40 ms·

I'd argue that instant-runoff is _worse than plurality_. It complicates the voting system, makes the wrong tradeoffs (lack of monotonicity is a very serious one

by pxx 5y ago

I'd argue that instant-runoff is _worse than plurality_. It complicates the voting system, makes the wrong tradeoffs (lack of monotonicity is a very serious one), and I'd really like to hear more about why you think approval voting is worse than runoff (instant or otherwise).

cpeterso 5y ago
IRV is also very hard to explain to average voters. Approval voting is easy to explain: you get to give a thumbs up or down to every candidate. In a local mayoral election that used IRV, I've seen some candidates tell their supporters to vote for them in every slot. Either those candidates didn't understand IRV themselves, or they didn't want to try to explain IRV to their supporters.
- mjevans 5y ago
  I prefer https://en.wikipedia.org/wiki/Schulze_method https://en.wikipedia.org/wiki/Schulze_method Though with a candidate modification: add 'none of them / no confidence' as a virtual candidate that can never be removed by the runoff processes. If 'NoT' is the winner then the entire election is thrown out and all of the candidates are banned from the re-run.
  - godelski 5y ago
    Condorcet methods are also hard to explain to voters. The tabulation is rather complicated. At least in comparison to something like cardinal methods. In light of recent recounts I think this is a really important factor to consider. While Condorcet methods have the best VSE, they aren't winning by a huge margin. Besides, VSE isn't everything.
  - ClayShentrup 5y ago
    Condorcet methods perform, at best, only a little better than approval voting in computer simulations. And they have only ever been used in one US city (Marquette, MI) which shortly thereafter repealed it. And they are radically more complicated and unlikely to ever see the light of day. https://electionscience.github.io/vse-sim/vse.html https://electionscience.github.io/vse-sim/vse.html
    - dragonwriter 5y ago
      > Condorcet methods perform, at best, only a little better than approval voting in computer simulations Computer simulations pretty universally paper over the main problem with approval compared to any ranked-ballot (whether forced or unforced preference) methods by pretending that “approval” has a consistent, cross-voter meaning. (This is an even bigger problem with score-based methods than approval.) Personally, for simplicity and good enough single winner methods, I prefer Bucklin which is both better than and simpler to explain than IRV, but more critically I prefer minimizing single-winner elections. 5-member legislative districts using a candidate-centric proportional method like STV or the method that has the same relationship to it as Bucklin as to IRV, which AFAIK isn't a named method, would be far superior to any possible system with single-member districts, and basically neutralize districting as a lever for partisan advantage. Paramount executive elections with a parallel successor (e.g., US Presidential tickets, if they were directly elected, or state Gov/LtGov pairs) can even be made into phased single-winner systems to get some of the advantages of multiwinner systems (not proportionality, but you do get some ability to separate partisan and personal preference in the general election, and to thereby weaken incumbency advantage: basically, use any ranked ballots [0] single winner election for the top spot, eliminate the winner, and rerun the election on the same ballots with the same method for the constitutional successor.) [0] while I am not partial to them, approval or score systems could be used to; simple FPTP fails because it mean those who voted for the top winner have no voice in the designated successor.
- Apofis 5y ago
  But I don't support every candidate I would vote for equally.
- godelski 5y ago
  People are actually very familiar with cardinal systems. You have your 5 star ratings (score) for restaurants, apps, etc. You have your thumbs up/down on Netflix. "On a scale from 1 to 10 how do you feel about x" etc. I always see people say that cardinal systems are hard to explain but the fact is that we use systems similar to them on almost a daily basis. (At least the front end is the same)
Apofis 5y ago
I don't support every candidate on the ballot equally because I believe every candidate's abilities to do the job vary. First choice, second choice, third choice is not hard to explain and I do not understand why some people think Americans have been lobotimized. NYC now has run off voting and it turned out marvelously, with Eric Adams being elected. https://vote.nyc/page/ranked-choice-voting https://vote.nyc/page/ranked-choice-voting

by pxx 5y ago

cpeterso 5y ago

IRV is also very hard to explain to average voters. Approval voting is easy to explain: you get to give a thumbs up or down to every candidate. In a local mayoral election that used IRV, I've seen some candidates tell their supporters to vote for them in every slot. Either those candidates didn't understand IRV themselves, or they didn't want to try to explain IRV to their supporters.

mjevans 5y ago
I prefer https://en.wikipedia.org/wiki/Schulze_method https://en.wikipedia.org/wiki/Schulze_method Though with a candidate modification: add 'none of them / no confidence' as a virtual candidate that can never be removed by the runoff processes. If 'NoT' is the winner then the entire election is thrown out and all of the candidates are banned from the re-run.
- godelski 5y ago
  Condorcet methods are also hard to explain to voters. The tabulation is rather complicated. At least in comparison to something like cardinal methods. In light of recent recounts I think this is a really important factor to consider. While Condorcet methods have the best VSE, they aren't winning by a huge margin. Besides, VSE isn't everything.
- ClayShentrup 5y ago
  Condorcet methods perform, at best, only a little better than approval voting in computer simulations. And they have only ever been used in one US city (Marquette, MI) which shortly thereafter repealed it. And they are radically more complicated and unlikely to ever see the light of day. https://electionscience.github.io/vse-sim/vse.html https://electionscience.github.io/vse-sim/vse.html
  - dragonwriter 5y ago
    > Condorcet methods perform, at best, only a little better than approval voting in computer simulations Computer simulations pretty universally paper over the main problem with approval compared to any ranked-ballot (whether forced or unforced preference) methods by pretending that “approval” has a consistent, cross-voter meaning. (This is an even bigger problem with score-based methods than approval.) Personally, for simplicity and good enough single winner methods, I prefer Bucklin which is both better than and simpler to explain than IRV, but more critically I prefer minimizing single-winner elections. 5-member legislative districts using a candidate-centric proportional method like STV or the method that has the same relationship to it as Bucklin as to IRV, which AFAIK isn't a named method, would be far superior to any possible system with single-member districts, and basically neutralize districting as a lever for partisan advantage. Paramount executive elections with a parallel successor (e.g., US Presidential tickets, if they were directly elected, or state Gov/LtGov pairs) can even be made into phased single-winner systems to get some of the advantages of multiwinner systems (not proportionality, but you do get some ability to separate partisan and personal preference in the general election, and to thereby weaken incumbency advantage: basically, use any ranked ballots [0] single winner election for the top spot, eliminate the winner, and rerun the election on the same ballots with the same method for the constitutional successor.) [0] while I am not partial to them, approval or score systems could be used to; simple FPTP fails because it mean those who voted for the top winner have no voice in the designated successor.
Apofis 5y ago
But I don't support every candidate I would vote for equally.
godelski 5y ago
People are actually very familiar with cardinal systems. You have your 5 star ratings (score) for restaurants, apps, etc. You have your thumbs up/down on Netflix. "On a scale from 1 to 10 how do you feel about x" etc. I always see people say that cardinal systems are hard to explain but the fact is that we use systems similar to them on almost a daily basis. (At least the front end is the same)

Apofis 5y ago

I don't support every candidate on the ballot equally because I believe every candidate's abilities to do the job vary. First choice, second choice, third choice is not hard to explain and I do not understand why some people think Americans have been lobotimized. NYC now has run off voting and it turned out marvelously, with Eric Adams being elected. https://vote.nyc/page/ranked-choice-voting https://vote.nyc/page/ranked-choice-voting