Riddle solved: Taimanov could not have defeated Fischer

About a week ago Karsten Müller invited readers of his latest riddle to take a close at the first game of the 1971 Candidates Match between Bobby Fischer and Mark Taimanov in Vancouver, Canada. The American defeated his opponent from the Soviet Union with "dry" 6-0 score.  But matters could have started differently. In the first game it was Taimanov who put on the pressure.

Taimanov played 27.h3, which nobody considered a bad move – until Kasparov pointed out its defect in 2004. Then Lakdawala found 27.Ba6 and subsequently Aagaard, Karolyi, and Timman found further improvements. Timman even concluded that "White has a winning advantage; Black's compensation for the exchange is absolutely insufficient."

Grandmaster Jonas Lampert agrees with our finding that 27.Ba6 does not give White a winning position. At first, the computer does think that White's advantage of the exchange is decisive. But deeper analysis shows that eventually, Black can hold a pawn-down rook ending.

After 27...Rf8 28 Ba6 Rb6 Taimanov continued to press for the win and chose 29.Rc7 instead of the interesting alternative 29.Bb5! White played the obvious check 31.Bxh6 (which should have lost), but new analysis has found Taimanov could have played 31.Nxf5+ and held.

As Kasparov pointed out, after 29.Rc7 Qa4 30.Rg7+ 31.Bxh6 Fischer should have played the winning 31...Kh7!  Instead, 31...Kf7 allowed Taimanov to again dream of victory.

Taimanov played 36.Nd4 and called it "capitulation."  Instead, 36.g4! would have led to positions highly favorable for White.  In fact, in one variation, Black has to play a temporary rook sacrifice to reach a drawn ending in which he is a full bishop down!

Fischer's 37...Qb4 completely turns the tables.  As Taimanov says, "A stunning manoeuvre, abruptly changing the character of the battle."

Here is the game with full analysis by Charles Sullivan [click on the notation to get a replay board, and on the fan button to get engine assistance]:

Taimanov, Mark E2620

0–1

Fischer, Robert J2740

Candidates Match Vancouver16.05.1971[Charles Sullivan]

1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5 7.0-0 Nc6 8.d5 Ne7 9.Bd2 Ne8 10.Rc1 f5 11.exf5 gxf5 12.Ng5 h6 13.Ne6 Bxe6 14.dxe6 Qc8 15.Qb3 c6 16.Bh5 Qxe6 17.Qxb7 Nf6 18.Be2 Rfb8 19.Qa6 Rxb2 20.Rfd1 e4 21.Qa3 Rb7 22.Bf4 d5 23.cxd5 cxd5 24.Nb5 Ng6 25.Nd4 Qd7 26.Qe3 Kh7?! Fischer makes a very risky move. Removing the bishop with 26...Nxf4 leads to a roughly equal game. 27.h3 Early commentators did not criticize this passive move. Taimanov (writing in 1992) says that he intended to play 27.Rc6, but decided that he needed to prevent 27...Ng4. Then, in 2004, Garry Kasparov suggested 27.Bb5 as an improvement and, in 2015, Cyrus Lakdawala was probably the first to show that 27.Ba6 is best. Next, Jacob Aaagaard (in the 2020 translation of Taimanov's "I Was a Victim of Bobby Fischer") improved upon the 27.Ba6 variation and concluded that "White has a serious advantage." Finally, in separate books published in 2021, Tibor Karolyi affirms Aagaard's analysis and Jan Timman claims that "White has a winning advantage." Kasparov's variation was 27.Bb5 Ng4 28.Bxd7 Nxe3 29.Bxe3 Rxd7 30.Nxf5 Rad8 31.Nxg7 Kxg7 32.Bd4+ Kf7 33.Rc6 Rd6 which is drawish. Aagaard's critical variation is 27.Ba6 Ng4 the only move that Lakdawala considers is 27...Rb6 28.Bb5 Qf7 29.Rc7 Ne7 30.Qa3 Nfg8 "when Black's position gets critically strained" [Lakdawala]. The computer says White wins after 31.Bf1! Qf6 32.Rd7 Qf7 33.g3 Qe8 34.Rc7 Kh8 35.Bd6 Qd8 36.Nb5 Bf6 37.Qc5 Rab8 38.Bf4 e3 39.fxe3 Qe8 40.Rb1 Rc8 41.Nd6 Rxc7 42.Qxb6 Rc6 43.Nxe8 axb6 44.Bb5 Rc2 45.Bd7 Bc3 46.a4 Kh7 47.Rxb6+- 28.Qg3 Nxf4! 29.Bxb7 Bxd4 30.Qxf4 Bxf2+ 31.Kh1 Bb6 32.Bc6 Qf7 33.Bxd5 Qf8 34.Rf1 (At this point, both Timman and Karolyi end their variations. Timman says, "White has a winning advantage; Black's compensation for the exchange is absolutely insufficient.") Rc8‼ 35.Rb1 The computer at first thought that White wins with 35.Rxc8 Qxc8 36.g3 Ne3 37.Qd6 The saving move is Bc5‼ 37...Qd8? 38.Qxd8 Bxd8 39.Rc1! Kg7 40.Bb7 Bb6 41.Rc8 Kf6 42.Ba6 Bd4 43.Rc6+ Kg5 44.Be2 f4 45.gxf4+ Kxf4 46.Rxh6+- 38.Rc1 Bxd6 39.Rxc8 Nxd5 40.Rd8 e3 41.Rxd6 e2 42.Re6 Nc3 43.Re3 Kg6 44.Kg2 h5 45.Kf3 Kf6 46.Re8 Kf7 47.Rxe2 Nxe2 48.Kxe2 Kg6= 35...e3 36.Be6 e2 37.Bxf5+ Kh8 38.Qxg4 exf1Q+ Aagaard gives 38...exf1R+ 39.Rxf1 However, the computer sees that Black can hold: Rc7 (GM Jonas Lampert reached this position and says "the resulting endgame is defendable.") 40.Bd3 Rf7 41.Re1 Bc5 42.Qc4 42.Rc1 Qe8! 43.Qg6 Qe3 44.Rb1 Qf4 45.Rb8+ Qxb8 46.Qxf7 Qg8 47.Qc7 Bb6 48.Qc3+ Qg7 49.Qc8+ Qg8 50.Qf5 Qg7 51.g3 Qa1+ 52.Bb1 Qg7 53.a4 Qe7 54.Be4 Bc5 55.a5 Bb4 56.Qg6 Qg7 drawish 42...Rf2 43.Qe4 Qg7 44.g3 Bd6 45.Qe3 Rf6 46.Re2 Bf8 47.Kg2 Qd7 48.Bc2 Qb5 49.Be4 Re6 50.Qf3 Qb4 51.h4 Qa3 52.Qg4 Qe7 53.a4 Re5 54.Qf3 Qe8 55.Rf2 Rxe4 56.Qxf8+ Qxf8 57.Rxf8+ Kg7 58.Rf4 Re5= drawish, according to Stockfish 27...Rf8 28.Ba6 Rb6 29.Rc7 Safer and also leading to balanced play is 29.Bb5! (Taimanov, Kasparov, Lakdawala, and Karolyi give this move) Rxb5 30.Rc7 Qe8 31.Rxg7+ Taimanov saw that 31.Bxh6 fails to f4 ; Stockfish says best play continues 32.Bxf4 Rb6 33.Bh6 Nh5 34.Rxa7 Rbf6 35.Ne2 Qe6 36.Qg5 Rf5 37.Nd4 Rxg5 38.Nxe6 Kxh6 39.Nxg5 Kxg5 40.Rxd5+ Be5-+ 31...Kxg7 32.Bxh6+ Kf7 32...Kh7 is also playable 33.Nxb5 Qxb5 34.Qxa7+ Nd7 also 34...Ke6 35.Bxf8 Nxf8∞ 35.Bxf8 Ngxf8∞ with approximate equality. 29...Qa4 30.Rxg7+ Kxg7 31.Bxh6+? Nobody has criticized this "obvious" move, but it should lose. White achieves salvation with 31.Nxf5+! Kh7 32.Be2 Nh5 33.Bxh5 Rxf5 34.Bxg6+ Rxg6 35.Rc1 Rg7 36.Bxh6 Rgf7 37.Bg5 Qxa2 38.Rf1 Qa6 39.h4 Qc6 40.Qe2 d4 41.Qh5+ Kg8 42.Qg4 Qg6 43.Re1 Re5 44.Rc1 Re8 45.Rc4 d3 46.Rc5 e3 47.fxe3 Rd8 48.Qd1 Qe4 49.Bxd8 Qxe3+ 50.Kh1 Qxc5 51.Qxd3 DRAWISH 31...Kf7? Taimanov actually praised this move: "Fischer's play in this stage of the game is irreproachable--his king would have been uncomfortable on h7." As Kasparov pointed out, the winning move is 31...Kh7! 32.Be2 Rf7 33.Bg5 Ne8 34.Bd8 Rd6 35.Qg5 Ng7 36.Nb5 Re6 37.Nd4 Qxa2 38.Kf1 Rd6 39.Nb5 Rdd7 40.Nd4 Qb2 41.Bf6 Qb6-+ 32.Be2 Rfb8 33.Nxf5 Rb1 34.Rxb1 Both Timman and Karolyi think White still has the superior chances after 34.Kh2 . Karolyi gives Qd7 35.g4 Rxd1 36.Bxd1 Rb1 37.Be2 (Transposing to the position reached after 36.g4 in the note to White's 36th move.) Rb2 38.Kg2 Rxa2 39.Bd1 Qc6 (For other continuations from this position, see the 36. g4 variation at White's move 36.) 40.Qb3 Ra1 41.Be3 a5 42.Be2 Stockfish finds this position very interesting, but ultimately drawish: Re1 43.Qb2 a4 44.Qd2 Rxe2 45.Qxe2 Ne7 46.Nd4 Qd7 47.Bg5 Qa7 48.Nb5 Qb6 49.Be3 Qb8 50.g5 Nf5 51.gxf6 Nh4+ 52.Kf1 Nf3 53.h4 a3 54.Nxa3 Nh2+ 55.Ke1 Nf3+ 56.Kd1 Qb3+ 57.Qc2 Qxa3 58.Qc7+ Kxf6 59.Qd8+ Ke5 60.Qe8+= 34...Rxb1+ 35.Kh2 Qd7 36.Nd4? "This is capitulation." (Taimanov) Taimanov suggested 36.g4! and play might continue Rb2 37.Kg2 Rxa2 38.Bd1 Qe6 Wolfram Schön points out that Stockfish also likes 38...Ra6 and the most interesting variation I can find is 39.Qd4 Qc7 40.Be3 Nf4+ 41.Bxf4 Qxf4 42.Qb4 Qc7 43.g5 Rb6 44.Qa3 Kg6!? Black can also reach a drawish ending by giving up his queen for a rook and knight: 44...e3 45.Qxe3 Ne4 46.Qd4 Kg6 47.Bg4 Nxg5 48.Qh8 Qh7! 49.Qf8 Ne6! 50.Nh4+ Qxh4 51.Bf5+ Kg5 52.Qe7+ Kh5 53.Bg6+ Kxg6 54.Qxh4 45.Ne7+ Kh7 46.gxf6 Rxf6 47.Nxd5 Rxf2+ 48.Kxf2 Qh2+ 49.Ke1 Qg1+ 50.Kd2 Qd4+ 51.Kc1 Qxd5 52.Qxa7+ . White has an extra bishop, but the endgame tables declare a draw. 38...Qc6 this position is discussed in the note to White's 34th move. 39.Qb3 Ra1 40.Be3 Black must find the problem-like Ng8‼ 41.Be2 Re1 42.Bb5 Rxe3! 43.fxe3 N8e7! 44.Be8+ Kf6 45.Bxg6 Kxg6 46.Nxe7+ Qxe7 47.Qxd5 Kh6‼ and, by some miracle, Black can draw this position: 48.Qf5 Kg7 49.g5 Qd6 50.Qxe4 a5 51.h4 Qd1! 52.Kg3 Qe1+ 53.Kg4 Qe2+ 54.Kf5 Qf2+! 55.Ke5 a4! and if 56.Qxa4 Qxe3+= the endgame tablebases say draw. Also playable is 36.Ng3 Qc7 37.Bg5 Rb6 38.Qd2= (Kasparov) 36...Qd6+ 37.g3 Qb4 "A stunning maneuver, abruptly changing the character of the battle. The threat of 38...Qe1 suddenly puts White in a hopeless situation." (Taimanov) 38.Nc6 Qb6 39.Nxa7 Qxe3 40.Bxe3 Re1 41.Bg4 The sealed move. "After 41.Bg4 Nxg4+ 42.hxg4 Ne5 resistance is pointless." (Taimanov) 43.Kh3 Rh1+ 44.Kg2 Rd1 45.Nb5 Nxg4 46.Bb6 Rd2 47.Kf1 Rxa2-+ 0–1