How to stop blundering in chess

A puzzle tells you two things before you look. There is a tactic, and it is your move. A game tells you neither, so a fork you would find in a puzzle can walk past unseen.

A one-move blunder needs no deep line to avoid. It is a move that works on the board in your head but not on the board on the table.

This guide covers three ways the two boards can differ. The first is the phantom defender, a piece that looks like a guard and cannot act as one.

Put a White bishop on b5, a Black knight on c6 and the Black king on e8, with d7 empty. The knight seems to guard e5, but it is pinned and cannot legally capture there.

The second is the forgotten piece, a bishop or rook that has not moved for many moves, so it is easy to stop looking at.

The third is the abandoned job. If the pawn on d2 is the only guard of your knight on c3, pushing d2-d4 leaves the knight loose although the knight never moved.

Treat board vision as a scan run in the same order before every move, not as a talent. A fixed order means nothing depends on where your eyes happen to land.

Set this up with White to move. White has Kg1, Qd1, Ra1, Rf1, Bc1, Nb1, Nf3 and pawns on a2, b2, f2, g2 and h2.

Black has Ke8, Ra8, Rh8, Bc8, Bb4 and pawns on a7, b7, c7, f7, g7 and h7. No Black piece defends the bishop on b4.

White plays Qa4+. The queen checks along b5, c6 and d7 to the king on e8, and attacks the bishop along the fourth rank.

Black must answer the check first. Blocking with ...c6, ...b5 or ...Bd7 does not defend b4, and neither does any king move, so Qxb4 follows.

The scan would have warned Black before the bishop ever reached b4. Run step three on the position after ...Bb4, and it finds the bishop loose, with no Black piece defending it.

A Memory Chess round checks your rebuilt board against the original, one square at a time. A piece scores only when the square, the type and the colour all match.

Placing more pieces than the position held is charged twice. Each piece beyond the count is counted as wrong and also takes ten points off your accuracy.

A wrong piece inside the count costs no more than an empty square. Either way one original piece goes unreproduced, so the extra charge lands only on invented pieces.

That makes a round a clean test of one blunder habit. A seventh piece on a six-piece board is a piece your memory added, the phantom defender in miniature.

Both drills use positions from your own games, because the blunders worth studying are the ones you actually make.

For the tally, a blunder is a move that loses a pawn or more to a reply one or two moves deep. A slow positional slide does not count, because no scan would catch it.

Write the number next to each result, wins included. Count the move even when your opponent missed the reply, because the habit being measured is yours.

Rating is a poor gauge of this one habit. It moves with your openings, your endgames and your pairings too, while the tally moves only with the scan.

On a low clock the full scan will not fit, so keep step one and drop the rest. A piece that just moved has fresh targets, and step one also catches any line it opened.

This is also why the advice to play slower fails on its own. Extra time is only as useful as the question you spend it on.