A little democracy can be a dangerous thing. When I opened up a poll asking which Gukesh victory I should analyze, I assumed the top pick would be the game that put him in the driver’s seat for a match with Ding Liren, Gukesh-Firouzja, round 13. Abasov-Gukesh from the previous round was interesting as well, a fine example of playing provocatively with black to induce the peacefully inclined Abasov to overextend and lose. The only game I didn’t want to analyze was the round two battle between Praggnanandhaa and Gukesh. The position felt ugly and strange to me, a sure sign that I didn’t really understand it. Of course, that’s the game that won the poll, so I gritted my teeth and started trying to figure it out.
To make things more interesting, I decided to play guess-the-move. I selected the game on The Week in Chess and moved the browser so that only the board and move arrows were showing; the game score was hidden. Once I got out of the opening I tried to figure out the best move in every position and then checked my work against what Gukesh chose and, eventually, the computer. This is a great activity for playing through top-level games because by taking an active role rather than a passive one you understand the strategic and tactical ideas on a much deeper level. The only problem was that I’d played through the game once or twice during the Candidates, and even though my memory isn’t what it used to be, I remembered a few key moments and ideas as well as the rather odd pawn structure.
I started guessing after the opening moves 1 d4 d5 2 c4 e6 3 Nf3 Nf6 4 g3 Bb4+ 5 Nc3 dxc4 6 Bg2 00 7 00 Nc6 8 a3 Be7 9 e4 a6 10 Be3 b5 11 Qe2:
As I said at the top, I find these kinds of positions to be ugly and strange, to the point where I tend to evaluate them as a little worse for both sides. That’s not really an evaluation, but to me it means that I would be uncomfortable and start playing bad moves with either white or black. When I look at black’s position, I see an extra pawn and an unpleasantly cramped life ahead. Where’s that Nc6 headed? Can I get my queen off the d-file before the center opens up? When I look at white’s position, I see perfect harmony but no obvious way to make progress or regain the sacrificed pawn.
My candidate moves here were … Na5 and … Bb7. I’m probably going to have to play both, but there didn’t seem to be any reason to move the knight before developing the bishop, and a glance at the amusing 11… Bb7 12 e5 Nd5 13 Nxd5 Qxd5 14 Ng5? Nxd4! helped reassure me that black’s position holds together tactically. My next few moves matched the choices of Gukesh: 11… Bb7 12 Rad1 Na5 13 d5 (I sort of thought I was stopping this, but clearly not) 13… exd5 14 e5 (I also underestimated 14 exd5 Nxd5? 15 Nxd5 Bxd5 16 Ne5!) 14… Ne8 15 e6:
A couple things to note: so far my guess-the-move score is at 100%, mainly because black hasn't had many good options (try moving the black knight to a square other than e8 on move 14, for example). Even though I remembered that Pragg played e5-e6 at some point in this game I was shocked to see it so early. I also remembered that Gukesh played 15… f5 in response, but I didn’t understand why he couldn’t have simply played 15… fxe6 16 Nd4 Bc8 (the computer endorses this idea, as well as 15… c6, which was played in the one game that reached this position in my database).
Pragg continued 16 Ne5, and here’s where Gukesh and I started to diverge. My candidate moves were … Bf6 (refuted easily by 17 Nxd5!), … c6, and … Nf6. I settled on 16… c6 because I wanted the flexibility of deploying the Ne8 to c7 (to round up the e6 pawn) or d6 (to exchange the white knight if it came to f7). The problem, and perhaps you noticed this right away, is 17 Nd7, trapping the rook and threatening Bb6. Gukesh played the more accurate 16… Nf6, and Pragg’s 17 Qc2 (another unexpected move, and not a good one, as it turns out) brings us to the position below:
We’ve reached a critical moment; there are many paths forward for black, some better than others. I looked at the following, from roughly worst to best:
17… Ne4? 18 Nxe4 fxe4 19 Bxe4 and white wins.
17… Qd6? 18 Qxf5 Ne4 19 Nf7 and again, black is toast.
17… g6 18 Nf7 Qc8 (better is 18… Rxf7!) 19 Nxd5! Nxd5 20 Bxd5 Bxd5 and here I missed the computer’s move 21 Qc3!
17… c5! 18 Qxf5 d4 19 Bxb7 Nxb7 with some further branches, none of which seemed very promising for white.
I remembered enough about the future pawn structure to know that Gukesh didn’t play 17… c5, but I chose it because it was a dynamic solution to black’s problems and I didn’t see an outright refutation. Gukesh played 17… c6, but in this rare instance the computer picked my move over his.
The game continued 18 Qxf5 Qe8 (I guessed 18… Bc8, since Bc8 followed by rerouting the Na5 through b7 is the most obvious reason to play 17… c6, but I got the move order wrong) 19 Nf7 Bc8 (a correct guess, although the other move I considered, 19… Rxf7, is also fine) 20 Rfe1 Nb7 21 Bg5:
It feels like black is one move away from consolidating, but the further I analyzed the more dangerous black’s position appeared. I considered:
21… Bxe6? 22 Qxe6 Rxf7 23 Bxf6 gxf6 24 Nxd5! cxd5 25 Bxd5 with a devastating attack.
21… Nd8 22 Bxf6 Bxf6? 23 Nxd5! Bxe6 24 Nxf6+ gxf6 25 Nh6+ with a significant advantage.
21… Nd6 22 Qe5! Nxf7 23 exf7+ Rxf7 24 Nxd5! cxd5 25 Bxf6! gxf6 26 Qc7! and black’s position is on the edge of disaster.
Therefore I arrived by process of elimination at 21… Ra7 (get that rook off the long diagonal), the move Gukesh played in the game. It’s a nice human solution, but both Pragg and I missed a brilliant resource that the computer is only too happy to demonstrate: 22 Rxd5! cxd5 23 Bxf6! Bxf6 24 Nxd5 with many looming threats, the nastiest of which is Be4 and Qxh7#. Fortunately for Gukesh (and me, since I got to guess some more moves that were entirely forced), the game continued with a less effective version of the sacrifice on d5: 22 Bxf6 Bxf6 23 Bxd5 cxd5 24 Nxd5:
I was pretty sure that I remembered the bishop dropping back to e7 in this position, but first I had to figure out why 24… Bxb2 doesn’t work. My idea was 25 Qg5 Bxe6! 26 Ne7+ Kxf7 27 Rxe6! Kxe6 28 Re1+! Kd7! (apparently the only winning move) 29 Qg4+ Kc7 30 Nd5+, picking up the queen, but the computer laughs and says that white is still much worse here; black’s exposed king and scattered pieces apparently don’t count for much. Instead the machine gives 25 Qc2! Bxa3 26 Nc7 Qe7 27 Nd5 Qe8 with a repetition.
Gukesh chose 24… Be7, leaving Pragg one last chance to save the game: 25 Nh6+! Kh8 26 Nf7+ Kg8. If instead 25… gxh6 26 Qg4+ Bg5 (26… Kh8? 27 Qd4+ picks up the rook) 27 Qd4 Ra8 28 e7! looked winning, but the once again the computer finds a way out: 27… Nd8! giving a rook for the e-pawn. Pragg either didn’t see the draw or didn’t want it and played 25 Qg4 Nd8 (pretty easy to guess, since both 25… Nd6 and 25… Bxe6 drop material) 26 Nxd8 Bxd8 (more natural than giving up the queen with 26… Qxd8 27 Nf6+) 27 Qd4 Rb7 28 Re4:
White’s position is superficially menacing, but black’s got everything covered. From here on out, guess-the-move is a good exercise to make sure we avoid blundering or giving white too much play as we try to bring home the full point. Gukesh and I continued 28… Bf6 (driving the queen away, as white has to keep the knight for any swindling chances) 29 Qe3 Be7 (the simplest way to stop the e-pawn) 30 h4 Qc6 31 h5 Bc5 32 Qg5:
Here, with Pragg on the verge of defeat, fully armed with the knowledge that the final move of the game was … Rxf2, I calculated 32… Rxf2 33 e7 Re2+? 34 Kh1? Rxe4 35 Nf6+ Kf7 and wins, missing that 34 Ne3! lets white back in the game. Gukesh had the sense to finish things up calmly: 32… Be6 33 h6 Rxf2 and white resigned.
What a fun game! There’s nothing like trying to figure out plans and tactics to make you appreciate the skill of these top guys. The moment that I’ll remember from this game is black’s choice on move 17. Gukesh opted to play conservatively, betting that his pieces could slowly swarm the advanced e-pawn, but in doing so he opened himself up to some nasty tactics. The more dynamic approach, 17… c5, would have done more to swing the initiative to black. That said, Gukesh played fairly conservatively in many of his victories and this approach paid off: he made very few mistakes along the way and took advantage of most of his opportunities. I’m excited to see how he fairs in the world championship match against Ding Liren later this year.