Ranked-Choice Voting and a noteworthy anomaly in Burlington Vermont in 2009
-
Just to let you know that, being a resident of Burlington Vermont who was around in 2006 and 2009 when IRV was used, that I have a paper about this that has been invited to be published in Constitutional Political Economy, but I do not know when. I think within a year.
I am just a simple advocate of One-Person-One-Vote and for Majority Rule: If more voters mark their ballots preferring Candidate A over Candidate B than the number of voters marking their ballots to the contrary, then Candidate B is not elected. So I think that Condorcet got it right and Hare not so much.
-
@rbj said in Ranked-Choice Voting and a noteworthy anomaly in Burlington Vermont in 2009:
If more voters mark their ballots preferring Candidate A over Candidate B than the number of voters marking their ballots to the contrary, then Candidate B is not elected.
But of course, that isn't always possible. I agree that it is vitally important for a method to choose the Condorcet winner if there is one, but the problem is that there isn't always.
STAR doesn't always choose the Condorcet winner, nor does Score. Approval ballots don't provide enough information to determine the Condorcet winner. (it may converge on the Condorcet winner, if voters have enough information and are properly strategic).
Do I understand your preference to be "“Bottom Two Runoff – Single Transferable Vote”? Basically a variation of IRV that chooses the Condorcet winner if there is one? I'm a bit confused as to the difference between "Condorcet winner" and "Pairwise champion" in your paper.
I personally prefer using Cardinal (i.e. score) ballots, since I think they are easier on voters, even if they are tabulated as if they are simply ranked. We use RCV here (San Francisco) and the ballots are awful.
I'm currently interested in a method with cardinal ballots like STAR, but that is a Condorcet method (so it only uses the full scores in rare cases) and is simple to explain: "Compare all candidates pairwise. Of those that have the most pairwise wins, the candidate with the highest total score wins." See https://www.votingtheory.org/forum/topic/130/star-like-method-reverse-star
-
Hi @rob,
If more voters mark their ballots preferring Candidate A over Candidate B than the number of voters marking their ballots to the contrary, then Candidate B is not elected.
But of course, that isn't always possible. I agree that it is vitally important for a method to choose the Condorcet winner if there is one, but the problem is that there isn't always.
True, but so far there hasn't been a cycle or "Condorcet paradox" in any single one of the 440 RCV elections analyzed by FairVote. Out of the 440, each RCV election did elect the Condorcet Winner, except for one, the Burlington 2009 election.
So "Pairwise Champion", "Consistent Majority Candidate" are neologisms intended to be a descriptive term for the Condorcet Winner.
So why use an election method that is not guaranteed to elect the Condorcet winner? Why not just use a Condorcet-consistent method instead of a method that gets approximately the same results?
-
Here are some templates for possible legislation. Hare RCV (a.k.a. IRV), BTR, and two versions of straight-ahead Condorcet are described.
.
.
.
Here is Hare RCV (a.k.a. IRV):
All elections of [office] shall be by ballot, using a system of ranked-choice voting without a separate runoff election. The presiding election officer shall implement a ranked-choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference. Lower ordinal preference shall be considered higher rank and the candidate marked as first preference is considered ranked highest. Equal ranking of candidates shall not be allowed. Any candidate not marked with a preference shall be considered as ranked lower than every candidate marked with a preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant runoff retabulation shall be performed by the presiding election officer. The instant runoff retabulation shall be conducted in sequential rounds. A "continuing candidate" is defined as a candidate that has not been defeated in any previous round. Initially, no candidate is defeated and all candidates begin as continuing candidates.
(4) In each round, every ballot shall count as a single vote for whichever continuing candidate the voter has ranked highest. The candidate with fewest votes is defeated in the current round.
(5) The aforementioned instant runoff retabulation, eliminating one candidate each round, shall be repeated until only two candidates remain. The remaining candidate then receiving the greatest number of votes is elected.
(6) The [governing jurisdiction] may adopt additional regulations consistent with this subsection to implement these standards.
.
.
.
.
Here is Bottom-Two Runoff:
All elections of [office] shall be by ballot, using a system of ranked-choice voting without a separate runoff election. The presiding election officer shall implement a ranked-choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference. Lower ordinal preference shall be considered higher rank and the candidate marked as first preference is considered ranked highest. Equal ranking of candidates shall not be allowed. Any candidate not marked with a preference shall be considered as ranked lower than every candidate marked with a preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, an instant runoff retabulation shall be performed by the presiding election officer. The instant runoff retabulation shall be conducted in sequential rounds. A "continuing candidate" is defined as a candidate that has not been defeated in any previous round. Initially, no candidate is defeated and all candidates begin as continuing candidates.
(4) In each round, every ballot shall count as a single vote for whichever continuing candidate the voter has ranked highest. The two candidates with the fewest votes in a round, herein denoted as "A" and "B", shall contend in a runoff in which the candidate, A or B, with lesser voter support shall be defeated in the current round. If the number of ballots ranking A higher than B exceeds the number of ballots ranking B higher than A, then B has lesser voter support, B is defeated, and A continues to the following round. Likewise, if the number of ballots ranking B higher than A exceeds the number of ballots ranking A higher than B, then A has lesser voter support, A is defeated, and B continues to the following round. In the case that the aforementioned measures of voter support of A and B are tied, then the candidate with fewest votes is defeated in the current round.
(5) The aforementioned instant runoff retabulation, eliminating one candidate each round, shall be repeated until only two candidates remain. The remaining candidate then receiving the greatest number of votes is elected.
(6) The [governing jurisdiction] may adopt additional regulations consistent with this subsection to implement these standards.
.
.
.
.
Here is straight-ahead Condorcet (version 1):
All elections of [office] shall be by ballot, using a system of ranked-choice voting without a separate runoff election. The presiding election officer shall implement a ranked-choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference. Lower ordinal preference shall be considered higher rank and the candidate marked as first preference is considered ranked highest. Equal ranking of candidates shall be allowed. Any candidate not marked with a preference shall be considered as ranked lower than every candidate marked with a preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, a Condorcet-consistent retabulation shall be performed by the presiding election officer. The candidate, who is the Condorcet winner, is elected if the rankings on all of the ballots indicate that this one candidate defeats, with a simple majority of voter preferences, every other candidate when compared in turn with each other individual candidate. A selected candidate defeats another candidate by a simple majority when the number of ballots marked ranking the selected candidate higher than the other candidate exceeds the number of ballots marked to the contrary.
(4) If no Condorcet winner exists in step (3), then the candidate with the plurality of first preferences is elected.
(5) The [governing jurisdiction] may adopt additional regulations consistent with this subsection to implement these standards.
.
.
.
.
Here is another straight-ahead Condorcet (version 2):
All elections of [office] shall be by ballot, using a system of ranked-choice voting without a separate runoff election. The presiding election officer shall implement a ranked-choice voting protocol according to these guidelines:
(1) The ballot shall give voters the option of ranking candidates in order of preference. Lower ordinal preference shall be considered higher rank and the candidate marked as first preference is considered ranked highest. Equal ranking of candidates shall be allowed. Any candidate not marked with a preference shall be considered as ranked lower than every candidate marked with a preference.
(2) If a candidate receives a majority (over 50 percent) of first preferences, that candidate is elected.
(3) If no candidate receives a majority of first preferences, a Condorcet-consistent retabulation shall be performed by the presiding election officer. The retabulation shall examine each possible pairing of candidates. If N is the number of candidates, then the number of possible pairings of candidates is N(N-1)/2. For each possible pairing of candidates, if fewer ballots are marked ranking a selected candidate over the other candidate than the number of ballots marked to the contrary, then the selected candidate is identified as defeated. After all candidate pairs are examined, the candidate who remains not identified as defeated is the Condorcet winner and is elected.
(4) If no Condorcet winner exists in step (3), then the candidate with the plurality of first preferences is elected.
(5) The [governing jurisdiction] may adopt additional regulations consistent with this subsection to implement these standards.
.
.
.
.
-
@rbj said in Ranked-Choice Voting and a noteworthy anomaly in Burlington Vermont in 2009:
So why use an election method that is not guaranteed to elect the Condorcet winner? Why not just use a Condorcet-consistent method instead of a method that gets approximately the same results?
I agree, 100%, that Condorcet winner should win, if there is one. Especially if the ballot allows for determining the Condorcet winner (which rules out Approval ballots).
From a marketing point of view, I think most Condorcet methods are too hard to understand. I think 90% of people would read your legislation templates and not understand them, or at least, wouldn't understand them quickly. I understand that the actual legislation isn't the way it would be presented to voters, but still. I also have concerns that most Condorcet compatible methods are hard to show results in a meaningful way. That's one reason I sorta got excited about the method I just posted since it is so easy to show the results in a way that is easy to wrap your head around. (admittedly, it is a tougher sell to, say, Burlington or San Francisco, because it uses a different ballot)
How would you feel about a ranked method that selected the Condorcet winner, but if none, just fell back on Hare? It might be easier to sell that to an electorate such as Burlington, since you could just add one line to the existing Hare RCV legislation.
-
@rob said in Ranked-Choice Voting and a noteworthy anomaly in Burlington Vermont in 2009:
How would you feel about a ranked method that selected the Condorcet winner, but if none, just fell back on Hare?
ick.
It might be easier to sell that to an electorate such as Burlington, since you could just add one line to the existing Hare RCV legislation.
no, you would have to take the "straight-ahead Condorcet" language above and for step (4) replace it with the Hare RCV language.
This might be one advantage of the BTR version. But personally I would rather see a "pure" Condorcet method, with a "completion method" to deal with the microscopically-unlikely case of a cycle.
-
I'm curious why, if a cycle is microscopically unlikely, Hare would be so bad as a fallback in that case?
Doesn't one of yours just fall back to plurality? (or do I misunderstand?)
I've been a fan of Condorcet forever (I participated in the election methods mailing list all the way back to 2003, I think I recognize your name from there), but just have become unconvinced that complex Condorcet completion methods are sellable to the public. Condorcet methods have gotten zero traction over the nearly 20 years I've been following this subject, from what I see. I suspect the complexity is the problem. Do you think there is a plausible path forward for a Condorcet method, at least one that isn't a complete rethinking?
Another reason I suspect Condorcet didn't get traction was a certain person (cough Clay cough) who aggressively and abrasively argued against them and told everyone that thought they were better than, for instance Score, that they were stupid.
I know some people think that "strength of preference" must be considered, and I disagree... the beauty of Condorcet is that it ignores it. I also very much disagree with some of the IRV people who say that Condorcet would elect weak candidates. I think the past 5 years have shown that polarization is 1000 times more of a problem than supposedly weak candidates.
Anyway, if you must use ranked ballots (as opposed to Cardinal ballots), what about just saying, resolve the pairwise tie via Borda count? I mean, sure, Borda is broken as a method in itself, but it's not terrible if it is used as a fallback for a very unlikely situation, so no one is going to be able to strategically manipulate it.
BTW, what do you mean by "straight ahead Condorcet"? What does "straight ahead" mean there?
-
I really like the Pairwise Sorted Margins methods that @Ted-Stern brought up (see: https://www.votingtheory.org/forum/topic/112/preference-approval-sorted-margins), although it probably introduces too many new concepts to voters to be effective for advocacy. There are two things that I like about it. First, it seems to be a logical extension of the runoff step in STAR. I expect that when there are more candidates, the value of the runoff step in STAR will be diluted, whereas the importance of the pairwise comparisons in Sorted Margins doesn't diminish with an increase in the number of candidates. Second, it does better against burying than most Condorcet methods, usually well enough to make it not worth attempting.
Score Sorted Margins[100]; STAR[90]; Score[81]; Approval[59]; IRV[18]; FPTP[0]
-
@Marylander Agree with regard to STAR having issues with larger numbers of candidates. Also agree that this one doesn't seem like one that would be easy to sell ("effective for advocacy"), at least unless it is presented more clearly.
-
Sorry @rob, I never saw this comment.
BTW, what do you mean by "straight ahead Condorcet"? What does "straight ahead" mean there?
It means a direct implementation of the Condorcet criterion on every possible pairing of candidates. Mark the loser in every pair with an "L" on their forehead. Then elect the only remaining candidate not marked with an "L".
-
@rb-j Right, no problem. I have since read your description/ proposed legislation and I understand you describe the method for resolving Condorcet failures as a simply choosing the plurality winner. (which is fine, in my opinion, because they are so rare)
