David — the bind was real; the blind spot was one move

The story of the game

This was your kind of position for sixteen moves. The Advance Caro-Kann handed you the bishop pair after 6.h3 Bxf3 7.Qxf3 and a centre James couldn't easily challenge, and by move 13 the engine had you near +2.5 — a clean, near-70% edge with no risk. Then 17.Bd2 undid all of it in a single tempo. This isn't a "you played the middlegame poorly" game; it's a "you were winning and stepped on a rake" game — which is actually the more fixable kind.

The key moment: 17.Bd2

James had just parked his rook on c3, hitting your knight on b3 and bishop on d3. You'd already solved that perfectly with 16.Qd1 — the queen guards b3 along the diagonal and d3 down the file at the same time. That move deserves a nod. The problem is what came next: 17.Bd2 attacks the c3-rook, but it plants your own bishop on d2 — the very square the queen needed to see through to defend d3. So 17...Rxd3 simply won the bishop; the queen could no longer recapture, blocked by its own piece.

The winning continuation was the knight jump Nc5 (+2.3): trade off James's good b6-bishop, keep the c3-rook offside, and grind the extra space. The habit worth taking from this — when your move makes a threat, finish the thought: "and what does this piece stop guarding?" 17.Bd2 created a threat and dropped a defender in the same motion, and only the second half was invisible.

The earlier slip: 14.Be3

Not costly, but instructive: 14.Be3 was the first easing of the grip. The engine wanted Qg4, leaning on g7 and the kingside while James's king still couldn't castle. When you have a bind like this, keep asking hard questions — the quiet developing move that merely improves a piece lets a cramped position breathe.

Phase notes

• Opening / early middlegame: excellent. The bishop pair and centre were textbook Advance Caro-Kann, and 16.Qd1 was a genuinely strong defensive resource. Keep doing this.
• The turn: a self-interference tactic, not a strategic drift — so the fix is different from your usual middlegame homework. See the drill.

Profile check

Your file flags the middlegame as where games leak, "usually a drift of small concessions rather than one blunder." This game refines that: not a drift, a single tactical self-block. Your #1 leak (knight forks) didn't appear, and your defensive resourcefulness (16.Qd1) landed on the right side of the ledger for once. The item to add is narrow and sharp: self-interference — check what your own move stops defending.

What to drill next

1. "What do I stop guarding?" — for one week, before every move that makes a threat, name the square your moving piece was defending. 17.Bd2 dies to that one question.
2. Removing-the-defender puzzles, both directions: sharpen the sense for when a piece is the only guard — you'll spot it for yourself as fast as you spot it for the opponent.
3. Convert-the-bind exercises: +2 positions where the job is to improve the worst piece and deny counterplay, not to force matters.