The story behind the game
In game six of the 1960 world championship in Moscow, Mikhail Tal created difficult decisions for Mikhail Botvinnik with a knight sacrifice. Modern engine evidence and the historical result answer different questions: distinguish an objectively best defence from the practical difficulty of finding it at the board.
Take these ideas to your next game
- A sacrifice can create difficult choices without being objectively forced.
- When defending, search for simplification as well as direct protection.
- Calculate forcing resources before rejecting an unfamiliar position.
- Use engine evidence to test a romantic story, not merely decorate it.
Read the complete annotated game
Full game transcript, including the result and the answers. Original explanations by chessgames.info; game score from the sources below.
Practical difficulty and objective advantage are different ideas. Tal’s 21…Nf4 invites gxf4 and creates a complicated position around the white king. The historical result does not establish that accepting the knight was a mistake. Compare the saved engine defence with what happened in the game, and describe the threats before judging the sacrifice. This distinction is central to learning from Tal without copying risks blindly.
Recaptures can matter because they open lines and create new threats. The game continues 22…exf4. Black’s advanced pawn becomes an important part of the complications, later reaching f3 and capturing on e2. This is not merely a recapture to recover material. Track the pawn’s threats and also look for White’s best defensive resources in the saved analysis.
Count the whole sequence before deciding who has won the exchange. After 25.Rxb2, Tal plays …fxe2. The advanced pawn captures a queen, and the struggle changes again. The earlier sacrifice created a position in which familiar material counts could be misleading for several moves. Reconstruct each capture from the exact board, then examine the defensive alternatives rather than assuming the final result was inevitable.
The position after a combination still needs a plan. With 31…Be4, Tal continues improving the bishops after the exchanges. The game has moved beyond the original knight offer; active pieces and passed pawns now determine the practical tasks. Name the new targets instead of replaying the earlier combination in your head. A successful attack often requires a second phase of ordinary, accurate chess.
Result: 0-1. The recorded score ends here; it does not imply checkmate.
Think through the turning points
Move 21: Create a difficult decision
How can Black increase pressure even if the knight can be taken? In the game: Nf4. Practical difficulty and objective advantage are different ideas. Tal’s 21…Nf4 invites gxf4 and creates a complicated position around the white king. The historical result does not establish that accepting the knight was a mistake. Compare the saved engine defence with what happened in the game, and describe the threats before judging the sacrifice. This distinction is central to learning from Tal without copying risks blindly.
Engine-supported alternatives
Nf6 g4 Bd7 Qd2 h5 g5 Ne8 a3 Qb3 Be4 Bf5 Bxf5 · +0.85 · depth 18
Nf4 gxf4 · +1.24 · depth 18
Bd7 a3 Qb3 Ne4 Rxc1 Rxc1 Rxc1 Bxc1 Nf6 Nxd6 Nxd5 Ne4 · +1.64 · depth 17
Move 22: Change the pawn structure
After White takes the knight, how can Black keep the position active? In the game: exf4. Recaptures can matter because they open lines and create new threats. The game continues 22…exf4. Black’s advanced pawn becomes an important part of the complications, later reaching f3 and capturing on e2. This is not merely a recapture to recover material. Track the pawn’s threats and also look for White’s best defensive resources in the saved analysis.
Engine-supported alternatives
exf4 a3 Qb3 Bxa7 Be5 f3 b6 a4 Rb4 a5 bxa5 Nd1 · +1.34 · depth 20
a5 fxe5 Bxe5+ Kg1 Rh4 a3 Qb3 Qb5 a4 Bf1 Bxh3 Bxh3 · +2.95 · depth 19
b5 fxe5 Bxe5+ Kg1 a5 a3 Qb3 Nd1 Rc2 Rxc2 Rxc2 Bd2 · +3.08 · depth 19
Move 25: Keep the sequence concrete
The queen has been captured. What forcing resource remains? In the game: fxe2. Count the whole sequence before deciding who has won the exchange. After 25.Rxb2, Tal plays …fxe2. The advanced pawn captures a queen, and the struggle changes again. The earlier sacrifice created a position in which familiar material counts could be misleading for several moves. Reconstruct each capture from the exact board, then examine the defensive alternatives rather than assuming the final result was inevitable.
Engine-supported alternatives
fxe2 h4 Be5+ Kg1 Rxc3 Bxc3 e1=R+ Rxe1 Bxc3 Ree2 Bxb2 Rxb2 · -2.79 · depth 20
fxg2 Kxg2 · +4.49 · depth 20
Rxc3 Bxf3 Rxc1 Bxc1 Be5+ Qxe5 dxe5 Be3 b6 Rb3 Kf7 Ra3 · +5.15 · depth 19
Move 31: Coordinate after the storm
Which move improves Black’s minor-piece coordination in the changed position? In the game: Be4. The position after a combination still needs a plan. With 31…Be4, Tal continues improving the bishops after the exchanges. The game has moved beyond the original knight offer; active pieces and passed pawns now determine the practical tasks. Name the new targets instead of replaying the earlier combination in your head. A successful attack often requires a second phase of ordinary, accurate chess.
Engine-supported alternatives
Be4 Ne2 Be5 f4 Bf6 Rxb7 Bxd5 Rc7 Bxa2 f5 gxf5 Kf2 · -3.69 · depth 20
Bb1 Ne2 Be5 f4 Bf6 Rxb7 Bxa2 Rxa7 Bc4 Ra8+ Kg7 Re8 · -2.74 · depth 19
Bc8 Kg2 Rc1 Ne2 Rc4 Nxf4 Rxf4 Bd3 b6 Ra3 a5 Rb3 · -1.14 · depth 19
Every piece has a part to play.
In game six of the 1960 world championship in Moscow, Mikhail Tal created difficult decisions for Mikhail Botvinnik with a knight sacrifice. Modern engine evidence and the historical result answer different questions: distinguish an objectively best defence from the practical difficulty of finding it at the board.
Look a little deeper
Play on the board or type your move. Strong engine-supported alternatives receive credit.
Future moves stay hidden until you continue.
Try a different path.
Select a key moment to explore legal continuations supported by saved Stockfish analysis.
Accessible evaluation values & provenance
Stockfish 17.1 · 75,000 nodes per position; 1,000,000 at key moments. Actual depth 12–20. Positive centipawns favour White; mate scores remain separate. Selective search is evidence, not proof.
Download complete engine evidence