Fairvote reports on instant runoff voting elections in 6 cities

Rob Richie and Dorothy Scheeline of Fairvote have written some interesting analysis of last week’s elections that used instant runoff voting, aka ranked choice voting. Instant runoff voting was used in San Francisco, CA, St. Paul, MN, and Portland, ME; in all three cities, Greens both helped enact IRV and ran in last Tuesday’s elections. For the mayoral elections in Portland and San Francisco, Fairvote has graphs that show the breakdown of votes round by round until someone takes a majority (in Portland, Greens David Marshall and John Eder finished 4th and 12th of 15; in SF, Green Terry Baum finished 11th of 16). In a Huffington Post article, Richie and Scheeline focus on the story of IRV’s success in Portland:

Repeatedly, we are seeing RCV winners being the candidates who do a particularly effective job at reaching out to voters, often with direct contact involving community debates, local events, and door-knocking. One Portland candidate, David Marshall, said he knocked on 20,000 doors. He didn’t win, but it was ballots from his supporters that provided a particularly strong boost to the new mayor’s win total.

Dave Schwab


  1. As a Green, the top fact about instant runoff voting that should concern you it that it, just like
    plain plurality voting, leads to 2-party domination in which Greens and other third-party members cannot win. Australia has used IRV to elect its house (not Senate!) for over 80 years. In their last 4 election cycles (600 seat-races in all) exactly one IRVseat was won by a third-party member. This is comparable to the USA House & Senate.

    You therefore (if you have a brain) should want a better voting system like Range Voting.
    Range voting is simpler than IRV, objectively better, and there is reason to believe it will
    not plunge us into the 2-party domination dead-end. For example in its over 500 years of consecutive use in Venice and Ancient Sparta, 2-party domination was not reported.


    • The reason there is not more party diversity in some countries is not because of IRV, but because those countries have parliamentary systems in which the executive branch is appointed by the party with the greatest plurality based on a one-time, post-election, behind closed doors negotiation to arrange a minimal majority legislative support for the appointed executives. This race for a plurality and the huge rewards for attaining it overpowers the natural diversifying effects of IRV. Luckily, IRV in the US does not have to overcome such problems.

      But when electing legislative bodies, the proportional representation methods that the Australian Senate uses, which are natural extensions of IRV, are better than any single-winner, winner take-all method.

      • David Cary,

        The issue here is “effective” complexity. IRV has led to around 7 times as many spoiled ballots as Plurality Voting (e.g. in San Francisco). It is more complex in that voters actually mess up their ballots more often, rendering them unusable. Approval Voting results in much LESS ballot spoilage than Plurality. You can argue that there was complexity involved in the sheer act of deciding how to vote, but that’s not a problem if it doesn’t lead to some side-effect like casting unusable ballots.

        A more important point is that you’re quite mistaken about the tactical voting issues regarding these systems, because:

        1) It has been shown through extensive Bayesian Regret calculations that Score Voting and Approval Voting produce better outcomes with 100% strategic voters than IRV does with 100% honest voters. (ScoreVoting.net/BayRegsFig.html) Therefore voters can just cast honest votes and not even be bothered with any tactical considerations, and STILL be better off than they would have with IRV.

        2) Tactical Approval Voting is essentially just to vote for the same candidate you would with Plurality Voting, plus everyone you like better. So the “hard part” of Approval Voting is something voters already have been doing for centuries. E.g. with Plurality Voting, you prefer Nader over Gore over Bush, and do the math to determine that a vote for Gore is your strategically best option. With Approval Voting you do the same thing, but THEN you can vote for Nader too, because it costs you nothing.

        3) IRV is EXTREMELY susceptible to tactical voting, which causes it to degenerate toward Plurality Voting. (ScoreVoting.net/irv-plurality)

        I don’t know what you mean by “indeterminant” in this sense, so it would be helpful if you could clarify that.

        Clay Shentrup
        The Center for Election Science

      • > The reason there is not more party diversity in some countries is not because of IRV, but because those countries have parliamentary systems..

        Wrong. Australia’s House of Representatives, for instance, is elected directly by voters. And of the last 600 seats that have been elected through IRV, only 1 has been a third party. That is two-party domination.

        Whereas if you look at most of the 27 or so countries which use the Top-Two (delayed) Runoff, most of them have 3 or more vital parties, even in single-seat elections. This was one of the 3 observations of Duverger. (ScoreVoting.net/DuvTrans.html)

        > This race for a plurality and the huge rewards for attaining it overpowers the natural diversifying effects of IRV.

        As far as I can tell, this argument makes no sense. The idea is that I’m a Green, but I vote for Labor because I believe a Labor rep would be more effective than a Green rep at preventing a NatLib from becoming Prime Minister. If that’s not boggling enough, let me also point out that the Australian PM is not elected, but appointed by the Governor-General. Forgive me if I’m misunderstanding your point.

        > But when electing legislative bodies, the proportional representation methods that the Australian Senate uses, which are natural extensions of IRV..

        Australia’s Senate is elected via Single Transferable Vote (STV). STV is not an “extension” of IRV. IRV is the single-winner case of STV. For multi-winner elections, STV seems to work pretty well. But for single-winner elections, IRV behaves quite poorly. It is widely understood to be the worst of the commonly discussed alternative voting methods, as you can see in this graph of Bayesian Regret figures:

        > [STV is] better than any single-winner, winner take-all method.

        That is certainly not a given. There are obvious advantages of proportional representation. But there’s no way to say that a legislature elected with a GOOD single-winner method (e.g. Score Voting or Approval Voting) couldn’t be even better. The biggest problem is that there’s no reliable way to preform Bayesian Regret measurements for multi-winner systems.

        And in any case, there are far better proportional methods than STV. Reweighted Range Voting and Asset Voting are two of them. And RRV is just Score Voting in the single-winner case, which is massively better than IRV.

  2. Score Voting (aka Range Voting) is much better than IRV, but not as politically viable because of limited prior use in government, which makes it hard to sell. But Approval Voting has most of its benefits and is incredibly simple. You just get an ordinary ballot, but you may vote for as many candidates as you want to. The one with the most votes wins.


    Pointing out to Greens that IRV is political suicide has never been particularly effective for some reason.

    • Range/score/approval voting are actually quite complicated for voters because they have huge problems with strategic voting. Unlike IRV, range/score/approval voting are indeterminate, so even sincere voting has so many options that strategic voting is usually necessary.

  3. I studied all of these systems in college, and I am still convinced that instant runoff voting is the best system for single-winner races. I see the biggest problem with our current system as strategic voting – the fact that voters do not vote for their actual preferences, because they understand that the system is likely to either not reward them, or actually punish them for it. Approval voting and range voting do not eliminate this problem, while IRV does.

    Another thing I studied was why no single-winner voting system can be perfect – all of them are flawed. For that reason, I support proportional representation and devolution of power to bodies that can be elected using PR. That is the best way to ensure that all votes count.

  4. If you want to participate on a team elected under pure proportional representation (PR) using the ranked choice voting combined with the Sainte-Lague parliament seat distribution system in a continuing effort of 16 consecutive years, be sure to sign up before 1/1/2012.

    We need volunteer vote counters (to observe “eballots” as they are cast), as well as members of Cabinet, ruling coalition members, and participants on four levels: 100-member county micro-state parliaments, 100-member mini-state parliaments (high populated counties and groups of counties), 100-member super-state parliaments (states and groups of states) and the 1000-member national American parliament.

    Sign up today, to be ready for the national US presidential elections of 2012.

    –James Ogle [Free Parliamentary]

    “Why do you THINK they called it Google?”

  5. I was thinking about a playoff format for elections. In the case of Maine, you had 14 candidates. What you could do is have a first round with 14 candidates, second round with 7 (1 bye), semifinals 4 and final 2.

Leave a Reply

Your email address will not be published.