3 ms·

OTOH, with ordinal methods you are forced to distinguish between candidates that you think are equally good. Say if you have three candidates, A, B, and C, wher

by jabl 5y ago

OTOH, with ordinal methods you are forced to distinguish between candidates that you think are equally good. Say if you have three candidates, A, B, and C, where A and B are very similar and you like both, though with a slight preference to A, whereas you dislike C very much. With approval voting you'd approve A and B, with score voting (say the vote is an integer between 1 and 5) you could vote A=5, B=4, C=1 (or A=5, B=5, C=1 if you think A and B are really close). Whereas with an ordinal system you'd vote A-B-C, but you have no way of 'telling the voting math' that the distance between A and B is much shorter than between B and C. Looking at the results of these simulations described in the article, and similar simulations using 'Bayesian regret', it seems that these issues tend to make cardinal systems better. Also, considering the various gotchas of voting systems, there is the famous Arrow impossibility theorem. However, it applies only to ordinal systems. There is no similar thing for cardinal systems.

vintermann 5y ago
> Whereas with an ordinal system you'd vote A-B-C, but you have no way of 'telling the voting math' that the distance between A and B is much shorter than between B and C. Kenneth Arrow is famous for his impossibility theorem, but he also rejected cardinal systems out of hand, because you can't honestly judge the strength of your own preferences compared to other voters. He argued (pretty well, I thought) that the extra information in cardinal systems was useless.
- jabl 5y ago
  > Kenneth Arrow is famous for his impossibility theorem, but he also rejected cardinal systems out of hand, because you can't honestly judge the strength of your own preferences compared to other voters. He argued (pretty well, I thought) that the extra information in cardinal systems was useless. Indeed, that is a strong argument in favor of approval voting; if everybody votes strategically under score voting it reduces to approval voting, so why bother with the extra complexity of score voting? FWIW, I recall Arrow himself said in some interview that his personal favorite was approval voting. Secondly, I'm not sure I understand the point about strength of preferences. Since everybody has the same range from within to assign preferences, it's all about the relative preferences among the candidates. Doesn't matter if I hate candidate X with the heat of a thousand suns, I can't give that candidate a score of -1000, only the minimum, just like everybody else.
  - notahacker 5y ago
    > Secondly, I'm not sure I understand the point about strength of preferences. Since everybody has the same range from within to assign preferences, it's all about the relative preferences among the candidates. Doesn't matter if I hate candidate X with the heat of a thousand suns, I can't give that candidate a score of -1000, only the minimum, just like everybody else Are the relative preference intensities expressed real? Everybody has the same range to vote Uber drivers, but apparently they're all nearly perfect. If a very successful campaigning organization convinces people that they have to give the Judean People's Front 0 for the People's Front of Judea to win even though the parties have similar policies and lots of Judeans strongly prefer them to the Romans, is the resulting distribution a real or artificial expression of the intensity of preferences?
  - jlokier 5y ago
    With approval voting, a voter faces an impossible to solve dilemma. If they hate one candidate a lot, and like one candidate a lot, should they approve all candidates except the one they hate, or only approve the one candidate they like? With strength of preferences, some people will give their scores as 2,3,4 (out of 5) because nobody is perfect and nobody is truly the worst imaginable. But other people will give out scores 1,3,5 if they feel that has more effect, or because the are comfortable using score extremes everywhere else in life - with no real difference in the strength of conviction between those two groups of people. And some people will use only 1,5 because they think giving every candidate except the one you hate a 5 increases the chance of the one you hate not getting in, even if they don't really like any of the others, or like one more than the others. It doesn't mean they are "pro" any of the candidates. And some will give their favoured candidate 5 and everyone else 1 or 2.
- godelski 5y ago
  1) Arrow's theorem isn't really about voting though. 2) Arrow's doesn't apply to cardinal systems. 3) Arrow said that for voting cardinal systems were probably superior. From wiki > Arrow originally rejected cardinal utility as a meaningful tool for expressing social welfare, and so focused his theorem on preference rankings, but later stated that a cardinal score system with three or four classes "is probably the best". Again note that he's talking about social welfare and not voting.
aragilar 5y ago
Can you point to a real example where you can't distinguish between candidates (I'm Australian, so vote with IRV, and I can't say it ever comes up, except perhaps when choosing who should be last vs second last ;)? Approval voting would seem to have the same convergence to a two party system as FPTP where you have a chamber composed of multiple members, a mayor or similar role would be different. Multi-member representation would be ideal, but I can see the appeal of having a single person represent you.
- jabl 5y ago
  > Can you point to a real example where you can't distinguish between candidates With hundreds/thousands(?) of ranked voting elections worldwide, with certainly millions of voters, you're statistically guaranteed to have that situation for some voter somewhere. > Approval voting would seem to have the same convergence to a two party system as FPTP where you have a chamber composed of multiple members, a mayor or similar role would be different. Not sure the voting method used for single-winner elections matter that much for the dynamics leading to a duopoly or not. I think convergence to a duopoly is most strongly influenced by winner-take-all districts. The solution to that, as you pointed out, is multi-member districts with proportional representation. > but I can see the appeal of having a single person represent you ... assuming that 'your' party wins. Otherwise you have no representation at all and you feel disfranchised? I'd say the major appeal of single-winner districts is that the districts are much smaller, and you have a better chance to get to know your own representative (again, assuming it's somebody you find broadly acceptable).
  - aragilar 5y ago
    > With hundreds/thousands(?) of ranked voting elections worldwide, with certainly millions of voters, you're statistically guaranteed to have that situation for some voter somewhere. One could argue the same thing applies when a voter chooses whether or not to approve a vote for a candidate they don't like in order to prevent an even more disliked candidate from winning. To me, it seems a false choice either way (to order or to approve). Possibly this a framing problem, given I've grown up in a system where IRV/STV is used, I naturally compare candidates, maybe someone growing up in a Approval voting system would naturally think about whether they like a specific candidate or not. > ... assuming that 'your' party wins. Otherwise you have no representation at all and you feel disfranchised? Having two houses where one uses STV and the other has single member districts does have its advantages as a "why not both" option.
jlokier 5y ago
I personally favour ranked voting with the feature that you can put multiple candidates at the same rank, and you can leave off candidates you really dislike. (Putting candidates in the "leave off" set is the same as ranking them below a "none of the below" pseudo-candidate because it means they won't inherit unused vote shares during counting). That solves your distinguish and C-dislike objections, and feels more honest when voting. It also lets you feel fair when there are two candidates you both like equally and don't want to add to signal-favouritism when the results come out. At a local club (where people are friendly so it's not meant to be competitive like in government), having to rank multiple loved and personally known candidates for a committee occupied by multiple people, when you like several equally, doesn't change the outcome at all, assuming they win, but it tends to greatly amplify the tiny, irrelevant biases fron the most petty things people have to draw on into very large differences in reported scores. This makes it looks like one person is very much preferred to be in charge of the rest of the committee or team, much more than is actually true, ie. it doesn't actually represent the voters intentions. The election outcome is representative, but the outsized score differences are interpreted wrongly. So this seeds ungrounded internal divisions after the vote, peculiar strategic voting at subsequent elections, and peculiar behaviours to chase after those "tiny, irrelevant biases". This isn't good for anyone involved. So it's good to allow voters to rank candidates equally for this reason too, if the voter genuinely feels that way.
zajio1am 5y ago
> OTOH, with ordinal methods you are forced to distinguish between candidates that you think are equally good. That is not true, ordinal systems (including IRV) may allow weak ordering.

by jabl 5y ago

vintermann 5y ago

> Whereas with an ordinal system you'd vote A-B-C, but you have no way of 'telling the voting math' that the distance between A and B is much shorter than between B and C. Kenneth Arrow is famous for his impossibility theorem, but he also rejected cardinal systems out of hand, because you can't honestly judge the strength of your own preferences compared to other voters. He argued (pretty well, I thought) that the extra information in cardinal systems was useless.

jabl 5y ago
> Kenneth Arrow is famous for his impossibility theorem, but he also rejected cardinal systems out of hand, because you can't honestly judge the strength of your own preferences compared to other voters. He argued (pretty well, I thought) that the extra information in cardinal systems was useless. Indeed, that is a strong argument in favor of approval voting; if everybody votes strategically under score voting it reduces to approval voting, so why bother with the extra complexity of score voting? FWIW, I recall Arrow himself said in some interview that his personal favorite was approval voting. Secondly, I'm not sure I understand the point about strength of preferences. Since everybody has the same range from within to assign preferences, it's all about the relative preferences among the candidates. Doesn't matter if I hate candidate X with the heat of a thousand suns, I can't give that candidate a score of -1000, only the minimum, just like everybody else.
- notahacker 5y ago
  > Secondly, I'm not sure I understand the point about strength of preferences. Since everybody has the same range from within to assign preferences, it's all about the relative preferences among the candidates. Doesn't matter if I hate candidate X with the heat of a thousand suns, I can't give that candidate a score of -1000, only the minimum, just like everybody else Are the relative preference intensities expressed real? Everybody has the same range to vote Uber drivers, but apparently they're all nearly perfect. If a very successful campaigning organization convinces people that they have to give the Judean People's Front 0 for the People's Front of Judea to win even though the parties have similar policies and lots of Judeans strongly prefer them to the Romans, is the resulting distribution a real or artificial expression of the intensity of preferences?
- jlokier 5y ago
  With approval voting, a voter faces an impossible to solve dilemma. If they hate one candidate a lot, and like one candidate a lot, should they approve all candidates except the one they hate, or only approve the one candidate they like? With strength of preferences, some people will give their scores as 2,3,4 (out of 5) because nobody is perfect and nobody is truly the worst imaginable. But other people will give out scores 1,3,5 if they feel that has more effect, or because the are comfortable using score extremes everywhere else in life - with no real difference in the strength of conviction between those two groups of people. And some people will use only 1,5 because they think giving every candidate except the one you hate a 5 increases the chance of the one you hate not getting in, even if they don't really like any of the others, or like one more than the others. It doesn't mean they are "pro" any of the candidates. And some will give their favoured candidate 5 and everyone else 1 or 2.
godelski 5y ago
1) Arrow's theorem isn't really about voting though. 2) Arrow's doesn't apply to cardinal systems. 3) Arrow said that for voting cardinal systems were probably superior. From wiki > Arrow originally rejected cardinal utility as a meaningful tool for expressing social welfare, and so focused his theorem on preference rankings, but later stated that a cardinal score system with three or four classes "is probably the best". Again note that he's talking about social welfare and not voting.

aragilar 5y ago

Can you point to a real example where you can't distinguish between candidates (I'm Australian, so vote with IRV, and I can't say it ever comes up, except perhaps when choosing who should be last vs second last ;)? Approval voting would seem to have the same convergence to a two party system as FPTP where you have a chamber composed of multiple members, a mayor or similar role would be different. Multi-member representation would be ideal, but I can see the appeal of having a single person represent you.

jabl 5y ago
> Can you point to a real example where you can't distinguish between candidates With hundreds/thousands(?) of ranked voting elections worldwide, with certainly millions of voters, you're statistically guaranteed to have that situation for some voter somewhere. > Approval voting would seem to have the same convergence to a two party system as FPTP where you have a chamber composed of multiple members, a mayor or similar role would be different. Not sure the voting method used for single-winner elections matter that much for the dynamics leading to a duopoly or not. I think convergence to a duopoly is most strongly influenced by winner-take-all districts. The solution to that, as you pointed out, is multi-member districts with proportional representation. > but I can see the appeal of having a single person represent you ... assuming that 'your' party wins. Otherwise you have no representation at all and you feel disfranchised? I'd say the major appeal of single-winner districts is that the districts are much smaller, and you have a better chance to get to know your own representative (again, assuming it's somebody you find broadly acceptable).
- aragilar 5y ago
  > With hundreds/thousands(?) of ranked voting elections worldwide, with certainly millions of voters, you're statistically guaranteed to have that situation for some voter somewhere. One could argue the same thing applies when a voter chooses whether or not to approve a vote for a candidate they don't like in order to prevent an even more disliked candidate from winning. To me, it seems a false choice either way (to order or to approve). Possibly this a framing problem, given I've grown up in a system where IRV/STV is used, I naturally compare candidates, maybe someone growing up in a Approval voting system would naturally think about whether they like a specific candidate or not. > ... assuming that 'your' party wins. Otherwise you have no representation at all and you feel disfranchised? Having two houses where one uses STV and the other has single member districts does have its advantages as a "why not both" option.

jlokier 5y ago

I personally favour ranked voting with the feature that you can put multiple candidates at the same rank, and you can leave off candidates you really dislike. (Putting candidates in the "leave off" set is the same as ranking them below a "none of the below" pseudo-candidate because it means they won't inherit unused vote shares during counting). That solves your distinguish and C-dislike objections, and feels more honest when voting. It also lets you feel fair when there are two candidates you both like equally and don't want to add to signal-favouritism when the results come out. At a local club (where people are friendly so it's not meant to be competitive like in government), having to rank multiple loved and personally known candidates for a committee occupied by multiple people, when you like several equally, doesn't change the outcome at all, assuming they win, but it tends to greatly amplify the tiny, irrelevant biases fron the most petty things people have to draw on into very large differences in reported scores. This makes it looks like one person is very much preferred to be in charge of the rest of the committee or team, much more than is actually true, ie. it doesn't actually represent the voters intentions. The election outcome is representative, but the outsized score differences are interpreted wrongly. So this seeds ungrounded internal divisions after the vote, peculiar strategic voting at subsequent elections, and peculiar behaviours to chase after those "tiny, irrelevant biases". This isn't good for anyone involved. So it's good to allow voters to rank candidates equally for this reason too, if the voter genuinely feels that way.

zajio1am 5y ago

> OTOH, with ordinal methods you are forced to distinguish between candidates that you think are equally good. That is not true, ordinal systems (including IRV) may allow weak ordering.