Round 4

Black has taken space with 3...d4 and disturbed White's set-up with 5...Bb4+. Almost two thirds of players would play 6...Bxd2+ here, because the bishop is offered and trading keeps things simple. Sebastian played 6...a5, and that is the better move.

The difference is not the bishop pair, it is the pawn on a5. It props up the bishop on b4, and if White trades anyway, Black recaptures towards the centre with the a-pawn. Which is what happened: 7.O-O Nf6 8.Bxb4 axb4, and suddenly Black has a pawn on b4, a half-open a-file and a rook already standing on it.

The idea travels. When a trade is offered, do not ask whether you want to keep the bishop, ask what you would recapture with. A move that prepares the trade for you instead of executing it often wins a file on the way.

10.a3 is a question, and it has only two honest answers: capture, or leave the b4 pawn and live with the exchange on b4. Best was 10...bxa3 11.Rxa3 Rxa3 12.bxa3 Qe7 13.Qb3. After that liquidation one pair of rooks is gone, White is left with an isolated pawn on a3, and nobody owns the a-file. Black is a shade more comfortable there.

The move played was 10...Qe7, and one move later, after 11.Qb3, again 11...Nd7 rather than 11...bxa3. Both are sensible, neither is an error, and still the hesitation costs the file: 12.axb4 Rxa1 13.Rxa1 Nxb4, and now White captures first, trades one pair of rooks and is left as the only side with a rook on the open a-file.

That is the practical rule this game teaches. When both sides can open the same file, what decides it is not who handles it better, it is who captures first. Wait, and you end up capturing with whichever piece your opponent leaves you.

Sebastian has stabilised with the knight manoeuvre 14...Nb8 15.Nc2 Nxc2 16.Qxc2 Na6 and in the diagram he is a quarter of a pawn better: the d4 pawn denies White the centre, the knight on b4 is secure and the bishop on g4 is active. The move to match that is 20...f5, taking space on the wing where the black pieces already are, for example 20...f5 21.h3 Bh5 22.c5 bxc5 23.Nb3.

The move played was 20...Bd7, and this retreat has a side effect that is easy to miss: the bishop steps between the queen on e7 and the pawn on c7. White saw it at once and played 21.Ra7, the c7 pawn is suddenly undefended, and Black has to move the bishop again with 21...Bc6 so that the queen can see the pawn.

Two moves for nothing, and with them the small edge is gone: from here it is White who has the extra idea. Check on every retreat which file or diagonal you are closing with your own piece. Pieces defend each other across the board, and a square that was covered a moment ago can be hanging after one harmless looking step back.

White is a third of a pawn better here, the rook on a7 is the better piece, and Black has no counterplay beyond a pinprick. Nine out of ten players would reach for 22...Bxg2 23.Kxg2 Nc6 in this position, because you take a bishop when you can take it. The difference to 22...Nc2 is small, a fraction of a pawn, but the knight move can do something the trade cannot.

It attacks the queen and forces her onto a square she has to leave again: 23.Qd1 Nb4 24.Qa1 Nc2 25.Qd1 Nb4, the position repeats, and neither side deviates. Draw. Whoever attacks the queen decides where she goes, and when she can only choose between two squares, he decides how the game ends as well.

Taking a draw is a decision like any other and is practised far too rarely as one. Sebastian was slightly worse, the clock was not on his side and the match was long since running on the other boards. In that situation a repetition is not a concession, it is the result that fits the position.