Test your chess: Reitstein problem 119

April 22, 2014

Black to play and win

J Ojermark v G de Marigny 1977

Solution

Quite an interesting one, and it showed the limitation of my calculating ability, or, at my limits compared to am engine. I found the line played in the game, but Stockfish (the app I typically use on my iPad) showed me that the line I couldn't quite get to work was actually better- better in the sense of having an even higher assessment (though, to be honest, it is not always true that an engine assessment of say -6 can really be compared with say -10: the -6 line can sometimes be 'more human'.

So, the first move is obvious: 1…Nf3+! 2 gf[]. Then what?

I tried and tried to make 2…Qh3 work, but couldn't crack 3 Bf4. Stockfish showed me 3…Re5! which, as soon as the engine showed it, I felt I should have been able to find. But Reitstein also didn't see it, or, more likely didn't bother to find: my solution was 2…Re2! which was the move played in the game, and it wasn't hard to see the resultant lines: for instance, 3 Qc1 Rc2! 4 Qe3 Re2! 5 Qc1 and now many things win. However, and this is why Stockfish prefers other moves (see below) after 2…Re2!, white can grovel with 3 Kg2!, when the positon is winning for black, but white could prolong the finish.

Finally, Stockfish's first move, which it rates higher than 2…Re2!, is 2…Nh5!, a move I never thought of, but again, when seen, is logical: it improves the positon of the N, threatening to come in to f4 either with the N or, better, the B, and just shows how tied up white is. Once Stockfish is given more time, its assessment changes, and the order is -6 for Re2, -9 for Nh5, -9+ for Qh3: essentially, many things win. The assessments are, to repeat, pretty meaningless. The lessons to draw are the recurrent advantage of remembering principles: 3…Re5! in the Qh3 line shouldn't have been hard, it is 'a rook lift' or 'bringing more pieces to the party'; and 2…Nh5! is also 'bringing more pieces to the party'.