Overview
The certified margin was 537 votes out of roughly six million. At that scale every administrative decision made before and during the count is large enough to have changed the result, which is the reason the argument has never resolved: almost any complaint about Florida in 2000 is, arithmetically, a complaint about the outcome.
The theories run from the procedural to the deliberate. At one end, a system too crude for a margin that small. At the other, a state administration run by the brother of one candidate, a secretary of state who co-chaired that candidate's Florida campaign, and a Supreme Court majority that halted the count.
What the record establishes
Some of the most-cited claims are not claims at all. The U.S. Commission on Civil Rights investigated the election and reported findings that are documented rather than alleged.
A contractor-built list intended to remove convicted felons from the rolls swept in people who were not felons. The list was assembled by matching names loosely, and county supervisors were given discretion over how to apply it; some checked it against other records, some did not, and there was no uniform way for a wrongly removed voter to be restored in time to vote.
The Commission found the process disproportionately affected Black voters, who made up roughly eleven per cent of Florida's registered voters and about forty-four per cent of the purge list. Estimates of how many eligible voters were removed vary, and the figure most often cited — around 91,000 — comes from analyses that later reviewers have both defended and disputed. The Commission described the outcome as a denial of the right to vote falling disproportionately on Black voters; it did not find that the purge had been designed to produce that result, and its members divided on how to characterise intent.
Ballot rejection rates were not evenly distributed. In precincts that were majority Black, spoiled-ballot rates ran far above those in majority-white precincts, in some counts by a factor of three. The equipment differed by county, and the counties using the least reliable equipment were not randomly distributed.
Palm Beach County's so-called butterfly ballot produced a pattern statisticians have studied since: an implausible vote for Pat Buchanan in precincts where he had no constituency, and roughly 19,000 ballots invalidated by two punches.
What the recount would have shown
After the election, eight news organisations paid the National Opinion Research Center at the University of Chicago to examine the uncounted ballots. Its results are the closest thing to an answer, and they are genuinely ambiguous.
Under the recounts actually being pursued in December 2000 — the partial, county-limited ones Gore had requested and the Florida Supreme Court had ordered — Bush finished ahead. Under a full statewide review of all uncounted ballots, applying any of several standards for what counts as a vote, Gore finished ahead.
Both of those sentences are true, which is why both sides quote the study. The recount that was stopped would probably have confirmed Bush; the recount nobody was asking for would probably have gone the other way.
The ruling
The recount ended by order of the United States Supreme Court on 12 December 2000, five votes to four. The majority held that counting ballots without a uniform standard for what marks counted as a vote violated equal protection, and that no constitutionally adequate recount could be completed in the time remaining before the deadline for certifying electors.
The opinion added that its consideration was "limited to the present circumstances" — a line quoted ever since by critics as an admission that the reasoning was not intended to function as law. That sentence does more work in the argument than any other single fact about the case, because a court declining to make its own reasoning precedent is unusual enough to invite the question of why.
Where the readings diverge
One reading holds that this is what a close election looks like when it meets imperfect machinery. Spoiled ballots, bad ballot design and flawed purge lists are failures of administration, and they fell hardest on Black voters because those voters lived in counties with worse equipment — an outcome of inequality rather than a plan. On this account the Supreme Court's intervention in Bush v. Gore was a defensible response to a state process that had become chaotic, whatever one makes of its reasoning.
The other reading holds that intent is visible in the pattern. The people running the count were not neutral parties; the purge contract, the equipment distribution and the decision to stop counting all cut the same way; and a ruling that expressly limited itself as precedent invites the suspicion that it was reached for the occasion.
A third position, held by some who accept most of the factual complaints, is that the decisive question was never the recount at all. On this view the election was determined before polling day by who was on the rolls and what machines they voted on, which makes the December litigation an argument about the smaller number.
Why it stays open
The 2000 election produced an unusual amount of documentation — a federal civil rights investigation, a commissioned academic ballot review, and a full evidentiary record from the litigation — and it still did not settle. The reason is that the documents answer factual questions and the dispute is mostly about intent, which the same documents support more than one reading of.