Solving expert classic sudoku

Expert classic sudoku is the tier where the X-wing finally earns its keep. At medium and hard the technique is theoretically available but rarely necessary; pairs and pointing pairs handle most stalls. At expert, X-wings appear in most puzzles, and swordfish — the three-row generalisation — appears in roughly half. Solvers who reach expert without learning these techniques find themselves stuck on puzzles they have no way through. Solvers who do learn them find expert satisfying — the tier where pattern recognition stops being a tutorial concept and starts being a daily tool.

Full pencil marks across the grid

Expert is the first tier where most solvers fill in candidate notes for every empty cell from the start, not just for the cells that haven't resolved. The reason is structural: the new techniques operate across multiple rows and columns, and missing a single pencil mark in the wrong place can hide the pattern entirely. Full notation isn't always strictly necessary — some expert puzzles solve without ever needing an X-wing — but the cost of skipping it is occasional unsolvable stalls. Most solvers who reach expert decide the discipline is worth the time.

The new techniques expert introduces

The X-wing is the textbook expert technique. Find a digit whose only possible positions in two rows are confined to the same two columns — the four cells form the corners of a rectangle. The digit must end up in two of those four corners (one per row, one per column), which means the digit can be eliminated from any other cell in those two columns. Our deep-dive on why the X-wing keeps tripping people up covers the recognition pattern in detail — most solvers learn the X-wing's definition long before they learn to spot it on a real grid, and the gap between knowing and seeing is where the tier's difficulty lives.

Swordfish is the X-wing scaled up. Three rows where a digit's positions are confined to the same three columns — the digit must occupy three of the nine cells in that 3×3 frame, one per row and one per column. Other cells in those three columns lose the digit as a candidate. Swordfish appears in roughly half of expert puzzles and in nearly every master puzzle; learning to spot it at expert is the right time. The pattern is harder to see than X-wings because the frame is larger and the empty positions look noisier, but the recognition habit is the same: scan by digit across rows or columns, looking for confined sets.

Locked candidates — both pointing and claiming — fire constantly at expert. They were available from medium on, but at expert they're the move that breaks most stalls before the X-wings and swordfish enter play. The piece on pointing pairs and the snake is worth re-reading at this tier; the patterns that looked occasional at hard now appear in nearly every box.

Hidden triples and naked triples

Hidden triples and naked triples appeared at hard and remain workhorse techniques at expert. Most expert puzzles contain at least one. The discipline is to scan for them before reaching for the more exotic patterns — solvers often jump straight to X-wing hunting and miss the naked triple in box 5 that would have unlocked the rest of the grid in two more placements. Our piece on naked and hidden pairs and triples covers the recognition habit.

When expert stalls

Expert stalls in two characteristic ways.

Pattern-blindness. The new techniques work by visual recognition, and recognition takes practice. A solver who knows the X-wing definition but hasn't yet trained the eye to spot one will look at a grid containing an X-wing and not see it. The fix is repetition — solving X-wings on puzzles known to contain them, until the four-corner pattern becomes visually salient. The piece on the puzzle that taught me the X-wing walks through one solver's version of that learning curve.

Single-axis fixation. Most solvers learn X-wings on rows first, and many never apply them systematically to columns. But X-wings fire equally often along the column axis, and a row-only scan misses half the pattern. When an expert puzzle stalls and a row-axis scan finds nothing, the next move is to scan columns by digit. The same applies to swordfish.

Two habits to drop

Bashing on a stall. Expert puzzles can stall for fifteen or twenty minutes, and staring harder doesn't help. The methods that work are systematic: re-pencil-mark, technique-by-technique scan (start with naked triples, then hidden triples, then locked candidates, then X-wings, then swordfish), then take a break. Bashing burns time and produces frustration without finding the move.

Skipping column scans. The single-axis fixation above earns its own bullet because it costs more time than any other expert habit. Build a routine that explicitly scans columns whenever you scan rows — it's the cheapest improvement most solvers can make.

When X-wings and swordfish start appearing on first inspection rather than on the third pass, you're ready for master, where the chain-based techniques (XY-wing, jellyfish, single-digit patterns) enter the rotation. Most solvers spend several months at expert before master feels accessible, and that's the right pacing. Expert puzzles reward repetition more than they reward studying new techniques.