beating the grünfeld.pdf

Beating

the Griinfeld

Beating the Griinfeld

ANATOLY KARPOV

Translated by John Sugden

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

Contents

Preface 7

1 Seville Variation 1 1

2 Exchange Variation with 7 �c4 - other systems 4 1

3 Modern Exchange Variation 52

4 Russian System 108

5 4 �f4 System 1 3 5

6 Fianchetto System 1 52

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.

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

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.

1 w

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.

1 2 13 fg 14 ..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.

2 w

Seville Variation 13

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 detail 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 o o o

"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 �g1 "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 19 �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 o· · "it"e6 16 eS! "it"c4! 17 �g1 J: d8 18 "it"e1 �c6 1 9 .tf2 a6 (wouldn't 1 9 . . . bS!? have been better?) 20 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 .. . 'lfc2 3 1 J: b7) 3 1 g6 11Vc2 32

14 Seville Variation

3 B

gh + .l:l. xh7 33 11fg4+ �h8 34 .l:l.fl 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 an even sounder answer to 14 . . . 11rd7 is 15 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 : 14 • • . 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. 1989, 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 1Wd_{5 (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 19 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.

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 19 h3 cd 20 cd -*.xeS 2 1 de 'ifxe5 22 -*.cl ! '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 : b1 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 "ife2 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 "iib3 + �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 • • • 'iff5, Henkin gives these variations : (a) 19 h4, and now :

(a1 ) 19 • • • "ifg4 20 �g1 h6 (or 20 . . . ltlc4 2 1 tt:Jg3! "ifxh4 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 "ifxa5, 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

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 J.xh6

20 �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:lg3! 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.

23 "it'e6

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 de! ..txeS 25 lle1 "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 llxd5 28 lt:lxe2 29 c6! 30 : xeS! 1 -0 llxel �f7 b6

Game No. 2

Karpov-Kasparov

Belfort 1988

Seville Variation 19

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:lc6 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 d8

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 . . . 'iith5, 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 'iltel, Black had the means to obtain a perfectly reasonable position.

6 B

17 'ilta4 (6)

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 <i>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 1989, went 19 de be ( 1 9 . . . l: f8? 20 cb ll:lc4 21 <i>g1) 20 l: d 1 'ilxe5 21 'ila4 l: f8 22 l: d3 c4 23 l: f3 'lidS 24 l: xf8 + <i>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 1988, 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 <i> g 1 and 2 1 g5) 21 . . . 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 <i>h2 W"e5+ 35 g3 'ile4 36 l: d8 + <i>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 1Wxc4 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 1Wd2 : 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 . •• 1Wc4 would be met b y 19 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 1Wxe2 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 1Wxe2 rid black of his worries, on account of 22 1We3 1Wxg4 23 de be (23 . . . .!tlc4 24 1Wd4) 24 1Wxc5.

21 h3 .!tlc4 22 ll'gS! (8)

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.

23 ll'cl 11'£7

8 B

Seville Variation ₂₃

(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 give Black counterplay. 24 • • • 11d5 (for example) has been 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 'flxa1 + 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 .tf2 b5 27 ll:lg3 .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

If White had wanted, he could have put a stop to Black's queenside counterplay with 28 : c t .

24 Seville Variation

28 b4

29 1Wg6 �f8

29 . . . be loses immediately to 30 o!Llf5 <i>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 1Wxg5, or 32 1Wh7.

30 .l:l. xfl

The exchange sacrifice is incapable of improving matters. All it demands from White is a modicum of accuracy.

31 �xf2 be 32 'tiffS+ �g8 33 1Wc8+ <i>h7 34 1W xc5 1Wf7+ 35 <i>g1 c2 36 o!Llg3 .i.f8 37 o!Llf5 �g8 39 .l:l.c1 1-0

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 1988 (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 15 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

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.

18 J.h6

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 illf3 .l:t d8 25 .td4 (25 g5! is also playable) 25 . . . J.f4 26 ¢>f2 1Fd5 27 'it' b l illc4 28 'ilt'e4 illb2 29 'ilt'c2 'it'b5 30 .l:t b l illd3 + 3 1 ¢>fl 'ilt'c4 32 'ilt'e2 b5 33 ille l ! illxe5 34 .l:t xb5 'ilt'f7! 35 ¢>g1 illc6 36 illf3 illxd4 37 cd .l:t c8 (but not 37 . . . J.d6 38 illg5 '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 . . . ille3 23 .!:ta l 'ilt'f7 24 J.xe3 J.xe3 + 25 ¢>h l 'ilt'f2 26 'it'b5 cd 27 illxd4!) 23 J.xe3 illxe3 24 .l:t d3 'ilt'f7 (or 24 . . .

.l:t fl + 25 ¢>h2 illxg4+ 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. 9

B

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 is already better) 26 .J:I.b1 .J:I.xc5 27 .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.b8+ 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+ 35 �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 .1:1.£8

20 g5!

The stock manoeuvre; here too it guarantees White a plus.

20 11d3

21 'ilbl!

Seville Variation 27

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

21 1i'e3

22 it'el R.g7 23 �gl

But not 23 lbc l ? cd and everything is fine for Black, as after 24 1t'xe3de, he wins. 10 w 23 24 lLig3! 1t' e4 (10)

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

24 1t'xh4

25 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.

25 l:txf2

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 lLixf2 cd 27 l:tdl ! 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.

28 1We3! lLic6 28 . . . lLic4 is well answered by 29 1Wd4.

28 §eville Variation 29 tl:lxd3 'lra4 30 'lrf3! 'lraS 31 e6 tl:ld8 32 tl:lf4 .i.eS 33 tl:ld5! 'IreS+

If 33 ... 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 + 14 �xfl 'ird6

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 �g1 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 . . . 'ifh4! 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..h6 comes too late : 1 9 �d2 l:l. f2 + 20 �d3 c4+ 21 .;;.xc4 J..xe3 22 'ii'd5+ .;;.g7 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..xh6 l:l. xe2 + 23 �xe2 'ifh3! 24 J..e3! 'ifxg4 25 l:l.fl ! h6! (a very strong move according to Henkin, who gives the following variations : 25 . . . cd 26 J..h6! 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..xh6 'ifxe4 27 l:l. f8 + �h7 28 .!.g5 'ife6 29 a4 cd 30 cd 'ifa2+ 3 1 .;;.el 'ifxa4 32 lH7 + .;;.g8 33 .J:I. xe7 'ii'b4+ 34 �f2 a5 35 J..f6 'ii'b6! 36 J..e5 a4 37 �f4 'ii'b2+ (Black loses after 37 . . . �f8 38 J..g7+ !, or 37 . . . a3 38 �e6!) 38 �g1 ! 'ifc1 + 39 �h2 'ii'b2+ 40 �g1 'ii'b 1 + 41 .;;.h2 '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'd₃

'ife6 1 7 g5!. However, White replied 16 'ifa4 'ifa6 1 7 .J:I.e 1 'ii'd3 1 8 J..f2 �c4 1 9 'ii'b5 l:l.f8 20 'ii'b l 'ii'd2 2 1 'ii' b3 b 5 2 2 'ii'xb5 �e3 23 h3, and now, in Gutman's view, 23 . . . J..h6! 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 llld5 .l: f8 23 i.g3 lllb6 (but not 23 . . . lllb2 24 'flc3 a3 25 h3, followed by i.e5, with advantage) 24 h4 (24 lllxb6 '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 llld6!) 22 . . . i.d2 23 a4 llld6!. 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.

17 lllxc4

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: e8 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 llla5 (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 lllc6 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 lllc1, with advantage) 25 lllxd4 lllxd4 26 .l: xb7 lllc6

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 <i>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.g7 24 :ldS a6 2S �f2 :le8 26 <i>f3 <i>f7 27 h4 �eS+ 28 .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 <i>f2 aS 22 thc 1 h6 (22 . . . �a3 23 :l b3 b4 24 cb cb 25 d6) 23 .i.e3 �xe3 (in Lputian's view, 23 . . . tLld6 24 <i>f3 :lf8 + 25 <i>e2 c4 26 <i>e1 �xe4 27 :lxbS �xc3 28 :lxaS :ld8 would have given equality; but not 23 . . . :lf8 + 24 <i>e2 �xe3 2S �xe3 :lfl 26 :la 1 ! and 27 �b3!} 24 <i>xe3 c4 2S �e2 .i.f8 26 tZ'lg1 .i.cS + 27 <i>e2 .i.xg 1 28 :lxg 1 <i>f7 29 a3 rj;e7 (29 . . . b4 would have retained some saving chances} 30 :l b 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 <i>fl! :l b4 (28 . . . b2 29 <i>e2) 28 .i.c5 :lxa4 29 .i.xd6 .i.xd6 30 :lxb3, and White has the better chances.

18 cd

Black can play 1 8 . . . eS at once, without opening the c-file. After 19 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!c1 !, heading for b3. Gutman quotes the following analysis by H. Wirthensohn : 23 . . . aS (23 . . . .t..xc1 24 'fJ.xc1 lt!d6 25 'fJ.e 1 a6 26 .t..g3 'fJ.e8 27 <;Pf2) 24 lt!b3 .t..b4 (24 . . . .t..c3 25 d6 a4 26 d7 rJ.d8 27 lt!c5 lt!b6 28 rJ.d6) 25 .t..c5 .t..xc5 26 lt!xc5 'fJ.c8 (26 . . . lt!d6 27 lt!d7 lt!xe4 28 lt!xe5 lt!d6 29 'fJ.c 1 ) 2 7 lt!b7 'fJ.c7 2 8 d 6 'fJ.d7 29 'fJ.d5 lt!b6 3 0 lt!c5, and White i s on top. 23 'fJ.c1 b5 Here 23 . . . lt!d6 is no good : 24 lt!g3 .t..b6 25 J.xb6 ab 26 'fJ.c7 'fJ.a4 27 'fJ.e7 lt!xe4 28 lt!xe4 'fJ.xe4 29 d6 �f8 30 'fJ.xh7 'iPe8 3 1 h5, and Black is doomed.

24 'fJ.c2 ltld6 25 ltlg3 ltlc4 26 ltlf1 ltld6 27 ltlg3 ltlc4 28 g5 �/7 White has consolidated his advantage. If 28 . . . a6 (28 . . . .t..b6 29 a4 lt!a3 30 'fJ.b2 ba 3 1 'fJ.a2), there follows 29 lt!fl lt!d6 30 'fJ.c6 lt!xe4 3 1 lt!g3! lt!c3 32 d6! �f7 33 d7 'fJ.d8 34 .t..c5 'fJ.xd7 35 'fJ.f6+ �g8 36 h5 gh 37 lt!xh5 'fJ.d8 38 'fJ.xa6 .t..c7 39 lt!f6+, and Black is in a bad way.

29 ltl/1 ltld6 30 ltlg3 ltlc4 31 �/1 �e7

Korchnoi suggested 31 . . . .t..b6 32 .t..xb6 ab 33 'iPe2 'fJ.a4, but a stratagem we have seen before - 32 a4! lt!a3 33 'fJ. b2 ba 34

'fJ.a2 - gives White the advantage. 32 .t..c5+ <;Pf7 (14)

33 J:.f2+ 14 w

Seville Variation 33

A critical moment in the game. After the obvious 33 �e2, Black would scarcely be able to save himself: 33 . . . i..b6 34 a4 i..xc5 35 ab J:. b8 36 �d3! lhb5 (36 . . . a6 37 �xc4 ab + 38 �b3 i..d6 39 J:. c6 �e7 40 �e2 J:. a8 41 �c3 J:. a3 + 42 �b2, or 40 ... b4 4 1 �c1, with a won position) 37 �xc4 J:.b7 38 lL!fl i..d6 (38 . . . a5 3 9 �d2, o r 38 . . . i..b4 3 9 �e3) 39 �d2, etc.

33 . . . �gl 34 J:.f6 i..b6 35 J:.c6??

By retreating his bishop to f2, White would preserve the better chances. Instead of this, I commit a fatal blunder. The sad fact is that the possibility of the black knight jumping to the edge of the board has escaped my attention.

35 . . . �a5 36 i..xb6 �xc6 37 i..cl J:.f8+ 38 �e2 J:.fl 39 i..d6 J:.dl 40 i..c5 �a5 41 �!1 .:tel 42 i..d6 J:.c2+ 43 �d3 J:.xa2 44 �e3 �fl 45 �g4 �c4 46 �xe5+ �xe5 47 i..xe5 b4 48 i..f6 b3 49 e5 J:.xg2 50 e6 + �f8! 0-1.

Now at last we return to the game Naumkin-Neverov, which, as it happens, was decided very quickly.

21 �g3 i..h6 22 i..cS

White proceeds ineffectively and is soon in a difficult position. The immediate 22 h4 was better.

22 23 h4 24 i..f2 25 �e2 26 J:.dl �c4 J:. c8! i..f4 i..d2 aS!

White's loss of time in moving his bishop about begins to tell: Black creates a passed pawn, which decides the issue.

34 Seville Variation

27 gS bS 28 ltlg3 b4 29 lllfl J.c3 30 lllh2 a4 31 d6 b3 32 ab ab 33 ltlg4 b2 34 d7 .IUS 35 :n <t;g7 36 ltlf6 llld2 37 ltle8+ <t;f7 o-1

So 1 2 J.xf7 + doesn't guarantee victory. One or two inaccur­ acies, and Black may assume a decisive initiative.

Game No. 5

Seirawan-Popovic

Manila 1990

1 d4 ltlf6 2 c4 g6 3 ltlc3 dS 4 cd lllxdS 5 e4 lllxc3 6 be J.g7 7 J.c4 cS 8 llle2 lllc6 9 J.e3 o-o 10 0-0 J.g4 1 1 f3 ltlaS 12 J.xf7 + ll xf7 13 fg ll xfl+

15 B

14 �xfl cd 15 cd (15)

Curiously enough, this precise position arose as long ago as 1 955, in the game Spassky-Korchnoi from the 22nd USSR Championship. Play continued : 15 ... 1i'd7 16 h3 1i'e6 17 1i'd3 1i'c4 18 1i'd2 (or 1 8 1i'xc4 lll xc4 19 J.g5 e6 20 ll d 1 b5, and despite White's extra pawn and the queen exchange, the initiative is with Black; Panteleyev-Prakhov, Bulgaria 1 970. An improvement was 20 .l:l bl !) 1 8 . . . 1i'a6! 19 1i'c2 ltlc4 20 1rb3 <t;h8 21 <t;g 1 ltld2 (21 . . . .l:l f8 is stronger) 22 J.xd2 1i'xe2 23 .te3, with equal chances. If Black wants, he can force a draw with 23 . . . .l:l c8 24 .l:lfl .l:l c2 25 ll f2 1i'el+ 26 ll fl 1i'e2.

The immediate pawn exchange on d4 at move 14 occurred only once in the Seville match - in the ninth game. Let us take quite a close look at it (recently some new thoughts about it have come to mind!):

Seville Variation 35 15 . . . flb6 16 �g1

With the queen on b6, 1 6 h3 can be strongly answered by 1 6 . . . �c4 1 7 ..t.f2 flb2! 1 8 �gl l:t f8 1 9 l:t b l flxa2 20 l:t xb7 .th6! with the initiative.

16 ... fle6

With the white king on g 1 , the line just mentioned no longer gives Black anything : 16 . . . �c4 1 7 .tf2 flb2 1 8 �f4! l:t f8 1 9 �d3.

17 fld3!

White returns the pawn while retaining all the positional trumps. The position had first been seen (it arose by transposition) in Alfeyevsky-Werner, corr. 1 984. That game went 17 �g3 l:t d8 1 8 l:t c l ( 1 8 l:t b l .txd4 1 9 .txd4 �c6 20 �e2 �xd4 2 1 �xd4 flc4) 1 8 . . . flb6 19 �e2 �c6 20 l:t b l �xd4! 2 1 l:t xb6 �f3 + 22 �f2 l:t xd l 23 l:t xb7 �xh2 24 g4 �g4+ 25 �f3 �h2+, with a peaceful conclusion by perpetual check.

1 7 ... flxg4

After this, I succeed in obtaining a substantial plus. After 1 7 . . . l:t d8 1 8 g5 �c4 1 9 .tf2 b5, the play would transpose into Chernin-Malishauskas which we have examined before (see page 3 1 ). (I should point out that the ideas and variations in Games 1 -5 have much in common, while sometimes being distinguished by small nuances; I hope the reader will not get tangled up in the 'undergrowth'.) Instead of the capture on g4, a better idea is 1 7 . . . flc4, aiming to exchange o ff the strongest pieces. This idea was employed in game 1 1 of the Seville match, with the difference that the exchange on d4 took place later (see notes to Game No. 4). White may avoid the queen exchange by retreating with 1 8 fld2 (exploiting the absence of the pawn from c3 and thus gaining a tempo). After 1 8 . . . fla6 19 flc2 �c4 ( 1 9 . . . flc4 is well answered by 20 :t e l ! - emphasising once again that the removal of the c­ pawns is favourable to White) 20 .tf2 .tf8 (20 . . . .th6 21 h4) 21 g5, and Black's position is devoid of counterplay.

18 l:t/1 l:tc8

The alternative 1 8 . . . l:t f8 is worse : 1 9 :t xf8 + .txf8 20 d5 b6 2 1 �d4!

19 h3! fld7 20 d5 �c4 21 ..t.d4

More precise than 2 1 �d4 �xe3 22 flxe3 l:t c4 23 �e6 flxe6 24 de .td4 25 flxd4 l:t xd4 26 l:t f7 l:t xe4 27 l:t xe7 b5, with equality; but 21 .txa7 b6 22 ..t.b8!? was worth considering.

36 Seville Variation 21 ... e5

If 2 1 . . . lbe5, then 22 ..txe5 .txe5 23 lbd4 .txd4+ 24 -.xd4 b6 25 e5 -.c7 26 -.g4! is not bad for White.

22 de -.xe6 23 .txg7 r;J;xg7 24 lb/4 -.d6 25 -.c3+ (16)

25 ... �h6 16

B

The only move. There would be a striking finish after 25 . . . -.es 26 lbe6 + ! �g8 (26 . . . r;J;h6 27 'ifc1 + g5 28 l:l.f5) 27 -.xc4 !

l:l. xc4 28 l:l. f8 mate. Another line that fails is 25 . . . r;J;g8 26 lbd5 lbb6 27 lbf6+ �h8 28 -.b2 -.c5 + 29 �h2 -.c3 30 lbe8! lbd7 (30 . . . �g8 31 -.bs l:l.c5 32 lbf6+!) 31 -.xc3 l:l.xc3 32 l:l.f7 l:l. d3 .(32 . . . l:l. c8 33 lbd6) 33 e5 r;J;g8 34 e6, and White wins.

26 lbd5?

Black's king is in quite a dangerous position, and White should have tried to approach it from the other direction with 26 lbd3! The threat of lbd3-f2-g4+ is then fairly serious (26 . . . lbb6 27 'ifd2+ �g7 28 'ifb2+ r;J;g8 29 -.b3 + �h8 30 l:l.f7, with a winning position).

26 . . . -.e5 27 Vd3

White could have maintained the tension with 27 'ife l . Many annotators recommended 27 -.b4 instead, but this is not dangerous for Black : 27 . . . 'ifxe4! 28 lbf6 aS! 29 -.bs (29 'ifc3 -.e3 +, or 29 'if a4 'ifd4+ 30 r;J;h 1 l:l.f8 3 1 lbg4+ �g7, with advantage to Black) 29 . . . 'ifd4+ 30 .l:tf2 (in the event of 30 �h1 lbe3 3 1 'if xb7 lbxfl 32 lbg4+, the black king fearlessly marches forward : 32 . . . r;J;g5 33 -.e7 + r;J;f4 34 -.f7 + r;J;g3 35 -.xn h5 36 -.f3 + �h4 37 g3 + �xh3 38 lbf2 + -.xf2! 39 -.xf2 l:l. c l + , and wins) 30 . . . -.at+ 3 1 l:l. fl -.d4+ 32 l:l.f2 -.a l +, and the contest could end in perpetual check.

Seville Variation 3 7 27 . . . �g7 28 !i:Jf6 11d6 29 11 c3 11e5 3 0 11d3 11d6 31 11c3 11e5 32 1ib3 l:l. c7 33 11 d3 l:l./7 34 11 xc4 l:l. x/6 35 l:l. dl b5

The position is drawn, but Black is playing somewhat recklessly. A simpler method is 35 . . . l:l. e6 36 l:t d7 + l:l. e7 37 l:l. xe7 + 11xe7 38 11d4+ �f7 39 11xa7 'lrxe4.

36 l:l. d7+ �h6 37 11e2 11 c5 + 38 �h2 11e5+ 39 g3 11 c3 40 �g2 11 c4 41 11e3 + g5 42 l:l. d2 11/1 + 43 �h2 11/3

The sealed move. Once again, just as happened in game 7 of the match, I persistently sought winning chances after resumption, and reached an endgame a pawn up - but alas, a drawn result was unavoidable. We will follow this interesting game to the end. 44 11d4 l:l. e6 45 e5 11/5 46 l:l. e2 a5 47 11d5 b4 48 11xa5 11d3 49 l:l.g2 11d4 50 11a8 11xe5 51 11/8+ �g6 52 11xb4 h5 53 h4 gh 54 11xh4 l:l. d6 55 11 c4 l:l. d4 56 11 c6+ �g7 5 7 11b7+ �h6 58 11 c6+ �g7 59 l:l. c2 l:l. h4+ 60 �g2 11e4+ 61 11xe4 l:l. xe4 62 l:l. c7+ �g6 63 l:l. a7 l:l. e3 64 �h3 l:tc3 65 l:ta8 l:l. c4 66 a4 �g5 67 a5

l:l. a4 68 a6 <i;h6 69 �g2 l:l. a3 70 �/2 �g7 t- t .

15 eS!?

This move had first occurred in the game Kir. Georgiev­ Ivanchuk, Reggio Emilia 1 989/90.

16 dS

The natural reaction - White creates a passed pawn. Despite the pawn extra, the endgame arising from 16 de i.xe5 1 7 11 xd8+ l:l. xd8 holds no danger for Black. Seirawan-H. Olafsson, Reykjavik 1 990, continued 1 8 l:l. c l !i:Jc6 19 g3 l:l. d3 20 i.f4 i.d4 21 !i:Jxd4 l:l. xd4 22 l:l. b l l:l. b4 23 l:l. xb4 !i:Jxb4 24 a4 !i:Jc6 25 �e2 a6 26 i.d2 b5 27 ab ab 28 i.c3 rj;f7 29 <i;e3 �e6. Subsequently Black tried to make something of his passed b-pawn, but without success, and a draw was agreed after another 25 moves.

Nor does White do any better with 16 l:l. cl 11d7 17 de 11xd 1 + 1 8 : xd 1 !i:Jc4 1 9 i.f2 i.xe5 20 l:l. d7 b 6 2 1 i.d4 i.xd4 22 !i:Jxd4 !i:Jd2+ 23 <i;e2 !i:Jxe4 24 !i:Jc6 a5 25 l:t b7 !i:Jc3 + 26 <i;d2 !i:Jxa2 27 : xb6 !i:Jb4, and the game is level; Schiissler-Kudrin, Saint­ Martin 1 990.

16 !i:Jc4

17 i.f2 11f6 1 s �g1 : rs 19 11e1

The position is unclear after 19 i.xa7 b6, or 19 i.c5 l:t c8 20

38 Seville Variation

19 R.h6

Threatening . . . R.d2.

20 lllg3 W'a6

20 . . . W'xf2+ would be over-hasty : 2 1 W'xf2 R.e3 22 W'xe3, followed by r:t cl.

21 <lo>h1

The threat was . . . r:t xf2, for example : 21 W'e2 r:t xf2! 22 <lo>xf2 R.e3 +; while if 2 1 lllfl, then 2 1 . . . lllb2! is decisive.

21 W'a4

2 1 . . . W'a3 is also worth trying.

1 7 B

22 R.gl! (1 7)

The white queen's problems can be solved later. In Georgiev-lvanchuk, White played 22 "ife2, and there followed : 22 . . . b6 23 h4 R.f4 24 lllfl (24 g5 llld6 25 r:t e 1 was more precise) 24 . . . llld6 25 l:l e 1 l:l c8 26 g3 r:t c2 27 W'f3 W'xa2 28 <lo>g1 R.h6 29 g5 R.g7 30 llle3 r:t c7 3 1 lllg4 r:t f7 32 W'e3 W'c2 33 h5 lllc4 34 W'c1 W'xc1 35 r:t xcl gh 36 r:t xc4 hg 37 r:t c8 + R.f8 38 R.e l ! (lvanchuk gives 38 <lo>fl ! b5 39 <lo>e2 b4 40 R.c5 a5 41 r:t a8 b3 42 <lo>d2, with a great deal of play left; or 38 . . . <lo>g7 39 <lo>e2 b5 40 r:t a8) 38 . . . <lo>g7 39 R.c3 R.d6! 40 r:t c6 R.c5+ 41 <lo>g2 r:t f2+ 42 <lo>h1 R.d4 43 R.b4 r:t f7! 44 r:t e6 r:t b7! 45 r:t c6 a5 46 d6? (lvanchuk considers this the decisive mistake, although even after the more stubborn 46 R.a3 b5 47 d6 r:t f7! 48 r:t c7 R.b6 49 .D. c6 R.d8 50 R.b2 r:t d7 5 1 R.xe5 + <lo>f7, Black has a clear plus; not however 47 . . . R.b6?? 48 r:t xb6!) 46 . . . ab! 47 r:t c7+ <lo>f8 48 r:t xb7 b3 49 r:t b8 + <lo>f7 50 d7 b2 5 1 r:t f8 + <lo>e6 52 d8lll+ <lo>e7 53 r:t fl �xd8, and White soon resigned.

Seville Variation 39

One other game is worth mentioning. In Naumkin-Miralles, Voskresensk 1 990, Black answered 22 11re2 with 22 • • • .i.g5, and only after 23 lOft did he play 23 . . . b6. (He could have brought about simplifications with the bold 23 . . . l0d6! 24 .i.c5 l0xe4! 25 .i.xf8 11rd4 26 .i.c5! 11rxa 1 27 11r xe4 11rxfl + 30 .i.g1 .i.d8, threatening . . . .i.b6.) Naumkin now seized the initiative: 24 h4 .i.e7 25 g5 .i.c5 (25 . . . l0d6 26 l0d2) 26 .i.g3 l0d6 27 l0d2 .i.d4 28 J: c 1 ! 1ixa2 29 J: c6! J: d8? (it was better to take the central pawn : 29 . . . l0xe4 30 11rxe4 11rxd2 3 1 J: xg6+ hg 32 11rxg6+ <i>h8 33 11rh6+, with a draw; sharp play would result from 3 1 <i>h2 11re3 32 11rg4!?) 30 <i>h2! aS (Naumkin gives these variations : 30 . . . 11r a 1 3 1 l0f3, or 30 . . . l0xe4 3 1 11rxe4 11rxd2 32 J: xg6+ hg 33 11rxg6+ ci>f8 34 11rf6+ ci>e8 35 11re6+ ci>f8 36 g6 .i.g 1 + 37 �xg1 11rxd5 38 11rf6+, with a substantial advantage to White) 3 1 h 5 11ra1 32 l0f3 gh (nor can he save himself with 32 . . . 11rb2 33 l0xd4 11rxd4 34 11rg4 11rxe4 35 11re6+ .!On 36 J: c8!) 33 l0xd4 11rxd4 34 11rxh5 l0e8 35 g6 1 -0.

22 b6

Before playing the knight on d6, it is essential to cover the c5-square.

23 11rc3 J: f7

23 . . . l0d6 is premature because of 24 11rxe5 .i.g7 25 11re6+.

2 4 J: bl .i.d2

24 . . . 11rxa2 is bad on account of 25 .1: b4, and the pawn on e5 falls. Seirawan recommends 24 . . . l0d2 25 J: b4 11r d 1 26 11rxe5 .i.e3 27 11re8 + with a draw, but instead White can maintain the pressure with 25 .1: e 1 .

25 11rd3

After 25 11rb3 11rxb3 26 ab l0d6, White's extra pawn is of no significance.

25 11rxa2

25 . . . .i.h6 (25 . . . .i.f4 26 l0e2 and l0c3) is answered by 26 a3, and the a-pawn is invulnerable : 26 . . . l0xa3 27 : a t .

26 d6! l0xd6

The only move. On 26 . . . J:t d7, Seirawan gives 27 11rd5 + ci>g7 28 J:t fl ! (28 11re6 J:t xd6 29 11re7+ ci>g8 30 J:t fl .i.f4) 28 . . . .i.f4 29 J:t xf4! ef 30 .i.d4+ <i>h6 3 1 g5 mate. (Editor 's note - it appears that Black can escape with 30 . . . �f8. Therefore White should prefer Stohl's 30 lt:lf5 + ! ! gf (30 . . . ci>f8 3 1 .!Oh6 �g7 32 g5!) 3 1 i.d4 + ci>f8 3 2 11rxf5 + ci>g8 3 3 11r g5 + , forcing mate).

40 Seville Variation

27 'irxd6 'irxb1 28 'irxd2 aS

28 ... 'irb5 29 h3 'ird7 30 'irc3 'ire? 3 1 'irb3 is more tenacious, although White's chances are still better.

29 'ird8+ �g7

30 'irg5 h6 White threatened the deadly 3 1 lDf5 + .

3 1 'irxe5 + �h7 32 h3! J: d7

White was intending to bring his bishop into play after 33 �h2. But now the denouement comes instantly.

33 lDh5!

1-0

2

Exchange Variation with

7

Jtc4 - other systems

12 .i.d3

12 .i.xf7 + attained wide popularity after the World Champion­ ship match in Seville, but capturing on f7 is, of course, not obligatory - the bishop may retreat instead. It is most securely placed on d3, although occasionally 1 2 .i.d5 is also seen.

12 cd

13 cd .i.e6 14 .:t el (18)

The once fashionable Sokolsky attack, 14 d5 .i.xa 1 1 5 "tl'xa1 f6, has now practically fallen into disuse. Black is the exchange up and can extricate himself without too much difficulty.

18 B

14 .i.xa2

15 "tlra4

42 Exchange Variation with 7 J.c4 - other systems

15 J.e6

16 d5 J.d7 17 'lfb4 (19)

An alternative is 1 7 'lfa3, whereupon, apart from the more conventional 1 7 . . . e6, Black has 17 • • • bS!?. After 18 :. fd1, the reply 18 . . . J:.b8 is rather slow, for example : 1 9 'lf b4 a6 20 l0d4 J:. e8 2 1 J.e2 J.e5 22 J:. a l ! J.c7 23 l0e6! J.d6 24 'lfxd6! fe 25 'lfe5 1 -0; Piskov-Lputian, Belgrade 1 988. In Ilic-Fercec, Kladovo 1 989, Black introduced the important innovation 18 . . . b4!, giving the pawn back but obtaining adequate counterplay : 19 'lfxb4 :. b8 20 'lfel l0b3 21 J:. b l a5 22 .tc2, and now 22 . . . a4 is logical, with a complicated game.

19 B

In this diagram we have a standard position in contemporary practice. Black can now choose between 1 7 . . . e6 (the game continuation) and 1 7 . . . b6. Before going any further, let us examine the latter.

1 7 . . . b6

Safer than 1 7 . . . b5, for example : 18 : fd 1 J.e5 19 J.c5 �b7 20 J.xe7 'lfb6+ 21 �h1 aS 22 'lfd2 J:. fc8 23 f4 J.g7? (the bishop should have been placed on d6) 24 e5 b4 25 J.c4 b3 26 J.a3 ! with a won position for White; Balashov-Sibarevic, Lugano 1988.

18 f4

Or 1 8 .ta6 J.c8 1 9 J.b5 a6 20 J.d3 b5 2 1 f4 e5! 22 J.c5 J:. e8 23 J.d6 'lfb6+ 24 J.c5 (24 �h1 J.g4! 25 J.c7 'ife3) 24 . . . 'lfd8 25 J.d6, with a draw; Savchenko-Lputian, Tallinn 1988.

18 . . . e5

Stronger than 18 . . . e6 ( 1 8 . . . J:. c8 1 9 l0c3 l0b7 20 J.a6 l:. c7 2 1 e5 l0c5 22 J.c4 'lfb8 23 'lf a3! J:. cc8 24 J:. fd 1 J:. fd8 25 .ixc5!

Exchange Variation with 7 J.c4 - other systems 43 and Black is in serious trouble; Vyzhmanavin-Ivanchuk, Tashkent 1 987) 19 d6 eS 20 f5 : c8 2 1 .!Llc3 J.c6 22 .!Llb5 1Wd7 23 f6! with a clear plus; Balashov-Hansen, Malmo 1 987/8. However, after 19

. . . .!Llc6 20 1Wb3 e5! 21 f5 : c8 22 .!Llc3 .!Lld4 23 J.xd4 ed 24 .!Lld5 : xc 1 25 .!Lle7 + �h8 26 : xc 1 gf 27 ef J.f6, Black has everything in order; Dolmatov-Gavrikov, Kiev 1 986.

19 :j2

In Vaiser-Gavrikov, Tallinn 1 988, Black seized the initiative after 19 fS?! :e8 20 J.a6? J.f8 2 1 1fc3 b5. At move 20, White should have played 20 d6 J.f8 2 1 J.c4 .!Llxc4 22 1fxc4 1Wf6 23 fg (23 .!Llc3? : ec8 24 gf : xc4! 25 gh + �xh7 26 : xf6 J.e6 27 .!Llb5 : xe4, with advantage to Black; Heinig-Gauglitz, Berlin 1 988) 23 . . . 1fxg6 24 : f3 1fe6, with complex play (Gauglitz).

19 ... : c8 20 : cf1 1Wc7

It is hard to give a preference to either side.

17 e6

This position occurrred as far back as 1 954, in the game Geller­ Liliental from the 2 1 st USSR Championship. Play continued 18

de J.xe6 1 9 : fd l b6 20 J.a6 1Wh4 21 .!Lld4, and now by means of 2 1 . . . J.c8 22 J.b5 J.e5 23 g3 1Wf6 24 f4 J.d6 25 1f a4 J.g4 Black could have kept his extra pawn with a sound position.

Nor does 18 d6 give White anything, for example : 1 8 . . . .!Llc6 1 9 1fxb7 : b8 20 1fc7 : b3! 21 : fd 1 .!Lle5 with the initiative; Razuvayev-Lputian, Sochi 1 987.

In the same Soviet Championship as the game we are annotating ( 1 988), Belyavsky-Kasparov continued as follows : 18 .!Llc3 ed ( 1 8 . . . b6 is also interesting: 1 9 f4 ed 20 .!Llxd5 J.e6 2 1 : fd 1 J.xd5 22 J.b5 1ff6 23 : xd5 : ac8 24 : xc8 : xc8 25 e5 1fe6 26 1We4 J.f8 27 J.d7 : c4 28 1Wd3 1fe7 29 e6, with approximate equality; Naumkin-Krasenkov, Vilnius 1988) 19 ed (Black similarly has a good game after 1 9 .!Llxd5 J.e6 20 : fd l J.xd5 etc.) 19 • • • : e8 ( 1 9 . . . b6 is also playable : 20 .!Lle4 : e8 2 1 J.d4 .!Llb3 22 J.xg7 .!Llxc l 23 1Wd4 : xe4! 24 fe .!Llxd3 25 J.h6 f6, with a good game; Utemov-Obodchuk, USSR 1 988) 20 J.f2 J.f8 2 1 1Wb2 (after 2 1 1Wf4 g5! 2 2 Wg3 .!Llb3 2 3 : b1 .!Llc5 24 J.c2 f5! Black would seize the initiative, so White is not justified in playing for a win here) 2 1 . . . J.g7 22 1W b4 J.f8 23 1Wb2 t-t.

The next two examples are identical with Belyavsky-Kasparov up to move 20.

44 Exchange Variation with 7 J..c4 - other systems

. • . J..e5 2 1 .!tJe4 .tfS 22 .tbS .C. f8 23 J.c5 b6! (not 23 . . . .txe4? 24 fe 11Vh4 2S g3 .txg3 26 .C. c2! and White has an obvious plus) 24 .txf8 11Vxf8 2S 11Vxf8 + �xf8 26 .C. fe 1 ! (after 26 .C. fd 1 ? the exchange on e4 is good for Black : 26 . . . .txe4 27 fe �e7) 26 . . . .!tJb3 27 .C. cd l .!tJd4 28 .ta6 .!tJc2 29 .C. e2 .!tJd4 30 .C. ee l .!tJc2 3 1 .C. e2 .!tJd4 _t-t.

One other game by Yusupov likewise ended in a quick draw :

20 . . . b5! 2 1 .C. fd l (or 21 .!tJe4 .!tJb7 22 .txbS aS 23 11Va4 .txbS 24 11VxbS .!tJd6, with equality; Naumkin-Mokry, Nemestovo 1 987) 21 . . . .!tJc4 22 J..xc4 aS 23 11Vb3 be 24 11Vxc4 .C. c8 2S 11Vd3 !-t; Yusupov-Smejkal, Munich 1 988.

18 .C. fd1

It seems to me that a very strong move here is 18 .!tJd4, but it doesn't appear to have occurred in practice yet.

18 ed

19 ed .C. e8

20 .tf2 b5!

This looks like a serious weakening, but Black secures for his knight (via c4) the shortest route to the centre, where the main action is going to take place.

21 .!tJd4 .!tJc4

22 .!tJc6

By exchanging the troublesome knight on c4, White would restore the material balance but could hardly count on an advantage : 22 J.xc4 aS 23 11t'cS be 24 11Vxc4 a4. Yusupov endeavours to extract the maximum from the position.

20 w

22 .txc6