One of the closest and most dramatic Portland City Council races in at least two decades has finally been decided, after a historic instant runoff tie, a ceremonial salad bowl name-drawing to determine a preliminary winner, a hand recount that flipped the ceremonial result and ultimately a concession of defeat.

Roberto Rodriguez won the at-large council seat Wednesday, 35 votes ahead of Brandon Mazer. But questions remain about why 36 more votes were recorded on election night than at this week’s hand recount. While three dozen ballots out of more than 21,000 cast may not seem like much, they could have mattered in such a close race.

Elections Systems and Software, the Nebraska-based company that helps run Portland’s ranked-choice elections is working with the city to determine the cause of the discrepancy, confirmed a spokesperson, who did not respond to follow-up questions.

“We have pledged our assistance in determining why the hand count audit yielded different results,” said ES&S Senior Public Relations Manager Katina Grander in a statement. “In addition to paper ballots, our systems contain a number of auditing features to verify the correct outcome of the elections. We are confident that the process will provide full accountability for the results.”

When the discrepancy was discovered following the hand recount Tuesday night, the city’s election administrator returned to City Hall to make sure that no ballots had been left inside the vault.

After determining that all ballots were on hand, the city expressed confidence that the hand recount included all of the ballots cast and speculated that the difference was likely an overcount by the voting machines, stemming either from rejected ballots or jammed machines. That would equate to an average of three overcounts for each of the city’s 12 voting precincts.

Interim Corporation Counsel Jen Thompson said in an email Friday that she was working with the city clerk on a joint report about the discrepancy to present to the City Council. Thompson said she expected to post the report online as well.

“We’ve begun working on that this week and are aiming to complete our report sometime next week,” she said.

Anna Kellar, executive director of the League of Women Voters of Maine, said the league is eager to learn more about why the discrepancy exists. The league has requested more detailed election data in an effort to conduct its own analysis, said Kellar, who uses they/them pronouns.

“We’re not that concerned that this means there’s a wrong outcome,” they said. “I think we’re confident the election has been decided, that it was well-run and the recount overall was well done. But we think it’s important to get that closure to what happened with every ballot.”

A national expert on vote scanners says that ballot scanners keep internal logs of events and that it should be relatively easy to determine what happened – whether it was a computer glitch, a miscalibrated ballot scanner, jammed or kicked-back ballots or human error.

Duncan Buell, the NCR chair in computer science and engineering at the University of South Carolina, said he has been analyzing ES&S voting systems for the last 11 years, primarily in South Carolina. He said it should take only a few hours to examine the event logs from the voting machines used in Portland’s election to determine how many jams, kicked-back ballots or other events occurred on each machine.

“If it were possible to get access to the event logs for the scanners, it would be enormously helpful in understanding what happened,” Buell said.

The city’s theory seems to suggest that a voting machine could record a ballot being cast, without actually registering the result. But Buell said that shouldn’t happen.

“If that’s what the software is doing, then the software is deeply flawed,” he said. “It should not check off the existence of a ballot until it actually finishes scanning and processing a ballot.”

Buell said there have been rare cases in which results from an entire voting machine were not included in the vote tally, but if that happened here, he would have expected to see a larger discrepancy.

“Thirty-six is kind of a weird number,” he said.

Sometimes, such discrepancies can be attributed to human error.

“At the end of a 15-hour day, tired people make mistakes,” Buell said.

Kellar said the League of Women Voters of Maine requested the “cast vote record” from the city, which includes detailed ranking information for each ballot, and the event logs from each voting machine.

Kellar noted that Portland’s ranked-choice voting rules, established in 2011, differ slightly from the state’s rules, especially regarding exhausted ballots and how ballots that skip more than one ranking are treated.

For example, the city’s rules for skipped rankings allow more ballots to be counted. That’s because the state deems any ballots with two or more skipped rankings – say, a voter ranks only a first and fourth place candidate – to be exhausted after the first round, while the city’s rules would allow the candidate marked fourth to be counted in the second round, if that voter’s first round pick was eliminated.

In addition to the discrepancy in the overall vote tallies, Tuesday’s hand recount resulted in 37 disputed ballots, but Rodriguez and Mazer were awarded two votes each and the remaining 33 were deemed exhausted.

Kellar said the cast vote record would allow the league to run its own instant runoff and see whether ES&S may have inadvertently used the wrong set of rules for the instant runoff.

“Mostly, it’s not close enough that those things would make the difference on the winner,” Kellar said.

It’s not unusual for voting tallies to fluctuate slightly during the certification process, as auxiliary ballots, which could not be read by the machines, are added manually and finalized. But this year’s totals, especially on the islands, seemed to vary greatly.

For example, the unofficial results from Nov. 2 didn’t show any absentee ballots cast for the Peaks Island Council or Casco Bay Lines director races, but they were included on the revised election results released on Nov. 8. City Clerk Kathy Jones told the Bangor Daily News that a programming error on the city’s side did not allow the city to publicly display them that night. But Jones told the paper she was confident that the absentee totals from the islands were included in the at-large city council race total that initially ended in a tie.

The Nov. 2 tally also showed 1,008 blank ballots being cast in the Peaks Island advisory council race. But the revised results showed only 75 blank ballots and added a second candidate, Jerzy Sylvester, whose name was not on the ballot and who likely ran a write-in campaign. Sylvester received 75 votes, but Peter McLaughlin still won with 367 votes.

Peaks Island has its own polling location and 1,038 registered voters, but is part of District 1, which includes the East End, even though it has its own advisory council.

Mazer’s campaign was focused during the recount on the 36-vote discrepancy between the revised unofficial results (21,287 ballots) and the final hand recount (21,251.) However, both of those totals differ from the instant runoff total that ended in a tie. A total of 21,284 ballots were counted in the instant runoff, compared to 21,251 ballots counted by hand this week, a difference of 33 votes.

Thompson, the city attorney, told both campaigns during the recount that a representative from ES&S assured the city it was “not at all uncommon” for the machine totals to vary from the hand recount.

In his concession speech, Mazer commended city staff for how they handled the recount. He said he still had questions about the discrepancy, but didn’t think it was worth drawing out the election, given that the odds were not in his favor.

“I do have some questions on the discrepancy of numbers. But at this point, to move on, to have the city start doing the work that needs to be done, and knowing it was a pretty high hill to climb, it makes the most sense to me (to concede),” he said Wednesday. “It wasn’t an easy decision, by any means.”

Related Headlines


Only subscribers are eligible to post comments. Please subscribe or to participate in the conversation. Here’s why.

Use the form below to reset your password. When you've submitted your account email, we will send an email with a reset code.