This coverage is made possible by a grant from The Saint Paul Foundation.

Tweaking ranked-choice voting: Minneapolis considering three changes

MinnPost file photo by Karl Pearson-Cater
A study of 2009 ranked-choice voting found that the biggest complaint about the process was the lengthy delay in knowing who the winners were on Election Night.

Members of the Minneapolis City Council got scolded by a member of the Charter Commission as they prepared to change some of the rules on how ranked-choice voting will be administered in this fall’s election.

“I would submit that 13 declared candidates for office, in an election year, five months prior to an election, have no business changing election laws,” said Devin Rice of the Charter Commission.

He also was critical of an earlier council decision to reduce funds available for voter education, given the incidence of voter error in the 2009 election.

That was the last time — and the first time — Minneapolis voters selected office holders by ranking their top three choices (PDF).

Errors in using ranked-choice voting showed up on 6.5 percent of ballots cast, Rice said.

“There is no question that the data that was published, and still resides on your website, identifies a dramatic increase in error in low-income, high-minority precincts than in white, affluent precincts,” he said.

Three ‘enhancements’ under review

Council members met Monday as the Election Committee to consider three “enhancements” (PDF) to city election rules (PDF) that involve:

• Determining voter intent when a ballot has been wrongly marked.

• Reporting results on Election Night.

• And implementing a new rule that would require write-in candidates to register with the city elections clerk.

The study of 2009 ranked-choice voting — by Hamline University professor David Schultz and then-Hamline student Kristi Rendahl — found that the biggest complaint about the process was the lengthy delay in knowing who the winners were on Election Night.

The study also noted that the low turnout (21 percent) might have made it a bad year to evaluate the new voting system.

“We had almost 10 percent of the ballots in that election that were incorrectly completed,” said Schultz. Of the 45,968 ballots cast in 2009, 1,888 were spoiled and 2,958 contained voter errors specific to ranked-choice voting.

“The worst-case scenario you get is you go into this election where it could be very close — a 3 or 4 or 5 percent error could be enough to affect the outcome,” said Schultz, looking at a mayoral race with seven candidates. “It could be the recipe for litigation.”

Complaints about slow counting

One problem that could be solved in this year’s elections is the ability to report winners in some of the races on Election Night.

In 2009, some candidates had to wait weeks for the ballots to be hand-counted.

City Council President Barb Johnson, for example, knew she had 47 percent of the vote on Election Night but waited three weeks to be declared the winner.

This year, the new Maximum Possible Threshold will determine the number of votes a candidate needs to be declared a winner, based on first-choice votes cast. That number will be based on ballots cast.

The old system — and the one currently used by St. Paul — based the threshold on votes (rather than ballots) cast (PDF) — which requires a complete counting and slows the reporting of winners.

In 2009, for example, even though Mayor R.T. Rybak had 73 percent of ballots cast, he could not be declared the winner until all of the ranked votes were counted.

“We should be able to process and declare winners where there are definitive winners,” said City Clerk Casey Carl, addressing the concern by some council members that vote counting would come to a halt once a candidate had reached the Maximum Possible Threshold and be declared the winner.

 “We can certainly go back and find statistical data that completes the calculations for each candidate,” he said.

“The first goal of this change was to declare as many races as possible on Election Night,” said Council Member Sandy Colvin Roy. “That will make people happy.”

In 2009, if a voter ranked too many candidates for an office — a situation called an “over-vote” — the ballot would have been rejected by the voting machine and the voter given a chance to recast the ballot.

New voting machines

This year, with new voting machines from Hennepin County, the ballot also would be rejected, but the voter could turn down the option to vote again.

In that case, assume the voter selected two first-choice candidates, those selections would be voided and the second-choice candidate would become the first choice. But over-voting is the only error the machines will reject.

This year, a voter who votes for the same candidate for all three choices, called repeat candidate voting, will give that candidate one first-choice vote but will not have the second and third choices counted.

This voting practice will not be detected by the voting machines and the voter will not have a chance to correct or change his or her ballot. This is unchanged from 2009.

Another practice, called skipped ranking, also will not be detected by the voting machines, so voters will not be have a chance to correct their ballot. In skipped ranking, for example, the voter could make a first choice but leave the second choice blank and then fill in a third choice. The first choice would be counted, but the third choice would move up to second choice in the counting.

In 2009, a skipped-ranking voter who made only a third-place choice in a contest would not have the vote counted at all. In 2013, such a voter would have that third choice counted as a first choice.

“It’s about voter education,” said Terra Cole, executive director for the Heritage Park Neighborhood Association in the 5th Ward, which had the highest number of voter errors in the 2009 election. “We tend to have a lot of folks who are voting for the first time or moving from other places. It’s a transient neighborhood.”

She used the example of explaining ranked-choice voting to her mother, who has a Ph.D., by equating the process to filling out your meal preference for a party or a conference: Do you want the beef, the chicken or the vegetable?

“When we educate our community of color about how to vote, it’s very easy,” said Cole, adding that potential voters in her community expect to rank more than three candidates because there will be more than three candidates in some races.

“There is no evidence that more rankings have imposed any greater difficulty,” said Jeanne Massey of FairVote Minnesota, a leading advocate for ranked-choice voting.

She told council members that in 2009, Portland, Ore., found that half of the voters ranked more than three candidates. “The easy instruction is to rank as far as you can and stop ranking.”

Council members expressed interest in expanding the rankings for the 2013 election to as many as five or six choices but decided to study the possibility until Thursday’s Committee of the Whole meeting.

New rule proposed for write-ins

The final change approved by the Committee would require write-in candidates to register with the city elections clerk at least seven days before the election if they want their votes counted and recorded.

In 2009, 3,221 write-in votes were cast, and each had to be hand-counted.

The registration requirement will still allow voters to cast their ballot for Mickey Mouse or Donald Duck, but there will be no count beyond the total number of write-in votes for non-registered write-ins.

It’s estimated that this change would cut the time required to count write-ins by 60 to 75 percent.

The changes, approved by the Committee, and the possibility of expanding the choices beyond the current three, will be discussed Thursday session, with a vote expected at Friday’s City Council meeting.

Comments (10)

  1. Submitted by Tim Saxton on 05/23/2013 - 11:05 am.

    With Ranked Choice, over 50% wins.

    If somebody as 73% of the vote, and the and the vote was counted, it’s over.

    Ranked choice ends when A) Somebody has more than 50% of the vote, or B) only one candidate is left.

    If nobody is over 50%, and there are more than one candidate, the lowest ranking candidate is dropped, and people who voted for him/her use their next ranked vote. This process repeats until winning conditions are met.

    If a candidate has over 50% of the vote, nothing could happen to reduce that count under 50%. It could go up as the ranked-choice process continues, but it can’t go down.

  2. Submitted by Connie Sullivan on 05/23/2013 - 12:10 pm.

    Just because IRV is more complicated than the “pick one” ballots we’re used to casting doesn’t mean that all votes shouldn’t be counted. If a candidate has more than 50% of the first-choice votes, fine, let her be declared the winner.
    But please: count all the votes. It is impossible to do any “study” of how Minneapolis elections work under IRV, and of how voters do with it, unless every part of the process is recorded, made public, and studied.

    With all the counting complications that IRV entails, now is not the time for its enthusiasts to get impatient with those of us who see little [or big] holes in it.

    • Submitted by Ken Bearman on 05/23/2013 - 01:08 pm.

      What a vote is

      A voter’s rank is a vote only if it’s the voter’s highest rank and is counted in a given round. None of the lower rankings are votes.

      If more than one candidate ranking were counted, that voter would have had more than one vote in the round. That would be a violation of the U.S. Constitution’s one-person, one-vote requirement.

      Furthermore, counting a voter’s lower rankings in a round (1) results in those working against that voter’s highest ranked candidate; and (2) results in more “votes” than voters. That’s been unacceptable in Minnesota for 98 years since our Supreme Court decided Brown v. Smallwood.

  3. Submitted by John Ferman on 05/23/2013 - 12:29 pm.

    With Ranked Choice

    The sad part of dropping the lowest vote tally candidate is the possibility of only one vote differentiate the lowest two candidates. There needs to be a much more thoughtful criteria for the dropping process.

    • Submitted by Ken Bearman on 05/23/2013 - 01:17 pm.

      One-vote difference

      In RCV, one more voter chose A than those who chose B, so A wins and B loses and is dropped as between the two of them. What’s sad (for B) is that two more people didn’t vote for or rank her or him. What do you suggest changing?

      It was the same in the old plurality winner system. If the count was a one-vote difference at the top, the winner was the one with the most votes. Was that always an unthoughtful criterion?

      • Submitted by Dan Hintz on 05/23/2013 - 02:03 pm.

        Not talking about the winner

        The problem is the difference between, say, the 5th and 6th place candidates could determine the winner of the election. If Candidate A comes in 5th place by one vote over candidate B, then candidate B’s votes get reallocated to other candidates. But if candidate B comes in 5th place by one vote, then candidate A’s vote’s get reallocated. The winner of the election could be determined by the less popular candidate’s (the 6th place candidate) second and third ranked choices, while the more popular candidate’s (the 5th place candidate) second and third choices aren’t counted simply because he or she got one more vote.

        There is no way to fix this. RCV is just fundamentally flawed and undemocratic.

        • Submitted by Ken Bearman on 05/23/2013 - 03:30 pm.

          More means more

          If 5th place candidate A has more votes, those votes continue to be counted for A in the next round. Those lower rankings don’t count against A’s voters’ highest ranks for A.

          If 6th place candidate B has the fewest votes, then B is dropped and those votes aren’t counted for B. You’re correct, her or his next highest rankings get reallocated for the next round. B lost to A for 5th place.

          More votes is more votes, whether it’s for 5th place or first place. You can’t have it two different ways unless you can explain why “more” sometimes isn’t “more”.

          • Submitted by Dan Hintz on 05/23/2013 - 04:12 pm.

            You completely missed the point

            Let me try to explain it again.

            The only race that matters is between the first and second place candidates. If no candidate has a majority, you have to reallocate votes. But the order that votes are reallocated can make a difference. Lets say that the second place votes from candidate A in 6th place put another candidate over 50 percent. In that event, the second place votes of candidate B aren’t counted. But if candidate B finshed 6th, then his or her second place votes would count and candidate A’s would not. If candidate B’s second place votes put a different candidate over 50 percent, then its the race between 5th and 6th that determines the winner, which is absurd.

            • Submitted by Steve Titterud on 05/24/2013 - 08:17 am.

              A fallacy

              What you don’t seem to understand is that within the algorithm of RCV, your statement that “The only race that matters is between the first and second place candidates” is a mistaken belief, a FALLACY – a vestige carried over from the prior system. It used to be close to true. It is no longer true.

              By its very rules, EVERY ranked choice has potential value – even 5th and 6th place preferences – and this potential value may, at some point, and depending on the circumstances and business rules, come into play to have an actual influence on the outcome. By the same rules, you may see the influence of an early Round voting choice elapse in a later Round.

              The individual voting choices’ importance shifts and changes from Round to Round as the algorithm seeks a winner based upon the preponderance of ALL voting choices, not just the 1st choice and 2nd choice (in your thinking, “The only race that matters…”).

              Your statement above (“The only race that matters..”) basically describes a rigid adherence of belief in the system that gave us the two party DFL/GOP duopoly. This is the very system that claimed if you voted other than DFL or GOP, you wasted your vote. It has delivered unresponsive, dysfunctional government – the closed loop that RCV supporters hope to open with this new voting regime.

              RCV may not be the cure – but we are trying to uproot a long entrenched, corrupt duopoly that does not serve the people well. It is worth a try.

              • Submitted by Dan Hintz on 05/24/2013 - 10:03 am.

                You completely missed the point too

                In saying that only the first or second choices matter, my point was only that there is one winner. And the problem with RCV is that the winner can be determines by which votes are counted depending on the unpopularity of losing candidates. Its about math, not ideology.

                RCV can alter election results. But it can do so by replacing democracy with a flawed formula that may result in a winner people did not want. You can’t fix a corrupt duopoly with a corrupt voting system.

Leave a Reply