Pointing pairs and the snake

There's a particular shape that recurs across mid-level Sudoku, and most intermediate solvers spend longer than they should before they start seeing it. Two of a digit's candidate-cells in a single 3×3 box sit in the same row or column — a small line of two cells, sitting inside one box but extending logically into the rest of the row or column. Some solvers describe the visual shape as a snake: a short body of two cells, with a tail of constraint reaching out into the rest of the line.

This piece is about the snake — what it tells you, why it's useful, and the perceptual habit that surfaces it without effort once it's natural.

What the snake actually tells you

When two of a digit's candidate-cells in a box sit in the same row, the digit must occupy one of those two cells. Even though you don't yet know which of the two, you know the digit can't appear in any other cell of the row outside the box — because if it did, the box would have nowhere to put it.

The constraint is small but useful. Eliminating that digit from the rest of the row often surfaces a hidden single somewhere down the line, or tightens a naked or hidden pair you'd already noticed but couldn't act on. The snake is rarely the move that finishes a puzzle, but it's frequently the move that breaks the half-finished-grid stuck moment.

The same logic works rotated ninety degrees: two of a digit's candidate-cells in a box sit in the same column, the digit can't appear elsewhere in that column outside the box. Same shape, same useful elimination, ninety-degree rotation.

Why it's easy to miss

Pointing pairs are easy to miss because they're a unit-spanning pattern. A box-row pointing pair lives in two units at once — the box, where the two candidate-cells are, and the row, where the elimination happens. Solvers who scan one unit at a time don't see the pattern, because the relationship between the two units is invisible from a single-unit perspective.

The other reason they're missed is that the candidate-cells often have other candidates listed besides the digit in question. A cell with candidates 9 doesn't visually announce that the 7 in particular has only one neighbour with the same candidate-cell-pattern. You have to be scanning for the digit across the box, not for the cell's full candidate list.

This is the same perceptual trap that hides X-wings and hidden pairs: the cell-first scan doesn't surface unit-spanning shapes. The fix is the same — unit-first scanning, one digit at a time, looking for where the digit could go in each box and noticing when the answer is "two cells in the same row" or "two cells in the same column."

The perceptual habit that surfaces them

Three small habits put pointing pairs in your peripheral vision.

Scan boxes by digit, not by cell. Pick a digit. For each box that doesn't yet contain the digit, ask where in the box the digit could go. If the answer is two cells, check whether those two cells share a row or column. If they do, you've found a snake.

Re-scan after every placement. Each placed digit changes the candidate sets in the affected box, row, and column. Pointing pairs that didn't exist before the placement often appear after it. Most solvers re-scan for naked singles after a placement; the same scan should include a quick check for new pointing pairs.

Read pencil marks as shapes, not lists. When you've pencil-marked a cluster, look at the candidates as a visual pattern across cells rather than as a list within each cell. The snake announces itself as a pattern — two cells side-by-side or stacked, both showing the same digit, no other cell in the box showing it. We covered this perceptual habit at length in reading pencil marks like a shape; pointing pairs are one of the patterns it makes visible.

When pointing pairs actually matter

In practice, pointing pairs surface in maybe one out of every two medium puzzles and almost every hard puzzle. They're rarely the only mid-level move — usually they appear in the back half, after pairs and triples have done early work — but they tend to be the move that breaks a half-finished-grid stuck moment, because the elimination they produce often unlocks a hidden single elsewhere on the row or column.

Solvers who can spot pointing pairs reliably tend to finish hard puzzles ten to fifteen per cent faster than solvers who can't, and the difference is mostly because pointing pairs prevent the stuck moments that other solvers spend three minutes hunting through. The technique itself is small. The compounding across a puzzle is meaningful.

The snake is one of the rewards of practising unit-first scanning. The eye picks it up over a few weeks of deliberate looking, and after that it becomes invisible — you don't notice yourself spotting them, you just notice that you stopped getting stuck on the same kinds of moments. That's most of mid-level Sudoku improvement: small perceptual habits becoming automatic, until the moves they surface arrive without effort. The snake is one of the cleanest examples.