KCN's King's Indian Defence Analysis Center.

The King's Indian Defence arises after 1 d4 Nf6 [the introduction to the so-called "Indian" Defences, in which Black restricts White's central buildup with pieces, rather than with the symmetrical 1 ... d5.] 2 c4 [In Queenside openings, the c-pawn is used by White to undermine or forestall Black from gaining a solid central foothold with a later ... d7-d5.] g6 [The name "King's Indian" refers to Black fianchettoing his bishop on his Kingside.] 3 Nc3 Bg7 [Black allows White to get in the move e2-e4. Black's counterplay will come from either ... e7-e5 or ... c7-c5, striking at the central dark squares, supported by the strong bishop on g7. If White then responds with the advance d4-d5, an unbalanced pawn structure creates attacking chances on opposite sides of the board.] 4 e4 O-O [Now White has a tremendous number of choices. The most important are the Classical 5 Nf3, the Four Pawns Attack 5 f4, the Saemisch 5 f3, and the King's Fianchetto 5 g3.

A great feature of the King's Indian is that Black can play the above moves in response to almost any move order, such as 1 Nf3, 1 c4, 1 d4 followed by 2 Nf3, etc., that White chooses.

Here is an example of the Classical King's Indian, but with the offshoot 6 h3. For some time, GM Boris Gulko has been exploring h2-h3 systems, creating various hybrids of this move with the Averbach, Classical, or Bd3 systems. But when Gulko essayed the King's Indian against the creative IM John Watson in the 5th round of the 1993 World Open, Watson decided to give Gulko a taste of his own medicine, playing the very line that Gulko had used against Larry Remlinger the round before.  Gulko was impressed with Remlinger's treatment and imitated it against Watson, getting a good game, but a careless slip allowed Watson to create dangerous complications, sacrificing the Exchange for a strong initiative. It turned out that Watson has also been playing the h3 systems for many years, but he plays it as a prelude to a kingside attack.

Watson - Gulko, 1993 World Open1 d4 Nf6 2 c4 g6 3 Nc3 Bg7 [Lately, it seems everybody is wheeling out the King's Indian. Kamsky and Yermolinsky have recently made it mainstays of their repertoires.] 4 Nf3 O-O 5 e4 d6 6 h3 [A general principle of the h3 systems is that whenever White normally plays Be2, he also could play h3. The move h2-h3 keeps Black pieces off g4, prepares to expand if desired with g2-g4, and allows for later deployment of the White bishop on d3, since Black can no longer pressurize the center with ... Bg4. A well-known version of this occurs without the knight on f3, i.e. 4 e4 d6 5 h3 e5 6 d5 Nh5 7 Be2 Nf4 8 Bf3, etc. In the 1992 U.S. Championship, after 4 e4 d6 5 h3, Fedorowicz and Sherzer each played 5 ... 0-0 against Gulko, and Boris responded with 6 Bg5, analogous to the Averbach Variation, but with h3, not Be2.]  Na6  [In Gulko - Shirov, Groningen 1990, black tried the immediate 6 ... e5 7 d5 (possible is 7 dxe5 dxe5 8 Qxd8 Rxd8 9 Bg5, with an Exchange Variation where h3 has been substituted for Be2) a5, reminiscent of the Petrosian Variation, although having h3 in instead of Be2, White is more aggressively placed on the kingside; after 8 Bg5 Na6 9 Bd3 h6 10 Be3 Nh5 11 Qd2 Nc5 12 Bc2 Nf4 Gulko rejected the pawn sac with 13 Rh2! and stood well. In this line Black could have dispensed with 7 ... a5 and played 7 ... Na6.] 7 Bg5  DIAGRAM [Previously 7 Be3 was the main move here. Black was able to find active play in Grunberg - Dydysko, Bratislava 1992 after 7 ... e5 8 d5 Kh8 9 9 g4 Ng8 10 Bd3 Nb4. In the previous round of the World Open, Gulko had tried the text against Remlinger. One of the points is that 7 ... e5? is unplayable because of 8 dxe5 dxe5 9 Qxd8 Rxd8 10 Nd5. Also, Black should be hesitant to weaken his kingside with 7 ... h6 8 Be3, as White can play g2-g4 at any time.] Qe8! 8 g4?!  [By radically gaining kingside space, Watson prevents Gulko from activating with 8 ... e5 9 d5 Nh5. In Gulko - Remlinger, rd 4, White also forestalled this plan, by menas of 8 e5 dxe5 9 dxe5 Nd7 10 Bf4, but Black equalized easily: 10 ... Ndc5 11 Nd4 Ne6 12 Bh2 f6 13 Nf3 b6 14 Exf6 exf6 15 Be2 Bb7 16 O-O Rd8 17 Qa4 Qxa4 18 Nxa4 f5 20 Rad1 1/2 - 1/2]  c5! [Usually one does not associate this move with the queen's position on e8, but given White's kingside stance, it is best to keep the center fluid.]  9 d5 e6 10 Qe2 Nc7?  [Inviting the following complications. Much better was 10 ... exd5. Then after 11 cxd5, (not 11 exd5 h6, and only Black can obtain the advantage) Black stands well in the unbalanced pawn majority position.] 11 e5! dxe5 12 d6 e4 [Black would stand badly after 12 ... Na6 13 Nxe5. The text is justified by Black's fifteenth.] 13 Bxf6!  [White would lose his grip on the center after 13 dxc7 exf3 14 Qxf3 Nd7, and 13 Nxe4 Nxe4 14 Qxe4 Bxb2 is just bad.] exf3 14 Qxf3 Bxf6 15 Qxf6 Qc6 16 dxc7 Qxh1 17 Rd1  [White has sacrificed the exchange to stunt Black's development and weaken the dark squares around his king. The major question was whether to castle or just play 17 Rd1. Castling would have allowed potential queen-exchanging checks on the c1-h6 diagonal, so Watson correctly leaves his king in the center. Also, the king needs to defend the bishop in case the rook leaves the back rank.] Qc6  [Too slow is 17 ... b6 18 Rd8 Ba6 19 Qe7.] 18 h4! [Watson picks the most ambitious choice, trying to win in the middlegame. Also very scary is 18 Qe7 (threatening 19 Rd8). On 18 ... Qe8 19 Qd8 Black just palys 19 ... Qc6! with the idea of 20 ... b6. Also not convincing are 19 Qh4 Qc6, or 19 Qxc5 b6. Best is 19 Qd6! Qc6 (what else?) 20 Qxc6 bxc6 21 Ne4 and Black, still the Exchange up, is in bad shape.] Qxc7 19 h5 Qh2  [Forced, to stop 20 h6. 19 ... h6 would have gotten blown out by 20 g5!] 20 Rd3 b5  [Since he cannot eject the White queen from f6, Black faces horrible threats of h5-h6 after White cuts off the Black queen by putting a piece on h3. However, Black can survive the immediate blocking of the h-file, e.g. 21 Rh3 Qd6 22 h6 Qd4, or 21 Bh3 h6, allowing 22 hxg6, but at least 22 g5 is refuted by 22 ... Qg1+. Meanwhile, it is impossible to improve the Black position. If 20 ... e5 21 Nd5 wins. Therefore, Black attempts to create a diversion.] 21 Bh3? [It seemed that 20 ... b5 was just a throwaway to distract White from the main arena. However, 21 Nxb5! was the right move (whether the b-pawn was there or not). The knight on b5 controls d6 and d4; therefore White threatens 22 Rh3. If after 21 Nxb5! e5, preparing ... Qf4 (this would have been a sufficient defense to 21 Ne4), then 22 Nc7! with the following variations: a) 22 ... Rb8 23 Nd5 Re8 24 Rh3; b) 22 ... e4 23 Rh3 Qxc7 24 h6, or c) 22 ... Bxg4 23 Nd5! Rfe8 24 Rh3! Bxh3 25 h6. Is it possible that Gulko used reverse psychology when he played 20 ... b5, in order to get Watson not to hop to that square?]  h6 22 hxg6 fxg6 23 Qxg6+  [Of course White is doing fine here, but the queen has been pushed off f6, and the Black queen is just barely able to come back to the defense.] Kh8 24 Qxh6+ Kg8 25 Qg6+ Kh8 26 Ne4!  [If White did not need to protect f2, 26 Bf1 would decide matters. The text is better than settling for the perpetual.] Qe5  [Of course not 26 ... Bb7? 27 Rd7.] 27 Bg2 Qg7 28 Rh3+ Kg8 29 Nf6+ Rxf6 30 Qxg7+ Kxg7 31 Bxa8 bxc4 32 Be4 1/2 - 1/2 ?! [A generous draw offer, as White's pieces are well placed and the g-pawn is dangerous, but Watson was in serious time pressure and felt unable to switch into endgame mode. At the inaugural event of the Intel World Speed Chess Grand Prix in the Kremlin, Vladimir Kramnik gave new breadth to Garry Kasparov's prophecy that Kramnik would someday replace him as World Champion. In the quarter-final round of the Game/25 knockout tournament, Kramnik eliminated Kasparov, with the same variation that he used to defeat Kasparov in Linares, the Petrosian Variation of the King's Indian Defence. As a quarterback would, several times during the action, Kramnik spread his arms to keep the 4,000 fans down to a simmer while he set the attack in motion. Kasparov fought back hard from a lousy position by forcing Kramnik to sacrifice. But a careless move allowed Kramnik to drive home his idea, and Kasparov's counterplay was not enough. Kramnik - Kasparov, Intel World Speed Chess Grand Prix  Quarterfinals M/2 1 Nf3 Nf6 2 c4  [In contrast to 1 d4 Nf6 2 Nf3, the move order chosen by White is designed to avoid the Grunfeld Defence and the Queen's Indian Defence.] g6 3 Nc3 Bg7 4 e4 d6  [Lputian has experimented with 4 ... e5 here. If White had sidestepped this possibility by playing 4 d4, then 4 ... d5 is the Grunfeld, and White's move order tricks have not accomplished anything.] 5 d4 O-O 6 Be2 e5 7 d5  [The Petrosian System. White plays d4-d5 "unprovoked", but keeps the Black queen knight off the active square e7 after the alternative 7 O-O Nc6 8 d5 Ne7.] Nbd7  [The modern main line is 7 ... a5 8 Bg5 h6 9 Bh4 Na6 10 O-O Qe8 11 Nd2 Nh7 12 a3 Bd7, as in Illescas - Frolov, Biel 1993, where 13 Kh1! (Kramnik) was introduced; White prepares for the opening of the kingside by both players. Previously played was 13 b3 h5 14 f3 Bh6 15 Kh1 Be3 16 Rb1 f5 17 ef gf 18 f4 ef 19 Bh5 Qe5 20 Ne2 Nc5 21 Nf3 Qg7 22 g3 fg 23 Ng3 f4 24 Re1 Nf6 with great complications, Naumkin - D. Wood, Londond 1993. In Kramnik - Kamsky, Melody Amber 1994, Kamsky tried 7 ... Na6 8 Nd2 (a completely different concept is 8 Bg5 h6 9 Bh4 g5 10 Bg3 Nh5 11 h4 Ng3 12 fg gh 13 Nh4 Qg5 14 g4 Bf6 15 Qd2, Fishbein - Dlomatov Philadelphia 1993) Kh8?! (instead Aleksandrov - Dydysko, Minsk 1992 continued 8 ... Bh6 9 h4 (better is 9 Nb3) Nc5 10 Qc2 Bg4! and Black was doing well) 9 a3 c5 10 h4 h5 11 Nf3 Ng4 12 Ng5 Nh6 13 Be3 f5 14 exf5 Bxf5 15 Qd2 and White had a significant advantage.] 8 Be3! [The exclamation is for varying from the Linares game, which Kramnik won brilliantly with 8 Bg5 h6 9 Bh4 g5 10 Bg3 Nh5 11 h4 g4 12 Nh2 Ng3 13 fg h5 14 O-O. Then Kasparov innovated with 14 ... f5 (Seirawan gives 14 ... Bh6 15 Bd3 Nc5 16 Bc2 a5 17 Qe2 as slightly better for White). The game continued 15 ef Nc5 16 b4 e4 17 Rc1 Nd3 18 Bd3 ed 19 f6!! preventing the bishop from getting to f5 and White won in 40 moves.] Ng4 9 Bg5 f6 10 Bh4 h5 [This was also played in Speelman - Je. Piket, Tilburg 1992 except that Black had played 7 ... a5 instead of 7 ... Nbd7. Kasparov probably remembered that Piket obtained a good position. But with the queen knight on d7, the g4 knight will have to retreat anyway after White's next. Even so, the gain of space may compensate for the potential weaknesses.] 11 Nd2 Nh6  [Suddenly Black is threatening to win with 11 ... g5.] 12 f3 Nf7 13 Qc2 Bh6 14 O-O-O  [It was also possible to play less sharply by castling kingside, but Kramnik is banking on the long-term weakness created by Black's advanced pawns.] c5 15 dxc6  [Since White is better developed, he opens the center. Otherwise, Black could roll on the queenside with no worries.] bxc6 16 Kb1 a5?! [Another idea was 16 ... Nc5 17 Nb3 Qe7! accepting doubled pawns, but preparing ... Nf7-d8-e6.] 17 Na4 c5? [Kasparov does not relish the figth for c5 after 17 ... Rb8 18 c5! A decent way to hold up White's plans was 17 ... Ba6! 18 Rhe1 Qe7, as then 19 c5 could be met thematically with 19 ... d5, as there is no sting in 20 exd5 cxd5 21 c6 Rfc8. The text cedes White a permanent positional advantage.] 18 Nc3 Be3 19 Nd5 Bd4 20 Nb3 Bb7  [Black needs to be able to eliminate the pressure against f6. For this purpose, ... g6-g5 will never be satisfactory, as it will just lead to a bigger kingside expolsion later. However, all kinds of light-square problems are associated with the inevitable trade on d5.] 21 Nxd4 cxd4 22 f4  [Kramnik begins to pull apart Black's position using the various "levers".] Rb8 23 Rhf1  [Natural, but an important inaccuracy. 23 g4! opens the g-file making Black's game very unpleasant.] Nh6!! [Lending additional support to f6, but the move contains a bonus, as suddenly this knight becomes quite functional. It controls g4 and will hop to f5 if the e-pawn is diverted.] 24 c5?! [A spectacular move breaking down Black's chain. If White does not take quick action, Black will take his chances with the 2 knights after ... Bb7xd5 and ... Nd7-c5. 24 h3 was a reasonable possibility.] Bxd5 25 exd5 Nf5 26 fxe5  [White can no longer withdraw from this course.] Nxh4 27 exd6 Ne5 28 Rxd4 Nf5  [The 2 knights are policing the whole board, and this cannot be tolerated.] 29 Rxf5!? gxf5 30 Qxf5 Kg7 31 Bxh5  [Kasparov has led his opponent totally astray, and it was high time to activate with 31 ... Rb4, which would have also been good on any of the last several moves. He decides instead to make one more consolidating move ...] Rh8?? 32 Rg4+!! Kf8  [Mate in two follows capture of the rook.] 33 Qe6 Rb7  [It was impossible to move any of the other pieces.] 34 c6! Rxb2+  [On other moves, 35 c7 would be decisive.] 35 Kxb2 Qb6+ 36 Ka3 Qc5+ 37 Ka4  [The rook on g4 is a sentry allowing safe passage for the king.] Qc2+ 38 Kb5 Qb2+ 39 Ka6 Qe2+ 40 Kb7 Rh7+ 41 d7  [After 41 ... Qb5+ 42 Kc7 Rxd7+ 43 cxd7 Qc5+ 44 Kb7 Qb5+ 45 Ka7 there is nothing left to do.] 1 - 0 BAD BISHOP BLUES Bill Goichberg's Chicago International, held April 4 - 12, promised GM norm opportunities, and delivered, as I. Gurevich and I. Ivanov each made norms. In such tournaments one must do well in the early rounds to play sufficiently strong opposition, in order to have norm chances. When norm hopefuls Jay Bonin and Alex Sherzer met in the first round of the event, a classic King's Indian confrontation of competing flank attacks developed. Both players must have been happy with the results of the opening. Bonin, whose penchant for knights over bishops is legion, exchanged off his light-squared bishop and constructed a light-squared pawn chain, while Sherzer, a much-feared attacking player, established his kingside pawn roller and wasted no time trying to hammer it home.  As Bonin's pawn chain came under attack, he must have missed the "bad" bishop which was needed for defence. The critical question became whether his queenside counterplay could stop Sherzer from bringing up decisive reinforcements.  Bonin - Sherzer, Chicago 1992, Round 1 1 d4 Nf6 2 c4 g6 3 Nc3 Bg7 4 e4 d6 5 Nf3 O-O 6 Be2 e5 7 O-O Nc6 8 d5 Ne7 DIAGRAM 9 Bg5  [A mysterious move championed by Eingorn, and played occasionally by Ftacnik. The variations are similar to those arising from 9 b4, a system in which Bonin is a well-known expert, in that both moves allow Black to play 9 ... Nh5, whereas the main lines, 9 Ne1 and 9 Nd2, prevent that maneuver. But while 9 b4 helps prepare the thematic break c4-c5, the text does not, which is one reason W. Watson felt confident enough recently to award 9 Bg5 with a "?!".]  Nh5 10 Ne1  [The usual reaction here has been 10 g3, as in Eingorn - A. Kuzmin, USSR championship 1991, which continued 10 ... f6 11 Bd2 f5 12 exf5 Nxf5 13 Bd3 (Not 13 g4? Nd4) Nf6 14 Ng5 Nd4 with equality. In Bern - Watson, Gausdal 1991 Black played more actively: 10 g3 h6 11 Bd2 Bh3 12 Re1 f5 13 Nh4 Nf6 14 exf5 g5! (an improvement over the 14 ... gxf5 given in the new E Encyclopedia) 15 Ng6 Nxg6 16 fxg6 Bf5. The text effectively redeploys White's knight, at the cost of ceding the two bishops.] Nf4 11 Nd3 Nxe2+ 12 Qxe2 h6 13 Be3  [Deciding to engage in a battle of pawn chains and competing flank attacks. Probably better was 13 Bd2 with a view towards opening instead of closing the center. DIAGRAM Then 13 ... f5 can be met by 14 f4! which either saddles Black with a weak pawn on e5 or forces him to exchange pawns and mobilize White's pieces after 14 ... exf4 15 Nxf4 fxe4 16 Nxe4. If Black snatches a pawn with 16 ... Bxb2, 17 Bc3! gives White strong compensation. But a more subtle method for Black is, from the Diagram, 13 ... g5!! which stops f2-f4 first, and looks forward to the usual kingside attack after f7-f5-f4. White could stop this radically by 14 g4?! but Black surely stands better after 14 ... Ng6. Or if 14 h4?! g4! (Not 14 ... gxh4? 15 f4) 15 f4 gxf3 followed by 16 ... f5 is strong. White could play modestly with 14 f3 f5 15 exf5 Nxf5 16 Nf2 with even chances, or aggresively with the paradoxical 14 g3!?, envisionsing the wild 14 ... f5 15 f4, or 14 ... Bh3 15 Re1 with the bizarre variations 15 ... f5 16 f3 g4 (the threat was 17 Nf2) 17 f4, or 15 ... g4!.]  f5  [Now this move is stronger than 13 ... g5 14 f3 f5 15 exf5 Nxf5 16 Nf2.]  14 f3  [I still prefer 14 f4, although now Black has 14 ... fxe4 15 Nxe4 Bf5 16 Ng3 e4! 17 Nf2 Qd7, so that on 18 Nfxe4 Rae8 Black has plenty of activity for the pawn.]  f4 15 Bf2 g5  [The classic King's Indian pawn roller. Although the competing pawn breaks, on g4 and c5, are quite familiar, White misses his light-squared bishop, which from e2 would slow down Black's "lever" g5-g4.]  16 c5 Ng6 17 Rac1 g4!  [Sherzer takes advantage of tactical opportunities to get the pawns moving: if now 18 fxg4 Qg5 19 h3 h5 20 gxh5? Bxh3.]  18 Kh1 Rf7 19 Nb5  [The g-pawn was again defended indirectly: 19 fxg4 Qg5 20 h3 h5! 21 gxh5 f3! 22 gxf3 Qxh5 and wins. But White could have tried 19 fxg4! anyway. After 19 ... Qg5 the solid 20 Ne1! recycles the knight back as a defensive bulwark, and after 20 ... Bxg4 21 Nf3 White will get serious queenside counterplay.]  Bd7!  [By drawing the c-pawn up to c6 Black gives up space but retains the solidity of his pawn chain.]  20 c6 bxc6 21 dxc6 Be6 DIAGRAM 22 Nb4?  [Heading in the wrong direction. 22 fxg4! Qg5 23 Ne1 was still a tough nut to crack.]  g3! 23 Bg1  [On 23 hxg3 Qg5 is too strong, e.g. 24 g4 h5 25 gxh5 Qxh5+ 26 Kg1 Bf6.]  Nh4  [Starting a new phase - the bombardment of g2.]  24 Rc2 a6 25 Nc3 Bh3 26 Qc4 Qg5 27 Nbd5  [If White attempts to fish in troubled waters with 27 Nxa6, then 27 ... gxh2 28 gxh3 hxg1(q)+ 29 Rxg1 Qh5 wins, for example 30 Nxc7 Nxf3 31 Qe6 Nd4, or 30 Nd5 Kh8 followed by 31 ... Nxf3. The text move stops this variation - 27 Nbd5 gxh2? 28 gxh3 hxg1(q)+ 29 Rxg1 Qh5?? 30 Nf6+ - but it has the disadvantage of unpinning the Black rook on f7.]  Nxg2  [The patient 27 ... Kh8! may be even better, as 28 Ne3 Raf8 does not help White, and 28 hxg3 fxg3 29 Be3 Qh5 30 gxh3 Nxf3 wins.] 28 Rxg2 Bxg2+ 29 Kxg2 gxh2+ 30 Kxh2 Kh8 31 Qe2?  [31 Kh1! saves an important tempo, and the position is unclear after 31 ... Qh5+ 32 Bh2 Rg8 33 Qe2 Bf6 34 Nxf6! Rxf6 35 Rg1.] Qh5+ 32 Kg2 Rg8 33 Bh2 Bf6+ 34 Kh1 Bh4 35 Nb1  [It's too late for 35 Rg1 Rxg1+ 36 Kg1 Rg7+ 37 Kh1 Bg3, or 37 Kf1 Bd8.]  Rfg7 36 Nd2 Bd8  [Now ... Qh3 and ... Rg2 cannot be prevented.]  37 Nc4 Qh3 38 Nce3 fxe3 39 Nxe3 h5 40 b4 Bh4 41 a4 Bg3 42 b5 axb5 43 axb5 Bf4 0 - 1 HOMEWORK Two young stars from the Baltics, Lembit Oll of Estonia, and Alexey Shirov of Latvia, faced off in a second-round match in the Tilburg, Netherlands knockout tournament. Of course, Shirov has already broken into the top elite of the chess world, while Oll is known as a fierce theoretician and tactician. The first game of their match followed a line I had suggested in the notes to Bonin - Sherzer, Game of the Month, July 1992. Oll forces us to recharacterize a variation that had seemed to be antipositional as, in reality, providing a dangerous attack. Even though Shirov creates an impressive-looking pawn chain, he is on the precipice. This game reminds us that bizarre ideas have to be treated with respect, because they are probably the product of rigorous home preparation. But like a term paper that takes forever to finish, Oll has trouble delivering the winning sacrifice. Although he does get the sacrifice in just before the "deadline", i.e. the completion of Black's development, Shirov is a harsh grader. Lembit Oll - Alexey Shirov, Tilburg 1992, knockout round 2 [E96] 1 d4 Nf6 2 c4 g6 3 Nc3 Bg7 4 e4 d6 5 Nf3 O-O 6 Be2 e5 7 O-O Nc6 8 d5 Ne7 9 Bg5  [For background information on this variation see my column in the July 1992 issue of CL.]  Nh5 10 Ne1 Nf4 11 Nd3 Nxe2+ 12 Qxe2 h6 13 Bd2 [Deviating from Bonin - Sherzer, Chicago 1992, which continued 13 Be3 f5 14 f3 f4, with a good game for Black. The idea of 13 Bd2 is to meet 13 ... f5 with 14 f4!] g5 14 h4! [I had considered and rejected this continuation in my notes to Bonin - Sherzer: "13 ... g5!! stops f2-f4 first, and looks forward to the usual kingside attack after f7-f5-f4. If 14 h4?! g4! (Not 14 ... gxh4? 15 f4) 15 f4 gxf3 followed by 16 ... f5 is strong. White could play modestly with 14 f3 f5 15 exf5 Nxf5 16 Nf2 with even chances, or aggresively with the paradoxical 14 g3!?, envisionsing the wild 14 ... f5 15 f4, or 14 ... Bh3 15 Re1 with the bizarre variations 15 ... f5 16 f3 g4 (the threat was 17 Nf2) 17 f4, or 15 ... g4!". But Oll investigated the situation more diligently and discovered that it is White who has the attacking chances in the first line above.]  15 f4 gxf3 16 Qxf3 f5 17 Qh5  [Black has the two bishops and the generally smoother pawn structure, but White's temporary lead in development and immediate attack on h6 is annoying. If now 17 ... Kh7, then 18 Bg5! (threatening the pawn on f5) f4 19 g3! is serious: 19 ... fxg3 fails to 20 Rxf8 followed by 21 Rf1, and 19 ... Bh3 20 Rg3 Qd7 21 Nf2 is no help.]  Rf6 18 g4! [Giving Black no time to organize with ... Bd7 and ... Qe8, etc.] f4 19 g5 hxg5!!  [Safer seems 19 ... Rg6, avoiding the opening of the h-file. But Shirov was afraid of 20 Bxf4! exf4 21 Nxf4 Qe8 22 Nb5! Then the defense of 22 ... Rb8 (what else?) 23 Nxc7 Qd8 24 Nxg6 Nxg6 25 Qxg6 Qxc7 would not be available because of 26 gxh6. Thus Shirov eliminates the h-pawns first.] 20 hxg5 Rg6 21 Kf2  [White prepares combined play on the h- and f-files by getting his king out of the way first. If he plays as in the above note -21 Bxf4 exf4 22 Nxf4 Qe8 23 Nb5 Rb8 24 Nxc7 Qd8 25 Nxg6 Nxg6 26 Qxg6 Qxc7, then after 27 Qe8+ Kh7, 28 Rf4 is met by 28 ... Bg4!, and various other attempts to land a rook on the h-file may be frustrated by diagonal checks. And on 21 Bxf4 exf4 22 Nxf4 Qe8 23 Kg2, then 23 ... Be5 24 Rh1 Qf7 25 Raf1 Qg7 is a sufficient defense.] c6 22 Rh1 cxd5 23 cxd5  [Not 23 Nxd5 Kf7, and Black's cramped position is relieved.] Qb6+  [If I were Black here, I might have played for maximum confusion with 23 ... b5 aiming for ... b4 and ... Ba6. 24 b4 or 24 a3 could be met by the consistent 24 ... a5. One goal of this diversion might be to exchange a pair of rooks, to lessen the chance of getting mated on the kingside. But Shirov stays cool.] 24 Ke2 Bd7 25 Raf1 Kf7!  [The threat of Bd2xf4 has become intolerable, so he heads for the hills.]  26 Bxf4 exf4 27 Rxf4+ Ke8 28 Rhf1 Kd8  [The threat was 29 Qh8+! On 28 ... Bb5 29 Nxb5 Qxb5 30 Rf7 too strongly menaces both 31 Rxe7+ and 31 Rxg7.] 29 Rf8+ Be8!  [The rook is not going anywhere, so Black does not allow White the extra possibility of 29 ... Bxf8 30 Rxf8+ Be8 31 Qh8 Kd7 32 e5 and White, a rook down, has plenty of chances.]  30 e5 Rc8!  [The tempting 30 ... Bxf8 31 Rxf8 does not lead to clarity after 31 ... Qg1 32 Qh8 or 31 ... Kc7? 32 Qh7.]  31 Rxe8+ [Trying to draw the king back to the more dangerous side.] Kxe8 32 Ne4 dxe5 [The king's bishop has done such a good job that Shirov, in time pressure, was reluctant to part with it. But 32 ... Bxe5 was winning, e.g., 33 Nf6+ Kd8 34 Nxe5 (or 34 Qh8+ Rg8) Qxb2+, etc.] 33 d6  [Now, if Oll can win the knight on e7, he will be only the Exchange down, with active pieces.] Rc4?? [Correct was 33 ... Qd4, which sets up ... Rc2+, undermining the knight on d3, forcing 34 Qf3. Then White can fight on after 34 ... Rxd6 35 Nxd6+ Qxd6 36 Qf7+ Kd8 37 Rd1! But 34 ... Rc2+ 35 Kd1 Nf5! wins - 36 Kxc2 Ne3+ 37 Kc1 Nxf1, or 36 Re1 Rxb2.] 34 Nf6+ Bxf6 35 gxf6  [The new f-pawn has turned into a monster, while the knight on d3 holds Black's major pieces at bay, leading to a problem-like conclusion.] Re4+ 36 Kd1 Qe3! [The best way to counterattack, as White is forced into offering the exchange of queens.] 37 f7+ Kd7 38 Qf3 Qxf3+ 39 Rxf3 Rg1+ 40 Kd2 Rg2+ 41 Kd1 Ng6 42 Nc5+  [Amazingly, White finally is able to equalize material. In view of 42 ... Kxd6 43 Nxe4+ Ke7 44 Rc3 Kxf7 45 Rc7+, the players agreed to a draw.] 1/2 - 1/2

Tate - A. Ivanov chicago, 1996 1.c4 g6 2.Nc3 Bg7 3.d4 c5 4.d5 d6  Ivanov declines the invitation to play the controversial 4 ... Bxc3+ 5 bxc3 f5, subject of extensive analysis by Dzindzihashvili in one of his "Roman Forum" videotapes.  5.e4 e6 6.dxe6  A sharp treatment, looking to get in an early Bc1-f4 if 6 ... Bxe6.  6...Bxc3+ 7.bxc3 fxe6 8.h4!? Tate has no problem playing this type of move.  8...Nf6 9.h5 Nxh5 10.Nf3 Nc6?!  [No good is 10...Qf6 11.e5 dxe5 12.Bg5, but one decent defence is 10...e5! 11.Bg5 Qc7! 12.Be2 Bg4! - in this manner Black makes sure that White does not get to hit the h5 knight with his bishop on e2 - in this line 11 ... Nf6 is weak because of 12.Nxe5 Qe7 13.Nxg6 Qxe4+ 14.Qe2]  11.Bg5 Qc7 12.Be2 e5??  [Things were already getting wobbly, but 12...Qg7 to shore up the kingside and induce White to castle (because of the threat to c3) would have been interesting. Then a strange possibility is 13 O-O Kd7!? 14 Qd2 Kc7 15 Rfd1 Qf8 and White is for choice.]  13.Nh4  Now if the h5 knight retreats, g6 falls. Rf8 14.Bxh5 gxh5 15.Qxh5+ Rf7 16.Nf3 Be6 17.Bh4  Crisply dealing with the situation. Bxc4 18.Ng5 Qa5 19.Rh3 Nd4 20.Nxf7 Bxf7 21.Qg5 Kd7 22.Qe7+ Kc6 23.Rb1 Nc2+ 24.Kd2  1 - 0