Nishio is single-candidate trial logic. Pick a candidate, place it tentatively, follow every forced move. If the chain breaks the grid, the candidate was wrong. erase it.
How it works
Pick an empty cell with a candidate worth testing. Place that candidate as a trial. From that placement, apply every forced move the grid allows: hidden singles, naked singles, pointing pairs, anything that does not require a second assumption. The chain runs until it stalls, produces a contradiction, or completes the puzzle.
A contradiction is two of the same digit in a row, column or box, or a cell with no remaining candidates. When that happens, the original trial candidate cannot be the answer. Erase it.
If the chain runs to completion, you have solved the puzzle. If the chain stalls without contradiction or completion, undo the trial and try a different candidate.
When to look for it
Last resort. Purists consider Nishio guessing; pragmatists use it when all other techniques fail. By the time you reach for it, coloring, XY-wing and forcing chains have already come up short.
Pick starting candidates with rich consequences. A candidate that triggers many immediate forced moves is more likely to produce a contradiction within a short chain.
Step-by-step example
- Pick a bivalue cell with candidates {A, B}. Choose one candidate to test.
- Tentatively place the candidate. Use a different pencil colour so trial moves remain distinguishable from real ones.
- Run forward. Apply every forced move (hidden singles, naked singles, pointing pairs). Stop when no more forced moves apply or when something breaks.
- If the chain hits a contradiction, the original candidate was wrong. Undo every trial placement and erase the candidate from the starting cell.
- If the chain stalls without conclusion, undo and try the other candidate, or pick a new starting cell.
Nishio vs forcing chains
Tips for spotting it
- Pick starting candidates that trigger rich forward chains. Bivalue cells in tight units are best.
- Use coloured pencil marks or digital trial mode so trial placements are obvious.
- Cap each trial at a reasonable depth. If you have written twenty trial placements with no conclusion, switch starts.
- Document the chain order. Erasing in reverse keeps things clean if the branch fails.
Common mistakes
- Forgetting to undo trial placements when the branch fails. Catastrophic.
- Calling it guessing. The elimination is logical: the contradiction proves the candidate impossible.
- Reaching for Nishio before simpler techniques. Most evil puzzles yield to coloring or forcing chains first.
- Mixing real placements with trials by using the same notation. Always distinguish.
Practise it
Reserve Nishio for the toughest puzzles on evil. When everything else has stalled, pick a candidate with strong consequences and follow the chain. The skill carries directly back to forcing chains; every Nishio trial is half of a forcing chain pair.