The butterfly ballot

Yes: a peer-reviewed study by Jonathan Wand and colleagues found strong evidence of systematic mistaken choices in Palm Beach County. Their estimate was that more than 2,000 Democratic voters selected Pat Buchanan unintentionally—enough to exceed the certified Florida margin. The conclusion is an estimate of behavior, not a list of identifiable people whose ballots can simply be transferred.[1]

The researchers tested competing explanations using several datasets. Buchanan’s election-day support in Palm Beach was about four times his absentee share; absentee voters used a different ballot. That discrepancy was unusually large compared with other Florida counties. Precinct patterns and other choices on the same ballots also supported the confusion explanation. An unusual result alone would have been weaker evidence than these comparisons.[1]

The legal challenge asked a different question. In Fladell, Florida’s Supreme Court held that the ballot’s form did not amount to the substantial statutory noncompliance required to void the election. It dismissed that challenge without deciding that the design was clear or that nobody had made a mistake.[2]

These findings can coexist. A ballot can satisfy the court’s threshold while still confusing voters in ways large enough to matter. The study supports that counterfactual; it does not establish that officials deliberately designed a trap, and a judge could not turn its estimated preferences into individually authenticated votes.

An overvote records more selections than the contest allows. An undervote records fewer than allowed, including no detected presidential choice. These categories describe what the voting system recorded; they do not explain why it happened. An intentional blank choice and a failed attempt to select a candidate can both appear as undervotes.[1]

On a punch-card ballot, the question is whether the mark left by the voter constitutes a valid choice. A partially detached paper fragment does not come with a reliable label saying “deliberate vote” or “accidental indentation.” Conversely, a machine’s inability to read a choice does not prove that the voter intended to abstain. This is why inspection rules and consistent standards matter: disagreement about classification can change a tally without anyone manufacturing a ballot. Saltman’s NIST lecture traces the long-known physical limitations of this technology.[3]

GAO’s nationwide statistical review found that state-level differences explained more variation in uncounted votes than county demographics or equipment alone. It identified administration, voter education and valid-vote rules among the possible contributors. The audit supports looking at the whole process rather than blaming a single machine or treating every residual vote as fraud.[2]

For this file, keep three questions separate: was a ballot cast, is a presidential preference discernible under a stated rule, and can that preference legally be counted? A fair retrospective comparison must state the rule and the ballots examined before reporting who benefits.