Not only will it take weeks before all results are known from the first attempt at ranked-choice voting in Minneapolis, but some of the election-night results will be misleading.

That was the picture painted Thursday by city election officials as they conducted a run-through for reporters of the complicated vote-counting scheme.

City officials expressed confidence that, on election night, they will be able to post how many people voted and the number of first-, second- and third-choice votes cast for each candidate. Trouble is, for each race they won't yet know the key number that determines who wins: the "threshold," the minimum number of votes a candidate needs to be elected.

Although it will be safe to say that candidates who post overwhelming margins are elected, the absence of the key number will make predicting a winner hazardous in closer races.

Candidates for single-seat races, which can have only one winner, need to amass one more vote than 50 percent to win. But that is 50 percent of the valid ballots, after errant ballots are removed. And the city won't start determining that threshold for winning -- which will vary by race -- until the day after the election. A lower threshold will be set for races such as Park Board where there will be multiple winners.

The removal of errant ballots before setting the threshold is mandated by the ordinance that governs the ranked-choice system approved by voters in 2006. Voters can have their vote thrown out in a given race by ranking two candidates as their first choice, for example. Jeanne Massey of FairVote Minnesota, which advocated for the new methods, said she expects an error rate of less than 1 percent of votes cast.

Another reason for the delay in tallying is that ballot counters will need to tally write-in votes before establishing the threshold.

The city plans to transmit periodic updates of results on election night to the Secretary of State's office for posting on a state website. Those will include the first-, second- and third-choice votes for each candidate. But the second- and third-choice votes won't mean much until first-choice votes alone don't put someone over the threshold.

That's the stage in single-seat races at which lower-ranking candidates on first-choice votes are dropped, and the second- or third-choice votes on those ballots are tallied. That was needed in more than one-third of the most recent ranked-choice races in San Francisco and Pierce County, Wash.

How much those added choices can shift an outcome can be illustrated by the race for county executive in Pierce County last fall, where the eventual winner ranked almost 9 percentage points behind after the first ballot.

The good news for voters is that it's easier to cast a ranked-choice ballot than to count them. A voter may rank up to three candidates in order of preference, or write in a substitute.

One result that voters should know election night is whether a charter amendment involving the Board of Estimate and Taxation passes or fails because that's a yes-no vote without ranked choices.

Unless there are clear-cut winners on first-choice votes, the races for the three citywide seats on the Park Board and two on the Board of Estimate and Taxation will be counted last. They feature a more complicated counting regimen in which both surplus votes over the threshold accumulated by winners and votes cast by supporters of dropped candidates are redistributed according to their second or third choices.

Steve Brandt • 612-673-4438