The Return, Part Four

In which my bad chess habits make their glorious return.

If you’ve read the first three installments in this series about my return to tournament chess, you’re probably aware that I’ve been aiming for positions with clear strategic ideas, eschewing opening theory and complex tactics. This has been an intentional choice, with the goal of ridding myself of a number of bad habits:

1. Seeking out complicated opening variations despite having only a superficial knowledge of the underlying ideas or theory.

2. Making decisions based on hubris about my tactical ability rather than objectively assessing risk in the middlegame.

3. Struggling to transition between the middlegame and endgame, particularly when it comes to converting a winning ending into a full point.

By shifting to offbeat but basically sound opening lines and emphasizing positional play I’m hoping to work on some of these weaknesses while preserving enough time and energy to play the endgame well. That didn't happen this week—all three bad habits are fully on display—and I was lucky to emerge unscathed. On the other hand, this game is far more interesting and entertaining than any of the games from the first three rounds, so dig in and enjoy the jumble of tactics and missed opportunities that follows.

William Gray (1918) - Andy Lee (2328), Tuesday Night Marathon (4), 7/30/24

1 Nf3 b6. If it works with the white, why not give it a go with black? I spent some time preparing for 2 e4 and 2 c4, but of course I didn’t bother to think about the variation the appeared on the board: 2 d4 Bb7 3 c4 e6 4 a3 Nf6. It’s hard to avoid opening theory, and we’ve just moved from a position that has occurred a few hundred times in the database to one that has occurred more than 6,000 times: the Petrosian Variation of the Queen’s Indian Defense. This would be a reasonable choice if I knew something about this line, but I don’t. Black most frequently transposes into a Dutch after 4 … f5, but I was worried about 5 Nc3 Nf6 6 d5, killing the bishop. 5 Nc3 Ne4:

Here we go, something stranger than the almost automatic 5 … d5 6 cxd5 Nxd5, and, fortunately for me, not objectively worse¹. What theory exists from this point says that white usually plays 6 Nxe4 Bxe4 7 Nd2 Bb7 8 e4 with the big center, but with a pair of minor pieces exchanged black doesn’t have to worry as much about running out of space and has the kind of flexible structure I’ve been aiming for all tournament long. White instead played 6 Qc2, and rather than playing the sensible 6 … Nxc3 7 Qxc3 Be7 8 g3 Bf6 (again, maximum positional flexibility), I conflated this variation with the main line (Bad Habit #1), and played the anti-positional 6 … d5?! 7 cxd5 exd5 8 Bf4 Bd6 9 Bxd6 Nxd6 10 e3 00 (I wanted to play 10 … Nd7 but was concerned that after 11 Nb5! Nxb5 12 Bxb5 00 13 Bxd7 Qxd7 white has a stable long term advantage that will be hard to break) 11 Bd3:

The impression that I get from this diagram is that black has been playing Fischer Random and white has been developing his pieces normally from the traditional starting position. The knight on d6 is usually good in these kinds of Carlsbad structures, but what about the passive Bb7? The lack of defenders on the kingside?

There are a few different ways to deal with the threat of Bxh7+, most obviously 11 … h6, but here comes Bad Habit #2 with 11 … g6?, creating the perfect conditions for a white attack: 12 h4 c5 (I spent some time on 12 … Qe7 13 h5 Ne4, but 14 Ne5! creates serious problems) 13 h5! and now I was sad to discover that 13 … c4 fails, not to 14 hxg6? cxd3 15 gxh7+ Kh8, but to 14 Bxg6! hxg6 15 hxg6 Qf6 16 g7! Qxg7 17 Rh3, winning material:

This was the first inkling I had that I might be getting crushed, but since pawns don’t move backwards there’s little choice other than to dig in and try to make the attack as difficult to conduct as possible. I played 13 … Qf6 to support g6, but after 14 hxg6 hxg6 15 Ne5 the black position is critical:

What’s to be done? 15 … c4 is met by 16 Bxg6! fxg6 17 Rh6. 15 … Nc6 drops the d-pawn or the exchange. 15 … Kg7 16 Rh6? is not good due to 16 … cxd4! (16 … Kxh6? 17 Ng4+) 17 cxd4 Nc6, but simply 16 000! brings the last white piece into play. All that’s left is some counterplay in the center: 15 … cxd4 16 exd4 Re8. An unhappy improvisation. I’d originally planned 16 … Kg7 but realized at the last second that since I’d already traded on d4 white can play 17 Qd2 and win on the spot. 16 … Ne4 17 Nxe4 dxe4 18 Bxe4 Rc8 19 Qd3 Bxe4 20 Qxe4 Nc6 was my bailout option, but I couldn’t quite bring myself to sacrifice a pawn, and the computer refutes it comfortably anyway: 16 … Ne4 17 Nxe4 dxe4 18 Qc7! Bd5 19 Ng4! Qxd4 (19 … Qe6 20 Be2! with either Qf4 or Qh2 next) 20 Nh6+ Kg7 21 Nf5+! is really nice:

The point is that 21 … gxf5 22 Qg3+ Kf6 23 Rh6+ Ke7 24 Qd6+ Ke8 25 Bb5+ is curtains. The cool Qc2-c7-g3 board geometry isn’t easy to see, but it’s a good reminder that even the positions that seem safe are actually on the brink of disaster.

The game continued 17 000 (17 Kf1 is possible, although not really necessary. During the game I thought I had a clever refutation: 17 … Ba6 18 Nxd5 Qf5 19 Nc7 Rxe5, only to notice 20 Nxa6, and all of black’s pieces are hanging.) 17 … Nc6 (Sacrificing a pawn and inviting all kinds of craziness, but black has to complicate matters. Note that 17 … Qf4+?! 18 Kb1 Qxd4 19 Nxg6! wins easily.) 18 Nxd5 Qg7 (the queen has to stay on the kingside: 18 … Qd8? 19 Qd2! is impossible to defend) and we’ve reached a critical moment. Take a look at the diagram below and try to find white’s winning move:

There are a lot of tempting knight hops. Let’s start with the worst and work our way up to the best:

• 19 Nc7? Nxd4 20 Nxe8 Qxe5 and black is winning.

• 19 Nxc6?! Bxc6 20 Qxc6 (20 Nc7 Rac8 21 Qxc6 Rxc7! transposes) 20 … Rac8 21 Nc7 Rxc7! 22 Qxc7 Rc8 23 Qxc8+ Nxc8 is a roughly equal ending, in which white’s material advantage is counterbalanced by the strong queen and knight duo.

• 19 Nd7?! Re6 was played in the game; we’ll return to it later.

• 19 Ng4! is similar to what was played in the game, but just a little more crafty. I assumed that 19 … Nxd4 20 Ngf6+ Kf8 21 Nh7+ Qxh7 held everything together, but it turns out that we both missed 21 Nd7+! Kg8 22 N5f6+, winning the black queen. This means 19 … Re6 is forced, and now 20 Nc7 picks up the exchange.

The problem with white’s choice in the game is that after 20 Nc7 Nxd4 21 Nxe6 fxe6 the Nd7 is vulnerable and black has some compensation. With the knight on g4 white’s attack still rages. White took a different approach, playing 20 Rde1?! Rae8 (Forced, since both 20 … Rxe1+? 21 Rxe1 Qh6+ 22 Kb1 Kg7 23 N5f6 and 20 … Nxd4? 21 Rxe6 fxe6 22 N5f6+ Kf7 23 Qa4 are quite horrible. I wanted to make the extravagant 20 ... Nb4? work, but 21 Rxe6! refutes.) 21 Rxe6 Rxe6:

Another inflection point in this wild game; we were both running low on time as well. I can understand white’s lack of enthusiasm for 22 Nf4 Re7 23 d5 Ne5 24 Nxe5 Qxe5; although the position is balanced, black’s position is much easier to play. 22 Qc3 Nxd4 23 N5f6+ Rxf6 24 Qxd4 is another interesting idea that I glanced at during the game.

The computer has a more compelling suggestion, again beyond the reach of most human imaginations: 22 N5f6+! Rxf6 23 d5! The point is that after almost all knight moves white has Qc7, resuming the attack in light of black’s badly tangled pieces and back rank problems. The computer assures me that the only correct defensive move is 23 … Ne5!, returning the piece immediately to gain a little more time.

My feeling at the board was that the tide had already turned; my opponent confirmed this by playing the shaky 22 Nb4? Nxb4 23 axb4 Qxd4 24 Rd1 Qxb4 and black now is in control. After 25 Qc7 I wasted a lot of time working out the details of 25 … Nc4² instead of playing the obvious 25 … Kg7 right away:

I’d already worked out 26 Qd8 Nc4! 27 Qf8+ (only move) 27 … Qxf8 28 Nxf8, but here the extra time would have come in handy; perhaps I would have stopped and noticed 28 … Rc6! 29 Be4 Rc7! 30 Bxb7 Ne3+ 31 Kd2 Nxd1 32 Kxd1 Rxb7. Nothing can save the white knight; black will emerge up a full rook. Instead, it was full steam ahead: 28 … Kxf8?! 29 Bxc4 Rc6 30 b3 b5 31 Rd7:

I’d seen this position quite a ways back in my analysis, which ended before Rd7 with the evaluation: black’s up a piece, the game is over. The problem is that black isn’t going to have many pawns remaining, so instead of moving the bishop, 31 … bxc4! seems like the most pragmatic solution. After 32 Rxb7 cxb3+ 33 Kb1 Rc2 34 f3 Rxg2 35 Rxb3 black is up two pawns and should win without any particular trouble.

A quick reminder about Bad Habit #3: this is where I run out of energy and focus after a long tactical melee and struggle to convert winning positions. I had the idea that I was winning a piece and the game stuck in my head, so I continued 31 … Bc8?! 32 Rxf7+ Ke8 33 Rxa7 bxc4 34 bxc4 Rxc4+ 35 Kd2:

There are some worthy options: 35 … Rg4 to force the white pawns to dark squares; 35 … Re4 to cut off the white king. My brain froze up instead; I knew that there was a lot still to do to convert this position and I couldn’t begin to figure out how: 35 … Rc6 36 Ke3 Rf6 37 f3 Bd7 38 g4 (white is aggressively placed to try to trade the last black pawn) 38 … Kd8 (38 … Ke7 39 f4!) 39 Ra5 Ke7 40 Ra7 Kd6 41 Ra6+ Bc6?! (an admission that I am out of patience, there’s no reason not to repeat moves, get a little time, and try to find something better) 42 f4 Ke6+ 43 Kf2! Kc5 44 Ra1 (for now, white chooses not to suffer in the theoretically drawn but very unpleasant R+B vs. R ending after 44 f5 Rf6 45 Ra7 gxf5) 44 … Re4 45 Rc1+ Rc4?:

I blocked the check as soon as the following thought popped into my head: a blunder! — now he’s got to trade rooks or the f-pawn hangs. The problem is 46 Rxc4+ Kxc4 and the careful 47 Kg3! Kd4 48 Kh4 Ke4 49 f5 liquidates the final black pawn. But after a valiant resistance, playing on the increment, my opponent made the last mistake: 47 f5? g5! White Resigns, the g-pawn survives.

That game doesn’t do wonders for my ego, but it was both exciting and interesting. Somehow I’m the last perfect score remaining the event, so the games will get more difficult the rest of the way. Time to do some endgame practice before next week rolls around!

1. Although I didn’t know the theoretical assessment of 5 … Ne4 during the game and blamed it for my later troubles.

2. It doesn’t work: 26 Bxc4! Rc6 27 Nf6+ Rxf6 28 Bxf7+! and white holds.