Nonprofit, nonpartisan journalism. Supported by readers.

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

Ranked-choice-voting reality: Theoretical ‘perfect case’ doesn’t happen

REUTERS/Derek Hauck

Second in an occasional series on ranked-choice voting.

If you haven’t experienced it or find it hard to grasp, here is how ranked-choice voting (RCV) works, in the theoretical perfect case.

Suppose there were five candidates running for mayor of your city. The ballot would enable you to rank them, one through five, in the order in which you would prefer each of them to be mayor.

In our perfect case, every voter would rank all the way from one to five. If a candidate was chosen as the first preference of a majority of those voting, he or she would win and there would be no need to do any further counting.

In the (fairly likely) case that no candidate was the first choice of a majority, the candidate who finished last (fifth, in our example) in the number of first-preference votes s/he attracted would be eliminated. If you had voted for that candidate, your support would be transferred to the candidate whom you had ranked as your second choice. If the counting of those (relatively few, since this candidate ran fifth) second-preference votes pushed one of the remaining candidates above the 50 percent mark — game over, a mayor is elected who was at least the first or second choice of a majority of voters.

If not, the candidate who finished on the bottom (fourth place, in our example) in the second round of counting would be eliminated and those votes would be credited to the highest-ranked candidate who was still in the race. If still no one had a majority, the bottom finisher would be eliminated, and his/her supporters would be counted as supporters of the remaining candidate who was ranked highest on those ballots. And so on.

RCV produces series of instant runoffs

 (Perhaps this makes clear why ranked-choice voting is sometimes called instant-runoff voting. It’s as if, in a theoretical sense, you had gathered all the voters and asked them whom they would support if these five candidates were their options. Then the last-place finisher would be eliminated and you would ask the still-impaneled electorate: “And now whom would you support if you had only these four choices?” And so on, with each round of counting functioning as a runoff election among the surviving candidates.)

Back to our theoretical perfect example: If no one achieved a majority after three rounds of counting, then only two candidates would be left standing and every ballot would be assigned to whichever of those two had the higher ranking on every single ballot.

 So, on the fourth round, unless those two tied exactly, one of them would have a majority and be the winner.

Video RCV demonstration

In case you prefer to visualize this, with music and narration, rather than just read my (oh-so-20th-century) typing, here is a charming, short (less than two minutes) film explaining the system. It was made by FairVote Minnesota, the chief group promoting ranked-choice-voting in our state.

In our theoretical perfect example, and in the Fair Vote film (even though the candidates there are running for best-president-ever) the new mayor would have at least a sort of majority of the vote — and whatever advantages that creates for the real or perceived legitimacy of his or her mandate to lead the city.

It is in the eye of the beholder whether a mandate achieved this way is more legitimate than one achieved through the pre-existing, first-past-the-post, plurality winner system. At the least, as I mentioned in the first installment of this occasional series, supporters of the runner-up would be less tormented by the belief that their candidate would have won if only third and fourth candidates hadn’t been drawing a portion of the vote.

Rice an outspoken opponent

But I keep calling the above example theoretical and perfect because, in the real world, there is no guarantee that the mayor-elect could claim a true majority of all the voters who turned out. This seems to be one of the key shortcomings of ranked-choice-voting in the eyes of Devin Rice, who sits on the Minneapolis Charter Commission.

Rice, one of the most outspoken opponents of RCV, has written for the Strib op-ed page criticizing the new system. He told me that although he has not tried to introduce a charter amendment to repeal RCV, he is open to the idea and will reassess after this fall’s big trial of the new system.

 One thing that sets Rice off is the claim (he calls it a “myth”) by proponents that RCV guarantees a majority mandate for the winner. In fact, in 2009, the first time Minneapolis used the new system, the race for Park and Recreation Board in the 5th District produced a winner who received support – any rank of support – on just 46 percent of the total ballots cast in the race. In San Francisco and Oakland, which have used RCV longer, Rice said there have been instances of races being won by candidates who received support on 43 (San Fran) and 42 percent (Oakland) of all ballots.

Effective DemocracyHow is this possible? There are two keys ways that RCV can fail to produce a winning candidate who received support on a majority of all ballots cast.

 The first way is that a voter could fail to rank all of the candidates. Then, if their ballots ran out of ranked preferences before the final round of counting, their ballots would be what, in RCV talk, is called “exhausted.”

Back to the example above. In the perfect case, five candidates on the ballot and each voter ranks them all the way from one to five in order of their preference. But in reality, many voters will not rank five choices. Perhaps it is a lot to ask of an average voter to know the field well enough to actually have ranked feelings across five choices.

In any event, a voter who ranked fewer than five will have his or her vote counted until the count reaches the point at which there is no candidate still alive in the counting for whom that voter has expressed any preference at all. That voter’s ballot is deemed to be “exhausted,” meaning there is no way to count it in the final determination of a winner. Once you start having ballots that can’t be counted in the final round, you can’t be guaranteed that the ultimate winner will have at least some measure of support from a majority of all the voters who participated in the election.

FairVote’s response

In a Strib op-ed responding to Rice’s op-ed, Jeanne Massey of FairVote Minnesota wrote:

 “As to the concern that RCV does not produce majority winners, the winner — in a single-seat race — is always the candidate with the majority of continuing ballots in the final round. In some situations, as in the case Rice points to, this is a plurality of initial ballots cast because some voters express only one preference and don’t have a candidate in the race in the final round of counting. This is the will of the voter.”

When I talked to him, Rice was particularly exercised about the “majority of continuing ballots in the final round” language, which to him is a fancy way of dismissing those whose ballots have been exhausted and fuzzing up the basic idea of a majority mandate.

“A majority means 50 percent of all the ballots,” he said. “No, Bigfoot isn’t real. Yes, our president was born in Hawaii. Global warming is real. We descended from apes. And no, ranked-choice voting does not assure majority outcomes.”

It is, of course, true that the traditional first-past-the-post plurality system is fully capable of producing winners with less than a majority of votes and, in Minnesota’s recent history, often has, especially in the high-profile statewide races for top offices. If you accept the validity of counting second- and third-preference votes, RCV seems substantially more likely than the traditional system to produce majority winners, even under Rice’s stricter (but reasonable) definition of what constitutes a majority.

But there is one last problem we should touch on before we wrap up for the day.

Massey correctly notes that a voter who ranks only one preference has (more or less, depending on what level of understanding you attribute to this theoretical voter) made a choice that if his or her first choice is eliminated, not to express a preference among the remaining candidates. That’s what she meant in the excerpt above by “This is the will of the voter.”

But in this year’s big Minneapolis mayoral race, it is quite possible (and very likely to happen) that some voters will fill out their ballots fully and still not have their preference counted in the final round.

Two key numbers

That’s because of two numbers: 3 and 35.

Thirty-five, rather amazingly, is the number of candidates who will appear on the ballot for mayor of Minneapolis. Three is the number of ranked preferences that the ballot will allow each voter to express.

At the very least, it is very possible that a voter (and perhaps more than a few) will actually rank three choices and — not because they chose not to rank any more but because they couldn’t — their ballot will nonetheless be “exhausted” because none of their three choice were still alive in the final round of counting.

True, Massey acknowledged when I asked about it. But this is not because of any inherent flaw in the RCV idea. Minneapolis has chosen to allow anyone to run who pays $20 and wants the fun or ego trip of seeing his or her name on the ballot. Most cities have higher requirements, such as the ability to gather a certain number of signatures, that would surely have discouraged some of the most frivolous of the Minneapolis 35, and Massey’s organization thinks Minneapolis should.

The administrators of the Minneapolis election have decided to limit voters to three choices. Massey wishes they had gone to at least six but sympathizes with the various contingencies that led to the decision to limit it to three, for this election. She said the experience of jurisdictions that have been using RCV suggests that almost no voters would have their choices involuntarily exhausted if they could rank six preferences.

The city’s new more-computerized election machinery will make it easier to handle six rankings and she hopes the city will expand the number of choices in future elections. “The trouble is,” she added, “there won’t be another mayor’s race like this for a long time.”

Effective Democracy is a year-long series of occasional reports supported by the Chicago-based Joyce Foundation, as part of a grant made to MinnPost and the Wisconsin Center for Investigative Journalism.

Comments (44)

  1. Submitted by Gregory Stricherz on 08/26/2013 - 08:49 am.

    What’s the answer?

    What if a voter is able to hold their nose and vote for only one candidate? Can that voter then make that candidate their first, second, third choice? Or would that disqualify their ballot?

    • Submitted by Charles Spolyar on 08/26/2013 - 11:02 am.

      Only the first place vote would count

      The ballot would still be qualified. However, the 2nd and 3rd choices would be invalid. It would be like you just voted the one, first choice only.

  2. Submitted by Steve Titterud on 08/26/2013 - 09:55 am.

    The opponents of RCV have their hands over their ears,…

    …their eyes shut, and, shaking their heads back and forth, are singing, “Nananananana”.

    They will not tolerate a *DYNAMIC* concept of “majority”. It’s as if “majority” is a holy word, a reserved word, and may not be applied to anything other than the ORIGINAL total of votes cast in Round 1 of RCV.

    So when the RCV system uses this holy word in an unapproved fashion, it drives them crazy.

    The legacy election system is a game with certain rules. RCV is another election system, another game with rules. You can’t properly invoke RCV rules in a legacy election system, and likewise, you can’t properly invoke legacy election rules in an RCV system.

    The legacy election system has failed us – in particular, it has installed and supported nearly ironclad control of elections by the two party system. And what have these two parties done for us here recently ? They invalidated the votes of the citizens of Minneapolis in the corrupt stadium mess, for one example. A government that delivers a corrupt result is a corrupt government.

    CHANGE the election rules !! Even though there are no guarantees it will have the desired effect, It’s worth a try !!

    • Submitted by Connie Sullivan on 08/26/2013 - 01:57 pm.

      Suddenly, we redefine “majority”?

      Very cogent arguments against the claim made by RCV proponents that their system assures that winners will have 50%-plus-one votes–an absolute majority of votes cast–have been made. So, proponents of RCV feel obligated to re-define “majority” as some slippery concept, some “dynamic” number or percentage that varies. . . well, it just varies, okay?

      The claim of majority-always was a primary selling point by RCV proponents. They now cannot hope to change the terms when that claim is shown to be false.

      You also can’t change the arguments, suddenly, to political parties and what-have-they-done-for-us-lately? We are discussing the value of the individual voter’s ballot, not third-party gains.

      Finally,we have the RCV proponents’ disdain for the discernment of the voters, stating or implying that those voters who cannot bring themselves to vote for [rank] all the candidates are unable to understand the system or cognitively unable to comprehend more than one candidate’s platform or positions/qualifications for office:
      “Massey correctly notes that a voter who ranks only one preference has (more or less, depending on what level of understanding you attribute to this theoretical voter) made a choice that if his or her first choice is eliminated, not to express a preference among the remaining candidates. That’s what she meant in the excerpt above by “This is the will of the voter.”

      Many of us may see two or three mayoral candidates as viable, but the most sophisticated voters out there, who follow politics closely, may only see one viable candidate out of 35 (two of my preferred viables have already dropped out of the mayoral race).

      RCV only works when the voter is at ease with compromising his or her political principles in order to see that every candidate running is worthy of the office, somehow. They are not viable mayors, for those in the know.

      • Submitted by Adam Miller on 08/26/2013 - 04:46 pm.

        I’d go a little further

        Voting is about preferring one (or more) candidate over the other. The voter whose ballot is exhausted may not have been choosing not to express a preference. They may have been expressing a dislike of the remaining candidates, and one of the stranger parts of RCV is pretending that a ranking below first implies a partial preference, rather than a partial dislike.

      • Submitted by Robin Garwood on 08/27/2013 - 03:25 pm.

        A few things…

        Dan and I have rebutted the “majority” definition point below. The key is that we always choose a denominator when talking about majorities: of continuing votes, of votes cast for a race, of total ballots cast, etc.

        I have to disagree that I’m seeing much “disdain for voters’ discernment” from the pro-RCV side of the argument. Jeanne correctly states that each voter gets to make a choice: rank one candidate or more than one? The system does *not* discriminate against voters who rank only one candidate, and advocates do *not* have the attitude that voters are “doing it wrong” if they rank only one candidate. Just as a voter does not necessarily have to vote for every race on the ballot, a voter does not have to rank every candidate for every race. (Indeed, in some races there will only be two candidates on this fall’s ballot – like Ward 2 City Council. A third ranking in that race would only matter to a write-in candidate.)

        In fact, I find this assertion somewhat odd, because in my experience it’s the ANTI-RCV folks, like Larry Jacobs, who continually – and with no actual evidence to back them up – state that voters are somehow too incompetent to understand RCV. As I showed in my last piece on MinnPost, that flies in the face of the evidence from right here in Minneapolis in 2005 – but that hasn’t prevented Jacobs and others from continuing to disparage voters’ cognitive abilities.

        Look, from my perspective there’s nothing wrong with voting for one candidate, if that’s one’s true preference. In the Ward 2 race this year, I’m ranking Cam Gordon number one, and leaving choices two and three blank. I may do something similar in the Mayor’s race.

        But for other offices this year, I will be taking advantage of the multiple rankings RCV offers me. For example, I’ll be choosing Annie Young as my first choice for At-Large Park Board, and John Erwin as my second. I can think of many past elections for which I would have felt absolutely blessed to have more than one choice. I’d have ranked Cam and then Paul Zerby in 2001. I’d have ranked Nader and then Gore in 2000. I’d have ranked Ken Pentel and then Roger Moe in 2002. I’d have ranked Jill Stein and then Obama in this past election. Et cetera.

        These are not examples of me “compromising my political principles.” In fact, it’s exactly the reverse! The *traditional* system demands that I hold my nose and vote for the least-worst person who can win, if I want to cast an “effective” vote. RCV frees me to vote for the candidate I most support – and then allows me to choose my least-worst fallback. It’s not a constraint on my freedom of action, it’s the *removal* of such constraints.

    • Submitted by Dan Hintz on 08/26/2013 - 02:38 pm.

      You’ve got it backwards

      First, Kudos to Eric Black for doing a good and even-handed job in explaining the “majority” dispute as it applies to IRV

      The only reason that we are talking about majorities in this context is that RCV’s main selling point was the claim that it eliminated the problem of plurality winners and produced majority winners. That claim is false. Well, its false unless you adopt a different or “dynamic” definition of majority. I don’t use a “dynamic” definition of majority, and not because I’m singing, “Nananananana.” I understand the RCV argument perfectly. I just happen to believe that a majority that depends on discarding valid votes isn’t a real majority.

      Let’s say you have an election with 100 valid ballots. 40 go to candidate A after reallocation and 35 go to candidate B after reallocation, with 25 votes going to other candidates and not A or B. Under RCV, A is a majority winner because he or she got 40 out of 75 votes after the 25 exhausted votes are discarded. The reality, however, is that out of the 100 votes, 60 of them did not choose candidate A. You can call it a majority using a “dynamic” definition, but the reason that majorities are supposedly important is that they provide a mandate and inspire confidence in elected officials. If you have to change the definition to get the majority, you haven’t solved the problem.

      Why is the RCV system any different than calling votes for Tom Horner and other 3rd party candidates in the “exhausted” votes and declaring that Dayton had a 51 percent majority if you only count the votes that he and Emmer received? And yes, again, I understand the run-off system in IRV. But for the people who voted only for candidates that did not make the run-off, telling them there is a majority and that their valid votes didn’t count isn’t going to do much for them.

      Steve, I understand your frustration with the outcomes traditional elections have produced. I share some of those frustrations. I just don’t think that RCV is the answer, and is something that makes things worse. It is never going to solve the two-party problem, because it will never be adopted anywhere other than one-party (one dominant party) cities. And if you look at the cities that have adopted it, a number of them have since repealed it or are in the process of doing so. RCV is not a solution to your problem. It is a solution to a problem that doesn’t exist.

      • Submitted by Ken Bearman on 08/26/2013 - 10:55 pm.

        Majority of votes, not of ballots

        The people who insist that “majority” applies to the number of people who cast a ballot don’t accept — or don’t understand — how RCV works. They want to apply the old rules to the current voting system.

        Devin Rice says, “A majority means 50 percent of all the ballots.” Mr. Hintz talks about “40 out of 75 votes after the 25 exhausted votes are discarded” and “a majority that depends on discarding valid votes isn’t a real majority.” Both, as well as Connie Sullivan (and others), have conflated and confused ballots with votes. They ignore the rules of RCV election in Minneapolis.

        A VOTE cannot be exhausted. If a candidate is marked in a round of counting, then there can be a vote for the candidate. A VOTE isn’t ever “discarded”.

        An exhausted BALLOT is not a vote because there is nothing to count. Not counting it isn’t discarding a vote.

        A majority of the ballots as Mr. Rice wants would mean that exhausted ballots — ones without a VOTE — would be counted in figuring the “majority”. But RCV is a series of runoffs, and his idea is equivalent to counting voters from a general election who don’t come back to cast a ballot in a traditional runoff election.

        If opponents of RCV reject the RCV rules, then they should clearly say so. But they shouldn’t apply one, old set of rules and thinking to a different, new election system.

        • Submitted by Dan Landherr on 08/27/2013 - 09:20 am.

          Mathematics of voting

          If you’re interested in the mathematics of voting as a means of reaching consensus I recommend Geometry of Voting by Donald G. Saari.

        • Submitted by Dan Hintz on 08/27/2013 - 09:41 am.

          No one has confused anything

          I understand exactly how RCV works, and I expect that Mr. Rice and others here do as well. You can argue about the terminology – “discarded” vs. “not counted” ballots, but the bottom line is that under RCV people who cast valid ballots (including full ballots) have their votes discarded or not counted so that an RCV “majority” is created. I will come out and say it – I don’t accept RCV’s definition of a majority.

          You make the analogy between exhausted RCV voters and primary voters who don’t vote in the general. That analogy only would work if primary voters who voted for candidates that did not advance were barred from voting in the general. The traditional system gives them a second chance. RCV disenfranchises them.

          • Submitted by Ken Bearman on 08/27/2013 - 12:06 pm.

            You have confused RCV with the Council’s ballot design

            Mr. Hintz’s issue seems to be with the three-choice ballot the City Council decided on. That has nothing to do with RCV elections or rules.

            RCV supporters, including this one, agree that three rankings is not enough. But if Mr. Hintz continues to insist that VOTES are “discarded or not counted”, then he’s dead wrong, and either he’s misrepresenting the way RCV works or he don’t understand it.

            A voter is disenfranchised if they’re deprived of the right to or prevented from voting. Nobody is disenfranchised by the RCV process in an RCV election. If a voter casts a ballot, whether leaving all lines blank or ranking three candidates, they’ve exercised their franchise in Minneapolis with the Council’s ballot design.

            The three-ranks decision wasn’t part of the RCV process. Read the Charter. And lobby the Council to redesign the ballot to have at least double that number of rankings.

            • Submitted by Dan Hintz on 08/27/2013 - 01:39 pm.


              “Mr. Hintz’s issue seems to be with the three-choice ballot the City Council decided on. That has nothing to do with RCV elections or rules.”

              I actually did not raise that issue at all, but I guess we can talk about it. Having more choices will probably reduce the exhausted ballot issue, but won’t necessarily solve it. The only way to do that would be to force every voter to rank every candidate on the ballot. If people can’t rank everyone, or choose not to rank everyone, you are going to have exhausted ballots.

              And if you don’t like the word “disenfranchise” in this case, that’s fine. But to get “majorities” under RCV, you may have to discard (or not count or exhaust) the voters who did not choose the RCV finalists. Using my example above, if 100 voters cast ballots and the winner only gets 40, I say there is no majority. Under RCV only 75 ballots are used in determining a majority – the 40 for the winner and the 35 for the runner-up. What about the other 25? They cast valid ballots, but are told that their votes did not count for purposes of determining there was a majority.

              • Submitted by Robin Garwood on 08/27/2013 - 03:59 pm.

                Just admit that “disenfranchise” is not the right word


                It’s funny that you raised the Princess Bride, because the word “disenfranchise” does not mean what you think it means.

                I find the high dudgeon about exhausted ballots really odd, because as far as I can tell, no one on the anti-RCV side finds anything wrong with down-ballot drop-off, which is exactly analogous. Many people who cast valid ballots don’t vote on every race on that ballot. That leads to lack of “majority” winners, even with the old nonpartisan primary, as I’ve shown – depending on one’s definition of majority, or majority of what. The only way to solve the “problem” of down-ballot drop-off would be, to use your language, to “force every voter to [vote for] every candidate on the ballot.” And short of doing that, the only way to ensure “majorities” under the old form of voting is to (again to use your language), tell people who “cast valid ballots… that their votes did not count for purposes of determining there was a majority.”

                It’s the same thing. But where you’re fixated on the one, the other seems to be perfectly fine. That doesn’t make sense.

                One last thing. You keep saying that votes “aren’t counted.” But they ARE counted! Look at the 2009 results! Every single vote that was cast was counted, minus exactly one vote in the whole city of Minneapolis that contained no valid rankings. Every vote was counted, and tabulated, and they’re right there on the internet for everyone to see. If your ballot becomes exhausted after your preference(s) are defeated, that is *exactly* analogous to voting for someone who doesn’t win in a standard election. Your vote was counted, but you didn’t get your way. Not being included in the denominator of continuing ballots does NOT mean your vote was invalidated, discarded, excluded, not counted, torn up, thrown away, burned or any other silly loaded word you might throw at it.

                To reuse Eric’s fantastic metaphor: being enfranchised is being invited to the table to choose an elected official. You sit down to participate in a series of runoffs, and you express a preference until you no longer prefer any of the continuing candidates, at which point you push away from the table. The fact that you got up from the table doesn’t mean you never had the opportunity to sit down.

                To throw the word “disenfranchise” at this situation mocks those who are *actually* disenfranchised – say, by restrictive voter registration rules. And again: the old nonpartisan primary system *actually* disenfranchised thousands of residents of Southeast Minneapolis by making them ineligible to vote in the nonpartisan primary that winnowed the field down to two, because they had lived at their current addresses for too short a time.

                • Submitted by Dan Hintz on 08/30/2013 - 02:19 pm.


                  I am pretty comfortable with my understanding of the term.

                  “To reuse Eric’s fantastic metaphor: being enfranchised is being invited to the table to choose an elected official. You sit down to participate in a series of runoffs, and you express a preference until you no longer prefer any of the continuing candidates, at which point you push away from the table. The fact that you got up from the table doesn’t mean you never had the opportunity to sit down.”

                  The problem is the idea that someone who cast a valid ballot somehow got up from the table in the middle of the vote counting. People just vote – they don’t “decline to participate” (as you put it in another post to which I responded). If I cast a valid ballot, you can’t say “sorry, you got up from the table and declined to participate further, so your vote doesn’t count in determining the majority”, because that isn’t what happened. If I cast a valid ballot, and my vote doesn’t count, then I have been disenfranchised.

                  Its also pretty dishonest to use the 2009 Minneapolis mayoral election as an example because Rybak had an actual majority of first place votes and RCV allocation wasn’t even necessary. The last Oakland and San Francisco mayoral elections had only “RCV majorities” – majorities when voters who “declined to participate” by voting for the wrong candidates were not counted.

  3. Submitted by Dan Landherr on 08/26/2013 - 10:24 am.


    If there are only 3 votes allowed in the ranked choice they need a primary to get it down to three candidates. I agree that a voter cannot be expected to know enough about 35 candidates to have a preference.

  4. Submitted by mark wallek on 08/26/2013 - 10:25 am.

    Ranked choice not workable

    If candidate A spends a billion dollars and candidate B spends 100,000 dollars on campaign advertising, I might not get a clear picture of either candidate. If we are going to do ranked choice, then the candidates should all stand on the same line so we can evaluate them fairly. In other words, lets do ranked choice, but let’s then take all the private billions out of the system and have elections structured around limited public funding. It might not be the hooker filled glamour of conventions and back room power plays, but the citizenry would be better served, and serving the citizen would be a nice change from the self service so popular today.

  5. Submitted by Kurt Anderson on 08/26/2013 - 11:50 am.

    We still need a primary

    It will not work to use ranked choice in lieu of a primary. We need a ranked choice primary and then either a ranked choice or 2-candidate general election.

  6. Submitted by John Ferman on 08/26/2013 - 11:52 am.

    RCV Fine Print

    What is poorly understood is which ballots get tossed out of the pool the defines the number that defines 50%+1. In the ordinance, the framers were careful to define two thresholds and then not make it clear which threshold governs. At one point, only ballots that have a first choice are counted for the threshold. So a voter who wants to vote “none of the above” is denied the vote. Mr Rice is correct – the ordinance should specify the number of ballots cast should define a one and only threshold. A very careful study of the ordinance is recommended, although that is like recommending castor oil for the headache.

  7. Submitted by Ray Schoch on 08/26/2013 - 02:19 pm.


    Something seems a little out of whack if RCV allows for only 3 candidates when there are 35 (!!) on the ballot.

    Personally, I think RCV an interesting experiment, and I’m certainly willing to give it a try, for all the reasons that proponents have used, but if only 3 are to be counted this time (and I certainly hope it’s 6 the next time around), it seems silly, as in *not serious,* or *worthy of ridicule,* to allow 35 candidates on the ballot, especially if the threshold for getting one’s name on that ballot is merely $20 and enough personal insecurity to require seeing one’s name on said ballot.

    I’d support an RCV primary of sorts to winnow the number of candidates down to a more reasonable and manageable half-dozen or so, and then go ahead with RCV for those that survived the primary. As it stands, the current RCV setup isn’t necessarily an example of voter probity and insight. Of course, the system that RCV is replacing is not without flaw itself, which is why we’re looking at an alternative. Let’s give it a try and see what happens.

  8. Submitted by John Ferman on 08/26/2013 - 02:52 pm.

    What was flawed by primaries

    In my 40 plus years of going to caucuses and conventions, I can only recall a few where the endorsee was elected by a landslide. Many times it was too close to endorse and the electorate had only campaign rhetoric to judge a candidate. One function of convention was to screen, to question, to scrutinize candidates to expose where a candidate stood with respect to Party principles. So it occurred to me that the Party could state that candidates A B C meet Party principles and candidates D E F do not. No winner take all; use the party function to help inform the electorate and help to take the spin out of campaigning. No more nice hair pictures.

  9. Submitted by Adam Miller on 08/26/2013 - 04:40 pm.

    I still don’t see the value

    In RCV, especially in a city like ours that typically has one party governance. Is the difference between getting 43% in a first pass the post and getting “51%” in RCV at all meaningful? Isn’t there going to be a mandate either way?

    So the only way I can see any particular “value” is if RCV let’s a candidate who isn’t first past the post win. Which in theory it can. But again, why is it better to have the person who was ranked third on the most ballots beat the person who was ranked first on the most ballots?

    And imagine if you really did have 35 ballots candidates to rank. Is someone filling in their tenth choice really expressing much of a preference?

    So, again, what’s the point?

    • Submitted by Robin Garwood on 08/27/2013 - 02:30 pm.

      Here are some values

      A few things:

      We don’t have “one party governance” in Minneapolis. There has been at least one Green on the City Council and one on the Park Board since 1997.

      One of the values I always saw – and continue to see – is that RCV replaced the old nonpartisan primary. I don’t think that having an extremely low-turnout election to winnow the field is particularly democratic. In some parts of town (the University area, especially), the primary is actually unavailable to thousands of voters because they move into their residences in early September.

      Another point, for me, is that RCV would work incredibly well for state offices. It’s been a long time since we elected a Governor in this state with a majority, and RCV would help with that (even if it’s “only” the majority of continuing votes). But the way things work in the real world is that a major reform like RCV cannot be adopted immediately at the state level; it must be proven to work at the local level. So Minneapolis is leading the state – followed by Saint Paul and others – in proving that RCV works.

      Which, by the way, it does.

  10. Submitted by David Cary on 08/26/2013 - 07:58 pm.

    The myth of *the* majority

    A well written, informative article. Kudos to Eric Black.

    There are a variety of traditional marjorities that are often used in various situations, even without RCV. Majority of votes counted in this round, majority of voters casting a ballot for this round, majority of eligible voters, etc. Recognizing that variety highlights just how selective, and dynamic, Rice’s definition is.

    We acknowledge the right of a voter to abstain, either by not casting a ballot or by not marking a preference for any of the (remaining) candidates. It can be a rational choice to abstain, even when the voter is informed and has definite preferences between the candidates. The prevalence of abstention as an effective a vote for none of the above is often ignored by a system that does not formally allow or recognize the expression of such sentiments. Including those abstentions would threaten the traditional myth of the majority.

    That Rice ignores all such abstentions in calculating his version of a majority for traditional elections, but insists that at least some of them be included when determining a majority for RCV is a clear double standard.

    The informed, rational choice of abstention, despite having preferences, also suggests Black’s “theoretical ‘perfect case'” is overly restrictive.

    • Submitted by Robin Garwood on 08/27/2013 - 01:12 pm.

      This is exactly right

      David, you’re absolutely right. Rice and others believe they are using the One True Definition of “majority,” but fail to apply that definition across the board. You’re also right that it’s absurd.

      Here are the data to back this up. In both the 2005 and 2001 municipal general elections in Minneapolis, two Council candidates were elected without a majority of the total ballots cast in their wards. But wait, how could that be? The old nonpartisan primary system ensured that only two candidates would be on the general election ballot! Didn’t that make it *certain* that the winning candidate won the majority of ballots cast in that ward?

      No. Why? Because the denominator always matters when defining majority. Note that I said “total ballots cast” rather than “votes for that office.”

      In 2005, Cam Gordon was preferred by 49.799% of voters who cast ballots in Ward 2, and Robert Lilligren won the votes of 49.469% of voters who cast ballots in Ward 6. In 2001, Paul Zerby won the Ward 2 seat with 48.917% of total ballots cast in Ward 2, and Natalie Johnson Lee was elected in Ward 5 with 49.051% of total ballots cast.

      How is this possible? Because a certain percentage of people decided to cast a ballot – for Mayor, say – but also decided not to cast a vote for their local City Council race (and an even smaller percentage of people wrote someone in). In a sufficiently close race, it’s common to have this down-ballot drop-off make the difference between an absolute majority of ballots cast and a majority of votes for that office.

      Do Rice and others own up to the fact that they have a nuanced understanding of “majority.” No. That they have chosen a denominator that suits their argument? No. Or even that there *are* other possible denominators? No.

      It is exactly, precisely as fair to say that RCV ensures a majority – meaning a majority of continuing votes in the final round of counting – as it is to say that the old nonpartisan primary ensures a majority – of votes cast for that office. In both cases, critics can argue that the denominator is “wrong.” But that’s not a technical argument, it’s a value-based argument.

      My personal view is that ballot exhaustion is perfectly acceptable. There are voters who will only rank a single candidate for a given office, even if there are 35 candidates running for that office (I think that’s what I hear Connie saying she plans to do this year). I think we all can agree that the voting system should not compel these voters to rank people they don’t want to see elected. So the outcome is that their ballots are exhausted for that race if and when their preferred candidate is defeated.

      Black’s description of RCV is very apt. It is exactly as if we keep the electorate impaneled for multiple rounds of runoff elections, until we find a winner. Imagine a huge table, with all of the voters sitting around it. I cast my ballot for my only preferred candidate, but not enough other people agree with me, and she is defeated. I don’t have a preference between the others running, so I get up from the table. I wouldn’t want to be compelled to stay and cast votes for candidates I don’t support, so I stand back. Depending on the dynamics of the race, a number of people may join me around the sidelines. But at the end of the series of runoffs, a candidate will be elected with a majority of votes of *people sitting at the table.*

      That majority is a majority. It’s not the same majority as the majority of people who voted in that race (which is a larger number), or the majority of people who cast ballots in that political subdivision (which is a still larger number), or the majority of people who were eligible to vote in that election (which is a still larger number), or the majority of people who live in that political subdivision (which is a still larger number). But it’s a majority nonetheless.

      Rice and the other anti-RCV folks have constructed, to use his so-often-misused word, a “myth”: the One True Majority, sacrosanct and inviolate. But in reality, they have done the same thing as the people they criticize – they chose a denominator to base their definition on. And then – dishonestly, in my view – pretended they didn’t.

  11. Submitted by Dan Hintz on 08/26/2013 - 11:16 pm.


    If I have this right, if you cast a valid ballot but don’t vote for the right people, you have abstained? Your vote doesn’t count because the candidates you chose did make the final RCV round?

    My guess is that the “abstaining” voters would not be too happy about being told that. And regardless of whether they get disenfranchised by RCV, those people cared enough to vote, and got a candidate they did not want. The purpose of achieving a majority is to provide a mandate and confidence in the elected official. A technical “majority” that relies on not counting certain voters doesn’t achieve that, or at least not any more than in a tradional plurality election.

    Finally, one RCV proponent (Steve) criticizes RCV opponents for not tolerating RCV’s “dynamic” definition of majority. You, however, call the traditional definition of majority “dynamic” and seem to use it as a pejorative term. I think Inigo Montoya might have something to say here.

    • Submitted by Ken Bearman on 08/27/2013 - 12:14 pm.

      Disenfranchised means prevented fron casting a ballot

      You can’t redefine words to suit your argument (unless you’re Humpty Dumpty in ‘Alice in Wonderland’).

      A voter is disenfranchised if they’re deprived of the right to or prevented from voting. Any voter who submits a ballot in an election, RCV or not, has exercised her or his franchise. They can hand in a totally blank ballot — no choices marked for any office or issue — or they can rank up to three choices in every RCV race. Then they’ve exercised their franchise … they’ve voted.

      • Submitted by Dan Hintz on 08/27/2013 - 01:25 pm.


        I have read a number of definitions of the term – none of which are as specific as yours – and I’m pretty comfortable saying that if you cast a valid ballot, but your vote isn’t counted, then you have been disenfranchised. In a traditional election, even if your candidate doesn’t win, your vote is counted as part of the total. Not so in an RCV election, which creates “majorities” where none exist by excluding (or discarding, or not counting, or exhausted or whatever term you want) the votes of the people who did not vote for the RCV finalists.

        • Submitted by Ken Bearman on 08/27/2013 - 03:36 pm.

          No votes are discarded

          Humpty Dumpty would be proud. In this redefined RCV process, valid votes aren’t counted and are discarded.

          In an RCV election, all valid votes are counted in every round. To claim that they weren’t — because ballots were exhausted before the final round of counting — is Looking Glass logic. It may happen somewhere, in some other place, but it doesn’t happen in Minneapolis (or St. Paul, by the way). And everyone can see that they were counted because the votes in every round are reported as counted.

          • Submitted by Dan Landherr on 08/28/2013 - 11:16 am.

            It isn’t true RCV if you can’t rank all the candidates

            In this specific case with 35 people on the ballot and 3 votes allowed you could have the winner receive less than 10% support and win. Either have a primary to trim it to a reasonable number of candidates or allow voters to rank all 35 candidates. Ranked choice only works when you rank all the choices.

            • Submitted by Robin Garwood on 08/28/2013 - 05:55 pm.

              “Could” does not equal “will”

              Or even “might.” We will not have a mayor elected with less than 10% support this year, full stop.

              I agree with Jeanne Massey that we should give voters six rankings instead of three. However, evidence from other places that use ranked choice voting (and other ranked ballot voting methods) is that voters’ use of rankings after six goes down precipitously.

              Out of the 35 candidates, perhaps five or six are actually viable. My sense is that it’s a three-tier race with two front-runners, three or four viable contenders in a pack behind them, and a bunch of also-rans.

              When the winner has significantly more than 10% support, will that be enough to convince you?

              • Submitted by Dan Landherr on 09/04/2013 - 12:45 pm.

                This is an issue that should be addressed

                20% support would still be pretty terrible. 10% support would inevitably have a recall campaign follow. Thirty five candidates are too many, just from a ballot marking perspective. Get it down to the the viable candidates (4-6) and then rank them all.

                This year might not have issues but what if next election there are over 100 candidates? An irritated citizen with a spare $2000 could make that happen.

  12. Submitted by Walt Cygan on 08/27/2013 - 04:32 pm.

    Two points

    To lay my cards on the table, I oppose RCV and have since I first heard about it.

    First point: The fact that there is such a tortured dissection of the word “majority” makes the point that RCV supporters do themselves a disservice of pointing to a winner having some mythical “majority” mandate as a benefit of RCV. I think such a mandate is an illusion that will melt away with the first highly popular or unpopular action taken by the winner. Either the “mandate” won’t matter because the official has done something well or it won’t matter because they have done something badly.

    Second point: In previous elections, the benefit of the primary was to reduce the field to a number (2) that most people could reasonably research. Having 35 candidates on any ballot for one office in my mind invalidates any benefit of RCV. If we had a system where if there were more than 6 candidates we would use RCV in a primary to whittle down the field to 4 and then use RCV in a runoff to pick the winner, I would probably support it. Instead some RCV supporters counter that we should make it more difficult for people to get on the ballot. Who is going to validate those signatures if a candidate wannabe needs to get 200? Do we really WANT a system that makes getting on the ballot more difficult?

    The mayoral election has almost (almost!) brought me to the point of deciding that my best vote would be to leave the mayor’s race blank. If 20-30% of voters voted for a council member and left the mayor’s race blank, that would be strong evidence that the system is not right yet.

    • Submitted by Robin Garwood on 08/28/2013 - 12:09 pm.

      Two rebuttals

      As may be obvious, I’m an RCV supporter, and have been for years.

      To your first point: I really don’t understand the hysteria about exhausted ballots. Let’s do a thought experiment together. In other parts of this country, they do runoff elections when one candidate doesn’t receive over 50% of the vote in the general. That runoff election is held on a later date, typically between the top two vote-getters from the general. The winner gets at least 50% of the votes of the people who show up for the runoff.

      This has some benefits. It allows people whose most-preferred candidate didn’t win to come in and support a fallback, if they have one. If we used this sort of typical runoff in Minnesota, it’s entirely possible that the Independence Party voters could have changed the results of at least the last three gubernatorial elections, for example. But it has one major drawback: the runoff election *virtually always* has lower turnout than the general. Often much, much lower.

      Now, here’s the experiment for you. Candidates A and B get the top two places in a general election, 40% to 35%, with all the rest of the field getting the other 25%. Turnout is [x%]. A runoff election is held. Candidate A wins with 52% of the vote, but turnout in the runoff is half what it was in the general (that’s fairly common – and the nonpartisan primary in 2005 had *less than* half the turnout of the general).

      So: did Candidate A receive a majority? Yes. But did Candidate A receive a majority of the votes cast in the general election? No. Candidate A actually received *fewer* votes in the runoff, or about 25% of the number of votes cast in the general. But according to the “traditional” definition of majority, did Candidate A get a majority in the runoff? Yes.

      RCV does *exactly* this same process, but holds the runoff on the same day as the general. This is a significant improvement, because the dropoff rate from exhausted ballots is actually *less* than the dropoff rate from lower turnout for a second election.

      To your second point: I could not disagree more.

      Let’s look at the reality of the nonpartisan primary. In 2005, fewer than *half* as many people voted in the primary as the general. Total turnout was an abysmal, embarrassing 15%. Getting rid of the low-turnout primary that kept qualified candidates from getting to the general election (an *incumbent* lost in a primary in 2001, after all) is not a bug of RCV, it’s a feature.

      Your argument, as far as I can tell, is that the majority of voters are simply too stupid to choose their preferred candidate from the whole field, and they need the help of a smaller, more “elite” group of voters to winnow the field to the two candidates that their little brains can handle. I think that’s a bunch of malarky. I think voters are perfectly capable of choosing between one and three people to rank in order of their preference from a crowd of also-rans. At its most fundamental level, your argument is antidemocratic: if people can’t be trusted to choose between 35 people, how can they be entrusted with the right to pick their leaders in the first place?

      I also completely disagree about the virtues of making it harder to get onto the ballot. Do you really believe that Captain Jack Sparrow belongs on a ballot in Minneapolis? Does he actually want to be mayor? If not, he shouldn’t be on the ballot. Right now, someone can run for mayor as a lark; it’s cheaper than a night out. The $20 filing fee that was put in place in 1967 was worth $140 in 2013 dollars. And as to who is going to validate signatures: it will be elections staff. You do know that independent and minor-party candidates have to get signatures to get on the ballot in state elections, right? There’s a long-standing, well-developed system for validating.

      Lastly, you can go ahead and leave the mayor’s race blank. But given that the chances of 20-30% of voters doing likewise is vanishingly small, I would happily use that as a measure of the voting system. My question to you: when this doesn’t happen, will you be prepared to say that RCV worked?

      • Submitted by Walt Cygan on 08/28/2013 - 04:06 pm.


        First point: I am not hysterical about exhausted ballots. I just don’t think the “majority” winner concept is very important.

        Second point: The reality is that we have the system we have. There are going to be a massive number of candidates on this Fall’s mayoral ballot. I think the test of this process will be turnout. If the turnout for this election is higher than the last highly competitve race (Rybak v. Sayles-Belton), then I will see it as a success. If it is significantly lower, as I suspect it will be, I will see it as a failure. I think the daunting nature of a 35-candidate ballot will be a turn-off to many voters.

        • Submitted by Robin Garwood on 08/28/2013 - 05:46 pm.

          Remember ’05

          Most people count the 2005 Mayor’s race between Rybak and McLaughlin as competitive as well. I think it would make sense to compare this election to both 2005 and 2001.

          You’re right that we have the system we have this year. There are 35 candidates for mayor. However, you also criticized attempts to raise the bar for filing, and that’s what I was rebutting. The filing fee and/or signature requirement should be raised to make folks like Captain Jack a little less likely to file for office on a whim or as a joke. I would argue for that whether we used the old nonpartisan primary method or RCV; people who file for office should actually want the job.

  13. Submitted by Robin Garwood on 08/28/2013 - 12:40 pm.


    I agree with the rest of the commenters that this piece by Eric Black is really, really good. Even-handed, accurate, informative, based on solid data.

    There is some additional context that would be helpful to get into this discussion, however, from analysis of past “traditional” Minneapolis elections with the nonpartisan primary, and the 2009 RCV election.

    First, in half of the races that went to multiple rounds in 2009, the number of “exhausted” ballots was *significantly smaller* than the number of “undervotes.” An undervote is when a voter votes for one race on a ballot (mayor, say), but doesn’t vote in downballot races (at-large park board, say). Here are the numbers:

    Board of Estimate and Taxation undervotes: 13,882
    Board of Estimate and Taxation exhausted ballots: 5,509 (less than half as many)
    Park Board At-Large undervotes: 9,313
    Park Board At-Large exhausted ballots: 4,412 (less than half as many)

    It’s curious that the 5,509 exhausted ballots for BOE generate hair-on-fire outrage from the anti-RCV folks, but the more than twice as many undervotes merit nothing more than a shrug.

    Second, the difference between the exhausted ballots and the total votes cast in each race is *much* smaller than the difference between turnout in the old nonpartisan primary and general.

    In 2009, Wards 4 and 5 went to multiple rounds. In Ward 4, 9.31% of ballots ended up exhausted. In Ward 5 that number was 6.73%.

    But in 2005, the difference between turnout for the primary and for the general in these wards was significantly worse. In both Wards, about a third as many people voted in the primary as the general (34% for Ward 4, 33% for Ward 5). Put another way, many *more* voters declined to participate in the nonpartisan primary in 2005 than in the rounds of instant runoffs in 2009.

    So which is worse? Having a third of the voters decide which candidates the other two-thirds get to pick between, or having less than a tenth of the voters decline to participate in all rounds of the instant runoff election? That’s a pretty easy question to answer, if you’re really interested in maximizing participation by voters.

    • Submitted by Dan Hintz on 08/30/2013 - 02:12 pm.

      Decline to participate?

      “or having less than a tenth of the voters decline to participate in all rounds of the instant runoff election?”

      I think that we have found the right euphemism for the RCV proponents. It’s not “exhaused” or “discarded” or “not-counted.” Its “decline to participate.” In their world, if you vote for candidates that don’t make the final RCV runoff, then you have “decline[d] to participate.” Your valid ballot doesn’t count in determining whether there is a majority because you have “decline[d] to participate” Never mind that you might not know which candidates would be finalists – if you don’t guess right, you have “declined to participate.” Awesome.

      I also really enjoy the hyperbolic descriptions of anti-RCV arguments like “hair-on-fire outrage.” Fairvote gets public money to provide voter education on RCV. Does that include insulting RCV opponents on blogs?

      • Submitted by Ken Bearman on 08/30/2013 - 08:29 pm.

        More redefining

        Humpty is absolutely beaming. Once again, Mr. Hintz has changed a meaning.

        In comparing the low-turnout 2005 primary to the later rounds of the 2009 RCV election, Mr. Garwood was talking about people who didn’t vote. In 2005, they didn’t vote — “declined to participate” — in the primary. In 2009, they didn’t rank three candidates — “declined to participate” after the first or second ranking.

        A ballot is exhausted if, in a round of counting, it has no remaining viable (still there) candidate ranked. That could mean they ranked only one or two and those were eliminated by having the fewest votes in a round. It also could mean that they ranked three candidates and all three were eliminated. In the former case, the voters declined to participate beyond one or two rankings.

  14. Submitted by James Hamilton on 08/30/2013 - 09:43 am.


    This is simply idiocy: 35 candidates, 3 ranked choice votes. I look forward to reading the name of the winner and the percentage of total votes cast for that candidate.

    By the way: does anyone really think it’s better to have more voters than to have informed voters? This ain’t Baskin-Robbins, folks.

  15. Submitted by Eric Paul Jacobsen on 09/01/2013 - 02:12 pm.

    Inflated Advertising May Still Refer to a Superior Product

    If at least a few voters for candidates in field of three or more understand how Ranked Choice Voting works, the winner in that three-way election is almost certain to represent a bigger plurality of voters than would exist under the single non-transferable vote system. Maybe the winner won’t always represent a majority, but the winner will nearly always represent a bigger plurality.

    The voters whose ballots an instant-runoff system eliminates first are not the traditional ones who vote for a Democrat or a Republican, but the diehard radicals who vote for Captain Jack Sparrow and will accept no other choice.

    I support Ranked Choice Voting because I am fed up with the two-party duopoly and I am convinced that RCV will do two things: (1) RCV will normalize the idea that it’s okay to vote for a third-party candidate, as long as you’re shrewd enough to vote for a back-up candidate, too, and (2) RCV will almost certainly represent the will of more voters – at least a bigger plurality among them, if not an absolute majority – in every election.

    I will not be satisfied with anything less than proportional representation, which would require us to allow more than one representative to be elected from each voting district. I favor RCV principally as a step in the direction of proportional representation. The worst disenfranchisement that presently exists is the system that virtually insures that either a Democrat or a Republican will always win in every district that comprises any area larger than a very small neighborhood – even though I and many voters like me consider both major parties to be irredeemably corrupt, or nearly so, and would like to give them some effective competition. Clearly, the present electoral system is not providing that competition, so I support any reform that makes third-party voting easier. I don’t expect any reform to be perfect, but I don’t think it’s hard at all to improve upon the horrible system we have now. I also have little patience for those who allow the perfect to be the enemy of the good, especially when this perfectionism is linked to ignorance or willful denial of how badly we are served by the status quo.

  16. Submitted by Steve Titterud on 09/02/2013 - 03:20 pm.

    One last reason I am glad we are trying out RCV,…

    …which hasn’t been discusses here, or at least not precisely, is that the legacy system seems so crude by comparison. You are allowed to cast ONE vote only per office, and that’s IT. There is no room in this system for any further expression of a more seasoned or complex voter opinion, which would span more than one simple vote.

    Prior to RCV, there has been no way – IN THE VOTING BOOTH – of indicating that yes, I like candidate A the best, but I really think candidates B and C are worthy candidates as well, and I’m fine with B or C gaining office. Never before have we been allowed to SPEAK – with our votes – about more than one candidate.

    I think voters will be more satisfied, and perhaps even better accept the result, when they are able to SPEAK in a more sophisticated voice in the voting booth – by voting for more than one candidate.

  17. Submitted by Tony Hill on 10/08/2013 - 03:12 am.

    The problem with IRV in Minneapolis

    The main problem with IRV in Minneapolis is that voters only have three choices.

    The second problem with the mayoral election isn’t the total number of candidates but the number of viable candidates. It doesn’t matter if there are 2 or 27 or 132 goofball candidates. In California in 2003, 135 people ran for governor and the winner was elected with a majority. The problem here is that there are at least eight serious candidates who have backing and money and name recognition and an arguable chance at winning the election and voters only get three choices among them.

    Most likely, in the end it will come down to two candidates. Any voter who did not specify one of those two among their choices will be stranded, or exhausted. Their ballot will not count in the denominator. The denominator will only include the ballots that have a choice for one of the two final candidates.

    In a contest for supervisor in San Francisco in 2011, there were 17,808 valid votes cast. If someone were to claim a majority, they would need 8905. However, the winner was elected with only 4,321 votes, and only 8,200 counted in the final tally. A majority of votes were stranded before the final tally. That is because voters ran out of choices. There were more serious candidates than the number of choices they had. 4,631 voters made no errors in exercising three choices but their votes were stranded anyway. That is more votes than the winner got.

    The same thing could happen in Minneapolis next month.

    Tony L. Hill, Ph.D. (political science)

Leave a Reply