Dump a thousand pieces on a table and every single placement decision becomes a scan of a thousand objects. Sort them into eight trays and each decision becomes a scan of roughly a hundred and twenty. You have not made the puzzle smaller. You have made every individual look-up cheaper, and you will do several thousand look-ups before the picture is finished. That is the whole argument for sorting, and it has nothing to do with tidiness.
Sorting Is a Search Cost Problem
The reason a smaller pile helps is not merely intuitive. Laboratory work on visual search has measured the relationship directly, and the standard result is that when a target shares features with the things around it, response times climb steadily with each additional distractor added to the display. Kristjansson’s review of visual search research describes conjunction searches, where the target shares features with each distractor type, producing response times that rise by roughly 10 to 50 milliseconds per added item. Puzzle pieces are about as conjunction-heavy as a search display gets: nearly everything shares colour, size, thickness, and general silhouette with nearly everything else.
Fifty milliseconds sounds trivial. Multiply it across a few thousand searches through a pile eight times larger than it needs to be and the arithmetic stops being trivial. Sorting is front-loaded cost that buys back time on every subsequent action.
The Primary Sorts That Actually Pay Off
Not all sorts return the same value. Three do most of the work, roughly in this order:
- Edges and corners first. This is cheap, fast, and unambiguous. A flat side is a binary property you can judge by touch without looking, which means you can do it while talking to someone.
- Colour or tonal blocks second. Not fine colour distinctions. Broad blocks: sky, foliage, water, building, text. Aim for categories a stranger could apply the same way you do.
- Piece shape third, and only where it helps. Sorting by tab and blank configuration is slow, so reserve it for regions where colour has stopped discriminating.
That last point is the one most people skip. On a low-contrast image, an autumn forest, a night sky, a bowl of pasta, colour sorting produces one enormous tray that has saved you nothing. Shape sorting still works there, because tab and blank arrangement is independent of the printed image. Two tabs opposite, two blanks opposite, one of each pair, and so on gives you a handful of genuinely distinct bins on a puzzle where the picture has gone useless.
How Many Categories You Can Actually Hold
There is an upper limit on how far to take this, and it is a cognitive one rather than a table-space one. Every sorting category you create is a rule you have to keep in mind while deciding where a piece belongs. Research on working memory capacity has spent decades arguing about the exact number, and a 2024 paper in the Journal of Cognition lays out the debate clearly: Cowan’s estimate puts the focus of attention at about four chunks in normal adults, while other measures using longer presentation times give figures closer to six or seven. The paper’s conclusion is that both numbers reflect the same underlying limit measured under different conditions.
For practical purposes, treat somewhere between four and seven active categories as the working ceiling. Beyond that, you start pausing over every piece to remember what the bins are, and the sorting itself becomes a search problem. If you genuinely need twenty categories, build them in two passes: sort coarsely into six, then subdivide one of those six when you get to it.
A Method You Can Repeat
Here is the sequence that holds up across most 500 to 1,500 piece puzzles:
- Turn everything face up in a single layer before sorting anything. Sorting a stacked pile means handling pieces twice.
- First pass: edges into one tray, everything else stays put. Do not sort colour on this pass.
- Second pass: four to six broad tonal blocks, plus one deliberate tray for pieces you cannot classify. That tray is not failure. It prevents you stalling on ambiguous pieces.
- Assemble the border, then work the highest-contrast region first, which drains your easiest tray and confirms your categories were sensible.
- Subdivide by shape only when a tray stops yielding placements at a reasonable rate.
The unclassifiable tray deserves a note. Every sorting scheme produces edge cases, and the instinct to force each one into a category is what makes sorting feel tedious. Let it be a bin. It shrinks on its own as the picture fills in and context tells you what those pieces were.
One last thing worth saying honestly: on a small puzzle, under about 300 pieces, sorting can cost more than it saves. The pile is already small enough to scan. Sorting earns its keep as piece count climbs, which is exactly what the set size research would predict.
