Week 112: how to count attackers and defenders in chess, the lesson of h4

Nine games from March 30 to April 5, 2026: 3 wins, 4 losses and 2 draws, rating 1233 → 1227 (−6). There were two draws, and both came from positions I should have won. But the moment I learned the most from was a simple capture on move 14 of a game I eventually won, and it answers a basic question I clearly hadn't absorbed: how to count attackers and defenders in chess before taking something. I counted wrong, and it cost me a bishop.

The week in numbers

With Black I scored 2–1–1, with White only 1–3–1. No game ended in checkmate. The best moment of the week was a 98.3 % win with Black against unknown-inf (1198), with no blunders at all — although my opponent made it easier by giving up the queen on move 4 (4.Qxf7+ Kxf7). Four of my six non-wins came from positions the engine had as winning for me.

How to count attackers and defenders: game 2027

With White against marcopaulo (1230), Black had pushed a pawn all the way to h4, where it attacked my bishop on g3.

r . q . k . . r
. p p b b p p .
p . n . p . n .
. . . p . . . .
. . . P . . . p
. Q P B P N B .
P P . N . P P P
R . . . . R K .

Here's the count I should have done. Attacking h4: my bishop on g3 and my knight on f3 — two pieces. Defending h4: Black's knight on g6, the bishop on e7 (along the diagonal e7–f6–g5–h4) and the rook on h8 (the h-file is open all the way down) — three pieces. Two against three means that when the captures stop, Black is the one who takes last. The engine's line confirms it: 14.Bxh4 Nxh4 15.Nxh4 Rxh4, and I've given a bishop and a knight for a pawn and a knight (−5.60, "black is winning"). In the game Black recaptured with the bishop instead of the rook, 15...Bxh4, with the same result. 14.Bf4 just moves the bishop away and keeps the game about equal (+0.29).

I came back in that game and won when my opponent resigned on move 47, but three more blunders on the way show how shaky it was (game).

The f2 fork, for the third week running: game 2026

I've written about this pattern in week 106 and week 111, and it came back. Against GoatNeemKaPatta (1260) Black's knight was on e4 and I castled queenside.

r . . q k . . r
p p . n . p b p
. . . . p . p .
. . . p . . . .
. . p P n B P .
. . P . P N . P
P P Q . B P . .
R . . . K . . R

The engine's reply to 13.O-O-O is the familiar one: 13...Nxf2, a knight that takes the pawn and attacks my rook on d1 and my rook on h1 at once (−2.49). With 13.Bxc4 my king stays on e1, still guarding f2, and I'm slightly better (+1.08). My opponent played 13...a5 and missed it, but the game was a long slide anyway: five blunders, and it was abandoned on move 68 (game).

An old habit with Black: game 2025

On the same Monday, as Black against Free_Palestine_Gaza (1271) in a Pirc, I repeated an idea from week 98.

r n b q k b . r
p p p . p p . p
. . . p . n p .
. . . . . . . .
. . B . . . . .
. . P P . N . .
P P . . . P P P
R N B Q . R K .

6...Bg4 pins the knight on f3 to White's queen, but it also takes the c8 bishop away from b7. The engine's answer is 7.Qb3, attacking the b7 pawn, which has no defender, with the knight on b8 behind it (+0.25). With 6...Bg7 I was clearly better (−1.55). White didn't play Qb3 at once, but on move 11 the queen took on b7 anyway, and the game was abandoned on move 27 (game).

The lesson of the week

Before a capture I now do the count out loud, even in a 10-minute game: attackers, defenders, and who takes last. Pieces on long lines are the ones I miss — the rook at the end of an open file, the bishop at the end of a diagonal — so I trace those lines to the end. And the same counting applies to my own squares: when my king leaves e1 or my bishop leaves c8, something it was defending is now defended by one piece less.

All games this week

Previous: week 111 · Next: week 113 · Back to the journey hub · Review your own games like this