The pain, the pain! At least that's how Vladimir Kramnik must feel, and it's a pain surely shared by his fans (if only to a lesser extent). The game between Evgeny Tomashevsky and Dmitry Andreikin was richer than yesterday's 16-move draw, but it was still relatively short and uneventful. In the other game, though, Kramnik ground away in a pawn-up ending with rooks and knights and all the pawns on the kingside.
Objectively it should have been a draw, but Maxime Vachier-Lagrave failed to handle the tension and blundered with 58...Rf1+?? What he should have played was 58...Nd6. Black's position would remain tenable, though Kramnik could keep trying indefinitely. Instead, after 58...Rf1+?? 59.Ke3 it was no longer possible to play 59...Nd6 (or 59...Nxg5, for that matter) on account of 60.Ng6+ Kf7 61.Rf8+, skewering the king and rook. At death's door, Vachier-Lagrave found the only way to continue: 59...g6! 60.fxg6 Kg7! 61.gxf7 Kxf7. With a rook, knight and pawn against Vachier-Lagrave's bare rook White's position was winning, but one bit of work remained. White's pieces were poorly coordinated, and his knight and pawn were a little vulnerable. If Kramnik could re-establish the harmony of his forces his opponent could resign, and then Kramnik would be in the final and qualify for the Candidates' tournament via the World Cup, leaving the second automatic rating qualification spot for the Candidates' to Sergey Karjakin.
I was watching the live coverage at this moment, and several thoughts ran through my mind. First, I remembered Kramnik's complaint during this tournament that he would have various lapses in concentration, when he would prematurely relax, and wondered if this too might be such a moment. Indeed, Kramnik has had such problems throughout his career, going back as far as the mid-1990s (several examples are mentioned in his best games book, co-written with Damsky). As a fan I hoped he would bear down and solve the final puzzle of the game, and was both spooked by and a made slightly hopeful by game 4 of his match with Kasparov. The same material balance existed there, and there too Kramnik's knight and pawn were rather awkwardly placed. In that game he failed to find the win - the spooky part - but I hoped that the process of having analyzed that game and a realization of the potential trickiness of the ending would place him in good stead for this game.
Nope. The key variation Kramnik needed to work through started with 62.Nd7 (natural enough, but the critical moment comes later) 62...Rf5 63.Rf8+ Kg6 64.Rg8+! Kf7 and now 65.Ke4!! This clever in-between move is most likely what he missed, with the further crucial point that 65...Ra5 66.Ne5+! saves the pawn, as 66...Kxg5 67.Rf5+! wins the rook: 67...Kh4/Kh6 is met by 68.Nf3+/Ng4+ and 69.Rxa5.
With more time on the clock or more energy, Kramnik (or any other strong grandmaster, and probably some "weak" ones too) would almost definitely be able to find that variation and work it out to the finish. There are some subtle moves, but in general the line is forcing enough that one won't get lost in a maze of variations. Whether from fatigue or a lack of time Kramnik didn't manage to find this one chance, and Vachier-Lagrave managed to achieve a fortress. White's forces were completely tied down, and the only way to attempt progress was to surrender the pawn and hope to win with rook and knight vs. rook. That ending is a fairly easy draw as long as the defender's king doesn't start out in a (very) bad position, and Vachier-Lagrave saved it without too much trouble. (The game is here, with the my comments above reproduced therein.)
Both matches are therefore going to tiebreaks, and we'll see if the "tie-break beggars" as Anish Giri labeled both Andreikin and Vachier-Lagrave (for their willingness to throw away the white pieces and then hold on with Black to try their chances at the faster time controls) manage to achieve success.