Counting rules

Every counting system begins with a threshold question: which marks on a ballot paper will be read as the voter's intention, and which will not? The answer is not obvious. A cross, a tick, a circle, a filled square, a number — different voters reach naturally for different symbols, and different electoral systems expect different things. In a simple plurality contest, a single unambiguous mark in one box is all that is needed; in a ranked system, a chain of numerals must be legible, sequential and unambiguous before the paper can be transferred at all.

Most jurisdictions codify the rule in two parts. First, a set of marks that are unambiguously valid: the clean cross in the correct box, the numbers 1, 2, 3 in clearly distinct cells. Second, a set of marks that are presumptively invalid: two or more boxes marked in a contest that expects one, a paper left entirely blank, a mark that cannot be attributed to any candidate. Between those two bands lies the disputed territory — the smudge that might be a cross, the 1 that wanders between two boxes, the tick accompanied by a scrawled note.

That disputed territory is where the counting rules do their real work. England and Wales, for instance, direct returning officers to count a paper if the voter's intention can be determined with reasonable certainty, and to reject it otherwise. The phrase "reasonable certainty" is not arithmetic; it is a judgment, and it is made by a human being under time pressure, in a counting hall, with candidates' agents watching over each shoulder.

In most systems the returning officer — or the equivalent presiding official — carries formal authority over disputed papers. Agents for each candidate may object to any ruling; the returning officer hears the objection and decides. Papers placed in the "rejected" pile are kept separately and, in many jurisdictions, their total is announced alongside the valid vote, so that the scale of rejection is on the public record.

The agents' role is adversarial by design. They are entitled to inspect each doubtful paper before a ruling is made, and their presence is supposed to guard against the accidental — or otherwise — disappearance of valid votes. Recount rules vary by jurisdiction: some re-sort and re-total only the papers already accepted, while others also allow doubtful or rejected papers to be re-adjudicated.

The deeper principle is temporal: the validity standard must be fixed before polling day, and preferably long before. A rule written after the fact — or interpreted in a way that shifts from one pile of disputed papers to the next — is effectively a way of choosing winners by choosing rules.

The 2000 United States presidential election made this legible to a wide public: the absence of a uniform statewide standard for what constituted a valid mark on a punch-card ballot meant that different county canvassing boards were applying different tests to the same category of paper. The United States Supreme Court's ruling in Bush v. Gore held that the varying standards violated the equal protection clause, which halted the recount rather than resolving it — a reminder that an unresolved rule question can stop the count altogether.

Ballot design interacts directly with validity rates: a confusing layout produces more ambiguous papers, not because voters are careless, but because the paper invites the wrong kind of mark. Ballot design and counting rules are, in practice, the two ends of the same problem.