 68 ltle2 b4 69 d5 b3 70 �g3 �c2 71 ltld4+ �c3 72 ltlxb3 �xb3 73 �f4 �c4 74 �e5) 67 ltlxb4 �xb4 68 �g3 �c4 69 �f4 �d3 (69 . . . �d5 70 �g5 �e6 7 1 �g6) 70 �e5 ltlf7 + 71 �e6 ltlg5 +, with a draw. ·

60 61

ltle2

a3

ltlcl (108)

108 8

In this position Timman sealed his move after nearly half an hour's thought. Many commentators were misled by Black's queenside activity, and in the local newspapers the following day there were even headlines like "Can Karpov survive?" In actual fact, adjournment analysis revealed that the only winning chances lie with White, whose pawns are much more dangerous than his opponent's. All the same, we shall see that the rightful outcome of this tense struggle would have been a draw. 61

�c7

A good sealed move, but of course we had analysed others too.

Fianchetto System 169 Let us look at the main variations discovered by Zaitsev and myself: (a) 61 -*.d7 62 -*.b3! �c6 63 �g3 �c4 64 �f4 �b2 (threatening the deadly . . . �d3 +) 65 �g5! �c4 66 -*.xc4 (given White's active king, the creation of another black passed pawn is not dangerous) 66 . . . de 67 �f6 b5 6S �a2 b4 69 �xb4 �b5 70 �a2 �a4 71 �c1 a2 72 �xa2 �b3 73 �c1 + �c2 74 �e7 �xc 1 75 f6 c 3 7 6 f 7 -*.xg4 77 fS(11t') c2 7 S 11t'f4 -*.f3 7 9 d5 � d 2 S O d6 cl(11t') S 1 d7, and Black is defenceless. (b) 61 ..te8 62 �a2 (after the careless 62 �g3, the black knight performs the same nimble triangulation that we have seen before : 62 . . . �c4 ! 63 �f4 �b2, and wherever the bishop goes, Black plays 64 . . . �d3 +) 62 . . . �c6 63 -*.b3 �c4 64 g5 �d6 65 f6 -*.f7 66 �g3 �c4 67 �f4 �d6 6S �c3 ..tg6 69 -*.xc4 de 70 �b5 + �c6 7 1 � xa3 �d5 (7 1 . . . b5? 72 �e5 c3 73 �c2 -*.f7 74 d5 + ! -*.xd5 75 g6) 72 �b5 �c6 73 �c3 b5 74 �e5 b4 75 d 5 + �d7 76 �a2 b3 77 �c3 b2 7S �d4, with a decisive advantage. (c) 61 �c6 62 �g3 �c4 (if 62 . . . -*.c4 63 �f4 �d7 64 g5 a2 65 �xa2 -*.xa2 66 g6, Black is mated after either 66 . . . �eS 67 -*.a4+ �e7 6S -*.xeS �xeS 69 �g5 b5 70 �h6 b4 7 1 g7 �f7 72 �h7 b3 73 gS(11t')+, or 66 . . . �e7 67 g7 �f7 6S f6 -*.c4 69 -*.h5+ �gS 70 -*.g4 -*.b5 71 -*.e6+ �f7 72 �f5 -*.eS 73 �g6 b5 74 -*.xd5) 63 -*.e2 �d6 (if 63 . . . �xe3?, then 64 -*.xb5 + �xb5 65 f6, and the pawn goes on to queen. Other variations are also unacceptable : 63 . . . �d6 64 -*.xb5 + �xb5 65 �f4 �b4 66 g5 �c3 67 g6 �e8 68 �e5 �b2 69 g7 �xg7 70 f6 �xc l 7 1 fg a2 72 gS(11t') a 1 (11t') 73 11t' g 1 + �b2 74 11t'xa1 + � xa 1 7 5 �xd5, o r 6 3 . . . �d6 64 -*.xb5 + �xb5 6 5 �a2! �d6 66 �f4 �c7 67 g5 �e8 68 g6 �f6 69 �g5 �e7 70 �b4 b5 7 1 �c6 + ! �d6 7 2 �xf6 a 2 73 g 7 a 1 (11t') 74 gS(11t') �xc6 75 11t'e6+) 64 �f4 -*.a4 65 g5 �b2 66 g6 �d3 + 67 �xd3! ed 68 ..txd3 �e7 69 g7 �f7 70 f6 �g8 (70 . . . a2 7 1 -*.h7 a 1 (11t') 72 g8(11t')+ �xf6 73 11t'f8 + �e6 73 -*.f5 mate) 71 �g5 a2 72 �h6 a 1(11t'), and a familiar mate follows : 73 -*.h7 + �f7 74 g8(11t')+ �xf6 75 11t'f8 + etc. Now let us see what happened in the game. � c4 62 �g3 62 . . . �d7? loses in a way which we have seen before : 63 �f4 -*.c4 64 g5 a2 65 �xa2 ..txa2 66 g6 �e7 (66 . . . �e8 67 -*.a4+) • • •

. . •

• • •

1 70 Fianchetto System 67 g7 �f7 68 f6 .tc4 69 .tg4 �g8 70 .te6 + t;)f7 7 1 �f5 .tb5 72 �g6 .te8 73 .txd5 b5 74 .ta2 b4 75 .tb3. 63 .te2 But not 63 �f4? t;)b2! threatening . . . t;)d3 + . .te8! 63 The strongest defence. 63 . . . li)xe3 is inadequate : 64 .txb5 t;)c2 65 g5 t;)xd4 66 f6 �d6 67 g6 �e6 68 g7 �f7 69 .te8 + �g8 70 �f4, and wins. 64 �f4 Podgayets revealed a striking variation culminating in a problem-like mate : 64 g5 t;)xe3 65 �f4 t;)c2 66 �e5 t;)b4 67 �f6 �d6 68 g6 t;)c6 69 ..tb5 e3 70 g7 t;)e7 7 1 ..txe8! e2 72 li)xe2 a2 73 t;)c3! t;)g8 + 74 �f7 t;)h6+ 75 �f8 a l (V) 76 t;)b5 mate (109) . 109 B

It would indeed be pleasant to finish the game like this, but unfortunately Black has a defence : 64 . . . �d6 65 ..txc4 de 66 t;)a2 ..ta4 67 t;)c3 .tb3 68 t;)b5 + �e7 69 li)xa3 c3 70 �f4 b5! 71 �xe4 b4 72 �d3 ba 73 �xc3 .td5 with a draw. 64 t;)b2 In my adjournment analysis I considered 64 . . . b5 to be more precise : 65 t;)a2 t;)b2 (but not 65 . . . �d6 66 g5 t;)b2 67 g6 b4 68 �g5! b3 69 t;)cl .ta4 70 g7 a2 71 g8(¥) a l (V), as White obtains an irresistible attack with 72 Vb8 + �d7 73 f6! Vxc l 74 ..tg4+ �c6 75 V c8 + ) 66 f6 t;)d3 + 67 �f5 b4 68 .td l . But the move played is not yet the cause of a catastrophe, either. 65 �g5 The sole winning attempt. It is true that after 65 t;)a2, the following long variation gives White good prospects : 65 . . . �d6

Fianchetto System 1 71 66 �gS �e7 (66 . . . lilc4 is a mistake in view of 67 ..txc4 de 68 �f6 ..ta4 69 �g7 ..tb3 70 f6 c3 7 1 lilxc3 a2 72 lilxa2 ..txa2 73 gS) 67 �h6 �f6 68 lilb4 ..tf7 (inadequate alternatives are 68 . . . ll:ld3 69 lilxdS + �f7 70 gS a2 7 1 g6+ �f8 72 g7 + �f7 73 ..thS+ �g8 74 lll e 7 mate, and 68 . . . ..tc6 69 ..tfl ! lildl 70 gS + �f7 70 . . . �xf5 7 1 ..th3 mate, is pretty - 7 1 g6+ etc.) 69 ..tfl ! lild l (it is too late for counterplay with 69 . . . lild3 70 gS + �e7 7 1 lilc2 a 2 7 2 �g7) 7 0 g5 + �e7 7 1 �g7 lilxe3 7 2 f6 + �e6 73 ..th3 + lilfS + 74 ..txfS + �xf5 75 �xf7 e3 76 �g7! e2 77 ll:lc2 a2 78 f7 e 1 (1i') 79 lilxe l a 1 (1i') 80 f8(1i')+. However, after the correct 65 . . . b5!, nothing can be found for White. lild3

65

65 . . . lilc4 comes too late : 66 �f6 lilxe3 67 �e7. al 66 lil b3! ? 67 68

ll:lal

69

-*.b3

bS

..tdl b4 (IJO) Bringing the king a little closer to the pawns does not work : 68 . . . �d6? 69 ..tb3 lilcl 70 �h6 b4 7 1 f6 �e6 72 �g7 lilxb3 73 lilxb3 ..ta4 74 lilcS + ! �d6 75 f7 a 1 (1i') 76 f8(1i')+ �c6 77 1i'c8 + �d6 78 'ifd8+ �c6 79 1i'd7 + �b6 80 1i'b7 + �aS 8 1 1i' a6 mate. 1 10 w

lilcl

After 69 . . . �d6 70 �f6, the g-pawn cannot be stopped. 70 ..t xd S �d6 71

..tc4

..tbS

Not 41 . . . �e7 42 f6 + �f8 43 �h6 ..tf7 44 dS b3 75 d6!, or 7 1 . . . ..ta4 72 f6 b3 73 �h6! b2 74 f7 b1 (1i') 75 f8 ('if)+ �d7 76 1Wf5 + �c7 77 1i'a5 + , but here 74 . . . ba (11') leads to a completely unclear position - G. Flear. An interesting possibility is 71 . . . b3 72 ..txb3 lilxb3 73 lilxb3

1 72 Fianchetto System �d5 74 �a1 �c4 75 .tf6 �c3 76 �e7 �b2 77 �xeS �xa1 78 f6 �b2 79 f7 a 1 (W) 80 f8(W) Wa4+ 81 �f7 �b3 + . We had reached this position in our analysis, and Zaitsev suggested 82 .tf6 W xe3 83 �e5 .tc2 84 Wf4 .td3 85 d5 with chances of success. But in the post-mortem, Timman pointed out the more effective 82 d5! Wxe3 (82 . . . W xd5+ 83 'Oftg6 with a technically won ending) 83 W b4 + ..tt c 2 84 g5 Wf3 + 85 �g7 e3 86 Wc4+ �b2 87 W b5 + �c1 88 d6 e2 89 Wc5+ �b2 90 Wb6+ �c1 91 d7 e l (W ) 92 d8(W) Wec3 + 93 Wdf6 etc. According to computer analyses, this ending is a win for White. � e7 72 ..tg8 73

�h6

After 73 f6 + �f8 74 ..te6 b3 (74 . . . ..ta4 leads to the same result as in the game) 75 .i.xb3 �xb3 76 �xb3 ..tc4 77 �at �f7 78 �f5, White wins. However, Black has the saving move 74 . . . .i.d7!, and if 75 ..txd7 (it is more sensible to settle for a draw) 75 . . . b3 76 �f5 b2 77 g5 �e2! 78 g6 � g3 + ! 79 �g5 �h5! 80 �xh5 ba(W) 8 1 g7 + �f7 82 .i.e8 + (82 ..te6 + �xf6 83 g8(W) 'if h 1 +) 82 . . . ..tt xf6, it is Black who unexpectedly wins. �f8? 73 It is only now that Timman commits the decisive error. He could have drawn with 73 . . . b3! 74 .i.xb3 �xb3 75 �xb3 .i.c4 76 � a l �f6!, as pointed out by his second, the Hungarian grandmaster Sax. .i. d7 74 ..te6! 75 76

gS g6

b3 1-0

Black resigned rather than be mated : 76 . . . ..txe6 77 fe b2 78 g7 + rJ;e7 (78 . . . �g8 79 e7 �f7 80 g8(W)+ �xe7 8 1 Wg5 + rJ;f7 82 Wd5+ rJ;f6 83 'ifc6+) 7 9 g8(W) ba(W) 8 0 W f 7 + �d6 8 1 W d 7 mate. After this victory, it became clear that another duel for the world crown between Kasparov and myself was inevitable. Game No. 24 Wojtkiewicz-Karpov

Haninge 1990 I have played the White side of the Griinfeld so often, especially the g2-g3 system, that when the symmetrical variation arose in

Fianchetto System 1 73 the present game, it was with some interest that I handled it for the other colour. lt:l f6 1 lt:lf3 2 3 4 5

c4 g3

.i.gl d4

g6

.i.g7 0-0 c6

I have rarely played either the King's Indian or the Griinfeld with Black. But given the choice between them, I prefer the latter. 6 7

8 9 10 11

lt:lc3 cd lt:l e5 0-0 f4 .i. e3

d5 cd e6 lt:lfd7 lt:lc6 lt:lb6 (1 1 1)

111 w

Kasparov had played this knight move against me in the first and third match game in Seville, and now I decided to use it myself. An alternative of equal value is 1 1 . . . f6 (see Game No. 23). In the present game, you might say that I combined the two ideas. Before proceeding with the main game, in which White played 1 2 b3, let us look at the two encounters from Seville, where I preferred 1 2 .i.f2. The first match game went as follows : 12 j_f2 .i.d7 13 e4 l0e7 14 l0xd7 In the game Drasko-Nikolic, Vrnjacka Banja 1 987, where it appears this position occurred for the first time, White opted for 14 ed lt:lbxd5 1 5 lt:lxd5 lt:lxd5 1 6 Wb3 .i.c6 1 7 : ac l 1Wa5 1 8 .:l. c5 'it'a6, and a draw was agreed.

1 74 Fianchetto System Another possibilty is 14 a4! de 15 aS .!Dbd5 16 .!Dxe4 .:t b8 17 'iVb3 ..te8 18 .l:tfcl .!Dc6 1 9 'iVa3 .!Dcb4. Nikolic-Hulak, Zagreb 1 987, continued 20 .:t c4? .!Da6 2 1 .!Dd6 .!Dac7 22 .:t ac l .!DbS 23 .!DxbS ..txbS 24 : cS ..te8 2S b4 b6, and White achieved nothing; but Gutman's suggestion 20 .!Dc3! gives him the advantage. At move 1 7, an alternative is 1 7 ..tc6 18 .!DeS .!Dc7 19 ..txc6! with the initiative; this is stronger than 18 .:t fc l a6 ( 1 8 . . . .!Dc7! at once is more precise) 19 .:t c4 ( 1 9 .!DeS! is correct) 19 . . . .!Dc7 20 .!Dc3 ..txg2 21 �xg2 .!bedS, with equality; Andersson-Hulak, Wijk aan Zee 1 987. 14 . . . 'iVxd7 15 e5 .:l.fc8 16 :t el ..t/8 17 ..tj3 .:l. c7 An accurate move which equalises. I would have answered the incautious 1 7 . . . a6 with 1 8 ..te2 .:t c7 1 9 g4 ! ..th6 20 'iVd2, and if 20 . . . .!Dc4, then 2 1 ..txc4 : xc4 22 .!De4 ! . But now this idea doesn't work : 18 ..te2 : ac8 19 g4 ..th6 20 'iVd2 .!Dc4 21 ..txc4 .:t xc4, and both black rooks are on the c-file. 18 b3 : ac8 19 'iVd2 lbc6 20 'iVb2 a6 21 ..te2 'iVe7! 22 l1Jb1 liJb4 23 lbc3 lbc6 24 l:lJb1 lbb4 25 .:l. c5 t:tJd7 26 : xc7 : xc7 27 lbc3 l:lJ c6 28 l:lJb1 t:tJb4 29 l:lJc3 lbc6 30 lbb1 i-J. In the third game, Kasparov varied with 1 2 . . . CfJe7! (a refinement on 12 . . . ..td7 of the first game) 13 a4 aS 14 'iVb3 (Black would have had more problems after 14 e4 de l S ..txe4 .!DbdS 16 'iVb3. Greenfeld-Birnboim, Tel-Aviv 1 988, continued 16 . . . f6 17 .!Dc4 �h8 1 8 .:t fe l CfJb4 1 9 .:t ad 1 .!bedS 20 .!De3 : a6 2 1 ..txdS ed 22 .!DexdS ..tg4 23 .:t d2 .:t e6 24 .!Dxb4 .:t xe l + 2S ..txe l ab 26 'iVxb4 : e8 27 ..tf2, and White gained the advantage) 14 . . . ..td7 1 S .:t fc 1 ..tc6 1 6 .!DbS .!Dbc8! 1 7 e 3 .!Dd6 1 8 .!Dxd6 'iVxd6 1 9 .tel .:t fb8 20 ..tfl f6 2 1 t:tJf3 'iVd7 22 'iVc2 .!DfS 23 ..td2 .!Dd6 24 b3 .:t c8 2S 'iV d l h6 26 .te l gS 27 : a2 'iVe8 28 : ac2 .U8 29 ..td3 g4 !-!. . . .

12

b3

This move was played for the first time in Portisch-Korchnoi, Reggio Emilia 1 987/8. It seems to me that if White is seeking the initiative, it is worth trying 1 2 .!Dxc6 be 1 3 ..tf2. 12 13

..td7

f6 'iV d2 In Portisch-Korchnoi, Black first played 1 3 . . . .:t e8 1 4 .:t fc l , and only then 1 4 . . . f6. After l S .!D d 3 .:t e7 1 6 � h 1 ..te8, the players agreed a draw. Evidently on that day they were just not in the mood for a fight; the result has nothing to do with the

Fianchetto System 1 75 position. In the present game, I carry out a regrouping in the centre as Black, without wasting time on rook manoeuvres. 14 15 16

IL!d3 .l:t acl

ILleS IL!d6

-*.f2 (1 12)

JJ2 B

The position is characterised by so-called dynamic equilibrium. The pawn chains are closed, and the forces cannot clash for some time yet. Some precise and none-too-obvious manipulation of the pieces is called for, and of course such operations are not to everyone's taste. But I personally have always liked this kind of game, based as it is on a wealth of subtle points. I had previously encountered such situations with White, but was also ready to try them with Black, especially since the configuration of pieces is nearly symmetrical. Here, by the way, is an illustration of the subtleties inherent in the position. Black just needs to play one careless move, 16 . . . ..te8?, and White immediately seizes the initiative : 1 7 ILleS! -*.f7 ( 1 7 . . . "W/e7 18 IL!xe6!) 18 e4 de 19 ILJ3xe4 IL!xe4 20 -*.xe4 "W/e7 21 IL!xb7! 16 17 18 19

IL!e5 �h1 g4!?

f5 "W/e7 .l:t fc8

1 9 IL!a4 or 19 a3 would have preserved equality, but White is trying for more. 19 20 21

IL!xe4 IL!xd7

IL!e4 fe

1 76 Fianchetto System The knight cannot hold out on eS for long, but perhaps it was worth exchanging it for the black one : 2 1 .l:. c3 �xeS 22 de .l:. xc3 23 'ii' x c3 .l:. c8 24 'ii' d 2 b6 2S .tg3, intending f4-fS. 2 1 .tg3!? at once is also interesting : 2 1 . . . �xeS 22 de .l:. xc l 23 .l:. xc l .l:. c8 24 .l:. xc8+ .txc8 2 S 'ii' c 3 (White's aim i s t o carry out f4-fS, but the situation does not yet call for it : 2S fS? ef 26 gf gf 27 'ii' xdS + ..te6, or 2S 'ii' d 4 b6 26 fS ef 27 gf gf 28 'ii' xdS + J. e 6 2 9 'ii' a8 + �f7, with a good game for Black i n either case) 2S . . . 'ii' d 7 (but not 2S . . . 'ii' d 8 26 .th4!, or 2S . . . .td7 26 'ii' c 7) 26 e3 b6 27 a4 .tb7, and again the chances are about equal. 21 22

'ii' xd7 a3 (1 13)

l /3 B

One thoughtless move, made on general grounds - and already Black has a chance to seize the initiative. 22 e3 was correct, avoiding weakening the queenside. 22 23

.th3

aS! 'ii' e7

Forcing the rook to abandon the c-file. l4

.l:. al

�d8

Black could have prevented the f4-fS break by playing 24 . . . .th6 2S gS .tg7. After 26 .l:. fcl, followed by e2-e3 and .th3-fl , it would be hard for either side to play for a win. Instead, I attempt to bring my knight over to the kingside, where the decisive events will shortly take place. 25

fS

gS

The sharp 2S . . . �f7!? is also interesting. 26

.tg3

�c6

Fianchetto System 1' 77 At this point I convinced myself that after 26 . . . lll f 7 27 fe'! 1Wxe6 28 : f5 h6 29 : an lll d 6 30 ..te5! lll xf5 3 1 gf 1Wd7 32 ..txg7 1Wxg7 33 f6 the initiative is with White, so I decided to bring the knight back again. 27 28 29 30

e3 fe .:. rs 1We2

.:t d8 1Wxe6 h6

30 .:. afl looks more consistent, yet White cannot successfully exploit his domination of the f-file : 30 . . . lll e 7 3 1 ..tc7 .l: dc8! (but not 30 . . . .l: e8 3 1 ..te5 lll xf5 32 gf 1Wa6 33 ..txg7 �xg7 34 f6 +, with an attack for the exchange), and now 32 ..te5 is not so effective : 32 . . . lll xf5 33 gf 1Wa6! 34 ..txg7 (34 f5 .:t f8!) 34 . . . �xg7 35 f6 + �h8! 36 1Wf2 .:t c7 37 ..td7 .l: d6 38 f7 .:t f8 39 1Wf5 �g7 40 1We5 + .:t f6 4 1 ..tb5 1Wd6, and White's attack is spent. On the other hand after 32 ..txa5 Black takes the exchange, and his chances must be preferable. 30 31 32 33 34

.:. n : act .:t xc6 .:t bl

lll e7 .:t dc8 .:t c6 1Wxc6

Rather a passive move. It was worth considering 34 .:t f2!?, and if 34 'ii' c l + 35 ..tfl 1Wxa3, then after 36 1Wb5 1W b4 (36 . . . a4? 37 1Wxb7 ab 38 ..td6) 37 1Wd7 White has a dangerous initiative for the pawn. Instead, 34 a4! leads to unclear play : 35 ba (better than 35 b4 1Wc1 + 36 'ii' fl 1Wxa3 37 ..td6 lll g6 38 1Wb5 'ii' c l + 39 ..tfl 1Wc6 with advantage to Black, or 36 ..tfl 1Wxa3 37 1Wb5 1Wxe3 38 1Wxb7 .:t f8 etc.) 35 . . . .:t xa4 36 1W b2. • . .

. . •

34 35 36 37 38

..tn 1W b2 .tel .tel

..tf8 lll c8 1Wd7 ..td6

The advantage of the bishop pair is illusory - it was better to eliminate the bishop on d6. After 38 :n (more precise than 38 ..txd6 lll xd6) 38 . . . ..txg3 39 hg, it would be difficult to breach White's fortress. 38 39

.:. cl

lll e7 : f8!

1 78 Fianchetto System 40

�g2

Not 40 J.xa5?, on account of 40 . . . l: f2. But 40 �gl, followed by J.e2-d l , was more tenacious. ll:l g6 40 41 42

b4 .-c2

a4 .-f7!

The threats on the f-file are becoming formidable. 43

.-dl

The black a-pawn is immune : 43 .-xa4 ll:lf4 + ! 44 ef .-xf4 45 J.g3 .-xc l 46 J.xd6 .-e l l 47 J.xf8 .-xe2 + 48 �gl e3 49 h3 .-f2+ 50 �h l e2, and wins. �h8!

43

Preparing to open the g-file. 44 l: c3 (1 14) 44 � g l would likewise be answered by 44 . . . h5!, intending to take the g-pawn; after 45 gh .-f5! 46 hg (46 �g2 ll:lh4 + 47 J.xh4 gh is crushing) 46 . . . .-h3 47 J.g3 J.xg3 48 hg .-xg3 + 49 � h l l: f2 5 0 l: c8 + �g7, i t i s all over. Nor does 44 l: c2 help : 44 . . . h5! 45 gh ll:le7 (45 . . . ll:lh4+ is premature because of 46 J.xh4 gh 47 J.g4) 46 h3 (46 l: c3 g4! and . . . .-xh5) 46 . . . ll:lf5 47 l: c3 ll:lh4+ 48 ..txh4 gh 49 .-el (49 .-gl l: g8 + 50 .tg4 .-r3 mate) 49 . . . .tg3 50 .-n .-f2+, and Black wins. 1 14 B

44 45 46

gh J.xh4

hS! tt:Jh4 + !

I f 4 6 �gl (46 �hl .-fl +), then 4 6 . . . .-fs, followed b y . . . .-f5-h3, is decisive.

Fianchetto System 1 79 46 47 48

•et

gh h3+ !

�hl

Other continuations are less stubborn : 48 � g1 • g7 + , or 48 �xh3 : g8 49 ..tg4 : xg4 50 �xg4 •B + 51 �g5 ..te7 + 52 �h6 •f6 mate. 48 49 50 51 52 53

• x£2 : cl ��· �hl ��·

•£2 : xfl : xhl+ : gl+ : fl

55 ..td3 is met by 53 . . . : f3. 53 54 55 56 57 58 59 60 61

�hl ��· �hl ��· �hl ��·

�fl :ci

..tg3 �h7 �h6 : hl + : £2 ..td6 : gl + : g8 : g3

6 1 . . . ..th2 was immediately decisive; so was 6 1 . . . ..txb4 62 ab h2 63 �f2 : g 1 64 : xg 1 hg(•)+ 65 �xg l a3. The game was now adjourned. Realising that Black's threats were unanswerable, White resigned without playing on. 0-1

Game No. 25 Vaganian-Kasparov

Barcelona 1989 1 2 3 4 5 6

10£3 c4 g3 ..tg2 10c3 cd

10£6 g6 ..tg7 0-0 d5 10xd5

The reader will recall that in the World Championship matches of 1 986-7, Kasparov preferred the symmetrical system, in which Black prepares . . . d7-d5 with . . . c7-c6 and recaptures on d5 with

180 Fianchetto System the pawn. In a game we played later (and which I shall examine further on), he chose . . .