GM Sebastian Bogner (SUI, 2545) – GM Peng Li Min (SUI, 2553)

Caro-Kann Advance, and White builds the pieces around the bishop on f5: 6.h4, 7.Ne2, 8.c3, 9.Ng3, and after 9...Bh7 the bishop has to go into the corner. That is not an attack, it is space. On h7 the bishop is not badly placed, it is simply out of the game.

Peng spends fourteen minutes on 10...f6 and then gets the game he wanted: castling long, the light-squared bishops traded with 12...Bxd3, and 13...c5 against the centre. 13.Qxd3 decides nothing, because every other move there costs a piece. 14.Qe2 does: most players put the queen on 14.Qd1, and Qe2 is worth a little more, because the queen stays on the e2-d3-c4 diagonal and therefore looks at the square the black pawn is about to occupy.

14...c4 costs Peng twenty minutes, the longest think of the game by either player, and it leaves him 53 minutes after the move against Sebastian’s 1 hour 19. The move takes space and takes the tension out of the centre, and it puts a pawn on a square a white pawn can reach from b2. That is exactly what happens six moves later. After 15.Nd2, the only sensible retreat, White has a position he knows and half an hour more on the clock. A pawn that gains space also fixes itself: before you push it, ask which enemy pawn can come and visit it there.

16.O-O takes eight minutes, then Peng takes a pair of knights off with 17...Nxg3 and shuts the kingside with 18...f5. That is well played: every exchange leaves less of White’s small edge in space, and by move 20 there is nothing left of it. The game will be decided on the queenside or nowhere, and there White has a plan: 19.Rfb1 after seven minutes, 20.b3, the pawn on c4 has to go.

21.a4 is the only move that holds the balance. After 21.Be1 or 21.Bf2 Peng has the slightly better game, and the difference is close to half a pawn. Sebastian finds a4 after five minutes, with 57 minutes left after the move.

Then 21...cxb3 22.Rxb3 Nc4, and the knight sits on a square no pawn can take away from it. 23.Nxc4 costs seventeen minutes, the longest think of White’s game, with 39 minutes left after the move, and it is the right one: the alternatives left the knight where it was and were about a fifth of a pawn worse. 23...Qxc4 24.Qxc4 dxc4: queens gone, knights gone, material level, and the position is what it was before. Thirty minutes for three moves that win nothing and hold something are thirty minutes well spent. A knight on a square no pawn can reach only gets more dangerous: if you can trade it, trade it.

25.Rb5, 26.Rab1, 27.Kf2, 28.Ke2: the king walks to the centre, which is where it belongs in an endgame. Peng defends with his rooks on the seventh rank and on the d-file, 26...Rh7, 27...Rc7, 28...Rcd7, 29...Rc7, and he does it on almost no time. Between move 20 and move 30 Sebastian spends 31 minutes and is down to 30:59 after his thirtieth, while Peng has 47:41 after his own thirtieth, a minute more than ten moves earlier. Credit where it is due: the shuffling looks like embarrassment and is precise waiting.

Behind it sits a concrete question: can White play d4-d5? While a black rook stands on c7, not to any purpose. The rook on d8 takes the pawn, the rook on c7 keeps covering b7, and the position stays level. The exchange is simpler still: 30...Rd5 31.Rxd5 exd5, and it is dead equal.

30...Rcd7 arrives after six seconds and is the mistake of the game. Both rooks are on the d-file afterwards, and neither of them can afford to take a pawn on d5. The rook on d7 also covers b7 and a7 along the seventh rank. If it leaves that rank to capture on d5, b7 stands alone, the white rook on b5 takes it with check, and the king cannot take back, because the second white rook stands behind it on b1; a7 goes the same way. The other rook, 30...Rdd7, would have occupied the same file and kept b7 covered. There is little to blame him for, this is the mistake a human makes here: two rooks, one square, and they look the same. When two rooks can go to the same square, the question is not which one belongs there but what the one that leaves is holding.

31.d5 was the move, and it gives the pawn away: 31.d5 exd5 32.Bd4, and White is a pawn down and about a pawn better. Three things on the board pay for it. The bishop reaches d4 and looks at a7. From there a7 and b7 are held by the king on b8 and by the rook on d7, which may no longer leave the seventh rank. And after the exchange the pawn on e5 is a passed pawn that wants to go to e6.

Sebastian looks at this position for 1 minute 07 and plays 31.Bg1, with 30:22 left after the move. This is not a question of move order and not a matter of taste, it is a missed break: three quarters of a pawn separate 31.d5 from the next best move. Peng could have switched it off again at once with 31...Rc7 32.d5 Rxd5 33.Rxd5 exd5, after which nothing of the advantage is left. He plays 31...Be7 instead, and 32.d5 is now worth more than it was a move earlier. 32.Be3 comes after four seconds. 33.d5 was still there, and 33.Bc1 came after not quite three minutes.

In practical terms the move that was played is the human one. A pawn sacrifice in a blocked position, on board one, against an opponent with twenty minutes more on the clock: almost nobody plays that, and 31.a5 stands next to it as the quiet continuation most players choose. It stays wrong all the same, three times over. After 34.Be3, with 25:27 left after the move, and 34...Rcd8 the game was given up as a draw. In a blocked position the break is the plan, and its value depends on where your own pieces stand: it was worth a pawn while the rooks stood on d7 and d8 and the bishop was one move from d4, and a fifth of that once the bishop stood on c1 and the rook was back on c8.