These guys know how to play chess that’s worth watching.
Once again, blood was the order of the day with three decisive games out of four, making it nine wins (five by Black) in total with just three draws. Please, gentlemen, keep it going.
The one draw was…kind of weird. The game Nils Grandelius - Dommaraju Gukesh repeated a couple of super-GM games through White’s 15th move, and then Gukesh more or less immediately played a new move, 15…Bb4. (He spent two minutes on it, which in this context was essentially immediate. It allows a very obvious Greek gift sac, which one does not allow unless it has been worked out in advance - almost certainly the case - or one spends a significant amount of time at the board making sure it doesn’t win.) Grandelius executed the sac after four minutes, and *now* Gukesh thought for almost 16 minutes. What?! Then, after a few very obvious moves, Gukesh thought for more than 33 minutes before playing the obvious 19…Kf6. It seemed like kabuki theater, but I think there’s a point to it and the time spent was genuine. Here’s my reconstruction:
Gukesh had prepared 15…Bb4, and knew that 16.Bxh7+ was only good for a draw, and also knew that White did not have to play the sac. So, he played his 15th move quickly, and then spent some time after 16.Bxh7+, surprised that White would so easily throw away a white game. Was there something he was missing? Not seeing a problem, he continued.
Then, on move 19, I think he realized two things.
First, White did not have to go for the repetition chosen in the game, but could meet 19…Kf6 with 20.c3, when it seems that Black should take twice on c3 - 20…dxc3 21.bxc3 Bxc3 - and now White has 22.Ne4+. Because the bishop is loose on c3, retreating the king to f7 won’t result in a repetition, but in a lousy position for Black where White has reestablished material equality with 23.Nxc3. Black must therefore take the knight (22…fxe4), allowing White to win the queen with 23.Bg5+ and 24.Bxd8. The resulting positions are sharp and materially imbalanced, and it would have taken Gukesh some time to work through the lines and assess that they are satisfactory for him.
Second, Black does not have to play the obvious 19…Kf6, but can also play 19…Qd7. This doesn’t result in any tidy repetitions, and entails significant risk for both sides. So it’s possible that he was analyzing this line instead, or in addition to, the move he chose in the game.
Indeed, it’s clear that both players were digging deep into the 19…Kf6 20.c3 line, because Grandelius spent 35 minutes in turn before deciding to repeat with 20.Nh7+ Kf7 21.Ng5+ Kf6 22.Nh7+ Kf7 23.Ng5+ and draw. I think it’s likelier than not that both players had this in their computers before the game (especially because one of the super-GM predecessors involved Jorden Van Foreest, who is playing in this tournament), but that doesn’t mean that they had looked at the line at any point in the recent past. Anyway, this short draw had its fair share of content, and guaranteed that Gukesh would remain with at least a share of the lead.
As it turned out, that’s all he wound up with, as both of the other players with a chance to catch him did so. Arjun Erigaisi blundered material against Peter Svidler with 12…Ne6, probably missing White’s zwischenzug 13.c5! (or at least missing its significance). The point was that after 13…d5 (more or less forced) the h2-b8 diagonal opened up for White’s bishop, and after 14.Nxe6 Bxe6 15.Bf4 Black was forced to cough up some pawns or give up the exchange. Erigaisi correctly chose the latter, but was unsurprisingly unable to keep the position closed forever. Svidler had to work a while, but there was never any reason to think the final result would be in doubt. Svidler thus joined Gukesh in first with 2.5/3.
Also joining the first place tie was the tournament’s bottom seed - by far - the American youngster Abhimanyu Mishra, who came back from a lost position to defeat Vincent Keymer. Keymer didn’t realize that 28.Rf1! Be2 29.Re1! was winning, so he played 28.Rd3 Qg1+ 29.Rd1 Qg1+ (repeating), and he didn’t find it the second time around either. This time he played 30.Bd1? Rc8 (=), and then erred again with 31.Qc7 (31.Kc1!=), when his position was lost after 31…Re8. It took a while for Mishra to finish him off, but he never allowed Keymer to reach a position that wasn’t lost from then on. So far, a great result for Mishra.
Finally, in the battle of the noughty boys Van Foreest and Boris Gelfand, it was Van Foreest who got on the scoreboard while Gelfand has now “castled queenside”. (In other words, three losses in a row - 0-0-0, for those who haven’t heard that quip before. Next up is the Audi logo followed by Olympic rings, but let’s hope for Gelfand’s sanity that it doesn’t get that far.) Gelfand seemed to be playing some deep preparation in a sharp Accelerated Dragon line that resulted in an unpleasant but “equal” ending. That is, it’s “equal” for the computer, but it’s the sort of “equality” where one player can easily lose while his opponent will never lose without making an overt blunder or a huge number of smaller errors. Sure enough, Gelfand’s seemingly small errors proved fatal, while Van Foreest’s one error only meant that Gelfand had one brief opportunity to find a brilliant draw (35…e4+!!). He didn’t find it, and there were no further chances.
No rest days yet, so round 4 is coming up today (Sunday), with these pairings: Erigaisi - Gukesh, Mishra - Grandelius, Gelfand - Keymer, and Svidler - Van Foreest. And here’s the Grandelius-Gukesh game, with my annotations.