Pencil marks without the clutter

Pencil marks are the small candidate digits you scribble in the corners of empty cells when you're not yet sure which digit goes there. They're meant to help you spot moves you'd otherwise miss. They can also turn a clean grid into a wall of scribbles that nobody — including you — wants to look at.

The trick is to write fewer of them than you think you need, and to write them later in the puzzle than feels comfortable.

When you don't need pencil marks

On easy puzzles, you usually don't. The naked and hidden singles of an easy puzzle can be spotted without writing anything down — your eye finds them as you scan, and the constraints aren't dense enough to need bookkeeping.

Writing pencil marks on every empty cell of an easy puzzle adds time and noise without adding moves. It also trains a habit — mark everything reflexively — that becomes a problem on harder puzzles, where indiscriminate marking actively hides the patterns you're looking for.

The strong version of this advice: solve easy puzzles entirely without pencil marks, until you can. The discipline of finding singles by eye sharpens the part of your brain that finds them, and you'll move faster on every harder puzzle as a result.

When pencil marks earn their keep

On medium puzzles, you'll hit cells where multiple digits feel possible and you can't easily remember which. That's the natural moment for marks: the cell whose options you'd otherwise have to recompute three times in the next five minutes.

The cells worth marking are the ones where you've already narrowed the options to two or three candidates by eyeballing the row, column, and box; where the cell sits in a heavily-constrained region (a row with four givens, a box with six placements); or where you suspect the cell is part of a pattern — a naked pair, a hidden pair — that you'll spot once you've marked a few neighbours.

The cells not worth marking are usually the ones in sparsely-populated regions, where seven different digits are still plausible. Marking these is doing the puzzle's bookkeeping for you when there's no useful pattern yet.

A minimum-marking workflow

A practical pattern for medium and hard puzzles:

1. Solve every naked and hidden single you can find without marking. Place them.
2. When the obvious moves run out, find the most-constrained empty cell on the grid — usually one in a row or column with five-plus givens — and mark its candidates.
3. Mark the immediate neighbours of that cell, the other empties in the same row, column, and box.
4. Look at the local cluster of marks. Pairs, triples, and pointing patterns often jump out from a partial pencil-marked cluster that wouldn't be visible on a fully-marked grid.
5. Erase as you place. Every digit you write becomes a digit you remove from neighbouring cells' marks.

Three properties of this pattern make it more pleasant than full-grid marking. It marks fewer cells, so the visual noise stays low. It marks them in clusters, where patterns live. And it updates as you go, rather than having you re-derive marks from a stale full-grid annotation.

Paper versus digital

The marking economics differ between the two media.

On paper, every mark and erasure is manual labour, and the cost of a wrong mark is a smudge or a hole in the page. The argument for minimum marking is strongest here.

On digital, marking is fast and erasure is free. There's no penalty for over-marking, and most digital interfaces also offer auto-candidate modes that fill in the candidates for every cell automatically. Auto-candidates are a different beast: they remove the cognitive work of marking but also the cognitive work of deciding what's worth marking, which means they can hide moves you'd otherwise notice on a sparser grid. Whether to use them is a personal call. Most experienced solvers we know turn them off; some swear by them.

The compromise that works for many people is to turn auto-candidates on for easy and medium puzzles where speed-of-solve is the goal, and turn them off for hard and expert puzzles where the practice of choosing what to mark is part of the technique.

The minimum version

If you remember nothing else about pencil marks: write them later than feels comfortable, write fewer than you think you need, and update them as you place. The cells that benefit most are the constrained ones, not the empty ones.

When that workflow stops yielding moves, your puzzle is asking for a harder technique. The marks that lead you to those — naked pairs, pointing pairs, the small patterns that medium puzzles hinge on — are easier to spot on a sparse grid than a dense one. Which is the whole argument for restraint.