The uniqueness trick, and why it feels like cheating

There's a class of Sudoku techniques that sit slightly off to one side of the rest. The unique-rectangle move and its relatives — sometimes grouped under uniqueness techniques or avoidable-rectangle logic — are perfectly valid in the sense that they always produce correct answers on properly-constructed puzzles. They're also a bit different in kind from every technique that comes before them, and a small number of solvers refuse to use them on principle. The disagreement is interesting, the techniques are worth knowing, and the question of whether to lean on them is more a stylistic one than a logical one.

This piece is the honest version. What the trick is, why it feels different, and the case both sides make.

What the trick is, briefly

The simplest version is the unique rectangle. Imagine four cells that form a rectangle on the grid — two cells in one row, two in another, sitting in the same two columns. Three of those four cells have been narrowed down so they each contain only the same two pencil-mark candidates, say {3, 7}. The fourth cell also has 3 and 7 among its candidates, plus possibly some others.

Here's the trick. If the fourth cell were also reduced to only {3, 7}, the rectangle would have a problem. The four cells could legally be filled as 3-7-7-3 reading top-to-bottom or as 7-3-3-7, and both fillings would respect every Sudoku constraint. The puzzle would have two valid solutions.

But the puzzle doesn't have two valid solutions. Properly-constructed Sudoku has exactly one. So the fourth cell cannot be reduced to only {3, 7} — there must be some third candidate left over to break the symmetry. That third candidate is what the cell ends up holding.

The conclusion is sound. Used carefully, the technique produces correct answers every time.

Why it feels different

The earlier techniques in the toolkit — naked pairs, X-wings, Y-wings, the bigger fish — all derive their conclusions from the rules of Sudoku alone. Each row, column, and box must contain the digits 1 through 9. From that, plus the givens, every elimination follows directly.

The uniqueness trick uses something else. It uses the constructor's promise that the puzzle has one solution. That's not a rule of Sudoku in the strict sense; it's a convention about how the puzzles are made. Most solvers don't notice the difference because every published Sudoku has had this property baked in since Nikoli's editorial standardisation in the 1980s — but it's a separate piece of information from the row-column-box rules, and the technique is using it explicitly.

That's what people mean when they say uniqueness feels like cheating. The deduction doesn't live in the puzzle in front of you. It lives in the meta-fact about the puzzle.

The purist position

A small contingent of strong solvers won't use uniqueness techniques on principle. The argument runs: a Sudoku is supposed to be solvable by direct logical reasoning from the rules and the givens, with no external assumptions. A solve that relies on this puzzle has only one solution is a solve that relies on knowing something the puzzle doesn't actually tell you.

The position has aesthetic appeal. There's a cleanness to a solve that goes from rules-plus-givens to a unique answer using nothing else. It also has a practical point: if you're working on a puzzle whose construction quality you don't trust — a hand-made puzzle from a friend, a printout with possible misprints, a generator from an unknown source — assuming uniqueness will mislead you when the puzzle isn't actually unique. The deduction can produce wrong answers on a malformed puzzle, where the rules-only deductions can't.

For these solvers, uniqueness is a shortcut that trades a small amount of soundness for a measurable amount of speed. Most of the time the trade is fine; some of the time it bites. They prefer not to make it.

The pragmatic position

The other side of the argument is also defensible, and it's where most experienced solvers land.

Sudoku as a published category has had uniqueness baked in for forty years. Any newspaper Sudoku, any major-publisher puzzle book, any reputable website is producing puzzles that are guaranteed unique. We've written about the editorial standards Nikoli established in the 1980s — single-solution-provable-without-guessing was one of those standards, and it's universal across reputable Sudoku now. Using that fact to solve faster is not, the pragmatic argument goes, fundamentally different from using the rule that each row contains the digits 1 through 9. Both are constraints that hold of the puzzles you're actually solving; both can produce valid deductions.

The more practical version of the argument: uniqueness techniques are useful in exactly the situations where you're stuck and need a move. They're rare, they require specific patterns to set up, and the pattern recognition for them — three cells already reduced to the same pair, fourth cell containing the pair plus extras — is a teachable skill that surfaces a real elimination. Refusing to use the technique on principle leaves a small but real edge on the table for the kind of solver who values it.

What this means for your style

There isn't a wrong answer. Both positions are coherent. What matters is being clear about which one you're choosing and why.

If you mostly solve published puzzles from sources you trust, and you care about speed or completion rate on hard puzzles, learning the unique-rectangle pattern and a few of its relatives is a reasonable expansion of the toolkit. The conclusions you'll get from them will be correct, and the technique will occasionally crack puzzles that would otherwise stall.

If you mostly solve hand-made puzzles, or care about the cleanest possible logical chain as part of what you enjoy in the activity, leaving uniqueness out is a perfectly reasonable choice. You'll occasionally get stuck on a puzzle where uniqueness was the intended move, and that's part of the deal.

The only position that doesn't really hold up is use uniqueness sometimes without thinking about it. The technique is meaningfully different from rules-only deduction, and applying it casually on the wrong puzzle (a misprinted one, a malformed generator output) can produce a wrong answer. Either you've decided uniqueness is a tool you accept, in which case use it knowingly; or you've decided it isn't, in which case keep it off your bench. The third option — apply by reflex without checking the puzzle's source — is the one that occasionally produces the surprising mistake.

A small wider point

The uniqueness debate is the rare place in Sudoku where a stylistic disagreement has a real and defensible substance behind it. Most Sudoku style debates reduce to I prefer pencil marks, you don't — different solvers, different rituals, no real disagreement. The uniqueness question genuinely is asking you to choose between two different conceptions of what counts as a solution: what the rules force, or what the rules plus the construction promise force. Either is fine. Neither is wrong. Picking one and being aware you've picked it is the move.