The Order to Try Candidate Eliminations in Sudoku
The optimal sudoku elimination order starts with unit-wide exclusions, then moves to box-line interactions, naked subsets, hidden subsets, and finally complex fish patterns and chains. This systematic approach minimizes scan time and prevents you from missing easy eliminations. Following a logical progression from simple to complex techniques is the most efficient way to solve any puzzle, as it ensures you apply low-cost scans before investing effort in intricate pattern searches. Our analysis of solving data confirms that players who adhere to this structured order solve puzzles faster and with less frustration.
Category 1: Unit-Wide Exclusions (Zero Scan Cost)
The first and fastest category involves checking each cell's row, column, and box to see which digits are already placed. This is the foundational step of candidate elimination and costs nothing because you must look at a cell's units to pencil in its candidates anyway. For example, if a row already contains the digits 1, 4, and 7, you can immediately exclude those three digits as possibilities for every empty cell in that row. This process, often called 'penciling in' or 'noting candidates,' is not a separate technique but a mandatory first scan that provides the raw data for all subsequent logic.
- Always perform unit-wide exclusion first when filling a cell's candidate list.
- This step has zero additional cost—you're gathering the information needed to proceed.
Category 2: Box-Line Interactions
Once all candidates are noted, the next most efficient scans are box-line interactions, which include Pointing Pairs/Triples and Box-Line Reduction. These patterns are easy to spot because they only require looking at the relationship between a single row/column and one 3x3 box. A Pointing Pair occurs when all candidates for a specific digit within a box are aligned in a single row or column. This alignment allows you to eliminate that digit from the rest of that row or column outside the box. Conversely, a Box-Line Reduction (or Claiming) happens when all candidates for a digit within a row or column are confined to one box, allowing you to remove that digit from the rest of the box.
- Scan each box for digits that are confined to one row or column (Pointing).
- Scan each row and column for digits confined to one box (Claiming/Box-Line Reduction).
Category 3: Naked Subsets (Pairs, Triples, Quads)
With basic interactions exhausted, shift focus to finding Naked Subsets within individual units (rows, columns, or boxes). A Naked Subset is a set of N cells that collectively contain exactly N distinct candidate digits. The classic example is a Naked Pair—two cells in a unit that only contain the same two candidates. Finding these allows you to eliminate those candidate digits from all other cells in the same unit. The scan cost is moderate; you check each unit for groups of cells with small, overlapping candidate sets. These patterns are more common than advanced fish and often reveal the next round of simpler moves.
- Look for pairs of cells in a unit sharing the same two candidates.
- Expand your search to triples (three cells with three shared digits) and quads.
Category 4: Hidden Subsets
Hidden Subsets are the logical counterpart to naked ones and are searched for in the same units. A Hidden Subset occurs when N digits are candidates in exactly N cells of a unit, and those cells may contain other candidates as well. Discovering a Hidden Pair, for instance, means two digits appear only in two cells of a unit, allowing you to remove all other candidates from those two cells. The scan cost is similar to that for naked subsets but requires looking at the distribution of each digit within a unit. It's efficient to check for hidden subsets immediately after naked ones, as they solve the same logical constraint from a different angle.
Category 5: Fish Patterns and Chains
The final category encompasses advanced, pattern-based techniques like X-Wing, Swordfish, and XY-Chains. These methods have a higher scan cost because they look for specific candidate alignments across multiple rows and columns. An X-Wing, for example, looks for a digit that is a candidate in exactly two cells in two different rows, with those cells also aligned in just two columns. This pattern allows for widespread eliminations. Because these patterns are rarer and harder to spot, they should only be sought after the simpler, high-probability categories are exhausted. For a full taxonomy of these advanced methods, see our guide on techniques for removing candidates.
- Search for basic Fish (X-Wing) before more complex ones (Swordfish, Jellyfish).
- If no Fish are found, look for shorter chains (XY-Wing) before longer, more complex ones.
How to Apply This Order in Practice
Treat this order as a dynamic checklist, not a rigid one-time sequence. After making any eliminations from a later category (like a Hidden Pair), always loop back to check for new Naked Singles or box-line interactions that may have been created. This back-and-forth is efficient; a complex move often reveals several simple ones. Most puzzles from easy to hard will be solved by cycling through the first four categories. Only the most difficult puzzles will require persistent searches in the fifth category. Using this ordered approach prevents you from wasting time looking for an X-Wing when a Pointing Pairs elimination is still available.
Key Facts
- ▪Unit-wide exclusion is the mandatory first step and has zero additional scan cost when noting candidates.
- ▪Box-line interactions (Pointing Pairs and Claiming) are the fastest dedicated elimination techniques to apply after basic notation.
- ▪Naked Subsets (Pairs, Triples) are searched for within individual rows, columns, and boxes.
- ▪Hidden Subsets solve the same logical constraint as Naked Subsets but from the perspective of digit positions.
- ▪Fish patterns like X-Wing and Swordfish require scanning candidate alignments across multiple units.
- ▪Chaining techniques (XY-Wing, XY-Chains) are typically the last resort before trial-and-error methods.
- ▪The optimal order is based on increasing scan cost and decreasing likelihood of occurrence in a typical puzzle.
- ▪After any elimination, you should always check if earlier, simpler techniques have become available again.
- ▪This structured order maximizes solving efficiency by applying low-effort, high-success-rate techniques first.
Frequently Asked Questions
Why should I try eliminations in a specific order?
It saves time. Simple techniques are faster to scan for and more common. Searching for a rare X-Wing before checking for a common Naked Pair is inefficient.
What does 'scan cost' mean in Sudoku?
Scan cost is the mental effort and time required to look for a specific pattern. Checking a single box for a Pointing Pair is low cost; scanning the whole grid for a Swordfish is high cost.
How do I know when to move from one category to the next?
Move on when a complete, careful pass over all relevant units (like all boxes for Pointing Pairs) yields no new eliminations for that technique.
Should I always go back to easier techniques?
Yes. After any elimination or placement, new opportunities for simpler techniques often appear. Always loop back to check for new Naked Singles or box-line interactions.
Do I need to know all these techniques to solve hard puzzles?
For very hard puzzles, yes. Mastering the first four categories solves most puzzles. The hardest puzzles require knowledge of fish and chains from the fifth category.
What is the biggest mistake in elimination order?
The biggest mistake is jumping to advanced patterns without exhausting simpler, faster checks first. This leads to wasted time and missed easy moves.