A Forcing Chain compares the logical consequences of a set of exhaustive alternatives.
Instead of following one AIC endpoint relationship, you ask:
If each possible starting alternative were true, what conclusion appears in every branch?
If all alternatives force candidate Z off, Z can be eliminated. If all alternatives force Z on, Z can be placed.
The deduction is logical because the starting alternatives cover every possible valid case.
Forcing Chains are not random trial and error
The word “if” can make Forcing Chains sound like guessing.
The difference is that a valid forcing proof:
- chooses an exhaustive set of alternatives;
- follows only valid Sudoku implications;
- compares all branches;
- keeps only a conclusion shared by every branch.
You never choose the branch you prefer.
If branch A and branch B lead to different unresolved states with no common conclusion, there is no forcing move yet.
Digit Forcing Chain
A Digit Forcing Chain often begins from a candidate that can be either on or off, or from the alternatives for one digit in a restricted structure.
Example concept:
- if
r2c4=7, then target 5 is eliminated; - if
r2c4<>7, a Strong Link forces another 7, and that chain also eliminates target 5.
Because the starting candidate is either true or false, target 5 is eliminated in all cases.
Cell Forcing Chain
A Cell Forcing Chain starts from all candidates in one unsolved cell.
If a cell contains {2,5,8}, one of those three digits must be true.
Trace each branch:
- assume 2;
- assume 5;
- assume 8.
If all three branches force the same candidate off/on elsewhere, that conclusion is valid.
The cell's candidate set is exhaustive, so no fourth possibility exists.
Unit Forcing Chain
A Unit Forcing Chain starts from all possible locations of one digit in a house.
If row 7 has candidate 4 only in c2, c5, and c9, then one of those three 4s must be true.
Follow each possible true position. If every branch reaches the same conclusion, that conclusion is forced.
This is the position-based counterpart of Cell Forcing Chains.
Worked Forcing Chain
Verity: every branch reaches the same result
A forcing verity is a result confirmed in every alternative.
For example:
- branch A →
r5c5=9; - branch B →
r5c5=9; - branch C →
r5c5=9.
Therefore r5c5=9.
The common result may also be an elimination.
Contradiction-based forcing
Another valid form shows that one starting alternative leads to an impossible state:
- cell with no candidates;
- two true digits in one cell;
- no location for a required digit in a house;
- repeated fixed digit in a house.
If the starting alternatives are binary, disproving one confirms the other.
This logic overlaps with Nishio and contradiction solving, but the proof must remain complete and explicit.
Forcing Chains vs AIC
An AIC is linear: each link depends on the immediately previous inference, with Strong and Weak Links alternating.
A Forcing Chain may compare multiple linear chains that begin from different exhaustive alternatives.
The individual branches can themselves be AIC-like implication sequences.
The defining feature is the branch comparison, not just the presence of chains.
Forcing Chains vs Forcing Nets
A Forcing Net allows consequences inside one branch to fork and interact/reconverge in ways that cannot be represented as a simple linear chain.
That is more general — and much harder to audit manually.
Forcing Nets are intentionally outside this freeze. The V2 library stops at linear forcing chains so the boundary remains teachable.
Forcing Chains vs guessing
Guessing says:
I will try 7 and continue unless it breaks.
A forcing proof says:
I will account for every possible value/state here, and I will make a move only if all alternatives prove the same conclusion or all but one contradict Sudoku.
The second is a deductive case analysis.
How to search for forcing chains responsibly
Do not branch on arbitrary nine-way possibilities.
Prefer constrained starts:
- bivalue cells;
- trivalue cells with strong downstream consequences;
- digits with two or three positions in a house;
- candidate states connected to many Strong Links.
And stop as soon as a simpler technique becomes available.
Forcing methods are expensive for humans and should not replace ordinary scanning/Chains when those suffice.
Common mistakes
Not covering every starting alternative
Then the common result is not guaranteed.
Using an invalid implication inside one branch
One bad link invalidates the branch proof.
Comparing only the branches that support the desired target
All alternatives count.
Calling a long guess a Forcing Chain
The conclusion must be derived systematically from exhaustive logical cases.
Confusing a net with a chain
Branching internal consequences may move beyond the linear Forcing Chain model.
FAQ
Are Forcing Chains logical Sudoku techniques?
Yes, when the starting cases are exhaustive and every implication is valid. They are formal case analysis, not arbitrary guessing.
What are the main Forcing Chain types?
Common classifications include Digit, Cell and Unit Forcing Chains.
Are Forcing Chains the same as AIC?
No. AIC is a linear alternating-inference structure; forcing techniques compare consequences of multiple exhaustive alternatives.
What is a Forcing Net?
A more general branching consequence structure. It is intentionally left for post-freeze expert expansion.
Should I use Forcing Chains often?
Usually only on very hard puzzles after simpler direct patterns and Chains have been exhausted.
What to learn next
For this V2 freeze, Forcing Chains are the end of the formal technique ladder. Later expert expansion can add Forcing Nets, Nishio variants, ALS Chains and other methods once data justifies deeper coverage.