23 July 2022

9:00 AM

23 July 2022

9:00 AM

Modern chess computers, like the program ‘Stockfish’, are treated as oracles. Plug in a position, start the engine, and within a fraction of a second it will identify the best move and its numerical evaluation (+1.27 pawns!).

So it is a natural misconception that an ambitious player must commit to endless clicking and memorising. On that theory, grandmasters are simply the ones who have set upon this treadmill with unusual fervour.

I have done my share of gawping at the screen, but in wiser moments, I remember that when Stockfish is running, my brain goes to sleep. The firehose of answers is unmemorable if you don’t articulate any questions.

The 17-year-old Andrew Hong is already a grandmaster, and he got the balance just right while preparing for a game at the US Junior Championships earlier this month. I’m sure he put in hours of computer time on this sharp line of the Sicilian Najdorf, but the deciding factor was the question he asked himself while ‘unplugged’. The critical moment arose after 17 Bd3 (see diagram). For tactical reasons, the e5 pawn can’t be taken, and White’s main threat is Nc3-e4. Hong knew that the silicon oracle recommended 17…d6-d5, but his analysis showed 18 Rg1 maintains a dangerous attack. Mulling this over in the shower, he wondered how to respond if Black plays 17…f5 instead.

To humans, this is very plausible, but a computer won’t flag it up. It loses by force, but the refutation is wildly complex. After his shower, Hong went back to the computer to fill in the gaps. As luck would have it, that was exactly the move chosen by his opponent, after more than 30 minutes thought.

Andrew Hong-David Brodsky,

US Junior Championship, July 2022

1 e4 c5 2 Nf3 d6 3 d4 cxd4 4 Nxd4 Nf6 5 Nc3 a6 6 Bg5 e6 7 f4 Be7 8 Qf3 Qc7 9 O-O-O Nbd7 10 g4 h6 11 Bxf6 Bxf6 12 h4 Nb6 13 g5 Bxd4 14 Rxd4 Qc5 15 Qf2 This is the first new move of the game. It looks surprising to place the rook in a pin, but as the game shows, the tactical tension can work in White’s favour. hxg5 After 15…Bd7 16 Qd2 the pressure on d6 cuts across Black’s intention to castle queenside. 16 e5 Powerful tactics. If 16…dxe5 17 Rd8+ wins the Black queen. gxf4 17 Bd3 Hong considered this more promising than 17 Ne4, which he had also investigated. f5 This is the key move that Hong refuted in his preparation. Another move to avoid was 17…Qxe5 18 Re1 and now 18…Qc5 (to prevent Rxd6) 19 Ne4! is devastating. In fact 17…d5 was the best defence. 18 Ne4 gives rise to fascinating complications: 18…dxe4 19 Rd8+ Kxd8 20 Qxc5 Nd7! 21 Qg1 exd3 22 Qxg7, when I suspect that White’s chances are better. Instead Hong intended 18 Rg1, keeping the pressure on. 18 exf6 e5 19 fxg7 Rg8 At first sight, Black has the upper hand, thanks to the pinned Rd4. 20 Bg6+ Ke7 21 h5! Uniting the passed pawns, but also creating space for a crucial check, so the queens remain on the board. Qxd4 22 Qh4+ Kd7 23 h6 Kc6 24 Be4+ The first move that Hong considered deeply, though at this point there are many paths to victory. White is a whole rook down, but the passed pawns are worth about a rook each! Kc7 25 Qf6 Bd7 26 Rd1 Qe3+ 27 Kb1 Kb8 28 h7 Ka7 29 hxg8=Q Rxg8 30 Qf7 Rc8 31 g8=Q Rxc3 31…Rxg8 32 Qxg8 is equally hopeless. 32 bxc3 Qxe4 33 Qff8 Nc8 34 Qg1+ Black resigns

Got something to add? Join the discussion and comment below.

You might disagree with half of it, but you’ll enjoy reading all of it. Try your first 10 weeks for just $10

Show comments