Chess Siberia
Home page News Openings Best players/games Software Donate Video Philosophy Interviews Reviews Literature Music Cinema FIDE Answers Old newspapers Correspondence Chess Downloads Links Chess

The Truth About the Decisive Game Between Deep Blue and Garry Kasparov, New York 1997, Game 6

by Boris Schipkov

Garry Kasparov, the 13th World Chess Champion. Photo by Boris Schipkov, 2005
Garry Kasparov, the 13th World Chess Champion. Photo © Boris Schipkov, 2005.

In 1997 World Chess Champion grandmaster Garry Kasparov fought in the second match against the big and strong supercomputer Deep Blue of IBM, but lost the last 6th game and therefore the match 2.5-3.5. A play "The Machine" by Matt Charman about the epic and dramatic battle was staged 4th-18th September 2013 in New York at the Park Avenue Armory. Newspapers and websites wrote in the end of September that the Hollywood company Walt Disney Studios plans to shoot a movie on this historic match Garry Kasparov versus Deep Blue.
After reading all these I thought that I and other people did not know all the truth about the match and especially about the decisive game 6. Before the game 6 the score was equal 2.5-2.5, and if this game would have ended in a draw, then the Man would have withstood against the Machine. So I decided to study game 6 with a little help from modern computer programs and with my human understanding of positions or heuristics to find the truth. How many mistakes are made by Kasparov in this game? Could he have equalized the game?

Deep Blue - Kasparov, Garry (2785) [B17]
Man vs Machine, New York (6) 1997

1. e4 c6
The Caro-Kann Defence.
2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nd7 5. Ng5 Ngf6 6. Bd3 e6 7. N1f3

A risky move. Frequently Black plays 7...Bd6.
8. Nxe6!
A very strong knight sacrifice. Now modern computer chess programs also choose this move. Probably Garry Kasparov thought that Deep Blue would have played 8. Ne4, because computers rarely sacrificed pieces in unclear positions in 1990s. Perhaps the move 8. Nxe6 was simply kept in the opening library of Deep Blue.
A mistake. Correct is to take the white knight at once, 8...fxe6, and Black has counterplay in all variations: 9. Bg6+ Ke7 10. O-O Qc7 11. c4 Kd8 12. g3 (12. Re1 Bb4 13. Bd2 Bxd2 14. Qxd2 Nf8) 12...Bd6 13. c5 (or 13. Qd3 c5! 14. Re1 b5, and Black attacks on the queenside) 13...Be7 14. Bf4 Qa5 15. a3 Nd5 16. b4 Qa6 17. Bd3 b5 18. a4 Qb7 19. axb5 cxb5. Though White has good compensation for the piece Black could make a draw. If we compare the lines after 8...Qe7? and 8...fxe6 with a help of heuristics or general principles of chess, we will see that in the positions after 8...fxe6 9. Bg6+ Ke7 10. O-O Qc7 11. c4 Kd8 Black can develop his kingside with ...Bd6, followed by ...Rf8 or ...Nf8. After 8...Qe7? the black queen stops the development of the kingside, Black cannot play the dark-squared bishop. One principle: Develop pieces quickly in the opening. Therefore according to calculations and general principles the move 8...fxe6 is better than 8...Qe7. E.g. 8...fxe6 9. Bg6+ Ke7 10. O-O Qc7 11. Qe2 Kd8 12. Rd1 Bd6 13. Ne5 Ke7 14. f4 Rf8 15. c4 c5, with counterplay in Mohr, S - Krizsany, A, Kecskemet op 1988, Black won in 30 moves.
Why did Kasparov prefer 8...Qe7? Probably he did not examine seriously the positions after the sacrifice 8. Nxe6 at home before the game 6, because expected the simple 8.Ne4.
9. O-O fxe6 10. Bg6+ Kd8 11. Bf4!
A winning move, as my investigation shows. White has a big lead in development, White's bishops are monsters, they control the important squares near Black's king, so Black cannot move the king on a flank to a safe place. Also possible is 11. c4!?.

Black tries to activate his forces on the queenside with the idea to move the king there, also he intends to place the knight in the center on d5. However this move and all others are lost, 11... Nh7 12. Qd3 Ng5 13. Nh4 Nf6 14. c4 b6 15. Rae1 Qd7 16. Bg3 Bb4 17. Rd1 Rf8 18. f4 Nf7 19. f5 Qb7 20. fxe6 Bxe6 21. a3 Be7 22. Rde1 Ng5 23. Nf5 Bxf5 24. Bxf5 Nh5 25. Be5 Nf6 26. h4 Nf7 27. Bh3,
11...Qb4 12. a3! Qxb2 (or 12...Qb5 13. b3 Be7 14. c4 Qa6 15. Qd3 b5 16. Rfe1 bxc4 17. bxc4 Nf8 18. Ne5 Nxg6 19. Nf7+ Kd7 20. Qxg6 Rg8 21. Rxe6 Kxe6 22. Re1+ Kd7 23. Qf5+ Ke8 24. Nd6+ Kd8 25. Qe5 Bd7 26. Nf7+ 1-0, Siklosi, Z - Nemeth, M, Lenk op 1990) 13. c4 (in the stem game Geller, E - Meduna, E, Sochi Chigorin Memorial 1986, White played 13. Qe2, and won in 36 moves) 13...Be7 14. Qd3 Qb6 (14...Ng4 15. Bg3) 15. Rfe1 Nf8 16. Ne5 Nxg6 17. Nf7+ Ke8 18. Qxg6,
11...a5 12. Bg3 b5 13. Re1 Bb7 (if 13...Ba6 then White wins with a nice pawn sacrifice 14. d5! Nxd5 15. Nd4 Kc8 16. Rxe6) 14. a4 Ra6 15. Nh4 Nd5 16. Bh5 Nf4 17. Bg4,
11...Nd5 12. Bg3 Qb4 13. Qe2 Nc7 14. c4 Be7 15. a3 Qb6 (15...Qb3 16. d5!) 16. Rfe1 Nf6 17. Qc2 Bd7 18. c5 Qa6 (18...Qa5 19. b4 Qa6 20. Ne5 Be8 21. Nf7+ Bxf7 22. Bxf7 Rf8 23. Bxe6 Nxe6 24. Rxe6) 19. Ne5 Be8 20. Nf7+ Bxf7 21. Bxf7, winning.
12. a4!
White storms the weak Black's queenside. Also good are 12. Re1, 12. Bg3 and 12. c3.
If Black advances the pawn with 12...b4, planning ...a5, ...Ba6 and the king manoeuvre ...Kc8-b7, then after 13. Re1 a5 White attacks with 14. c4! bxc3 (14...Ba6 15. d5!) 15. bxc3 Nh7 16. Qb3, winning.
13. Re1! Nd5 14. Bg3

Other moves are hopeless too: 14...a6 15. Re2 Kc8 16. Qe1 and 14...a5 15. Bh4 N7f6 16. Ne5 Kc8 17. Nf7 Rg8 18. Bg3.
15. axb5 cxb5 16. Qd3
Or 16. Rc1 N7b6 17. Qd3 Nc7 18. Ne5.

Black's king is in danger. After 16...Nc7 17. Ra3 b4 18. Ra5 Qd8 White has a powerful attack in the centre with 19. c4! Be7 20. d5. Or 16...Qb4 17. Bf7 Nc7 18. Bxe6 Nxe6 19. Rxe6 Be7 20. Ne5 Nxe5 21. Rxe5 a6 22. Qf5+, and White wins.
17. Bf5!
Also possible is 17. c4 Nb4 18. Qc3 bxc4 19. Qxc4 Nb6 20. Qe2 Kb7 21. Ne5 Bd5 22. Rec1 Rc8 23. Nf7 Rg8 24. Nd6+, but after the text the game finishes quicker, because Black must give his queen.
17...exf5 18. Rxe7 Bxe7

19. c4!
Again the best move: White opens files for his queen. After 19...bxc4 20. Qxc4 Nb4 21. Re1 Re8 22. Bd6 Bxd6 23. Rxe8+ Kb7 24. Rxa8 Kxa8 25. Qe6 Bb8 26. d5 White grabs a piece and wins.
Black resigned. 1-0

Now we have the answers. How many mistakes are made by Kasparov in this game? Garry Kasparov made only one mistake, the eighth move 8...Qe7?. It was enough for the supercomputer, Deep Blue played well and won.
Could Kasparov have equalized the game? Garry Kasparov had real chances to equalize with the move 8...fxe6, if he would have taken White's knight at once.

October 20, 2013


Deep Blue - Kasparov, Garry (2785) [B17]
Man vs Machine/New York (6) 1997

1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nd7 5. Ng5 Ngf6 6. Bd3 e6 7. N1f3 h6 8. Nxe6! Qe7? 9. O-O fxe6 10. Bg6+ Kd8 11. Bf4! b5!? 12. a4! Bb7 13. Re1! Nd5 14. Bg3 Kc8 15. axb5 cxb5 16. Qd3 Bc6 17. Bf5! exf5 18. Rxe7 Bxe7 19. c4! 1-0


© 2013 Boris Schipkov