Top two runoff is the best we can do
Given the referendum vote in Minneapolis for IRV for city elections, I thought this is my last chance to test my arguments against a fundamental failing of IRV as proposed. (I should add that I live in a suburb and work in Minneapolis, so it won't effect me directly)
After significant research and analysis I have concluded I only unconditionally support IRV if it is implemented as a top-two runoff process, rather than the bottom-up elimination as usually described.
I'd appreciate feedback here. I see my argument as fully defendable, but when I've tried explaining this in the past, I find IRV promoters don't want controversy, and most people who study election methods want FURTHER reform away from plurality like Condorcet, so my poor top-two "option" seems unduly, and in my opinion dangerously neglected.
Well, the time is NOW when IRV is on "firm ground", but not yet planted to discuss elimination rules to be used.
I base my reasoning on a principle I call "plurality fairness". By that I mean under a plurality (single vote) system voters are rewarded by working together and uniting behind a single candidate if the they want a good chance to win. Candidates are only rewarded by becoming the first choice of as many voters as possible.
Given this pre-election work, it can be assumed that on election day "voters have compromised as much as they're willing", and no more voting rounds will make a difference. By that reasoning, the plurality winner is sufficient, no matter how small a portion of votes he got.
It is a false assumption that "no one will compromise further", but that fear of "last chance" is what strengthens voters to unite and this reward system shouldn't be tampered with lightly. But it may be tinkered at with care.
Our "tool' is a "harmless" mathematical principle called "majority rule" that says "When a candidate gains over a majority (50%) of the vote, then no other candidate can win." I say "harmless" because any self-respecting winner ought to believe he or she can beat any competitor head-to-head. And a winner who gains a majority is further honored by knowing a majority is in support.
The top-two runoff (or top-two primary or top-two IRV) offers this confirmation step. And as a consequence, it also allows the plurality-second candidate a chance to stand as an equal with a chance to win. This candidate might be considered the "wildcard team" by sports analogy - being rewarded for being "almost on top".
Well, that's very good, but some reformers will go further. They'll worrifully notice that in a 4 candidate contest, sometimes a candidate who "would have been second" falls to a close third due to a "weak spoiler" who divided the support. Those divided votes on the bottom will ask for fairness and claim the right to a 3-round runoff where the top-three get a chance to compete as equals.
Now we've already got the Majority-rule principle covered by the Top-two process, so why should we try a Top-three runoff? (Via two elimination rounds)
In my judgment there's little support for this step - just ask the candidates themselves. The plurality first and plurality second candidates want a guarantee for their shot in the final round - the only round they're treated fairly as individuals. Supporters of these plurality top-two candidates ALSO want their candidates their shot in the final round. There's not going to be much pity by these voters on top to risk losing their shot in the final round.
Now again remember the (brutal but sincere) principle of plurality is to reward candidates to compromise BEFORE the election, and the assumption that candidates (and voters) who don't compromise will lower their chances of picking the winner. A top-two runoff is EXACTLY like plurality - it encourages voters to compromise early, and directly reward candidates for this success.
Bottom-up IRV doesn't support this goal. Bottom-up IRV can treat TWO candidates fairly, but it chooses those two candidates through a convoluted (recursive elimination) process where a small change in votes on the bottom and elimination order may change the final-two and hence the winner. There's no defendable justification for this, and actual "harm" here to a "plurality fairness" principle which a "one vote" system must always promote.
I'm not against IRV, and I'm not against giving a "plurality third" candidate a fair shot to win. However I want this done RIGHT. If a plurality third candidate gets a shot head-to-head in the final round, then I believe the plurality-second candidate also deserves equal treatment. That is, if we want this, we need a full Condorcet pair-wise (head-to-head) competition among all three, and then we can be sure which deserves to win. (shushing unlikely rock-paper-scissors cyclic preferences away)
So for now I conclude a "top-two IRV" is the best we can support with "plurality fairness" - rewarding voters fairly by those who compromise early.
An example, using political parties:
Plurality vote: GOP=34%, IP=30%, DFL=28%, Green=8%
The top-two runoff would be between GOP and IP. The IP supporters EARNED "playoff" status here under top-two.
The bottom-up runoff final-two are yet uncertain, but we imagine the Green transfer votes would push the DFL to second and take away the IP candidate's chance to compete head-to-head against either competitor.
All we know here is that ANY one of the top three could win depending on transfer votes, and only two are allowed to compete in the final round.
Please note that I'm not saying top-two is better than bottom-up. I accept statistically that bottom-up IRV will more often pick a Condorcet winner (the candidate who can win head-to-head against all others) than a top-two process. I'm just saying "usually better" is insufficient to justify more than two rounds that SOMETIMES eliminate the Condorcet (head-to-head) winner who had been in the original plurality top-two.
Hopefully I've argued my points reasonably clearly. It's not easy to express concern over the sides-effects of recursive elimination in 4-way contests. Examples that try to clarify issue entice people to take sides.
My point is "when there's three strong sides", there's no "fairness for all" in a one-vote system. And my conclusion "If we must choose two, take the plurality top-two".
It is a "small difference" and that's why it is hard to demonstrate. However when it is different, it's always disagreeable to a candidate who will likely have a good fraction of the vote, and who will be angry at being displaced, and that's the whole problem in a nutshell.
Maybe in less competitive nonpartisan city elections, bottom-up IRV is acceptable and candidates will gladly accept elimination (Even party endorsements are "cooperative" for allowing multiple elimination rounds), but I expect it'll never be accepted in partisan elections when candidates see this subtle effect explained clearly. They may accept against their will and will simply try to undermined the reform as circumstances allow.
Comments are welcome. Thanks!
After significant research and analysis I have concluded I only unconditionally support IRV if it is implemented as a top-two runoff process, rather than the bottom-up elimination as usually described.
I'd appreciate feedback here. I see my argument as fully defendable, but when I've tried explaining this in the past, I find IRV promoters don't want controversy, and most people who study election methods want FURTHER reform away from plurality like Condorcet, so my poor top-two "option" seems unduly, and in my opinion dangerously neglected.
Well, the time is NOW when IRV is on "firm ground", but not yet planted to discuss elimination rules to be used.
I base my reasoning on a principle I call "plurality fairness". By that I mean under a plurality (single vote) system voters are rewarded by working together and uniting behind a single candidate if the they want a good chance to win. Candidates are only rewarded by becoming the first choice of as many voters as possible.
Given this pre-election work, it can be assumed that on election day "voters have compromised as much as they're willing", and no more voting rounds will make a difference. By that reasoning, the plurality winner is sufficient, no matter how small a portion of votes he got.
It is a false assumption that "no one will compromise further", but that fear of "last chance" is what strengthens voters to unite and this reward system shouldn't be tampered with lightly. But it may be tinkered at with care.
Our "tool' is a "harmless" mathematical principle called "majority rule" that says "When a candidate gains over a majority (50%) of the vote, then no other candidate can win." I say "harmless" because any self-respecting winner ought to believe he or she can beat any competitor head-to-head. And a winner who gains a majority is further honored by knowing a majority is in support.
The top-two runoff (or top-two primary or top-two IRV) offers this confirmation step. And as a consequence, it also allows the plurality-second candidate a chance to stand as an equal with a chance to win. This candidate might be considered the "wildcard team" by sports analogy - being rewarded for being "almost on top".
Well, that's very good, but some reformers will go further. They'll worrifully notice that in a 4 candidate contest, sometimes a candidate who "would have been second" falls to a close third due to a "weak spoiler" who divided the support. Those divided votes on the bottom will ask for fairness and claim the right to a 3-round runoff where the top-three get a chance to compete as equals.
Now we've already got the Majority-rule principle covered by the Top-two process, so why should we try a Top-three runoff? (Via two elimination rounds)
In my judgment there's little support for this step - just ask the candidates themselves. The plurality first and plurality second candidates want a guarantee for their shot in the final round - the only round they're treated fairly as individuals. Supporters of these plurality top-two candidates ALSO want their candidates their shot in the final round. There's not going to be much pity by these voters on top to risk losing their shot in the final round.
Now again remember the (brutal but sincere) principle of plurality is to reward candidates to compromise BEFORE the election, and the assumption that candidates (and voters) who don't compromise will lower their chances of picking the winner. A top-two runoff is EXACTLY like plurality - it encourages voters to compromise early, and directly reward candidates for this success.
Bottom-up IRV doesn't support this goal. Bottom-up IRV can treat TWO candidates fairly, but it chooses those two candidates through a convoluted (recursive elimination) process where a small change in votes on the bottom and elimination order may change the final-two and hence the winner. There's no defendable justification for this, and actual "harm" here to a "plurality fairness" principle which a "one vote" system must always promote.
I'm not against IRV, and I'm not against giving a "plurality third" candidate a fair shot to win. However I want this done RIGHT. If a plurality third candidate gets a shot head-to-head in the final round, then I believe the plurality-second candidate also deserves equal treatment. That is, if we want this, we need a full Condorcet pair-wise (head-to-head) competition among all three, and then we can be sure which deserves to win. (shushing unlikely rock-paper-scissors cyclic preferences away)
So for now I conclude a "top-two IRV" is the best we can support with "plurality fairness" - rewarding voters fairly by those who compromise early.
An example, using political parties:
Plurality vote: GOP=34%, IP=30%, DFL=28%, Green=8%
The top-two runoff would be between GOP and IP. The IP supporters EARNED "playoff" status here under top-two.
The bottom-up runoff final-two are yet uncertain, but we imagine the Green transfer votes would push the DFL to second and take away the IP candidate's chance to compete head-to-head against either competitor.
All we know here is that ANY one of the top three could win depending on transfer votes, and only two are allowed to compete in the final round.
Please note that I'm not saying top-two is better than bottom-up. I accept statistically that bottom-up IRV will more often pick a Condorcet winner (the candidate who can win head-to-head against all others) than a top-two process. I'm just saying "usually better" is insufficient to justify more than two rounds that SOMETIMES eliminate the Condorcet (head-to-head) winner who had been in the original plurality top-two.
Hopefully I've argued my points reasonably clearly. It's not easy to express concern over the sides-effects of recursive elimination in 4-way contests. Examples that try to clarify issue entice people to take sides.
My point is "when there's three strong sides", there's no "fairness for all" in a one-vote system. And my conclusion "If we must choose two, take the plurality top-two".
It is a "small difference" and that's why it is hard to demonstrate. However when it is different, it's always disagreeable to a candidate who will likely have a good fraction of the vote, and who will be angry at being displaced, and that's the whole problem in a nutshell.
Maybe in less competitive nonpartisan city elections, bottom-up IRV is acceptable and candidates will gladly accept elimination (Even party endorsements are "cooperative" for allowing multiple elimination rounds), but I expect it'll never be accepted in partisan elections when candidates see this subtle effect explained clearly. They may accept against their will and will simply try to undermined the reform as circumstances allow.
Comments are welcome. Thanks!
posted by Ares Olympus at 11:33 PM
0 Comments:
Post a Comment
<< Home