Chess Opening Theory/1. e4/1...e5/2. Nf3/2...Nc6/3. Bb5/3...f5/4. Nc3/4...fxe4/5. Nxe4/5...d5/6. Nxe5/6...dxe4/7. Nxc6/7...Qg5/8. Qe2/8...Nf6/9. f4/9...Qxf4/10. Ne5/10...c6/11. d4/11...Qh4/12. g3/12...Qh3/13. Bc4/13...Be6
Ruy Lopez, Classical Schliemann Defence
For the first time since Black played 3...f5, White has no material hanging! So she has a major decision to make about where to develop the last remaining minor piece.
This is going to be followed by either 14...Bd6 15.O-O-O O-O-O or 14...O-O-O 15.O-O-O Bd6 (see diagram left). The move ...Bd6 carries a vague threat of ...Bxe5 and ...Bg4 winning an exchange, or at least relieving the pressure on f7.
The game Carlsen-Nisipeanu now continued with 16.Rhf1 attempting to exploit the pin on the f6-knight, but Nisipeanu was able to ignore it with 16...Rhe8!? and draw the ending that resulted after Carlsen accepted the pawn. Don't capture your opponent's weak pawns, capture your opponent's strong pawns!
The most common line is 16.Qf1. Qf1 not only piles up on the f6-knight, it also introduces an immediate threat of Nf7 forking the rooks. Black can't just capture on f1 because e6 drops with check. So the standard reply to that is again 16...Rhe8, side-stepping the fork. (16...Rhf8 is less accurate as 17.Nf7! forces Black into playing 17...Rd7 anyway, and White can then close the f-file with Bxf6.)
Unfortunately after 16...Rhe8 White isn't simply winning a pawn on f6, because Black has a queen fork on h6 that adds a second defender. In fact, it's Black who's threatening to win a pawn by capturing twice on e5, so White has to accept the trade of queens with either 17.Bxf6 Qh6+ 18.Kb1 Qxf6 19.Qxf6 gxf6 or with 17.Qxh3, which at least preserves the advantage of the bishop pair for the ending.
That's all a bit ho-hum, so can we look again at the diagram position and find a more energetic plan for White? Here's White's essential problem: the unopposed d-pawn would be a great source of energy if it could advance to d5, but right now it's frozen. Its job is to make the knight on e5 look pretty, which means White is playing with 3 pawns against 3 on the queenside instead of 4 against 3. What's the knight contributing to White's position to justify this lavish expense? If the answer was anything at all, Black could exchange it off any time she wanted, and the resulting isolated pawn on e5, although it's "passed", would have a lot less potential energy than it had on d4.
So how about making that knight earn its keep? White already has a perfect square for it, which is e3. The knight is the best blockader of a passed pawn, and from e3 it supports the pawn advances c4 and d5. If White could play the knight directly from e5 to e3, it would be a good move. The knight would be better placed on e3 than on the "outpost" e5.
Since the rules of chess don't allow that, White has to first exchange light-squared bishops (...Bg4 is still a threat) and then send the knight via c4. And since 16.Bxe6+ Qxe6 17.Nc4 would leave the knight pinned to the pawn on a2, White is also better off swapping queens first with 17.Qc4. Black will probably decline to exchange on c4, but the queen exchange on e6 can't be avoided due to the weakness of f7.
White doesn't have to pull the trigger on Bxe6+ straight away; instead the idea of Ne5-c4-e3 can be combined with the idea of pressurising Black's own frozen knight on f6. But after 16.Qf1 Rhe8 the exchange on e6 would leave the queen misplaced on f1, the square the h1-rook really wants to be on. So as presumptuous as it is to improve on Magnus Carlsen's moves, his idea of 16.Rhf1 Rhe8 may hold more promise if followed up by an exchange on e6 and then Ne5-c4-e3.
This is a less forcing move than 14.Bg5. White wants to indirectly cement the knight on e5 by ensuring that if Black plays Bf8-d6xe5, the recapture by the bishop will leave the 4-vs-3 queenside pawn majority intact.
With no threats on the board, both players can make some developing moves: O-O-O, Kb1 and Rhf1 for White, ...Bd6, ...O-O-O and ...Rhe8 or ...Rhf8 for Black. A future ...h6 would threaten to drive back the White bishop with ...g5. ...Bc7 and ...Bb6 would force White to think about the defence of the d-pawn. But ultimately, Black's pawn on e4 is more liability than asset, as it can't go anywhere and it ties down the f6-knight - whereas White's queenside majority is healthy and d5 break will come with gain of tempo on the e6-bishop.
Both White and Black win more games (at the expense of draws) after 14.Bf4 than after 14.Bg5. This seems paradoxical, but it's quite a common effect: the more forcing a line is, the more scope there is to analyse it to a draw with computer assistance.
Sadly for Schliemann fans, that's still only a 19% win rate for Black compared to 49% for White, which is a bad score even for the Black side of the Ruy Lopez - and White has a potential improvement with 9.Nxa7+. Realistically, Black needs to make 7...Qd5 or 5...Nf6 work in order to bump the Schliemann up the list of Ruy Lopez defences.