For presidential elections, and other single-winner elections, it’s a bit of a toss-up between 3-2-1 and STAR voting. I am partial to STAR, personally, because (as you can see in the graph) it outperforms 3-2-1 by quite a bit in the 100% honesty case, while only slightly underperforming 3-2-1 in the worst voter model. Quinn argues that 3-2-1 is better because its efficiency depends less on the voter model; it squeezes the different scenarios together more, making for more reliable quality.
It’s also important to consider which is less confusing. Again, I personally find STAR pretty intuitive, but Quinn tends to think 3-2-1 is simpler, so “more research is needed” I guess.
Here’s how STAR works: everyone gives a ‘star rating’ to each candidate, from 0 to 5 or 0 to 10 (whatever granularity you like), much like score voting (aka range voting). (If you leave someone blank they get 0.) The top two candidates are chosen by total score. Then, these two candidates are compared in a virtual runoff election (a lot like instant runoff): each candidate gets a “vote” for each time they are rated higher than the other candidate on a ballot. (This ensures that everyone gets a strong voice in the choice between the final two candidates, so long as they rated them any differently on their ballot.)
Here’s how 3-2-1 works: a voter rates each candidate “approve/neutral/disapprove”. Take the three candidates who each have highest approval. Then eliminate the one candidate out of these with highest disapproval. Then compare the remaining two candidates in a virtual runoff, like in STAR, where people are assumed to vote for candidates who they rated higher.
For multi-winner elections, IE legislative elections, it would be great to use methods which aim for proportional representation; but, I don’t really have anything to say about what differentiates those methods from each other. I have the impression that they’re all pretty good. If you made me choose right now, I’d go with Allocated Score (AKA proportional STAR voting). If STAR voting is already in use for single-winner elections, this has the advantage of presenting voters with a familiar ballot. (That’s not my reason for picking it—my reason is (a) it’s the simplest to understand out of the options I just now looked over, (b) it avoids center-squeeze-like problems associated with the more popular single-transferable-vote.)
We have had a multi-winner election system in Tasmania since 1896. It is called the Hare-Clark system and has its drawbacks. 1. Candidates compete mainly with members of their own party. 2. Urban areas are favoured over rural areas in the same electorate because that is where the votes are. 3. There are no bi-elections. 4. The parties are usually very evenly balanced in numbers leading to hung parliaments. 5. The balance of power is often held by extremist groups or eccentric individuals. 6. Electorates can be very large geographically making campaigning difficult and with little commonality of interest between remote areas.
Yeah, I haven’t looked at any statistics, but anecdotally it seems like I can’t support the claim that better voting methods are that much better in practical terms. Many places have systems which seem better than that of the USA, without markedly better political results.
However, in my country, where we use FPTP, we have frequently got extremist parties in power. If seats had been given proportionally, they might not have got power, or would have to operate in Big Tent coalitions.
In the last general election, 486(!) parties fielded candidates. Of course, most of them would not reach the threshold for representation in PR. But some of them will. That would almost always cause hung parliaments.
Upside is, minorities like Muslims will get better representation. They are heavily underrepresented in legislatures due to their dispersal.
IRV has the problem that (a) many votes would get exhausted. However, if to control for that you impose a minimum number of choices, you would get a higher number of invalid votes. Or even donkey voting—just filling out at random.
Score voting, in my opinion, better captures the reality of human preference.
For presidential elections, and other single-winner elections, it’s a bit of a toss-up between 3-2-1 and STAR voting. I am partial to STAR, personally, because (as you can see in the graph) it outperforms 3-2-1 by quite a bit in the 100% honesty case, while only slightly underperforming 3-2-1 in the worst voter model. Quinn argues that 3-2-1 is better because its efficiency depends less on the voter model; it squeezes the different scenarios together more, making for more reliable quality.
It’s also important to consider which is less confusing. Again, I personally find STAR pretty intuitive, but Quinn tends to think 3-2-1 is simpler, so “more research is needed” I guess.
Here’s how STAR works: everyone gives a ‘star rating’ to each candidate, from 0 to 5 or 0 to 10 (whatever granularity you like), much like score voting (aka range voting). (If you leave someone blank they get 0.) The top two candidates are chosen by total score. Then, these two candidates are compared in a virtual runoff election (a lot like instant runoff): each candidate gets a “vote” for each time they are rated higher than the other candidate on a ballot. (This ensures that everyone gets a strong voice in the choice between the final two candidates, so long as they rated them any differently on their ballot.)
Here’s how 3-2-1 works: a voter rates each candidate “approve/neutral/disapprove”. Take the three candidates who each have highest approval. Then eliminate the one candidate out of these with highest disapproval. Then compare the remaining two candidates in a virtual runoff, like in STAR, where people are assumed to vote for candidates who they rated higher.
For multi-winner elections, IE legislative elections, it would be great to use methods which aim for proportional representation; but, I don’t really have anything to say about what differentiates those methods from each other. I have the impression that they’re all pretty good. If you made me choose right now, I’d go with Allocated Score (AKA proportional STAR voting). If STAR voting is already in use for single-winner elections, this has the advantage of presenting voters with a familiar ballot. (That’s not my reason for picking it—my reason is (a) it’s the simplest to understand out of the options I just now looked over, (b) it avoids center-squeeze-like problems associated with the more popular single-transferable-vote.)
We have had a multi-winner election system in Tasmania since 1896. It is called the Hare-Clark system and has its drawbacks. 1. Candidates compete mainly with members of their own party. 2. Urban areas are favoured over rural areas in the same electorate because that is where the votes are. 3. There are no bi-elections. 4. The parties are usually very evenly balanced in numbers leading to hung parliaments. 5. The balance of power is often held by extremist groups or eccentric individuals. 6. Electorates can be very large geographically making campaigning difficult and with little commonality of interest between remote areas.
Yeah, I haven’t looked at any statistics, but anecdotally it seems like I can’t support the claim that better voting methods are that much better in practical terms. Many places have systems which seem better than that of the USA, without markedly better political results.
However, in my country, where we use FPTP, we have frequently got extremist parties in power. If seats had been given proportionally, they might not have got power, or would have to operate in Big Tent coalitions. In the last general election, 486(!) parties fielded candidates. Of course, most of them would not reach the threshold for representation in PR. But some of them will. That would almost always cause hung parliaments. Upside is, minorities like Muslims will get better representation. They are heavily underrepresented in legislatures due to their dispersal.
IRV has the problem that (a) many votes would get exhausted. However, if to control for that you impose a minimum number of choices, you would get a higher number of invalid votes. Or even donkey voting—just filling out at random. Score voting, in my opinion, better captures the reality of human preference.