And so another edition of Dortmund comes to a close, and for a change Vladimir Kramnik was not the winner. (He came close and had his chances, but for now he's "stuck" at 10 wins there.) Officially, the title goes to Fabiano Caruana for having more wins than Sergey Karjakin (I'm not sure why that should be a tiebreaker - it's far from obvious that it's intrinsically better to have a win and a loss than a pair of draws - but no one asked me!), but the two both won their last-round games and finished with identical 6-3 scores.
Caruana's game was in fact the first to finish. Against Mateusz Bartel he chose his favorite, the Gruenfeld, and Bartel replied with the Rb1/Be2 line, but with 11.Qd2 rather than the more enterprising (but deeply worked-out) 11.Bd2. Teimour Radjabov had played 11.Qd2 against Caruana at the Tal Memorial earlier this year and obtained an overwhelming advantage, so Bartel must have been hoping for a little of that action himself. Obviously Caruana had prepared, however, and varied with 12...e6 from his earlier 12...Rd8.
On move 15, 15.Bb4 is both more popular and more successful in the database than Bartel's 15.Rc7 (played after nine minutes), but there haven't been any high-level tests of either move. (I note that Houdini 2 prefers 15.Bb4 as well, but prefers the untested 15.Rfe1 as its #1 choice.) The engine isn't enamored by Bartel's 16th move either; essentially, the bottom line in this variation is that White's pieces look like they're swarming all over Black's position, but Black always has sufficient resources - not to mention an extra pawn and targets on d4 and a2.
White's 18th move was inaccurate as well, according to the oracle; it prefers 18.Be7 Re8 19.Bb5 Nc6 20.Bxc6 Bxc6 21.Nxf7 Rec8 22.Rxc8+ Rxc8 23.Nd6 Rc7 24.Rc1 Rxe7 25.Rxc6, "thinking" that after 25...Bxd4 Black's extra pawn will be difficult-to-impossible to exploit. Bartel's 18.Bb5 (still in the databases) cost him 12 minutes, but he was still following a couple of old (1996 and 2000) correspondence games. Only after Bartel's 23rd move did they reach new ground, and by then Bartel was clearly worse, and his new move was a disimprovement. Black was winning, and with perfect play Caruana finished him off after just 27 moves.
That guaranteed that Kramnik could not join in a tie for first, but he still finished on a high note, crushing Georg Meier with a mating attack. (Very sportingly, Meier allowed Kramnik to deliver the mate - not everyone would do that, especially when his king is sitting on d5!) An English turned into a Semi-Tarrasch where White had the isolated d-pawn; a structure which saw Kramnik win a string of beautiful games (including two over Anand) around the turn of the century. Meier played an unusual move, 11...Nde7, and Kramnik invited him to go pawn-grabbing.
Maybe Meier should have played the circumspect 16...Qe7, because 16...e5(?) gives the knight access to d5, opens the Bc4's diagonal and doesn't even inconvenience the Bc7 because of f4. Maybe 20...Qg5 would have minimized the damage, and on move 23 Meier had to give up the exchange. 23...Re8 is simply a blunder, but one that will live on for a few generations of tactics anthologies. Kramnik played the remaining 10 moves with the accuracy of a computer, and delivered the aforementioned mate.
Now let's turn to the event's co-winner, Sergey Karjakin. Jan Gustafsson played the Caro-Kann against him, which is a double surprise: "Gusti" is renowned as a 1...e5 specialist, while Karjakin has a tremendous track record of bludgeoning the Caro-Kann with the Advance Variation. As it turns out, both players abstained from e5, as Karjakin chose the Classical lines with 3.Nc3. The players were in well-worn territory for a long time, though 17.Be3 isn't as usual as 17.Ne5.
Gustafsson's 18...Nf6 was a new move and a tacit draw offer, but obviously with first place on the line, the white pieces and a big rating edge that wasn't going to happen. After 19.Ne5, it looks like Karjakin was gifted with a better version of the 17.Ne5 line, as the bishop is more usefully tucked away on c1 - on d2 it blocks the d-pawn's defense by the rook. After 17.Ne5 Qe4 White doesn't play 18.Qf1, as Black can safely take on d4. Here, 20.Qf1 was perfectly sensible, and White gained a useful tempo a little later with 21.f3.
Gustafsson's 21...Qh7 was another questionable decision. It's a nice diagonal and the queen helps secure the kingside - maybe - but White enjoyed a significant edge occupying the abandoned center. The final consequential error was 26...f5. Black desperately wanted some activity for his pieces, but White's pieces were better prepared for an open position. A catastrophe quickly ensued on g7, and Black did not make it to the time control.
Ruslan Ponomariov seemed a likely candidate to join the tie for first, with White against Daniel Fridman, but it was not to be. They were in new territory in a Tartakower QGD as early as move 13, as they reached a normal position but where White had gotten in e3 for free. (Fridman apparently thought this was a neutral change or maybe slightly detrimental to White's prospects.) The play sharpened when Black played 18...dxc4, creating serious pawn majorities for both sides. 18...bxc4 was safe and easy, but Fridman showed some ambition at this point. 19...b4 was another sharp decision, enabling White to eliminate Black's majority at the cost of an exchange - but when one side gets two pawns for the exchange can we really speak of the "cost"?
The question was whether White could succeed in advancing his central majority. Ponomariov managed to push a pawn to d7, but at the cost of his a-pawn, and with Fridman's a-pawn on the loose he wisely decided to force Fridman to force a perpetual. (The "alternative" was 29.Qc6, but only the most naive computer user could consider this a real winning try. Black makes what are generally obvious moves and it's still dead even: 29...a3 30.Qc5 Qd2 31.Qxa3 Qd1+ 32.Kg2 Qd5+ 33.f3 Rb8 [of course not 33...Qxb5?? 34.Qxf8+ with an extra queen now, after 34...Kh7 35.d8Q, or inevitably, in the winning pawn ending after 34...Qe8 35.Qxe8+ and so on] 34.Qa6 Rd8 35.Qc8 Qa2+ etc.)
Finally, two of pre-round co-leaders faced off in the battle between Arkadij Naiditsch and Peter Leko. As if to make up for Karjakin-Gustafsson, they played an Advance Caro-Kann; amusingly, Naiditsch's 8.Na3 was preferred by Leko himself last year against Roiz, as opposed to 8.dxc5 as chosen by Karjakin against Grischuk and Carlsen (also last year). Leko deviated from Roiz's play with 11...d3, and Naiditsch's 12.b4 was a new move. White wound up with an extra pawn, but his unwieldy mass on the queenside and the ...a5 break kept the chances balanced.
The assessment remained equal throughout, but not for want of trying or a lack of imbalances in the position. In the end, Naiditsch had decimated Black's queenside, but Black's counterplay led to a fairly simple but spectacular perpetual check combination that saw Leko sac a bishop, then the exchange and then a rook. (One might joke that Leko's drawing mojo is so powerful he can draw even when he's two rooks down!)
So that brings this edition of the Dortmund super-tournament to a close, one of the hardest-fought I can recall. Here are the final standings:
1-2. Caruana, Karjakin 6 (of 9)
3-6. Ponomariov, Naiditsch, Leko, Kramnik 5.5
7. Meier 4
8. Fridman 3.5
9. Bartel 2
10. Gustafsson 1.5