Reaching 700: the game where I didn't punish the blunders
My rating first went above 700 with game #2 of my journey: on January 10, 2024, the rating Chess.com shows for me in that game is 781, after the 631 of game #1. I won it with Black against jon_maik (697), by checkmate on move 36. It's also the first game where Stockfish shows me how to punish blunders in chess by showing what I didn't do: my opponent blundered twice in three moves, and each time I gave back about half of the advantage.
This is the second story of my journey from 631 to 1385. The full game is public on Chess.com. All evaluations come from the engine analysis of that game, always from White's side: a minus sign means Black (me) is better.
The game in numbers
| Date | January 10, 2024 (game #2) |
| Colour | Black |
| Opponent | jon_maik (697) |
| My rating | 781 (631 in game #1) |
| Time control | 15|10 |
| Opening | Van 't Kruijs Opening (1.e3) |
| Result | Win, checkmate on move 36 |
| Accuracy (Stockfish) | 92.6% me, 86.9% my opponent |
| My inaccuracies / mistakes / blunders | 1 / 3 / 0 |
| My opponent's | 2 / 2 / 2 |
I started this game at 16:53 UTC, four minutes after game #1 ended, and it lasted about half an hour. A 92.6% accuracy with zero blunders looks like a clean game. It wasn't. All three of my mistakes came right after my opponent's errors, and that is what this page is about.
A quiet opening
My opponent opened with 1.e3, a modest first move that the opening books call the Van 't Kruijs. I answered 1...e5, and after 2.d4 I played 2...d5, which Stockfish marked as my one inaccuracy (it preferred taking on d4). White took my pawn on e5, I got it back on move 6 after trading my bishop for his knight on f3, and his pawns on the king's side were doubled. Twelve moves in, the position was about level. On move 12 my knight jumped to b4, where it attacked his queen on d3.
Blunder 1: the queen goes to c3
My opponent answered the knight with 13.Qc3. From c3 the queen attacked both of my knights, the one on b4 and the one on f6, which probably looked active. Stockfish called it a blunder: from -0.06 to -1.95. The engine's move for White was the modest retreat to d1.
I played 13...c6, and the engine counted it as a mistake: the evaluation went back to -0.76. Its move was 13...Be7, bringing out my last minor piece. The engine's line (its idea, not the game) continued 14.Qd2 Ne4 15.Qe2 c5, with Black clearly better. One move later the game gave me that bishop move anyway, but under different circumstances: White checked my king with 14.Qe5+, and I blocked with 14...Be7.
Blunder 2: the knight goes to a3
On move 15 White developed his knight to the edge, 15.Na3, and Stockfish called it a second blunder: from -1.24 to -5.75. That's not "slightly better" any more: the engine had Black winning. And I castled.
Castling is the move every beginner is told to play, and here it was a mistake: the evaluation went to -3.04. The engine's move was 15...c5. Its line was 15...c5 16.Ba2 O-O 17.Qg5 Nxa2, and the idea is visible on the board: one more step, to c4, and the pawn would attack White's bishop on b3. In the engine's line the bishop steps back to a2 first, and my knight takes it there. I'm not going to invent a deeper explanation than the engine gives. What the numbers say is simple: the safe move kept me better, the pawn push would have kept me winning.
The double attack that wasn't the point
Stockfish wanted 16...c5 again on the next move; I played 16...g6. Then White put his queen on g5, a mistake (from -3.43 to -5.78), and I finally found something that looked like punishment.
My knight jumped to e4. It attacked the queen on g5, and by moving it opened a line for my bishop on e7, which now attacked the queen too. A double attack, two of my pieces hitting the same target at once, sounds strong. But the target was a queen, and a queen can just leave: 18.Qh6. Stockfish rated 17...Ne4 my third mistake (back to -3.55) and wanted, for the third move running, the pawn push 17...c5.
So three moves in a row, at moves 15, 16 and 17, the engine's first choice was the same quiet pawn move, and three times I played something else: castling, a pawn move to g6 and a flashy attack on the queen.
How the game was won anyway
My opponent kept making mistakes, and the clock started to matter. After move 20 he had 6:47 left; after move 25, 3:06. I still had over ten minutes. The game opened up: I took a pawn on e4 on move 20, traded my knight for his bishop on b3, and on move 25 my knight took his knight on a3. On move 27 he put a rook on c5; I took his bishop on d4 with check and then the rook. The finish was a long walk for his king: my queen checked it from d3, d2 and f2, a pawn check on f5 drew his last rook into a capture, and the queen chased the king to g5. On move 36, 36...Qg4 was checkmate. He had 1:20 left on his clock.
That's the honest summary: I won because my opponent made more mistakes than I did, not because I punished the first ones.
How to punish blunders in chess: my lesson from game 2
Looking at the three positions together, I don't see three different errors. I see one habit. When my opponent blundered, I played a move that was fine in general (develop, castle, attack the queen) instead of asking what had changed. So this is the routine I'd suggest to any beginner, and the one I wish I'd had that evening:
- Notice the blunder. If your opponent's last move looks odd, assume it might be a gift, not just a different plan.
- Ask what changed. Which piece is now loose, which square is weaker, which of your moves just became possible?
- Prefer the move that keeps the pressure. A safe developing move is not wrong, but in my game castling turned "Black is winning" into "Black is better".
- Don't confuse an attack on the queen with a punishment. A queen that can step away costs your opponent nothing.
The engine can't tell you why it likes a move unless someone puts it into words. That's the part I care about most now, and it's what Chess Coach does when it reviews a game inside Claude. If you're wondering whether an assistant can actually do this, I wrote about whether Claude can analyze chess games.
What came after
My next game, #3, was rated 868 in the same week, and by the end of that first week I was at 875 after 30 games. The rating attached to my games never went below 700 again: the lowest point after this game was 771, in game #5. The first week's diary, week 1: how to analyze chess games as a beginner, has the full table of those 30 games. And the journey from here to 1385 is on the journey page.
Frequently asked questions
How do you punish a blunder in chess?
First ask what the blunder changed: which piece is now loose, which square is weak, which move became possible. Then look for the most forcing or most restricting move, not just a safe developing one. In my game, Stockfish wanted the same pawn push three moves in a row, and I played other moves each time.
Is it a mistake to play a safe move after your opponent blunders?
Not a disaster, but it can be costly. After my opponent's blunder on move 15, castling was a normal, safe move, and Stockfish still counted it as a mistake: the evaluation went from -5.75 to -3.04, so about half of the advantage disappeared.
Does attacking the queen always punish a mistake?
No. On move 17 my knight attacked my opponent's queen and opened my bishop's line to it as well, a double attack on the queen. The queen simply moved away, and the engine rated my move a mistake compared with its quiet pawn push.