Make your own free website on

Garry Kasparov vs. Deep Junior Analysis

The FIDE "Man versus Machine: World Championship" was held between the worlds number one rated human player Garry Kasparov and the world micro-computer chess champion Deep Junior. It was an exciting match that brought world attention including CNN news overage and the final game had 3 hours of coverage on ESPN2 the first live televesion coverage of a chess game since the 1972 Fischer-Spassky match.

The match ended in a 6 game draw. Kasparov winning the first game while losing the 3rd with the rest drawn.

On this page I give anlaysis from the Chessbase web-site as well as extensive computer-assisted analysis of game 5 and 6. Game 5 was revolutionary for the advance of computers as Deep Junior played like Mikhail Tal offering a piece in the opening as black without a clear win, a real speculative sacrifice! White was under enormous pressure and decided to bail into a draw. Game 6 is also interesting as Kasparov was clearly better and settled for a draw preferring not to lose the match on a human error against the tireless machine.


(from chessbase site)

1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.e3 e6 5.Nf3 Nbd7 6.Qc2 Bd6 7.g4 dxc4 8.Bxc4 b6 9.e4 e5 Junior was surprisingly out of book and thought about 10 minutes now.

10.g5 Nh5 11.Be3 0-0 12.0-0-0 Qc7 13.d5N [ RR 13.Be2 exd4 14.Nxd4 Nf4 15.Kb1 Be5 16.h4 Nc5 17.h5 Nce6 18.g6 Nxd4 19.Bxd4 Be6 20.gxh7+ Kxh7 21.Bf3 Rad8 22.Be3 Rxd1+ 23.Rxd1 Rd8 24.Ne2 Rxd1+ 25.Qxd1 c5 26.Qd2 Nxe2 27.Bxe2 Qd6 Ward,C-Gausel,E/Copenhagen 2002/CBM 88/[Wells]/1/2-1/2 (58); RR 13.Kb1 g6 14.Be2 exd4 15.Nxd4 Nf4 16.h4 b5 17.Bxb5 Bb7 18.h5 Be5 19.Bc4 Nb6 20.Be2 Rae8 21.hxg6 fxg6 22.a4 Nxe2 23.Qxe2 Qf7 24.f4 Bh8 25.e5 c5 26.Ndb5 Qb3 27.Nd6 Bxh1 Hillarp Persson,T-Borgo,G/Batumi 1999/CBM 74/1-0 (37)]

13...b5?! 14.dxc6 bxc4 15.Nb5! Kasparov's knight will be a tower of power on d6. [ 15.cxd7 Bxd7 16.Qd2 Bg4 17.Qxd6 Qxd6 18.Rxd6 Bxf3 19.Rg1+/= ]

15...Qxc6 16.Nxd6 Kasparov would win this position against any human. Does he succeed against a computer as well?

16...Bb7 Black hopes to get play against the pawn e4.

17.Qc3! [ 17.Nf5 leads to nothing: 17...Qxe4 18.Ne7+ Kh8 19.Qxe4 Bxe4 20.Rxd7 Bxf3= ]

17...Rae8?! sacrifices the exchange to keep the pawn e5. Junior probably evaluates the knight d6 to be as strong as a rook. [ 17...Rab8 18.Nxe5 Nxe5 19.Qxe5 Qa4 looks better.]

18.Nxe8 Rxe8 19.Rhe1 Qb5 20.Nd2 Rc8 21.Kb1 Nf8 22.Ka1 Ng6 23.Rc1 Ba6 24.b3! cxb3 25.Qxb3 Ra8 26.Qxb5 Bxb5 27.Rc7 The resignation is a bit early, but there is no doubt about the final result anyway. (white has a winning material and positional advantage)1-0


1.e4 c5 2.Nf3 e6 [ Normally Kasparov prefers the Najdorf variation: 2...d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 ]

3.d4 cxd4 4.Nxd4 a6 5.Bd3 Bc5 [ Another set up is 5...Nf6 6.0-0 d6 7.c4 Be7 8.Nc3 ]

6.Nb3 Ba7 7.c4 Nc6 8.Nc3 d6 9.0-0 Nge7 10.Re1 [ RR 10.Qe2 0-0 11.Be3 e5 12.Bxa7 Rxa7 13.Qe3 Be6 14.Rfd1 Qb8 15.Be2 Rd8 16.Rd2 b5 17.cxb5 Bxb3 18.axb3 axb5 19.Rxa7 Nxa7 20.Qd3 Nec6 21.Nxb5 Nxb5 22.Qxb5 Nd4 23.Qxb8 Nxe2+ 24.Rxe2 Rxb8 Ramesh,R-Oral,T/Bled 2002/EXT 2003/1/2-1/2 (43)]

10...0-0 11.Be3 e5 12.Nd5N [ RR 12.Bxa7N Rxa7 13.Qd2 Be6 14.Bf1 b6 15.Red1 Rd7 16.Rac1 f5 17.f3 f4 18.Nd5 g5 19.Qf2 Nc8 20.c5 bxc5 21.Bxa6 Bxd5 22.exd5 N6e7 23.Bd3 Ra7 24.Qc2 Kg7 25.Bxh7 c4 26.Nd2 Nb6 1-0 Kogan,A-Bezold,M/Wuerzburg 1996/EXT 99 (38)]

12...a5!? Kasparov has managed agian to reach a position, which is not easy to play for the computer.

13.Rc1 a4 Kasparov fights for the square d4.

14.Bxa7 Rxa7 15.Nd2?! the knight is not well placed here. [ 15.Na1!? Nxd5 16.cxd5 Nd4 17.Nc2 should be OK for White.]

15...Nd4 16.Qh5?! more a gesture than a threat.

16...Ne6! aims for the very strong square c5.

17.Rc3 [ 17.b4 axb3 18.axb3 Nd4 19.Ra1 Rxa1 20.Rxa1 Nxd5=/+ ]

17...Nc5 18.Bc2 Nxd5 easier than trying to ignore and play around the Nd5 with [ 18...Nc6]

19.exd5?! The computer play's for an attack, which just isn't there. The knight c5 is now a giant. [ GM Karsten Muller likes 19.cxd5 better. ]

19...g6 the thematic ...f7-f5 does not run away. ...g6 allows the introduction of the Ra7 into the defence of the kingside with b7-b6, if neccessery.

20.Qh6 f5 Black's cental pawn mass is very impressive.

21.Ra3 Qf6!! A devilish trap into which Deep Junior falls. It just likes the material too much. [ 21...Ra6 and; 21...Qb6 came into consideration as well.]

22.b4 axb3 23.Rxa7 bxc2 White has to lose time now to neutralize the pawn c2. The queen h6 and the rook a7 are far away.

24.Rc1 e4 25.Rxc2 [ 25.Qe3? f4 26.Qa3? Qd4-+ ]

25...Qa1+? Kasparov played this move too fast. [ 25...f4! threatens e3 and cuts the queen off from the defence of White's first rank 26.Nf1 ( 26.Ra8 e3 27.Nf1 Qf5 28.Rxc8 Rxc8 29.Qh3 Qxh3 30.gxh3 exf2+ 31.Rxf2 g5-/+ ; 26.h4 Qa1+ 27.Nf1 e3 28.h5 Bf5-+ ) 26...e3 27.Nxe3 ( 27.fxe3 fxe3 28.Qxf8+ Kxf8 29.Ra8 Ke7 30.Rxc8 Qf5-+ ) 27...Bf5 28.Rc1 Qb2 29.Rd1 fxe3 30.Qxe3 Bd3-/+ ]

26.Nf1 f4 27.Ra8! The only defense with the idea Qxf8+ followed by Rxc8+. Probably Kasparov had missed this resource.

27...e3 [ 27...Nd3!? was interesting, e.g. 28.Re2 Qd4 29.Qh4 Qxc4 with initiative.]

28.fxe3 fxe3 29.Qxf8+! Kxf8 30.Rxc8+ Kf7 [ 30...Kf7 31.Rc7+ Kf8 ( 31...Ke8? 32.Re2 Kd8 33.Rxh7 Qc1 34.Rxe3 is dangerous for Black.) 32.Rc8+ Kf7 with perpetual check.] 1/2-1/2


1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.e3 e6 5.Nf3 Nbd7 6.Qc2 b6 The programmers of Deep Junior obviously hope to circumvent g2-g4 like this. But Kasparov finds a way nevertheless. [ In the first game 6...Bd6 was played.]

7.cxd5 exd5 8.Bd3 Be7 9.Bd2!? (novelty) Kasparov chooses a new path. [ RR 9.0-0 0-0 10.Bd2 h6 11.Ne2 Bb7 12.Ng3 Re8 13.b4 Rc8 14.Bc3 Bd6 15.Nf5 Bb8 16.Nd2 c5 17.bxc5 bxc5 18.Qb2 Qc7 19.g3 c4 20.Be2 Nb6 21.Qc2 Qd7 22.a4 Bc6 23.a5 Na4 Akesson,R-Eriksson,J/Skara 2002/CBM 91/0-1 (39)]

9...0-0 10.g4!? He plays courageously for attack against the machine!

10...Nxg4 11.Rg1 sharpens the play further. [ 11.Bxh7+ Kh8 12.Bd3 came into consideration as well.]

11...Ndf6!? A very surprising move for the commentators. Junior makes the play even sharper. Why not? It is the computer! [ 11...Ngf6 12.0-0-0 with white initiative.]

12.h3 Nh6 13.e4! dxe4 14.Bxh6?! [ 14.Bxe4!? Nxe4 15.Bxh6 Bf5 16.Rxg7+ Kh8 17.Nxe4 Qd5 18.Rg4!? with a very complicated position.]

14...exd3 15.Rxg7+ [ Obviously Kasparov did not like 15.Bxg7!? Ng4 ( 15...dxc2? 16.Bxf6+ Bg4 17.Rxg4# ) 16.Qxd3 Kxg7 17.hxg4 Bg5! gefiel Kasparov offenbar nicht.]

15...Kh8 16.Qxd3 Rg8! [ 16...Qe8? 17.Ng5 Ba3+ 18.Kd2 Bxb2 19.Rxh7+ Kg8 20.Rg1+- ]

17.Rxg8+ Nxg8 [ 17...Qxg8? 18.0-0-0 Qg6 19.Qe3+- with the idea Rg1]

18.Bf4 f6 Black has to control e5. The german GM Rainer Knaak on was now sure that Kasparov will not win this game.

19.0-0-0 Bd6 Black allows that its pair of bishops is halved. The Bc8 is the strong one anyway as it has no counterpart and the light squares in White's camp are weak. [ 19...Be6!? ]

20.Qe3!? sacrifices the pawn h3. [ 20.Bxd6 Qxd6 21.h4 came into consideration as well.]

20...Bxf4 21.Qxf4 Bxh3 22.Rg1 [ 22.Nd2!? -- 23.Nde4 ]

22...Qb8 23.Qe3 Qd6 24.Nh4 [ 24.Ne4 is answered by 24...Qf8]

24...Be6 25.Rh1 Rd8 26.Ng6+ White's attack looks dangerous, but Deep Junior has everything under control.

26...Kg7 27.Nf4 Bf5 28.Nce2 Kasparov directs his Queen's knight into the attack, but he has just not enough firepower left to win.

28...Ne7 29.Ng3 threatens Qxe7+ (which will not work against a machine of course)

29...Kh8 30.Nxf5 Nxf5 31.Qe4 Qd7! The game should now end in a draw, but in time trouble Kasparov blunders.

32.Rh5? Kasparov overlooks the mating attack in the 35th move. [ 32.Ng6+ Kg7 33.Nf4 Kh8 ( 33...Re8 34.Qg2+ Kf8 35.Ng6+ hxg6 36.Qxg6 Qd8 37.Rh7 Re1+ 38.Kd2 Qxd4+ 39.Kxe1 Qe4+= ) 34.Ng6+= ]

32...Nxd4! 33.Ng6+? [ Retreating with 33.Rh1 was best.]

33...Kg8 [ 33...Kg7?? 34.Rxh7+ Kxh7 35.Nf8++- ]

34.Ne7+ Kf8 35.Nd5 [ Kasparov want to go for 35.Rxh7 but he had overlooked 35...Nb3+ 36.Kc2 ( 36.axb3 Qd1# ) 36...Na1+ 37.Kc3 Qd2+ 38.Kc4 b5+ 39.Kc5 Qd6# when making 32.Rh5?.]

35...Qg7 [ White was hoping for 35...cxd5 36.Qxd4 ]

36.Qxd4 Rxd5 [ 36...Rxd5 37.Rxd5 cxd5 38.Qxd5 Qg5+ 39.Qd2 Qxd2+ 40.Kxd2 h5 and the distant passed pawn on the h-file decides, e.g. 41.Ke3 Kf7 42.Kf4 Kg6 43.a3 b5 44.b3 a6 45.b4 Kh6 46.Kf5 Kg7 47.f3 Kh6 48.f4 ( 48.Kxf6 h4 49.f4 h3-+ ) 48...Kg7 49.Ke4 Kg6 50.Kd5 h4 51.Kc5 h3 52.Kb6 h2 53.Kxa6 h1Q-+ Kasparov's courage was not rewarded again. Probably he did not have enough time left to master the complications in the 32nd move.] 0-1


1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 e6 5.Nb5 d6 6.c4 White has installed the Maroczy-bind. He has more space, but Kasparov just sets up a hedgehog.

6...Nf6 7.N1c3 a6 8.Na3 this knight is not well placed here and has already moved quite often. Black should have equality.

8...Nd7!? (novelty) to throw Deep Junior out of its opening book. [ 8...Be7 is usual.]

9.Nc2 Be7 10.Be2 b6 11.0-0 Bb7 12.h3?! one of those weakening pawn moves the white players make so easily in the hedgehog. It does not fit into the following white play.

12...0-0 13.Be3 Rc8 14.Qd2 Nce5 the Nc6 was a small problem in Black's position. Normally it is redeployed to d7 via e5 and thanks to h3 Kasparov does just this now.

15.b3 Nf6 16.f3 does not fit to h3 (the dark squares on the kingside are now weak), but signales that Junior wants to play on the queenside, which is right here, I think.

16...Qc7 17.Rac1 Rfe8 [ 17...b5?! plays with the fire due to 18.cxb5 Qxc3 19.Qxc3 Rxc3 20.Bd4 Rc7 21.bxa6 ]

18.a3 Ned7 Black has reached a typical hedgehog position

19.Rfd1 Qb8 leaves the vis a vis of the Rc1 and increases the harmony in Black's camp [ 19...Nc5!? 20.Rb1 d5 came strongly into consideration.]

20.Bf2 [ 20.Nd4?! allows 20...d5 ]

20...Rcd8?! invites Junior to play on the queenside, which is not good as the computer follows a strong plan then. [ I like 20...Ba8 more.]

21.b4 [ 21.Bg3 is easily fended off with 21...Ne5 ab.]

21...Ba8 22.a4 White takes the initiative on the queenside.

22...Rc8 23.Rb1 Qc7 24.a5!? According to Anand a correct manoeuvre of the machine.

24...bxa5 25.b5! Bb7 [ 25...axb5? 26.Nxb5 Qd8 27.Nxd6+/- ]

26.b6 [ 26.Nd4 came strongly into consideration as well.]

26...Qb8 27.Ne3 Nc5 28.Qa2 Nfd7 29.Na4 Ne5 30.Nc2 good prophylaxis [ 30.Nxc5?! dxc5 31.Qxa5 Nc6 is not, what White wants.]

30...Ncd7 31.Nd4 Red8 32.Kh1 Such a computer-move always shows that the machine does not see a way to improve its position.

32...Nc6 33.Nxc6 Rxc6 34.Kg1 h6 35.Qa3!? Junior fights again hard!

35...Rdc8 Kasparov did not have much time left and had to play quickly.

36.Bg3 Bf8?! now Bd8 is no option anymore. [ 36...Ne5?? 37.Bxe5 dxe5 38.Qxe7+- ; 36...Kf8!? Anand]

37.Qc3 Ne5 38.c5! opens the floodgates

38...Nd7 39.Qxa5 Nxc5 40.Nxc5 Rxc5 41.Qa4 R5c6 [ 41...a5 42.Bb5 is very good for White as well.]

42.Bf2 [ 42.Bxa6? Qa8 43.Bb5 Rxb6 is wrong.]

42...d5 43.Bxa6 Bc5 44.Bxc5 Rxc5 45.Bxb7?! [ 45.exd5 exd5 ( 45...Bxa6 46.Qxa6 exd5 47.b7 and the b-pawn is a powerful force.) 46.Bd3 (Knaak) was much better. Now it is very questionable, if Black can survive.]

45...Qxb7 46.exd5 exd5 47.Qa7 R5c7 Kasparov sets a very deep trap. [ 47...Rb8 48.Re1 Rc7 came into consideration as well.]

48.Qxb7?! [ 48.Qa5!? ]

48...Rxb7 49.Rxd5 this double rook ending is completely drawn.

49...Rc6! 50.Rdb5 h5 Black has an impregnable fortress as the whote rook's are tied to the defence of the b-pawn. Deep Junior does not underastand the very nature of this conception and plays on.

51.Kf2 Re6! assures that White's king remains out of play.

52.f4 g6 53.Kg3 Kg7 54.Kh4 Kh6 55.R1b4 Rd6 56.g3 f6 (threatens Rbxb6 Rxb6?? g5 followed by mate.)

57.g4 hxg4 58.hxg4 Kg7 59.Rb3 [ 59.f5 g5+ leads to nothing as 60.Kh5?? is refuted by 60...Rb8 61.b7 Rh8#]

59...Rc6 60.g5 f5 61.Rb1 1/2-1/2


1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.e3 O-O 5.Bd3 d5 6.cxd5 exd5 7.Ne2 Re8 8.O-O Bd6 9.a3 c6 10.Qc2 Bxh2+ 11.Kxh2 Ng4+ 12.Kg3 Qg5 13.f4 Qh5 14.Bd2 Qh2+ 15.Kf3 Qh4 16.Bxh7+ Kh8 17.Ng3 Nh2+ 18.Kf2 Ng4+ 19.Kf3 Nh2+ 1/2-1/2

Garry decided playing for a win was too risky here as he would be under serious pressure and one slip would lose. He would have to play extremely accurately to obtain winning chances and not get crushed.

Let's examine Kasparov's best winning chances and see just how accurate he had to be. The key move to play for a win is 16. g3!

Using Fritz 8 (Bahrain engine) after 4 and one half hours of thought we get the following line:

(0.84): 16.g3 and now:

A. 16...Nh2+ (computers seem to want to play this move)

Fritz 8 now considers that black should take a draw by repetition with 24...Qh2 25. Bb4 (or 25. Bc3) 25...Re3 26. Bd2 Re8 27. Bb4 Re3.

However this is not whites best chance to play for a win. White can improve in this line with 22. Nd6! (instead of 22. Nxf6+)

White can play for a win but its hair raising stuff that requires complete accuracy by the human player.

Here is that analysis:

22. Nd6 Re6 23. Nc4 Nd5 (diagram below)

And now White has 2 main choices:

  1. 24. Be4!? (Fritz liked this move after 30 minutes analyzing the above position) And here: 24...Qh2 25.Rh1 Qf2+ 26.Kd1 Rh6 27.Re1 b5 28.Nc3 Bd7 29.Bxd5 cxd5 30.Ne5 Nxe5 31.dxe5 Rc8 32.Qd3 And now after a further hour gives 32...Bg4+ as best for black. 33. Ne2 Rh2 And again white has 2 main choices:
    1. 34. a4 which gives black an advantage after 34...Rc4 35.Qe3 Qg2 36.e6 fxe6 37.Rc1 d4 38.Qe5 d3 39.Rxc4 bxc4 40.f5 dxe2+ (-0.69) and black may be winning here. This analysis after 9 hours of thought from 34. a4.
    2. 34. b3 (to stop the Rc4 invasion)this is whites best chance in this line after (24. Be4!?) Best play is now: 34...d4 35.e6 fxe6 36.Rc1 Rd8 37.Rc7 Qf3 38.Qxf3 Bxf3 39.Ba5 d3 40.Rxg7+ (-0.72) black is better but white has drawing chances after: 41. Bxd8 Kg6 42. Ba5 Rxe2+ 43.Rxe2 dxe2 44. Bc7 Kf5 45. Be5 Kg4 46. Ke1 black has an extra pawn, a passed pawn on the 7th rank that needs constant attention, and the superior king position. Whites has compensation however in the bishops of opposite color ending that results. For example 46...a5 (46...Kxg3 47. f5+=) 47. b4 axb4 48. axb4 h6 49. Kf2 Kh3 50. Bg7 h5 51. Bc3 and white should draw. I give the remainder example analysis for instructive purposes 51...Bg4 52. Bd2 h4 53. gxh4 Kxh4 54. Bc3 Bh5 black can make no progress 55. Be5 Kh3 56. Bd6 Bf3 57. Be5 Kg4 58. Bd6 Kf5 59. Be5 Ke4 60. Ke1 Ke3 61. Bd6 Be4 black can never make progress. It's a bishops of opposite color ending draw.

  2. 24. Ne5! this was suggested to me by NM Sheldon Gelbart as best. And indeed the computer analysis confirms this as whites best chance for a win. After over 23 hours of thought here is Fritz 8 analysis:
    1. 1. (1.16): 24...Qh2 25.Bxh7+ Qxh7 26.Qxh7+ Kxh7 27.Nxg4 Rxe2+ 28.Kxe2 Bxg4+ 29.Kf2 Kg6 30.Rae1
    2. 2. +- (1.47): 24...Re8 25.Be4 Qh2 26.Rf1 Ndf6 27.Bf3 Be6 28.Qd3 g6 29.Rh1 Qf2+

    White has a clear advantage here and my be close to winning

B. 16...Qh2

  1. 17.Rae1 f5 [17...Nxe3? 18.Bxe3 Qh5+ 19.Kg2 Bh3+ 20.Kf2 Bxf1 21.Rxf1 Qh2+ 22.Kf3 Qh5+ 23.g4 Qh3+ 24.Ng3 Rxe3+ 25.Kxe3 Qxg3+ 26.Kd2 and white is winning] 18.Bxf5 Qh5 19.Bxh7+ Qxh7 20.Qxh7+ Kxh7 21.Rh1+ Kg8 22.e4 and White is better) 17.Kf2 Ng4+ 18.Ke1 Qh3 19.Rg1 Nd7 20.e4 dxe4 21.Nxe4 Ndf6 22.Nxf6+ Nxf6 23.Kd1 Bg4 24.Re1 (line from Chessbase)
  2. 17. Bf5 Qh5 18. Bxf5 and here analysis by Fritz 8 (3 hours):
    1. = (0.00): 18...Bxg4+ 19.Kf2 Qh2+ 20.Ke1 Bh3 21.Rg1 Bg4 22.Qb3 b6 23.Qa4 Nd7 24.Rf1 Bh3 25.Rg1 Bg4
    2. 2. (0.97): 18...Qxg4+ 19.Kf2 Bf5 20.Qb3 b6 21.Ng1 Bd3 22.Rfc1 Qf5 23.Qa4 Qe6 24.Nf3 Nd7 25.Rh1
  3. 17 f5 (IM Malcom Pein considers this best) and here analysis by Fritz 8 (after 8 hours):
    1. 1. +- (1.62): 17...Nd7 18.Kxg4 Qg2 19.e4 Nf6+ 20.Kf4 dxe4 21.Bxe4 Nxe4 22.Nxe4 Qxe2 23.Rae1 Rxe4+ 24.Qxe4 Qxd2+
    2. 2. +- (2.06): 17...Qh3 18.Rh1 Nh2+ and in this last line:
      1. 1. +- (2.09): 19.Kf2 Nd7 20.Rag1 Nf6 21.Nf4 Nfg4+ 22.Ke2 Nxe3 23.Nxh3 Nxc2+ 24.Kd1 Ne3+ 25.Bxe3 Rxe3 26.Kd2
      2. 2. (1.34): 19.Rxh2 Qxh2 20.Rf1 b6 21.Qd1 Qh5+ 22.Kg2 Ba6 23.Rh1 Qg4 24.Rh4 Qg5 25.e4


    1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be2 e5 7. Nb3 Be7 8. O-O O-O 9. Kh1 Bd7 10. Be3 Bc6 11. Bf3 Nbd7 12. a4 b6 13. Qd3 Bb7 14. h3 Rc8 15. Rad1 h6 16. Rfe1 Qc7 17. g3 Rfd8 18. Kh2 Re8 19. Re2 Qc4 20. Qxc4 Rxc4 21. Nd2 Rc7 22. Bg2 Rec8 23. Nb3 Rxc3 24. bxc3 Bxe4 25. Bc1 Bxg2 26. Kxg2 Rxc3 27. Ba3 Ne8 28. f4 1/2-1/2

    This was a very exciting game. The crowd ROARED and cheered wildly as Kasparov played the thematic Sicilian sacrifice 23...Rxc3! to mortally wound whites pawns.

    Kasparov knew he was certainly not worse in the final position but decided to take a draw in the match rather than risk a defeat.

    In the final position black has 2 central pawns for the exchange sacrifice which must be good. Most GM's consider black to be much better here. Let's look a bit deeper.

    After whites last move 28. f4 GM Lev Alburt gives 28...f6 as best intending to centralize the king by marching it to e6 and unraveling the black pieces. Fritz was inclined to try and exchange rooks in analyzing this position, something black certainly does not want to do. For example: 29. Rd3 Rc4

    Here is that analysis:

    1. (-0.44): 30.a5 Kf7 31.axb6 Nxb6 32.Na5 Ra4 33.Nc6 Nc4 34.Bb4 a5 35.Bc3 Nc7 36.fxe5 fxe5
    2. (-0.62): 30.fxe5 fxe5 31.a5 Kf7 32.Rf2+ Ke6 33.Re2 g6 34.Nd4+ Kf7 35.Nb3 e4 36.Rf2+ Nef6 37.Rd1 d5

    Black is better but theres a long way to go