Adithya Sambamurthy/The Bay Citizen

Voters fill out their ballots, an hour before polls closed on Tuesday, November 8,2011

Sixteen percent of San Francisco voters who filled out their ballots correctly and completely -- more than 31,500 people -- did not have a say in the final outcome of the city's mayoral race, according to The Bay Citizen's analysis of election results.

Their ballots were discarded or exhausted, because they did not list either Ed Lee, the eventual winner, or runner-up John Avalos as one of their top three candidates. Unlike other cities, San Francisco does not allow voters to rank all the candidates on the ballot.

The analysis renews questions about the fairness of the city's ranked-choice voting system at a time when supervisors are considering repealing it.

Critics of the system say it quietly disenfranchises a significant percentage of voters, because those voters cannot fully express their preferences. San Francisco's voting machines only allowed voters to select three of the 16 candidates on the ballot.

“When you have more candidates like this and you constrain the number of choices to three, you have a lot of exhausted ballots and a less than ideal election,” said Corey Cook, a political science professor at the University of San Francisco, who is working with The Bay Citizen to analyze the election results.

Under ranked-choice voting, if one candidate earns a majority -- 50 percent plus one vote or more -- he wins. That is what happened in 2007, when Gavin Newsom was re-elected with 74 percent of the vote.

But if no one wins a majority, the candidate with the fewest votes is eliminated, and his votes are redistributed. The process continues for as many rounds as it takes until one candidate wins a majority of the remaining ballots. According to the Department of Elections, it took 12 rounds to determine Lee won the mayor's race.

As more ballots are exhausted with each round, the number of votes used to determine a winner decreases, allowing candidates to gain office with support from a minority of voters. In the final round, Lee had only 43 percent of all votes and Avalos had 29 percent.

(Click here to see the round-by-round results.)

According to Cook, every exhausted ballot amounted to half a vote for Lee, because it reduced the likelihood of someone catching up to him. In every round, Lee's support grew as more ballots were discarded, reducing the number of votes available to his competitors.

Critics of ranked-choice voting point to the results of San Francisco's mayoral race as evidence of what is wrong with the system: It favors incumbents, and it rarely requires a majority of voters to determine a winner.

“It’s the disingenuousness of the system,” said Terry Reilly, who has helped several campaigns across the U.S. and abroad fight or repeal measures to implement ranked-choice voting.

On Election Day, Supervisor Mark Farrell introduced a ballot initiative to eliminate ranked-choice voting. Supervisor Sean Elsbernd, who supports Farrell's proposal, said that under the system nine of 11 of the city's supervisors were elected without a clear majority from the outset -- including Farrell, who overcame a first-round deficit to Janet Reilly last year to win the District 2 seat.

But supporters of ranked-choice voting say that the problems with San Francisco's implementation of the system can be easily fixed. Rob Richie, executive director of the ranked-choice voting advocacy group FairVote, described the city's problems as a “short term technology issue.” Richie said that other cities, like Portland, Oregon, have ballots allowing voters to rank as many candidates as they choose. If someone chooses not to rank a candidate in Portland's system, that is about equal to declaring no preference for that candidate, he said.

Despite the large number of candidates, very few voters were confused by ranked-choice voting. Only 2 percent of voters made a mistake on their ballot, and more than 73 percent of voters filled out their ballots completely, selecting three different candidates for their top three choices.

More than 52,000 ballots -- 26 percent of all ballots cast -- were exhausted. That stands in stark contrast to last year's ranked choice mayoral race in Oakland last year, where just 11 percent of ballots were exhausted, even though voters could only select three candidates.

But in Oakland, there were only three strong candidates: Don Perata, Rebecca Kaplan, and Jean Quan, the eventual winner. In San Francisco, the field was stronger, with ten widely-known candidates.

"Mathematically when you’re limited to three choices, the more candidates you have the more strong candidates you have, voters are more likely to spread their rankings around," Richie said.

When San Franciscans approved ranked-choice voting in 2002, proponents of the system said it would increase voter turnout and reduce costs by eliminating a December runoff election.

In November's mayoral race, turnout was 42.5 percent, a bit below average for a mayoral election.

Steven Hill, the so-called father of ranked-choice voting, said the turnout and the high percentage of exhausted ballots is misleading. He said exhausted ballots are like people who show up for a November election, but stay home if there’s a December run off.

In an email, Hill wrote that the number of voters who don't cast a ballot in a traditional runoff election is significantly larger than the number of exhausted ballots in ranked-choice elections.

But Reilly said the most accurate measure of whether the system works is the number of people who can participate in the final round of the process.

As long as San Francisco only allows voters to rank three candidates, the system may be deficient, Cook said.

“If people had a strong preference and didn’t get to articulate it, then you have a problem with our voting system,” he said.