How to solve a 20×20 nonogram

Why your usual opening stops working

A clue of length k in a line of n cells forces 2k − n cells to be filled, wherever the block finally sits. On a ten-cell line a clue of 8 hands you six cells for nothing. On a fifteen-cell line the same clue gives you one. On a twenty-cell line it gives you zero — you need a clue of 11 before overlap forces even two cells, and clues that large are uncommon unless the subject spans most of the board.

So the first pass on a 20×20 typically fills almost nothing, and that is normal rather than a sign you have missed something. Solvers who learned on smaller grids often read the empty board after their usual opening as evidence the puzzle is broken. It is not; the opening simply does not apply at this length.

What does still pay is the edges. A clue that is large relative to the line still forces cells, and the outermost rows and columns of a picture are where large clues live — a subject that spans the board produces a long run somewhere near its widest point. Start by scanning every line for its largest clue rather than working in order, and place the few forced blocks that exist before doing anything else.

Cross-referencing is the actual method

On a small grid you can treat rows and columns as two separate passes and alternate between them. At 400 cells that is too slow, because each pass mostly rediscovers what the last one already established. The productive loop is narrower: take one cell you have just proved, and immediately ask what it forces in the line crossing it, then what that forces in turn.

Concretely, if you fill a cell in row 7 and the column through it has a clue of 4 with only five candidate positions left, you have just eliminated several of them. Follow that chain until it stops rather than returning to a systematic sweep. A single deduction on a large board often cascades through six or seven lines, and chasing the cascade is far faster than finding the next independent starting point.

This is also why the large boards feel qualitatively different rather than merely longer. The work stops being "scan for a clue I can resolve" and becomes "follow the consequences of what I already know", which is a more demanding kind of attention and the reason a 20×20 is an evening rather than a longer break.

Empty cells carry most of the information

On a sparse board — and most 20×20 subjects leave a good deal of the grid blank — you will prove far more cells empty than filled. Marking them is not optional housekeeping; it is where the deductions come from. A run of proved-empty cells splits a line into segments, and a clue of 6 in a line whose remaining segments are 4, 5 and 3 cells long has just been pinned to one of them.

That segmentation is the single most useful habit at this size. It converts a twenty-cell line, which is too long for any clue to constrain, into two or three short lines, which are exactly the length the overlap rule works on again. In effect you spend the early part of a large solve manufacturing the small grids you know how to handle.

The corollary is that an unmarked cell and a proved-empty cell must look different, both on screen and on paper. If they do not, you will re-derive the same emptiness repeatedly and the board will feel like it is not progressing when in fact you keep solving the same square.

Structuring a long solve

A 20×20 does not need to be finished in one sitting, and treating it as though it does is how people abandon them. Reaching a natural stopping point — a completed region, a fully resolved band of rows — and leaving it is fine. What is not fine is stopping mid-cascade, because the chain you were following lives in your head rather than on the board, and picking it up cold costs more than finishing it would have.

Work outward from whatever resolves first rather than left to right. Large pictures tend to have one region that gives way early — a solid mass, a straight edge, a symmetric feature — and expanding from there keeps every new deduction adjacent to something known, which is where cross-referencing is cheapest.

If a board is symmetric, exploit it deliberately. Read the column clues inward from both ends before placing anything and note where they mirror; on a symmetric subject a deduction on the third column usually transfers straight to the matching column on the far side. That is one of the few genuine shortcuts in nonograms that is not a restatement of the overlap rule, and at this size it can halve the real work.

When you have made a mistake

Errors behave differently on large grids. On a 10×10 a wrong cell usually produces a contradiction within a few deductions. On a 20×20 it can sit quietly for twenty minutes and only surface when a line at the far side refuses to resolve, by which point a great deal of correct work depends on it.

The instinct is to hunt for the error by re-checking everything, which is slow and usually fails. The faster approach is to find the line that produced the contradiction and check only the cells in it that you cannot independently re-derive. A cell you can prove again from scratch is not the problem; the one you placed because it "had to be" three steps ago probably is.

It helps enormously that this is possible at all: every puzzle here has exactly one solution reachable by logic alone, so a contradiction is always your mistake rather than the puzzle's ambiguity. On a board that permits multiple solutions there is no way to tell the two apart, which is why an ambiguous 20×20 is not a harder puzzle but a broken one.

Which boards to practise on

The catalog has thirty-one boards in the 20 bucket, twenty-three of them true 20×20s. Twelve are rated hard and eleven of those twelve are the square ones, so shape and difficulty travel together here more than anywhere else on the site.

The theme you pick matters as much as the size. All seven animals boards at this size are rated normal, which makes them the sensible way in — and side-on animals are near-predictable in outline, useful when a wrong guess costs real time. At the other extreme, every everyday objects board at 20×20 is rated hard, all four of them, and both alchemy boards are too. Food and drink and hats and masks are the joint-biggest sets here at nine boards each and pull in opposite directions: only two food boards are hard against four of the hats.

Frequently asked questions