Mishra-Pawar, World Junior Championship
Chess autopsy: what was the mistake that took down the top seed?
The World Junior Championship ended just a day or two ago; the list of winners of this event includes many of the best players of the past seventy years. Despite the emergence of many highly rated players age 20 and younger, this year’s event was perhaps a little less strong than it has been in the past: guys like Eragaisi, Praggnanandhaa, and Abdusattorov are looking for tournaments with their peers in the world top ten, and Gukesh is busy training for his world championship match! This meant that Abhimanyu Mishra, who became the youngest grandmaster in history at a little under 12 1/2 years of age (surpassing Sergey Karjakin’s 2002 record by a couple months), was the top seed, youthful even by World Junior standards, at only fifteen years of age.
By the time I tuned in, after round five, Mishra wasn’t to be found at the top of the leaderboard. What happened to him? It didn’t take long to find out—he was the victim of a first round upset at the hands of 2132 rated Harshit Pawar, one of many Indian juniors representing their home country in Gandhinagar. No matter how underrated Pawar is, he pulled off an impressive feat by winning with the black pieces after being more or less lost right out of the opening.
Before reading the annotations below, I recommend skimming through the whole game with the following questions in mind:
When does the position shift from advantage white to advantage black?
At what point is white’s position beyond saving?
I split the game into four chunks for easier digestion. My hypothesis was that black gained the advantage in the third chunk and was only winning for good in the fourth. This evaluation wasn’t particularly accurate but perhaps sheds some light on why Mishra also misevaluated the position and lost such a brutal game against a much lower-rated opponent.
Chunk One: A Dubious Opening
Move three marked the start of the Bogo-Indian Defense, an opening that does not have a reputation for fireworks: 1 d4 Nf6 2 c4 e6 3 Nf3 Bb4+ 4 Nbd2 b6 5 a3 Be7 (This is a strange move, played about 1% of the time in comparison to the 99% favorite, 5… Bxd2+, which prevents white from establishing a huge center.) 6 e4 Bb7 (Eight previous games reached this position in the lichess database; white won all eight of them.) 7 Bd3 d5 8 cxd5 exd5 9 e5 Ne4 (If black could maintain a knight here this variation would be a whole lot more playable.) 10 Qc2 Nxd2 11 Bxd2 c5 (An earlier game continued 11… g6 instead, but it didn't go any better for black.)
Normally I would wait until the kings were castled before declaring the opening complete, but that’s not going to happen. Black’s king can’t safely go to the kingside, and white becomes (understandably) possessed by a furious bloodlust and can’t be bothered. Let’s sum up white’s advantages: ahead in development, excellent attacking chances on the kingside, potentially much safer king. Any of these factors on their own might make for a small edge; combined they mean that white is on the verge of winning.
Chunk Two: White Goes For the Kill
Mishra decided to put black to the sword right away: 12 e6! is really excellent, highlighting black’s weakened kingside, particularly on the light squares. The game continued 12… fxe6 13 Ne5?
The knight hops ever so naturally into the square vacated by white’s pawn and takes aim at d7, f7, and g6. It’s so thematic that it’s hard to believe that this move squanders white’s advantage. The problem is that white’s position is so very, very good. There are a lot of good looking moves that all appear to bring us closer to the desired win, but they aren’t actually equivalent; some turn out to be much stronger than others. In this case some combination of Bxh7 and 00 continues the assault on the light squares and brings white’s rooks into the game. On the other hand, Mishra’s knight move doesn’t finish white’s development and allows his opponent counterplay against the d4-pawn.
Black continued with a move that I assumed was forced, 13… Bf6, but it turns out that 13… Nc6 is fine: 14 Bg6+ Kf8 15 Nf7 Nxd4 16 Qd3 Qe8! 17 Nxh8 hxg6 18 Nxg6+ Kg8 19 Nxe7+ Qxe7 and black’s missing exchange is no problem:
One of the themes of this game is that black can give up an exchange without being worse; note here the monster Nd4 and the mobile pawn center versus white’s drowsy rooks.
Back to the game, where white continued playing the most natural attacking moves: 14 Bg6+ (14 Qa4+ or 14 Bb5+ get the nod from the machine, as either would have forced 14… Kf8) 14… Ke7 15 Bxh7 Nc6 16 Ng6+ Kd7 17 Nxh8.
Without the help of the computer, I was a believer in white’s position. The d4-pawn is falling, but black will have only a single pawn for the exchange, and his king is oddly placed on d7. But take a look at the previous diagram. The comparison is all in black’s favor, as he has maintained his strong dark-squared bishop, whereas white has no good way to open up the black king. The advantage has swung to black both on the board and psychologically, as it’s very difficult to admit (or even notice) that instead of winning smoothly it’s time to think about how to fight for equality.
Chunk Three: Black’s Counterplay Arrives
We get an important moment right away: 17… Nxd4 18 Qd1. I preferred 18 Qd3 in my analysis, aiming to bring my king to safety after 18… Qxh8 19 00. The computer has a fun suggestion here, 19… g5!, that aims to maximize black’s advantage:
The point is that black gains space on the kingside to reorganize the heavy pieces: Qg7 and Rh8 are coming, and the g-pawn can be potentially used as a lever against h2-h3. The bishop on h7 is really uncomfortable, so Mishra decided to reserve more options for its escape on the diagonal.
There followed 18… Qxh8 19 Bg6 (19 Bd3 is the sort of move you prefer if you want to castle, but it’s not clear that white’s king is going to be safe on the kingside, and now 19… c4 now comes with tempo.) 19… Ba6 20 Be3 (I looked at 20 Bc3 to make black think twice about 20… Nc6, but either Qh6 or Be2 are rather strong.) 20… Nc6 21 Rb1:
When I first saw this game, I suspected that this little rook move was the start of white’s undoing. The computer reveals that black’s advantage predates Rb1 by a fair bit, but white’s utter disregard for black’s initiative is shocking: how is there time for such a move? That said, what I tried showed a different kind of misunderstanding: 21 Bd3 Bxd3 22 Qxd3 Bxb2 23 Rd1, and although white is a little more active, it’s not healthy to keep giving up pawns. The computer suggests a totally different solution that is difficult to see because it cuts against the grain of the position: 21 Qc2 Qh4 22 000!
Castling queenside is an admission that white’s king is never going to be fully secure. But having accepted this truth, white can use his king to guard b2 (instead of condemning his rook to a life of passivity). Now black’s Ba6 looks a little silly, there’s no attack to worry about on the h-file, and white’s heavy pieces can be deployed to the center, quite a transformation!
Black’s problems grow over the next few moves. Alex Yermolinsky wrote about the difficulty of reversing a negative trend in his excellent book The Road to Chess Improvement, and it’s oddly comforting to see that negative trends affect the play of GMs and amateurs alike . After 21… Qh4 22 Bd3 Bxd3 23 Qxd3 c4 24 Qe2 Rh8 white still cannot castle, whereas the black center pawns are already rolling and the threat of d4-d3 looms large:
Chunk Four: Rated M for Mature
Can white’s position still be saved? It’s certainly uncomfortable, and if I were playing black I would expect to win. That said, I think that white can still try to pose problems, although Mishra did not manage to do so. He played 25 h3, the other move (along with 21 Rb1) that initially seemed to me to be much too slow. That said, what else is to be done about the problem of getting the white king to safety? The pawns continued to roll in the center: 25… d4 26 Rd1 Ke7.
This position represents, at least in my analysis, the final critical moment. White has to choose between muddying the waters or passively acquiescing. Mishra played 27 Bc1, essentially sealing his defeat. I wanted to give up a piece to change the character of the position: 27 00 dxe3 28 fxe3. It’s a little silly to hope for salvation here, but if black doesn’t find either 28… b5 29 Qf3 Rc8 or 28… Ne5 white starts to creep back into the game. The computer, as usual, has a far better suggestion: 27 Bd2! d3 28 Qf3, and now the only move that keeps a serious advantage is 28… Qh5!
It’s a non-obvious way to deal with the threat on the Nc6, both because pinning the white queen to mate on e2 is a little unusual and because forcing a trade of queens goes against the narrative that black will win by attacking the white king. The nature of black’s advantage is such that there are paths to victory in both the middlegame and endgame.
Back to the game; there’s not much more to say: 27… d3 28 Qf3 (I wanted to try 28 Qe3 when 28… Nd4 is fine, but so is 28… Bd4 as long as black realizes that these endgames are all winning.) 28… Nd4 29 Qb7+ Kd6 (the king is perfectly safe here) 30 Kf1 Rf8 31 Qxa7 (no more pretensions of resistance) 31… Nc6 32 Qa6 Bd4 33 g3 Qxg3 34 Be3 Rxf2+ White Resigns.
Final Thoughts
I saw some of the same problems that plague my own chess affect Mishra in this game. He initially had trouble choosing between a couple variations that looked winning right out of the opening; as the position turned against him, he failed to readjust to fight for equality or seek counterplay. The negative trend swallowed him up, and he went down without much of a fight.
In addition to the objective superiority of black’s position throughout much of the middlegame, black’s moves were much easier to play than white’s. I struggled to find improvements to Mishra’s moves and was struck by how much better the computer was at finding non-obvious ways to fight back. Obviously the computer is better than me in all sorts of positions, but the problems facing white in this game seemed particularly suited to the kind of petite tactics, those precise two or three move calculations, that the computer is so good at.
I’d attribute white’s loss to the fact that he cashed out early and won a very small amount of material in exchange for some long-lasting positional problems. You can play for a long time down a pawn or the exchange in the right kind of middlegame without being affected by the material deficit. This game is an excellent example of the notion that positional gains and the initiative often trump material concerns, and a call to sacrifice material early and often!