Sometimes no direct technique fires, yet the puzzle is far from ambiguous. The contradiction test — what-if analysis — converts a stall into a proof: tentatively assume a cell is filled, propagate only the forced consequences through its row and column, and watch for an impossible state. A clue with no room left, two runs forced to merge, a count that overshoots the line. If the assumption leads to impossibility, the cell is provably empty. Assume empty and break the line instead, and it is provably filled.
Here is a complete one-line example. Width 6, clues 2 2, and the third cell is already known filled. Is the first cell filled? Suppose it is: then the first run of 2 occupies cells 1-2, and a run of exactly 2 must be followed by an empty cell — but cell 3 is filled. The runs would merge into a block of at least 3, which no clue allows. Contradiction; cell 1 is empty:
Notice what happened after the proof: with cell 1 marked, the first run must cover the known cell as 2-3, punctuation seals cell 4, and the second run has exactly cells 5-6 left. One contradiction unlocked the entire line. That payoff pattern is typical — use the test on cells whose state would ripple, usually where a row and column are both nearly resolved.
Two rules keep the technique honest. First, propagate only forced moves while testing; if you find yourself choosing between options inside the what-if, back out — that is guessing wearing a disguise. Second, keep the chain short. On this site every puzzle is generated and verified to have a unique, logic-reachable solution, so a one- or two-step look-ahead is always enough; if a deep chain seems necessary, a simpler deduction is sitting unnoticed in a crossing line. The contradiction test is the last tool in the box precisely because the others are cheaper — but when you need it, it is a proof, not a gamble.