Manko vs Krush

After 6...a6 White has two decent roads. The quiet one runs 7.dxc5 Bxc5 8.Qxd8+ Kxd8 9.Be2 Ke7 into a queenless middlegame where nobody risks anything. The other is 7.e4, and that pawn lands on a square the f6 knight already attacks and no white piece defends. Barely anything separates the two, so this is taste rather than correctness. Against an opponent rated a hundred points higher, taste is a strategic decision.

The game went 7.e4 b5 8.Be2 Nxe4 9.a4 b4, and there the account stands. 9.a4 asks the b5 pawn the question, 9...b4 takes c3 away from the b1 knight, and that is precisely why the knight goes to d2, where it attacks the knight on e4. Almost nobody at this level plays either move; most would reach for 9.dxc5, and that is the real error: 9.dxc5 Qxd1 10.Rxd1 Bxc5 takes the queens off and leaves White a pawn down with no compensation at all. After 10...Bb7 11.Nxe4 Bxe4, by contrast, the knight that held the extra pawn together is gone.

11.Nxe4 was the only move that holds on to anything at all: after 11.a5 the position is dead level, and after 11.Nc4 Black is already the more comfortable side. Anyone who gives up a pawn should be able to name two things, what for and until when. Here the answer is a lead in development and a king in the centre, and the deadline runs for as long as Black needs to castle. Krush never castled in this game.

After 13...Nc6 the black knight attacks the one on d4, and more than half the players at this level recapture here. 14.Nxc6 Qxd1 15.Raxd1 Bxc6 is what that gives: queens off the board, the pawn still Black’s, the pressure gone and the position merely level. Manko played 14.Bf3 and offered the trade of the light-squared bishops instead, a move about one player in seven finds at this level.

14...Bxf3, and the same question returns at once, with three possible answers. 15.Nxf3 Qxd1 16.Raxd1 leads to the same relief as one move earlier; 15.gxf3 tears open White’s own king and costs more than a pawn. 15.Qxf3 keeps the queens on, and after 15...Nxd4 16.Bxd4 the bishop stands on the long diagonal looking at g7.

16.Bxd4 is not the same thing as the queen move to e4. It attacks the rook on a8 and the h7 pawn at once and looks strong for that reason, but 16.Qe4 Ne2+ 17.Kh1 Qd5 18.Qc2 Bc5 19.Qxe2 Bxe3 20.Qxe3 takes the shine off it: the knight leaves d4 with check, White has to fetch it back on e2, and by the end every minor piece is off the board while the extra pawn is still there. White is then more than a pawn worse off. The rule behind it is short. A pawn down, you do not trade queens, because compensation is made of pressure and pressure needs pieces.

The g7 pawn has stood there without a defender since 16.Bxd4, and more than half the players at this level take it right here. 18.Bxg7 Rg8 19.Qe5 Qxe5 20.Bxe5 Rg4 shows why that earns nothing: the bishop on g7 needs protection, the protection costs the queens, and the black rook ends up in the game via g4. A small advantage has turned into a small disadvantage.

18.Rfd1 leaves the pawn alone and puts the rook behind White’s own bishop. Rook, bishop and black queen now stand in a row on the d-file, so any bishop move would be a discovered attack. Krush left the file at once with 18...Qg5 and offered the queen trade herself. That way round it is good: 19.Qxg5 Bxg5 20.Bxg7 Rg8 21.Be5, and the pawn is back without a single black piece living off it.

Count it up: two rooks and a bishop each, five pawns against five. Black was a pawn up for twelve moves, and from move twenty nobody was. The same capture was worth less than nothing two moves earlier and is now the best move on the board, and all that changed is that the queens have gone. An undefended pawn does not run away. Take it at the moment your opponent gets nothing for it.

22.f4 attacks the bishop on g5 and gains a tempo. The more ambitious try was 22.Rd6, where a rook on the sixth rank attacks a6 and e6 at the same time: 22.Rd6 f6 23.Bg3 Bc1 24.b3. Almost nothing separates the two moves, and that is precisely what the position is telling you. The pressure that kept Black busy for twenty moves has run out.

The account opened on move seven came with a deadline, and the deadline expired here: 22...Be7 23.Rac1 Rxc1 24.Rxc1 f6 25.Bd4 Kd7. The black king, who never castled and who was the whole justification for the gambit for twenty moves, now walks into the centre of his own accord. With the queens gone he is no longer a target there but the most mobile piece on the board; from d7 he travelled via c8 and c6 all the way to b6, guarding the entire queenside on the way.

Manko read that deadline correctly. She did not let the compensation run until it was worth nothing, she exchanged it for material on moves nineteen and twenty. That is the real skill in a pawn sacrifice, and it has nothing to do with attacking: the position does not tell you when to cash in, the question of what your advantage will still be worth in five moves does.

28...Rc8 offered the trade of the last rooks, and accepting with 29.Rxc8 was right: after the alternative 29.Rd1+ nothing at all is left of White’s small edge. Then the white king set off: 29.Rxc8 Kxc8 30.Ke3 Kd7 31.Kd4 Kc6 32.Kc4, and the two kings faced each other on the c-file with one square in between.

32...a5 shuts the queenside. The pawn plants itself in front of a4 and defends b4 at the same time, which no pawn had done until then; and because the black king on c6 controls both b5 and d5, the white king has no entry square left. Whatever can still be moved here changes nothing.

Krush did the rest with the bishop trade: 33.Bd4 Bd6 34.Be5 Bxe5 35.fxe5, and the pawn ending is fixed solid. After 35...h4 36.Kd4 Kb6 37.Kc4 the draw was agreed. If you want to hold an endgame, do not look for the pawn that runs furthest, look for the move that takes the last entry square away from the enemy king. 32...a5 attacked nothing and settled the game.