The puzzle that taught me the X-wing

Most experienced Sudoku solvers can name a specific puzzle where their relationship with a technique changed. Not the first hard puzzle they finished, not the moment they read about a technique in a book — the one specific grid where the abstract knowledge became real, where the move that lived in their head finally lived on the page.

For a lot of solvers, the technique that does this is the X-wing. The puzzle is rarely impressive in the abstract; what makes it the right puzzle is the timing, and the small accumulated frustration of the previous twenty puzzles that hadn't quite cracked the move open. This piece is about a puzzle like that — composite rather than literal, but true to the experience.

The setup

The puzzle was an expert classic, somewhere around the fortieth puzzle into a stretch of expert solving where most of them had been finishing in twenty-five to thirty-five minutes. This one had been forty minutes already and was sixty per cent done — the classic half-finished grid problem that intermediate solvers know intimately.

The remaining empty cells were spread across three rows: rows two, four, and seven. Each row had four or five empties, each empty had three or four candidates, and none of the obvious moves were working. Naked pairs: scanned, none. Pointing pairs: scanned, none useful. Hidden singles: re-scanned, all already placed. The puzzle had stopped yielding to the toolkit, and the temptation to guess was rising in a particular way.

The solver had read about X-wings. They'd seen the diagrams in puzzle books — the four corners of a rectangle, the digit confined to two columns by virtue of having only two candidate-cells in two different rows, the elimination running down the columns. The technique made sense in the abstract. It had never appeared, in any of the previous forty puzzles, in a way that felt findable in real time.

This time it was going to.

What changed

The solver started scanning by digit instead of by cell. This wasn't deliberate strategy; it was a small adaptation born of running out of ideas. Where could the 4 go in row two? Two cells, in columns three and seven. Where could the 4 go in row seven? Two cells — and they were in columns three and seven.

The shape registered slowly. Not as "an X-wing" — the technical name was still an abstract label — but as a small geometric pattern across the grid. Four cells, in two rows, in two columns, all of them showing 4 as a candidate. The eye saw it as a rectangle before the brain named it.

Then the brain caught up. If the 4 in row two is in column three, then the 4 in row seven has to be in column seven (because columns can't repeat). If the 4 in row two is in column seven, then the 4 in row seven has to be in column three. Either way, the 4 in row two and the 4 in row seven occupy columns three and seven between them.

Which meant the 4 couldn't appear elsewhere in either column. Which meant five other cells, scattered across columns three and seven, could have the 4 eliminated from their candidate lists. Which meant — checking — three of those five eliminations were sufficient to surface naked singles in their cells. Which meant —

The puzzle started solving itself.

What the moment was actually about

Five minutes later the grid was finished, and the solver sat looking at the empty paper for another minute, trying to decide what had just happened.

The X-wing wasn't new. The solver had known what an X-wing was for months — could have described it in plain English, could have drawn the rectangle on a whiteboard, could have explained the elimination logic to anyone who asked. What had been missing wasn't the knowledge. It was the recognition — the eye's ability to see the shape as it occurred in a real puzzle, in a real cluster of pencil-marked candidates, with a dozen other digits and patterns competing for attention.

What changed in that puzzle wasn't the addition of a new technique. It was the conversion of an abstract technique into a perceptual one. The technique had moved from the part of the brain that knows things into the part of the brain that sees things. After that puzzle, X-wings weren't the technique-that-keeps-tripping-people-up. They were a small recognisable pattern that the eye knew how to find.

What this is true of

The same pattern is true of most mid-level and advanced techniques. The first time they appear in a real puzzle is the puzzle that teaches them, regardless of how many books or articles you'd read beforehand. The reading puts the technique in your head; the puzzle puts it in your eye.

Most experienced solvers can name the puzzle that taught them swordfish, the puzzle that taught them simple coloring, the puzzle that taught them XY-wings. Some of them have decades of solving behind them and can still locate those specific moments. The puzzles aren't impressive on their own — the X-wing puzzle described above was a fairly ordinary expert from a not-particularly-celebrated puzzle book — but they're the moments when the abstract becomes the actual. That's most of what improvement is, after the first few months: the slow conversion of read techniques into seen patterns, one specific puzzle at a time.

If you're at the stage of knowing the X-wing in theory and not seeing it in practice, the puzzle that will teach it to you is somewhere in the next twenty or thirty hard or expert solves. There's no faster route. There's only the looking, until one day the looking finds the shape, and the shape finds the move.

You probably won't remember the puzzle the next morning. But you'll remember the moment. And every puzzle after it will be slightly different.