Georgiadis vs Gurel

For eighteen moves very little happened in this structure, and that is exactly what the Opocensky is played for: White clamped the queenside with 13.a5, Black sat behind e5 and d6 and waited. With 18...b5 Gurel made the position concrete, and after 19.axb6 Nxb6 20.Rxa6 White had an extra pawn. 20.Rxa6 is the strongest move in the position, and the pawn is still worth less than it looks.

The price is on the board two moves later. 20...Nc4 attacks the bishop on e3, and after 21.Qa1 Nxe3 22.Rxe3 Black has the bishop pair. With pawns on both wings and open diagonals that is real compensation: what is left of the extra pawn is not quite half a pawn, and against an opponent of this strength that is nothing to work with. 21.b3 was just as good as 21.Qa1, and the difference between the two is a matter of taste.

More important than the pawn was the choice of recapture on move 22. 22.Rxe3 keeps the pawns intact and the rook mobile on the third rank. 22.fxe3 doubles the pawns on the e-file, and after 22.fxe3 Rb8 23.Be2 d5 24.Bd3 Bc5 Black is already the slightly more comfortable side, a pawn down. The rule behind it: before you take a wing pawn, work out what your opponent gets for it. Two bishops on an open board are nearly always worth more than a pawn on the edge.

Between move 24 and move 32 almost everything came off: a pair of rooks on a2, the queens on a6. What is left is in the diagram, and the question is no longer what the extra pawn brings but whether two knights hold their own against two bishops. They just about do: even with the best move White is no longer better here.

33.Bc4 is that best move, and it offers the trade of the light-squared bishops: after 33.Bc4 Bxc4 34.bxc4 h4 35.Nf1 Rc8 36.Nd5 Nxd5 37.cxd5 Rxc2 Black takes the pawn back, but the position is level and the bishop pair is gone. The move played was 33.Bf1, a retreat to the starting square, where the bishop guards nothing and attacks nothing. After 33...Ra1 34.Rd3 Black, a pawn down, is about half a pawn better.

With two knights against two bishops there are two healthy plans: trade one pair of bishops, or put the knights on squares no pawn can ever attack. 33.Bc4 follows the first, 33.Bf1 follows neither. And the wider lesson for any ending: an extra pawn is not an advantage in itself, it is one only when your pieces stand at least as well as your opponent’s.

The moves 38...Ba6 39.Rd2 Bc8 40.Rd3 Ba6 41.Rd2 Bc8 look like clock management and are more than that: Black twice offered the trade of the bad white bishop, White twice declined, and the position stood on the board for a second time. With 42.Kg2 White stepped out of the repetition, and that was objectively the best move.

Then came 42...Bxh3+, and the point lies on f1: that bishop is defended by its own king and by nothing else, and the check drags the king away from g2. 43.Kxh3 Rxf1 44.Kg2 Ra1, and the material is level again. Taking was obligatory, because 43.Kf3 Bxf1 and 43.Kh2 Bxf1 both cost a whole piece.

The pattern is deflection of the defender, and it happens more often in endings than in openings, because there the king is one of the defenders himself. Before every king move, count which of your pieces he alone is holding. The larger lesson, though, lies ten moves earlier: a bishop that stands on f1 with no diagonal will cost you exactly one pawn in the end.

After the bishop sacrifice the material was level, and Georgiadis regrouped patiently: 46.Nb5, 47.Nec3 and 48.Nd5 put a knight on the best square in the position, 50.c4 and 51.b4 set the queenside going. Black released the tension with 52...fxe4 53.fxe4 and stood about a pawn worse. 53...Bd8 and 53...Nf6 would have kept it there, as both send a piece towards b6, the square where the passed pawn gets stopped. 53...Ke6 does not, and after 54.b5 Bd8 White was a good two pawns better.

55.Rb2 is the move, and it is worth almost three pawns. The rook steps in behind the passed pawn, and then the pawn runs: 55.Rb2 Nc7 56.b6 Na6 57.b7 Nb8 58.Ra2 Kd7 59.Ra8 Rxc3 60.Nxc3 Kc7, and White is the exchange up with a pawn on b7. The move played was 55.Ra2, the rook to the a-file instead of behind its own pawn, and after 55...Nf6 56.Ra8 less than one of those three pawns was left.

The best known rule about rooks comes from Siegbert Tarrasch: rooks belong behind passed pawns, your own and your opponent’s. The reason is simple. Behind the pawn the rook covers every square of its path and gets freer with every step; beside it the rook has to find a new post each time the pawn moves. Georgiadis had a good five minutes on the clock at this moment, and in five minutes you play the move that looks active, not the one that looks like a retreat.

Two of the three recaptures come with check, and those two are the wrong ones. After 57.exd5+ Ke7 58.Ra7+ Kf8 59.Ne4 Rxc4 Black has the c4 pawn and the position is level, and 57.cxd5+ Kd7 58.Ra7+ Kc8 59.Na4 Rb1 is the same story. 57.Nxd5 was played, and here that is the difference between a pawn and a half and nothing at all.

The reason fits into one sentence: the knight on d5 is the best piece in the position, and nobody defends the pawn on c4. Recapture with a pawn and you put a pawn on the square the knight occupied and hand c4 to the black rook. A check is not an argument, it is only a loud move. With every recapture, ask first which piece stands best after the exchange, and only then which move looks most forcing.

The rest of the game was spent on rook checks: 58.Ra7+ Kc8 59.Ra8+ Kd7 60.Ra7+ Kc8, with 61.Kf3 and 64.Kg4 walking the white king towards the kingside. About a pawn of pressure remained to the end and no more: the bishop on d8 covered b6, the black king stood in front of the passed pawn, and the rook stayed active. On move 71 the two of them agreed a draw, with the position on the board for the second time.