Tal frequently played sacrifices that he couldn’t calculate to a clear conclusion; he wrote that his reasoning was “If I can’t see through it all, neither can he!” and he would proceed to throw away his knight or rook or queen or whatever under the assumption that he would subsequently out-calculate his opponent.
I kinda played that way in college.(Not like Tal, obviously, duh.) Sometimes I didn’t know what to play so I would just pick the most complicated move I could come up with. At that time it usually worked.
Nowadays, not so much.
However, maybe you can fake it. I recently saw an idea from a GM about “intuition” in such circumstances, which sounds a little mystical. But the GM put forward a reasonable substitute for mysticism: counting. How many pieces are attacking the king? And how many are defending? If A outnumbers B, you’ve got decent odds that the attack will win out.
Here’s an example, or a game which COULD have been an example.
Slater (2129) – Pan (2101)
Mechanics Institute G45
1. d4 d5 2. Nf3 c6 3. c4 e6 4. Qc2 Bd6 5. Bg5 Be7 6. Bf4 Nf6 7. e3 Nh5 8. Bg3 O‑O 9. Nc3 Nd7 10. Bd3 g6 11. O‑O‑O dxc4 12. Bxc4 b5 13. Bd3 Bb7 14. Kb1 a5 15. Ne4 Nxg3 16. hxg3 a4 17. Rh6 Re8 18. Rdh1 Nf8 19. Ne5 Qd5
After Black’s move, the Stockfish evaluation jumps from +1.5 (White’s got a very good position) to +5 (White’s completely winning).
Of course Stockfish doesn’t sit with you at the board.
Now any good little boy who grows up playing the Blackmar Diemer Gambit wants to play Nxf7 in this position. Unfortunately Black has that knight on f8 holding everything together. So, yeah, I poked at the idea a little bit and then abandoned it.
In fact, thank you Professor Stockfish, White’s position is so overwhelming that he can play 20.Nc3 Q-whereever 21. Nxf7, or even go 21.Rxh7 straight away. (20.Nc3 Qxg2 21.Be4 is a cute sideline.) It’s like any wanton kingside sacrifice other than Rxg6+ wins here.
Against Black’s best defense the lines are sorta long, and hard to calculate if the little Blackmar Diemer Gambit boy’s goatee has turned gray.
But let’s forget precise calculation for a second and just reason through the position. Or count.
White has five pieces attacking the light squares around Black’s king, f7/g6/h7. Two rooks, the monster Ne5, Bd3 and Qc2. Black has only one piece, the Nf8, protecting those squares. His other pieces are offside or blocking each other from affecting the light squares.
What’s that you say? You say White’s queen and bishop aren’t actually attacking those squares? They’re blocked by their own knight on e4?
Ah yes. That optically wonderful centralized Knight on e4 is actually in the way. It’s my only piece that isn’t attacking light squares.
Ergo, Nc3 to get out of the way. With tempo. And then an avalanche on the light squares. White can sack a rook and still have remaining attackers easily outnumbering the defenders.
20. Nc3 Qd8 21. Nxf7 Kxf7 22. Rxh7+ Nxh7 23. Rxh7+ Kf8 [23…Kg8 24.Bxg6 Rf8 25.Bf7+, I missed that mating pattern] 24. Bxg6 Bf6 25. Rf7+ Kg8 26. Bh5 Bg7 27. Qg6 and 27…Qf6 is the only way to delay mate.
Could I calculate that line with its necessary branches? Evidently I could not, not right now. But could I have counted 5 attackers versus 1 defender? Yes, I can still count.
I just need to sharpen up my attacking intuition.
[FWIW we continued with 20.f4? f5? 21. g4! which Stockfish says is not strongest, but completely sound. And kinda cool. After …f5 weakened g6, THEN I could see 21…fxe4 22.Bxe4 Q-whatever 23.Bxg6 was overwhelming.]