5 ms·

Condorcet methods performed well in this analysis, even under the various strategy models. Although it says that STAR performed slightly better under strategic

Condorcet methods performed well in this analysis, even under the various strategy models. Although it says that STAR performed slightly better under strategic voting, its scores are still very high. The main reason given in the text for preferring to highlight STAR was a (common) belief that the Condorcet methods are too difficult to explain and that explainability is important. Personally I think the Condorcet methods are not that difficult to explain. The various ways of resolving not perfectly clear situations are a little involved, but everything else makes perfect sense. I built a visualisation a few years ago that created a directed graph of results, and then analysed the strongly connected components. Having said that, I hadn't come across STAR before and I think it makes a fair amount of sense. Any of these high quality methods are enormously better than those used in lots of countries.

godelski 5y ago
> Personally I think the Condorcet methods are not that difficult to explain. To the voter, no. It is just a ranked choice/ordinal system (which is why people get mad at IRV being called RCV). Just rank your candidates in order of preference. People naturally understand this, yes. BUT that's not what people typically mean when they says the system is hard to understand. What matters is transparency and how easy it is to tabulate the votes. Actually this is where all ordinal systems fall in comparison to cardinal systems. In cardinal systems just just sum some columns (in STAR you have to do this twice). But in ordinal systems you have to do many rounds of counting and eliminations. Look at the current NYC mayoral race. We have only one round counted and they have like 10 more days before round 2 (which is funny because IRV results typically follow the first round results, we expect Adams to win). So look into tabulating Schulze or Ranked Pairs. Condorcet methods require you to run n^2 election because you have to run every pair. Now consider contested elections like what is going on in Arizona. The more complex of a counting system is the more either end can be abused (the original vote or the verification). If it is sufficiently complex a group will easily be able to justify foul play regardless of actual proof because the evidence against the claim is nuanced. So yes, it isn't hard to explain to voters, but it is difficult to explain how you count the votes. But VSE is only one metric. Like you pointed out max(STAR) is only slightly less than max(condorcet) but min(STAR) has a decent gap over min(condorcet), which also matters. The smaller range on STAR suggests it is more resilient to strategic voting (or voter manipulation). But cardinal systems are not only easy to explain to voters but it is also easy to explain how votes are counted. I should also note that Approval/Score/STAR are monotonic and will not spoil elections (IRV increases spoilage where cardinals decrease). I'll also suggest that if you're a data person you should be in favor of Score and STAR because they are more expressive than any ordinal system because the distance between your preferences isn't constant and thus we can get more fine grained data on voter preference. If we only look at the VSE metric, condorcet wins, but what makes voting complicated is that this isn't the only metric.
- kybernetikos 5y ago
  I do like STAR after having read this article, and I agree that it's easier to explain, but I'll defend my saying that condorcet isn't that hard: Those virtual elections are not difficult to run (electronically) or explain though: "we use the ballots to work out for every person how they would do in a one on one election against each other person". And the process for running the elections is straightforward too - count the number of ballots where one person appears higher in the list than the other person. There's no room for argument here, unless you're arguing about the ballots, which is something you can do with any system. All of that stuff seems pretty straightforward to me. Now, it is the case that the various ways of resolving situations where you have cycles is somewhat complex. I think that's just because those situations are genuinely complex, but there are things that can be done here to choose a relatively explainable one and work out good ways to visualize and explain it. Finally, while cardinal systems can express the distance between preferences, there are ways an ordinal system can express things hard for a cardinal system. In an ordinal system you can allow a ballot not to include someone and have that mean 'no preference' between that person and all other candidates (perhaps because of lack of knowledge), while in an cardinal system an unscored person would have to be counted as having an actual score. There's also the question whether preferences really are consistent enough for a cardinal system to accurately reflect them - you might like X three times more than Y, and two times more than Z, but does that necessarily mean that you like Z one and a half times more than Y? Personally I also think there's a big difference between preferences between two candidates you'd be happy to win and candidates you actively don't want to win, and I don't think that difference can be captured by just making a scalar larger - not that ordinal approaches capture this either.
  - godelski 5y ago
    For the most part, yeah, Condorcet methods aren't that hard. But also consider who you and I are. We're talking on a very niche form where there's an expectation that everyone here can program and do calculus. We are not the average people. But what I do want to stress is that cardinal systems are _easier_ than condorcet. I actually use approval frequently (it is also extremely natural). One case is how to solve where to eat as a group. It is infinitely scalable and you can just keep adding places as you think of them. If you reach a unanimous decision you stop, if you exhaust the list you take the maximal (baring strong preferences, which makes it kinda score but this works out naturally). I think from this example it should be pretty clear that it is much harder to do this sort of voting using any ranking system (nor is it scalable and is very computationally expensive to add new items). It is often missed, especially by nerds (I fall in this camp too, I used to prefer Condorcet over cardinal), how important the transparency aspect of the voting system is. Arizona nailed the lid shut for me. If we're looking at it we're really not losing much VSE by using STAR over RP, but we do also gain a lot on every other metric (including spoilage, ease, transparency, and resistance to strategy). VSE isn't the only metric that matters and we shouldn't make large sacrifices in other metrics to gain a minute optimization in it. > while in an cardinal system an unscored person would have to be counted as having an actual score. This isn't necessarily true. You can modify cardinal ballots just like you do ordinal (the exact example you gave is not the theoretically perfect ordinal so I'm not sure why you compare to theoretically perfect cardinal). In practice people don't rate every candidate. In ranking you could effectively say that unranked candidates have a minimal score (and you are able to reduce the number of rounds through this, but comes with added complexity). As to making the scale larger, yes this is correct that an infinitely large scale better encodes information. But there is a balance between that and simplicity. 10 is extremely effective. 5 is more than enough. But even 2 (approve/disapprove) seems to work out pretty well (aka "good enough").
ClayShentrup 5y ago
Condorcet is just needlessly complex and has worse behavior under high levels of strategy. Score voting (STAR, approval, etc.) is roughly as good and maybe even better, and dead simple, and very transparent. The lesson of voting theory is to avoid ranked voting methods.

godelski 5y ago

> Personally I think the Condorcet methods are not that difficult to explain. To the voter, no. It is just a ranked choice/ordinal system (which is why people get mad at IRV being called RCV). Just rank your candidates in order of preference. People naturally understand this, yes. BUT that's not what people typically mean when they says the system is hard to understand. What matters is transparency and how easy it is to tabulate the votes. Actually this is where all ordinal systems fall in comparison to cardinal systems. In cardinal systems just just sum some columns (in STAR you have to do this twice). But in ordinal systems you have to do many rounds of counting and eliminations. Look at the current NYC mayoral race. We have only one round counted and they have like 10 more days before round 2 (which is funny because IRV results typically follow the first round results, we expect Adams to win). So look into tabulating Schulze or Ranked Pairs. Condorcet methods require you to run n^2 election because you have to run every pair. Now consider contested elections like what is going on in Arizona. The more complex of a counting system is the more either end can be abused (the original vote or the verification). If it is sufficiently complex a group will easily be able to justify foul play regardless of actual proof because the evidence against the claim is nuanced. So yes, it isn't hard to explain to voters, but it is difficult to explain how you count the votes. But VSE is only one metric. Like you pointed out max(STAR) is only slightly less than max(condorcet) but min(STAR) has a decent gap over min(condorcet), which also matters. The smaller range on STAR suggests it is more resilient to strategic voting (or voter manipulation). But cardinal systems are not only easy to explain to voters but it is also easy to explain how votes are counted. I should also note that Approval/Score/STAR are monotonic and will not spoil elections (IRV increases spoilage where cardinals decrease). I'll also suggest that if you're a data person you should be in favor of Score and STAR because they are more expressive than any ordinal system because the distance between your preferences isn't constant and thus we can get more fine grained data on voter preference. If we only look at the VSE metric, condorcet wins, but what makes voting complicated is that this isn't the only metric.

kybernetikos 5y ago
I do like STAR after having read this article, and I agree that it's easier to explain, but I'll defend my saying that condorcet isn't that hard: Those virtual elections are not difficult to run (electronically) or explain though: "we use the ballots to work out for every person how they would do in a one on one election against each other person". And the process for running the elections is straightforward too - count the number of ballots where one person appears higher in the list than the other person. There's no room for argument here, unless you're arguing about the ballots, which is something you can do with any system. All of that stuff seems pretty straightforward to me. Now, it is the case that the various ways of resolving situations where you have cycles is somewhat complex. I think that's just because those situations are genuinely complex, but there are things that can be done here to choose a relatively explainable one and work out good ways to visualize and explain it. Finally, while cardinal systems can express the distance between preferences, there are ways an ordinal system can express things hard for a cardinal system. In an ordinal system you can allow a ballot not to include someone and have that mean 'no preference' between that person and all other candidates (perhaps because of lack of knowledge), while in an cardinal system an unscored person would have to be counted as having an actual score. There's also the question whether preferences really are consistent enough for a cardinal system to accurately reflect them - you might like X three times more than Y, and two times more than Z, but does that necessarily mean that you like Z one and a half times more than Y? Personally I also think there's a big difference between preferences between two candidates you'd be happy to win and candidates you actively don't want to win, and I don't think that difference can be captured by just making a scalar larger - not that ordinal approaches capture this either.
- godelski 5y ago
  For the most part, yeah, Condorcet methods aren't that hard. But also consider who you and I are. We're talking on a very niche form where there's an expectation that everyone here can program and do calculus. We are not the average people. But what I do want to stress is that cardinal systems are _easier_ than condorcet. I actually use approval frequently (it is also extremely natural). One case is how to solve where to eat as a group. It is infinitely scalable and you can just keep adding places as you think of them. If you reach a unanimous decision you stop, if you exhaust the list you take the maximal (baring strong preferences, which makes it kinda score but this works out naturally). I think from this example it should be pretty clear that it is much harder to do this sort of voting using any ranking system (nor is it scalable and is very computationally expensive to add new items). It is often missed, especially by nerds (I fall in this camp too, I used to prefer Condorcet over cardinal), how important the transparency aspect of the voting system is. Arizona nailed the lid shut for me. If we're looking at it we're really not losing much VSE by using STAR over RP, but we do also gain a lot on every other metric (including spoilage, ease, transparency, and resistance to strategy). VSE isn't the only metric that matters and we shouldn't make large sacrifices in other metrics to gain a minute optimization in it. > while in an cardinal system an unscored person would have to be counted as having an actual score. This isn't necessarily true. You can modify cardinal ballots just like you do ordinal (the exact example you gave is not the theoretically perfect ordinal so I'm not sure why you compare to theoretically perfect cardinal). In practice people don't rate every candidate. In ranking you could effectively say that unranked candidates have a minimal score (and you are able to reduce the number of rounds through this, but comes with added complexity). As to making the scale larger, yes this is correct that an infinitely large scale better encodes information. But there is a balance between that and simplicity. 10 is extremely effective. 5 is more than enough. But even 2 (approve/disapprove) seems to work out pretty well (aka "good enough").

ClayShentrup 5y ago

Condorcet is just needlessly complex and has worse behavior under high levels of strategy. Score voting (STAR, approval, etc.) is roughly as good and maybe even better, and dead simple, and very transparent. The lesson of voting theory is to avoid ranked voting methods.