Successive Rank Voting



  • @Keith
    Thank you Jack & Keith for humoring me while I educate myself and float ideas on voting methods. Let me explain my background. I am working on an open primary ballot initiative in Oregon, and our group is at loggerheads in deciding how run an election. We did agree that since primaries tend to have low turnout, more than two candidates should be advanced to the general election (we settled on Top-4 like Alaska).

    We are constrained by the state constitution which says that voters “may vote for one person under the title for each office. Provision may be made by law for the voter’s direct or indirect expression of his first, second or additional choices among the candidates for any office.”

    I’m not too worried about the primary election itself. Even plurality voting would likely produce four acceptable candidates. But looking at the popular menu of election methods for a 4-way single-winner election, none of them is appealing. Range and approval voting (and variations thereof) have the same flaw as plurality voting: if voters only pick one candidate, then the candidate who is the most opposed by a majority of voters could end up winning. For that reason I favor methods that ensure some level of majority support for the winner. However ranked-choice voting with instant runoff (RCV/IRV) does not reward a candidate for being nearly everyone’s second choice (proponents rationalize this flaw by claiming it to be a desirable feature). I also don’t like the complexity of RCV/IRV in a time when many voters are being told that elections can’t be trusted.

    I came up with the “Vote for 2 then Instant Runoff” method as a way to include voters’ second choices in determining who would reach the runoff, while allowing voters to choose between “friend betrayal” and “later no harm” as a resolution of Arrow’s impossibility theorem. I mentioned this idea to someone promoting a different method, and they suggested that I post it on this forum. I still think it is not a bad way to run an election, but it is a bit awkward formatting it into a ranked-choice ballot.

    I then started wondering if there was a way to guarantee majority support without having a runoff. That led me to the “Successive Rank Voting” proposal that I now realize is just a form of Bucklin voting. I’m surprised that Bucklin voting never popped up in any of my internet searches. Evidently it is not good to force voters to choose candidates they don’t like, so I’m abandoning that idea.

    However, Bucklin voting does seem like a logical way to count successive ranks, so I now favor combining Bucklin voting with an instant runoff if no candidate achieves a majority. The runoff can be achieved using a ranked choice ballot by distinguishing between approved and non-approved candidates in the rankings. See http://www.classicalmatter.org/Election Science/BAIR Voting.pdf . This admittedly wouldn’t qualify as a “new method”, except perhaps in the details of vote tabulation. But I still appreciate feedback.
    That’s my story. Thanks for indulging me in this discussion.
    -Robert



  • Just on ballot design: I think this would be easier for voters (less confusion and mistakes):
    ballot_approvol_cutoff.jpg



  • Range and approval voting (and variations thereof) have the same flaw as plurality voting: if voters only pick one candidate, then the candidate who is the most opposed by a majority of voters could end up winning. For that reason I favor methods that ensure some level of majority support for the winner.

    This is an incorrect understanding of the situation. If a voter gives one candidate full endorsement and all others none (ie bullet vote) that is their choice. If they truly feel that way then there is no reason to prevent them from voting that way. If this happens then it means that there is no other candidate which challenges that one. There are long term game theory effects which imply that would not happen in repeated elections. Also, there are systems which disincentives strategic voting of this nature. You seem to prefer Majoritarian methods so choose one of those.

    My suggestion would be to go with STAR voting. It is Majoritarian and disincentives strategic voting of the nature you do not like.

    I would also suggest that 4 is too few. The magic number in research is 7



  • @multi_system_fan Thanks for the suggestion. I was thinking that 1st rank votes should be treated as if they were "approved" even if they were not. But I suppose you are right that some voters might not want to help their 1st choice at all until the runoff stage.



  • @Keith STAR does not guarantee any sort of majority support for the winner. If voters bullet vote, the winner will simply be the plurality winner. And not bullet voting decreases the chances of the first choice winning.
    Also, I am concerned that STAR would not pass constitutional muster since giving two candidates a score of "5" seems to me like "voting for more than one person".



  • @robertpdx while not approving the 1st ranked might be preferred by probably a small group of voters, that was not my intended contribution ( I hadn't realised this). I just found the original ballot very complex visually and expected many voters would get confused or needing unnecessary amount of "cognitive load". I would expect many invalid ballots and uneasy feelings with voters about their choices and support for this method. So no change in method or tabulation intendend, just a different visual ballot-design that is less demanding on "cognitive load" for (below) average voters.



  • STAR does not guarantee any sort of majority support for the winner.

    This is explicitly the pupose of the runoff. I am unclear how you can argue otherwise

    If voters bullet vote, the winner will simply be the plurality winner.

    This is what I mean by "This is an incorrect understanding of the situation.". Your presuppositions about strategy and expression are wrong so you are drawing the wrong conclusion. What you claim is better is to force people to rank everybody to give the illusion of expression. It is an illusion because ranking contains less information in the strict information theory sense so any benefit you see is not real. Empirically it has been checked a number of times if people behave the way you are claiming and they do not. You have a concern about something which is based on a misunderstanding of the information content and is used to justify rejection of the system even though both theoretically and empirically the evidence does not support it.

    And not bullet voting decreases the chances of the first choice winning.

    I am not sure what you are referring to here. Can you give an example? Do you mean that in some cases STAR voting will take a more generally liked candidate over a polarizing candidate. Yes this is true. While it is a Majoritarian system it is on the Utilitarian side of Majoritarianism. Most people view this as good not bad. Why would we not want more happy voters?



  • @Keith The STAR "runoff" is not a true runoff. It simply checks each ballot to see which candidate scored higher. With bullet voting, each candidate either received a 5 or a zero on each ballot with no overlap. So the winner is just the candidate who received the most "5" scores to begin with, and that support could be much less than 50%.
    If voters choose to give nonzero scores to additional candidates, it increases the likelihood that the preferred candidate will not make the final runoff at all. That is a small risk if the score is a 1, but it is still a risk. Simple example with three voters and three candidates (A,B,C):
    Voter 1: (5,0,0)
    Voter 2: (0,5,0)
    Voter 3: (1,1,5)
    Candidate C missed a tie for the runoff because Voter 3 gave points to candidates A and B.
    I don't object to the election of a more popular candidate. I just think think that since bullet voting is a winning strategy, voters will use it.
    You seem to be claiming that bullet voting won't happen with STAR (or presumably with range or approval either). I am skeptical of that claim, but I can't say that I've done a lot of research on it. I have read that Bucklin voting historically had little impact because of a predominance of bullet voting. I don't see any reason why scored voting would have a different result.
    I said that I abandoned coercive Bucklin voting. I would support non-coercive Bucklin voting with an instant runoff between the two most approved candidates if no majority is reached (a true runoff in which every voter may express a preference without risk of harm to their favored candidate).



  • For your purposes, what do you think of my KP Voting?



  • @robertpdx In Score, some people will bullet vote. However, a sufficient proportion of them may do otherwise to affect the outcome. Predicting that everyone will bullet vote is predicting that they will voluntarily give up power, which experience says people don't do in an individual sense (they do in a collective sense when individual interest and collective interest are at odds -- Prisoner's Dilemma).

    Consider a Nader/Gore/Bush type election and the predicament of the Nader 1 Gore .1 voters. If they plan to bullet vote for Nader, they can count on the Duopoly and its usual tactics to convince all but a fringe to vote for the Duopoly. But if they vote Nader 1 Gore .99, then after a few elections like that in a row, supporters of Nader-type ideas will see they have a majority and will use it to get their candidate.



  • @robertpdx said in Successive Rank Voting:

    We are constrained by the state constitution which says that voters “may vote for one person under the title for each office. Provision may be made by law for the voter’s direct or indirect expression of his first, second or additional choices among the candidates for any office.”

    What systems does this exclude? Take for example Approval Voting. It provides for the voter's expression of her first, second, and additional choices, even if it does not distinguish first from second in the tally.

    Would the constitution conflict with a system saying you can list so many of the candidates as you want, and each one on your list except the last one gets a point, and the last one gets 0.99 point? You could leave other candidates unlisted, and they would get zero from you.



  • This post is deleted!


  • @robertpdx

    The STAR "runoff" is not a true runoff.

    Yes it is an automatic/instant runoff not a true runoff like with the French presidential election. True runoffs are expensive and France is considering getting rid of theirs.

    I just think think that since bullet voting is a winning strategy, voters will use it.

    Exactly, you think bullet voting is a winning strategy with STAR but it is not. This has been shown theoretically and empirically with surveys. There is a case that MAX/MIN voting is a good strategy for score but not bullet. STAR was invented to remove the MAX/MIN incentive from score. People do not and will not behave as you think. Those that do will be hurting themselves. It has been studied a lot.

    I have read that Bucklin voting historically had little impact because of a predominance of bullet voting.

    I do not know of any studies on this. Do you have a reference? The best Bucklin type system is likely Majority Judgment do you have any references for that?

    I would support non-coercive Bucklin voting with an instant runoff between the two most approved candidates if no majority is reached (a true runoff in which every voter may express a preference without risk of harm to their favored candidate).

    Can you describe your proposed system in more detail?



  • Separate runoff elections are expensive. But it is possible to have an instant runoff in which every voter may express a preference without risk to their preferred candidate. RCV/IRV is one example. Ranked voting could also be used to perform an instant ordinary runoff between the two candidates with the most first-rank votes. The method I proposed is at http://www.classicalmatter.org/Election Science/BAIR Voting.pdf .
    The distinguishing feature of my proposed method is that voters are allowed to choose between "friend betrayal" and "later not harm".
    The criticism of Bucklin voting that I read was the FairVote site: http://archive.fairvote.org/?page=2077
    I don't claim to know how voters will vote. But it is quite clear to me that bullet voting maximizes the chances the chosen candidate will win, although it may not maximize voter satisfaction. And even if voters do score multiple candidates, STAR has no guarantee that the winner would have majority support since a majority of voters may have given equal low scores to both finalists. The winner need only have more support than the runner-up. Majority support relative to the runner-up could be guaranteed if equal scores were not allowed, but then it would be a form of ranked voting.



  • @robertpdx You are referring to the Later-no-harm criterion. Later-no-harm is incompatible with the Condorcet criterion. It is basically the most extreme version of majoritarianism. I do not think this is a good property. In fact I would think that passing this criteria should eliminate a system from consideration. We want to represent the public. Representation is not about tyranny of the masses.

    The criticism of Bucklin voting that I read was the FairVote site

    I would not expect to learn much from reading propaganda from a lobby group.

    it is quite clear to me that bullet voting maximizes the chances the chosen candidate will win

    If all you care about is that your favorite wins and you do not care about any compromise then you should bullet vote. Your argument seems circular to me.

    STAR has no guarantee that the winner would have majority support since a majority of voters may have given equal low scores to both finalists.

    I feel like you are just playing with semantics here. Majority support is that the candidate who is preferred by the majority of voters wins. It is implied that this is the majority of voters who have a preference. This is one of the ways that STAR discourages people from bullet voting. There is a lot of incentive to express your true preference.

    The winner need only have more support than the runner-up.

    Yes, as they should per the definition of majoritarianism in a 2 person race

    Majority support relative to the runner-up could be guaranteed if equal scores were not allowed, but then it would be a form of ranked voting.

    That is not just a bad idea because ranking is bad but because you are forcing people to show FALSE preference. As I said in a prior message forcing people to express something so that you get the illusion of an outcome you want is antithetical to democracy.



  • @Jack-Waugh The mechanics of it look good, but I suspect it is too complicated for the general public. I had to read the instructions very carefully to interpret your example. I worry that even adding an "approval" button to ranked-choice voting might be too complicated.



  • @Jack-Waugh As far as I know, there is no case law establishing a precedent for what is acceptable and what isn't. But I wouldn't want to jeopardize an open primary initiative with any method that looks like "voting for more than one candidate". Since ranked choices are specifically allowed, the safe thing to do is to have voters rank choices and treat those choices as discrete ranks.



  • @Keith
    If all you care about is that your favorite wins and you do not care about any compromise then you should bullet vote.

    I am glad we agree on that point. The reason I am interested in open primaries is that there is not enough compromise in politics these days.

    Majority support is that the candidate who is preferred by the majority of voters wins. It is implied that this is the majority of voters who have a preference.

    This is where we disagree. I interpret “majority” as meaning over 50%. The “majority of voters who have a preference” would be a plurality. I worry about the worst-case scenario of the election winner being the most opposed candidate of a majority (>50%) of voters. STAR does not prevent that.

    That is not just a bad idea because ranking is bad but because you are forcing people to show FALSE preference.

    Unless the candidates are clones, I expect all voters to have some preference, whether they express it on the ballot or not. If they really don't have a preference, then they could randomly rank one candidate above the other without any regret. I see no harm in forcing voters to express a preference between near-equally-supported candidates. Requiring unequal ranks or scores also prevents voters from “voting for more than one candidate”, which is unconstitutional in my state.



  • @robertpdx said in Successive Rank Voting:

    Since ranked choices are specifically allowed, the safe thing to do is to have voters rank choices and treat those choices as discrete ranks.

    Given that constraint, I wonder whether one of the answers that would work best against spoiler effect might be Coombs.



  • Or how about a hybrid of Borda and Coombs and Baldwin?

    Voters rank the candidates.

    For each ballot, the tallying algorithm arrives at a rating of the candidates by assigning numbers linearly to the ranks (or with a logistic function instead??). If a voter leaves candidates unranked, those receive the score that would be given to the bottom rank.

    Each round eliminates the candidate with the lowest total score. When one sole candidate remains, that one wins.

    When going from one round to the next, for any ballots that don't use the full range, they are expanded linearly to fill it.


Log in to reply