How to stop blundering in chess: a check that survives the clock

You stop blundering by running the same three questions before every move: what did their last move change, what can they capture after my move, and what can they check. Then play a time control that leaves time to ask them. Puzzles train recognition; the check trains the habit of looking.

Blunders are a looking problem, not a seeing problem

Take the last piece you hung in a game, set the position up as a puzzle with the label "find the best move," and see how long it takes you. For most players it takes seconds. The position was never hard. You did not fail to see it; you failed to look.

That diagnosis changes what the fix should be. If blunders were a knowledge gap, the answer would be more patterns. But the player who hangs a knight to a one-move capture knows exactly what a hanging piece is. What they lack is a procedure that checks for one before the hand touches the piece, and a clock setting where that procedure has room to run.

87654321

abcdefgh

The position comes from our hanging pieces lesson, and from a real Lichess game: 3+2 blitz between two players rated above 2300, one of them titled. The move before this diagram, White's king stepped to d4 and attacked the knight. Black, with seconds on the clock, pushed the a-pawn. The knight had seven squares to go to, one of them with check. Black was already in a bad way, but a full piece down the game was over.

A 2300 does not need to learn what a hanging piece is. They needed one second of looking that the clock did not give them. Blunders are what the brain does when it skips the looking step, and the brain skips it whenever it is allowed to.

Why does "checks, captures, threats" fail?

Ask anyone how to stop blundering and you will hear the same answer: before every move, look at all checks, captures and threats. It is good advice, and it does not work for most people who try it. There are two reasons.

First, as usually taught, it is a search for your moves: checks you can give, captures you can make, threats you can create. That is a decent tool for finding tactics. But blunders are not failures to find your own tactic. They are failures to notice their reply. The piece you hang is hanging after your move, on their turn, and a list of your own candidate moves never visits that moment.

Second, it is too wide to survive the clock. "All checks, captures and threats" in a middlegame can be fifteen moves to look at. With ten seconds a move, the brain drops the procedure after three games and goes back to intuition, which is where the blunders live. A check you only run when you have time is a check you run when you do not need it.

So the fix has to be about the opponent's reply, not your own options, and short enough that you run it on every move, including the boring ones. Three questions is about the limit.

How can I stop blundering in chess? The three questions

Before you move, in this order:

1. What did their last move change? Not "what is their plan." Concretely: which squares does the piece that just moved now attack, and which lines did it open by leaving its old square? A move arrives and departs, and the departure is the one everyone misses.

2. After my intended move, what can they take? Imagine your candidate played. Look at every piece of yours and ask if it is attacked and whether it still has a defender. Pay special attention to the piece that just moved and to whatever it was defending before it left.

3. After my intended move, what can they check? Checks are the forks, skewers and loose-piece grabs with teeth, because a check cannot be ignored. If their reply is a check, trace it one move further: where does my king go, and what falls when it does?

If any question turns up something, pick a different move and run the three again. If none does, play it. On most moves this takes a few seconds, because on most moves the honest answer is "nothing."

Run it on the first diagram from Black's side. Question one: White's king went to d4, and from d4 it attacks c5. That is a new attack on my knight, so the knight has to move; the pawn push never gets considered. That is the question Black skipped.

Now a position where the blunder is subtler, because the piece that gets lost is not the piece that moved.

12345678

hgfedcba

This is from our fork lesson, again from a real 10+0 rapid game between two players around 1800. Black's king was on f4, in check from the knight on d5. Five squares were legal; three kept the bishop. Black chose e4, the most natural of the five: central, forward, where a king "should" go in an endgame.

Question one does not help here; the check was the change, and Black saw it. Question two finds nothing: after ...♔︎e4, nothing of Black's can be taken. Question three catches it. After ...♔︎e4, what can White check? ♘︎f6. Where does the knight stand when it gives it? On f6, which also attacks e8. One move further: the king steps away, ♘︎xe8. The bishop was never in danger until the king walked onto a square a knight's jump from it.

That is why the third question is specifically about checks, and why it traces one move further. A fork with check is the one tactic the opponent cannot answer by moving the threatened piece, which is why the fork and the pin win so much material at club level. You do not have to know every fork pattern to avoid them. You have to look at the checks.

Making it automatic

A procedure you have to remember is a procedure you will forget on move 23 when you are winning and relaxed. The goal is a reflex, which means repetitions under slightly artificial conditions.

Say it out loud for twenty games. Rapid, 15+10 or slower. Before every move, quietly voice the three answers: "their rook now hits c7; after ♘︎d5 nothing of mine is loose; the only check is ♕︎b3 and my king has h1." It feels silly, and that is the point: you cannot skip a step you have to say. Around twenty games it stops feeling like a chore and starts feeling like thinking.

The hand-hover rule. Online, the blunder happens in the gap between deciding and dragging. Put a rule in that gap: the move is not played until you have said "nothing" to questions two and three. A hand already on the mouse is the signal you skipped them. Over the board, the same rule is "sit on your hands," and it has worked for a century.

Run it on the boring moves most of all. Your blunders are not in the sharp positions where you were already thinking hard. They are in the recaptures, the "obvious" developing moves, the king walk in the endgame.

Check your losses for which question you skipped. When you analyze your own games, stop at the move where the evaluation dropped and ask which of the three would have caught it. Most players find it is the same one every time: hanging pieces means question two, walking into forks means three, being surprised by the opponent's idea means one. That turns a vague "I blunder too much" into one specific thing to fix.

Blunders that aren't blunders

The engine marks a move that drops the evaluation from +0.3 to -0.8 with the same red glyph it uses for hanging a rook. Those are not the same event, and treating them as the same makes the problem worse.

Losing a pawn to a tactic you missed is a mistake, and often a recoverable one. Losing a piece or walking into a mating attack is what actually ends games at club level. The three questions are built for the second kind. They will not stop you from misjudging a pawn structure or picking the second-best plan, and they are not meant to.

So when you review a loss, separate the two. The moves that lost a piece, allowed a fork, or let a check decide things get the procedure. The moves the engine disliked by a pawn: note them and move on. Panicking about -0.8 trains you to distrust your judgement everywhere, which slows every move and leaves less time for the three questions on the moves that matter.

One more honest point. The procedure reduces blunders; it does not eliminate them. You will still hang things, less often and in harder positions. That is what improvement looks like from the inside: the same kind of mistake, moved a level up.

Are blitz blunders a time-control problem?

Yes, and it is the part of the fix most people would rather not hear.

The three questions take a few seconds on a quiet move and perhaps twenty on a complicated one. In a 15+10 game you have that on every move. In 3+0 you have it on almost none. A blitz blunder is not evidence that you blunder; it is evidence that you chose a time control in which the looking step does not fit. The 2300 in the first diagram was not worse at chess on that move. They had seconds.

That makes blitz blunders a scheduling choice. If you play mostly blitz and want to stop hanging pieces, the first change is not a thinking method but moving a share of your games to rapid, enough that the method can become a habit. A habit survives in faster games in compressed form, but it cannot form at 3+0. Most players stuck around 1600 play mostly blitz, because blitz fits a lunch break. I plateaued there on Lichess for a long time while playing a lot, and the plateau post describes what that looks like from the outside: no single bad move per game, just a slide of small ones with no time to look at any of them.

Keep the blitz if you enjoy it. Just do not expect it to be where the blunders stop.

If you want the pattern explained to you

The three questions are the method. Puzzles are the practice that makes the answers come faster. Disclosure: I built LearnChess, so read this paragraph as a founder describing his own product.

LearnChess puzzles are rated and adapt to your level, so the fork or the loose piece you are checking for shows up at the difficulty where you still miss it. On the paid plans, a second mode builds puzzles from the mistakes in your own imported games, the fastest way to find out which of the three questions you keep skipping. After each puzzle, Knighty, the coach, explains the pattern you missed in plain language, grounded in the engine evaluation rather than a generic tip. The tactics lessons cover the motifs behind questions two and three, the fork and the pin among them, with you making every move on the board. None of it replaces running the check in your own games. It makes the check faster, which, under a clock, is the same as making it happen.