About a decade ago a friend proudly showed me his new wooden board and set. There was just one problem: it had shipped without any knights. Rather than find some bottle caps or poker chips, we decided that this called for an experiment: what is chess like if you play without knights? We quickly learned that the no-knights variant is not very fun. The long-range pieces were exchanged for each other quite rapidly, space advantages weren’t worth much, and neither of us could create useful imbalances.
The game I want to analyze today does not suffer from a dearth of knights. Rather, it epitomizes the chaos that only the knight can bring. It pits one of the stars of the new generation of teenage super-GMs, Nodirbek Abdusattorov, against Sweden’s top GM, Nils Grandelius. Here’s the position after move 20, when the waters are still calm:
Get out a board or check out this link to the game, because what happened next is a lot of fun: 21 Nh4 Nf6 22 d5 Bxg2 23 Qxg2 c6 24 Nd4 Qb6 25 Nhf5 cxd5 26 Ne6 Rf7 27 Rf1 Nd3 28 Nexg7 dxc4 29 Nh5 Kh7 30 Ne7 Qd4 31 Nd5 Nxd5 32 Rxf7+ Kh8 33 Raf1 Ne5 34 Rxb7 Ne3 35 Rf4 Qd3 36 Rh7+ Kxh7 37 Qb7+ Kg6 38 Qg7+ Kxh5 39 Rh4#:
That’s not the typical finish to a game played by a couple of strong GMs!
My approach these days to reviewing the games played in top-level tournaments is not about the latest opening novelty or checking on typical middlegame plans. I just skip past all the draws and skim through the decisive games until I find something like this, a game that makes you feel (apologies to Emily Dickinson) like the top of your head has been taken off.
After enjoying the aesthetic finish to this game, I was curious to know more about it. How did white transform an equal position (where if anything, black was pressing) to a crushing attack? When did black err? Was there a critical moment (or moments) where the game could have been saved?
To do this analysis I split up the remaining 19 moves of the game into four chunks and worked backwards, starting with chunk four:
Chunk One: 21 Nh4 Nf6 22 d5 Bxg2 23 Qxg2 c6 24 Nd4 Qb6 25 Nhf5 cxd5 26 Ne6
Chunk Two: 26… Rf7 27 Rf1 Nd3 28 Nexg7 dxc4 29 Nh5 Kh7 30 Ne7
Chunk Three: 30… Qd4 31 Nd5 Nxd5 32 Rxf7+ Kh8 33 Raf1 Ne5 34 Rxb7 Ne3 35 Rf4
Chunk Four: 35… Qd3 36 Rh7+ Kxh7 37 Qb7+ Kg6 38 Qg7+ Kxh5 39 Rh4#
Before getting into the analysis that follows, try making an educated guess about where the critical moment(s) will be found: early on, in chunk one? A mistake towards the end, in chunk four? Or somewhere in between?
Chunk Four: The final combination
White’s last move was Rf1-f4, attacking the black queen and bringing the rook into play along the fourth rank. This would normally be a major decision, whether or not to trade queens, but it turns out that neither choice brings black any joy.
Grandelius chose to keep his queen, playing 35… Qd3 which seems designed to defend h7. Ironically, it loses to a sacrifice on that very square: 36 Rh7+! Kxh7 (also bad is 36… Qxh7 37 Qxa8 Qg8 38 Rf8) 37 Qb7+ with forced mate to follow. That leaves trading queens as the only reasonable way to try to improve black’s play, but both 35… Qxf4 36 gxf4 Nxg2 37 Nf6 and 35… Nxg2 36 Rxd4 Rf8 (to stop Nf6, as in the previous variation) 37 Kxg2 leave white with forced mate or up a rook. By move 35 black was already lost.
Chunk Three: The exchange falls
This section begins just after white’s clever Nf5-e7. Let’s see if this knight sortie is more shocking or substantive:
Grandelius played 30… Qd4, protecting the Nf6 and centralizing the queen. It’s hard to find fault with this, given that the alternatives fail to save the exchange. Both a) 30… Rxe7 31 Nxf6+ Kg7 32 Nd5 and b) 30… Raf8 31 Rxf6 Rxf6 32 Nxf6+ Rxf6 33 Nd5 leave white comfortably ahead. Abdusattorov was not to be denied, and after 31 Nd5 (anyway!) both 31… Nxd5 32 Rxf7+ Kh8 33 Raf1 (as in the game) and 31… Nxh5 32 Rxf7+Kg8 33 Raf1 are bad for the same reason: white is up material and still attacking. 31… Qxd5? 32 Nxf6+ is noteworthy only for losing on the spot. There are only two places left to find improvements. Instead of 33… Ne5 black could have traded queens, but 33… Qxb2 34 Qxb2 Nxb2 35 Rf6! leads to one mate or another. After 34 Rxb7 activating the rook with 34… Re8 makes sense, but black’s position is too creaky to withstand the assault:, 35 Qc2 (not 35 Rf6? Ng4+! 36 hxg4 Nxf6) 35… Nd3 36 Qd2! Ne3 37 Rff7 Qe4 38 Qxe3! Qxe3 39 Rh7+ Kg8 40 Rbg7+ Kf8 41 Rh8#:
This chunk concludes with 34… Ne3 35 Rf4. I think you could make the case that black has made some subpar choices that will help white finish things quickly, but there was no saving move here.
Chunk Two: Under pressure
As we move backwards again, the white center is falling, but with after white’s last move, Nd4-e6, black’s kingside is starting to look very shaky:
The game continued 26… Rf7 (what else?) 27 Rf1 Nd3. I wanted to avoid some of the tactics that appeared in the game by playing the prophylactic 27… Kh7, but 28 Nfxg7 is now very strong. The computer has a much better idea: 27… Rc8! 28 Nexg7 Rxc4! with sufficient counterplay. It should come as no surprise that activating black’s undeveloped rook greatly improves the quality of his position. That’s one critical moment; now all that’s left is to see if black had any other other chances to save the game rather than 28 Nexg7 dxc4 29 Nh5 Kh7. One idea is 28… Rxg7 29 Nxh6+ Kh8 30 Rxf6 dxc4 where I struggled to find anything promising. The computer says that you have to keep sacrificing: 31 b3! cxb3 32 Qe4! b2 33 Qh4! with a position worthy of a diagram:
The threat of Nf7+ and Qh8# leaves black no time to take on a1, so white’s winning, but this would have been more difficult than the game continuation. There are other ideas instead of … dxc4 as well, such as 28… Kh7 29 b3 dxc4 30 Qd2 Ng8! Therefore, I think it’s safe to say that the game was lost on moves 27 and 28.
Chunk One: A double-edged battle
We’ve found a couple critical moments where black could have saved the game, but one question lingers. Black’s position looked so solid at move 20—is it really possible that he got into trouble so quickly without making a mistake? Let’s take a look at the position after: 21 Nh4 Nf6 22 d5 Bxg2 23 Qxg2 c6:
When I initially skimmed through this game, I wondered if black should have tried to keep his light-squared bishop. As later developments demonstrate, the white knights were delighted to dance around on f5, e6, h5, and d5. But upon further reflection (aided by some computer ideas), I’m really impressed by black’s play to this point. He’s fighting back in the center, activating his queen, and trying to win an unbalanced and complicated position. The game continued 24 Nd4 Qb6 25 Nhf5 and here, rather than continuing 25… cxd5 26 Ne6, Grandelius could have played in the same style with 25… g6! 26 Nxh6+ Kh7. The point is that both white knights are hanging, and white has a difficult decision to make: either invest some material (27 Ne6) or trying to hold everything together with 27 Qd2, which runs into 27… Ne4 28 Qe3 Rf2+ 29 Kg1 Re2!
I don’t think that you can blame Grandelius for missing this sequence, but it does validate his earlier play and my initial feeling that black was doing well at move 20. For that matter, he was probably happy with his decisions in the game until 30 Ne7 appeared on the board and his position started to unravel.
I’m sure there’s a lot more lurking in shadows of this game. Even with just the notes here, it’s a good reminder that the critical moments of a chess game usually appear earlier than you might think and that the vast majority of games are decided by who calculates more deeply and accurately. A triumph for Abdusattorov, who also ended up winning the tournament, and a tough loss for Grandelius, who fought hard but couldn’t quite hold on as the position got more irrational and challenging.