The Algebra of Election Certification: An Equation That Can't Close
Every certified election rests on one assumption: that the voter roll, R, is identical to the set of true qualified voters, Q. If |Q| = X, certification implicitly assumes |R| = X — that every record on the roll corresponds to exactly one real, eligible voter, and vice versa.
Direct analysis of voter rolls across twelve-plus states shows this assumption is false. In practice, |R| = X N, where N is a measurable surplus of records that do not correspond to genuine qualified voters: clones (duplicate registrations for one person), ghosts (deceased voters never removed), ineligibles (non-citizens, out-of-state movers, and others disqualified), and fictitious entries.
N is never zero, and it is not trivial:
Kansas and Oklahoma (~2 million voters each): 7,500–15,000 cloned records identified.
New York (~21 million voters): approximately 1.5 million cloned or excess IDs.
Even the cleanest of these rolls falls outside the error tolerances that any bank, hospital, or corporation would be permitted under standard data-integrity audits.
Here is why N > 0 breaks certification itself, not just data hygiene. When officials certify an election, they verify that ballots cast (B) correspond to records on the roll — B ⊆ R. But the legal requirement is B ⊆ Q: that ballots came only from real, qualified voters. The inference from B ⊆ R to B ⊆ Q holds only if R = Q, which requires N = 0. Since N > 0, and no real-time system exists to identify and subtract N at scale, that inference cannot be made. Certification asserts a conclusion the underlying data cannot support.
This is not a demand for perfection — it is a demand that the equation close. Until rolls achieve N = 0, through unique identifiers, aggressive maintenance, and zero tolerance for clones and ghosts, no certification can be said to reflect Q rather than Q N.