How the fastest solvers think

The fastest competitive Sudoku solvers in the world will finish a hard 9×9 grid in two or three minutes. Untimed recreational solvers, including reasonably good ones, take fifteen to twenty-five minutes for the same puzzle. The order-of-magnitude gap is real, and it raises a fair question: what are they actually doing in their heads that takes so much less time?

The answer is less mysterious than the gap implies. The fastest solvers aren't using techniques the rest of us don't know. They're using the same techniques with much faster lookups — and a small set of habits around chunking, scanning, and notation that don't take years to learn but do take deliberate practice. Most of those habits are visible to a recreational solver, and some of them are even worth borrowing. Some of them aren't.

Same techniques, faster lookups

The first thing that's true about competitive solvers is what isn't different about them. They use naked singles, hidden singles, naked and hidden pairs, pointing pairs, X-wings, Y-wings, and the wider chain repertoire — the same techniques in the standard recreational toolkit. Nothing on a normal hard 9×9 grid requires anything they have access to that an intermediate solver doesn't.

What's different is the speed of the perceptual lookups. A recreational solver scans a row and thinks about which digits are missing; a competitive solver glances at a row and knows which digits are missing in something close to single-digit milliseconds. Same operation, different cost. Both are arriving at the same conclusion; one is arriving so much faster that the rest of the solve doesn't have to wait for it.

The lookup speed comes from sheer practice. Tens of thousands of hard puzzles' worth of pattern exposure builds a perceptual catalogue — this configuration → this technique applies → this elimination follows — that fires below conscious deliberation. The recreational solver is running the same logic on the same patterns; they're just running it slowly because they haven't seen each pattern enough times to make it automatic.

This is also the unglamorous answer to how do I get faster. Mostly: do many more puzzles, deliberately, until the patterns stop being effortful. Speed in Sudoku is, more than anything else, a story about pattern fluency.

Chunking the grid

A second visible difference is how competitive solvers carve up the grid as they work. A recreational solver tends to think in single cells, sometimes single units. The competitive solver thinks in digit-shaped fields — where can the 7s go in this stack of three boxes? — and updates that field-shape view continuously across the solve.

This is the scanning shift from cell-first to unit-first thinking, pushed further. Once you've done enough hard puzzles, the scanning isn't unit-first either; it's digit-shaped-region-first. A competitive solver looking at a hard grid is essentially watching nine separate sub-puzzles unfold simultaneously, each tracking the candidate space for one digit, with eliminations propagating across them.

The chunking matters because it lets the same number of lookups produce more eliminations. A unit-first scan finds one constraint; a digit-shaped-region scan often finds two or three at once, because the digit's behaviour across multiple units is being tracked together. The bottleneck moves from "find moves one at a time" to "find moves in batches," which is most of where the speedup comes from past the basic-fluency stage.

The notation and the pencil-mark trade-off

Most competitive solvers do less pencil-marking than ambitious recreational solvers, not more. The reason isn't that they don't need pencil-marks; it's that pencil-marking is mechanically slow, and a pattern they could catch by mental candidate-tracking is faster than the same pattern caught from a marked grid.

The trade-off is real for the recreational solver too. Pencil-marking on every cell is the right move when you're learning a technique, working at hard or expert tier under no time pressure, or solving a puzzle that's stalled. Pencil-marking everything during a fluent solve of a medium-difficulty puzzle is over-engineering — the marks take longer to write than the moves they would have surfaced.

The competitive version of this isn't don't pencil-mark. It's pencil-mark only the cells where the constraint is interesting, and let the rest of the grid run on mental candidate-tracking. That habit is teachable and worth developing if you care about speed at all; it isn't a habit competitive solvers have access to that you don't.

What's transferable

A few of the competitive habits are worth borrowing for the recreational solver who wants to get a bit faster without changing what the activity feels like.

Pattern fluency through volume — solving many hard puzzles deliberately, ideally with attention to which technique is doing the work each time — is the highest-leverage practice habit. Speed comes from this more than from anything else.

Digit-shaped-region scanning, taught explicitly on a few practice puzzles, transfers reasonably quickly into solving habit. It's a small mental reframe with a real return, and it doesn't require a stopwatch to develop.

Selective pencil-marking — marking the cells where the constraint is informative and not the cells where it's obvious — is a habit that compounds over months. Recreational solvers who develop it tend to find their hard-puzzle times drop noticeably, even without any other changes.

What isn't transferable, honestly

Most of the rest is the kind of capacity that comes from years of full-time engagement with the puzzle as a serious activity, not as a daily fifteen minutes. The lookup speed of a championship-tier solver isn't borrowable in any meaningful sense; it's the product of a different relationship to the puzzle than most readers have or want.

That isn't a knock. We've covered the broader question in the stopwatch problem and the trade-off speed makes against the better move, and the honest answer is that the recreational solver and the competitive solver are doing slightly different activities. The recreational solver is solving for the experience of solving; the competitive solver is solving for the speed of solving. Both are real, both are valid, and the recreational version doesn't gain much by aspiring to the competitive version's metrics.

What it can gain — and what this piece is mostly about — is borrowing the few competitive habits that improve the recreational experience without changing its character. Faster pattern recognition, smarter chunking, lighter pencil-marking. None of those require a stopwatch, and all of them make the same daily fifteen minutes feel slightly more fluent over the months you keep doing it.