The people all arguing how approval is better than ranked choice remind me a lot of people here arguing that functional programming is better than imperative, and being constantly flabbergasted that their preference isn't reflected in real world implementations.
Approval voting has advantages when you get into the weeds, but ranked choice has one big advantage right out the gate: people like it more. If you present the two systems to regular voters with simple one paragraph descriptions, people will choose ranked choice nearly every time, because they want to be able to rank candidates if they're voting for more than one person. Nobody wants to have to say that the guy there willing to hold their nose for is just as worthy of their vote as the guy they really want in office.
Ranked choice has a lot of advantages over the alternatives that are often underestimated by people too worried about theory.
The big advantage of ranked choice is that it is simple to understand and implement.
In elections, it is essential that the average person understand exactly how the system works. An election result that people don't understand/trust is worse than useless.
No it isn't, technically. I can guarantee you that most people in Norway can't explain the Sainte-Laguë divisors method which delivers proportional representation.
What is true, is that simplicity is a propaganda advantage for actually getting it passed, since "It's too complex so we can't trust it" is a good attack line even if it isn't true.
I'm not sure that advantage is worth it, given that 1. it's a terribly long shot anyway, and 2. It's the same argument which will be used to defend the status quo.
> I can guarantee you that most people in Norway can't explain the Sainte-Laguë divisors method which delivers proportional representation.
I expect most could if they spent 5 minutes on Wikipedia researching it. They don't, because they don't want to; most probably regard it as an unimportant implementation detail.
You don't need to know the method, though. You just need to know that in Norway getting above or under the 4% line drastically affects how many representatives a party will get, and vote according to that.
Regular voters will basically always prefer ranked choice over approval imo, because they really want to be able to rank their preferences. A leftist who wants a green party candidate but will hold their nose for a Democrat doesn't want them to be "equal" in how they vote.
Approval voting makes much more sense to me. Consider ranked-choice if it were applied to the presidency: Given that some states are slow to report their results, there are scenarios where you can't fall back to people's second choices until all the votes are in and finalized to determine who lost the first round.
With approval voting, you can declare a victor without running a state machine.
Ironically, although no individual would want it, it is probably a good thing that it is hard to tell what a ballot's effect will be.
If the situation is close the best outcome may be a somewhat random outcome. Having 1,000 extra votes determine the outcome is just as arbitrary as anything else.
> I think ranked choice is in that sweet spot where it is reasonably simple to use and simple to count.
"Ranked choice" is a range of methods and usually people mean instant-runoff. Recognizing that no system is perfect, I nonetheless have significant worries about instant runoff.
Imagine we have a society that's predominately split into two religions, where members of each religion would love a theocracy of their flavor, would settle for secular tolerance, and is vigorously (sometimes violently!) opposed to theocracy of another flavor.
If we have three candidates, one representing each of these positions, and everyone votes their true preferences then we see maybe 40% theocracy A, 40% theocracy B, 20% secular tolerance. The very first thing instant-runoff does is throw out the compromise, and we chose violence and strife.
If we generally believe that "a good compromise is when both parties are dissatisfied" then IRV will never choose a good compromise. I worry that that's a bad way of choosing what we do as a society.
I would really like to see open primaries with approval voting and the top two going to the general election. I think that is much simpler than RCV and I'm not convinced RCV would ever provide a better outcome.
It seems strange to give people worse election outcomes that they have to live with for 4 years at a time just to make their 5 minutes in the voting booth feel a little more pleasant.
Approval voting may not seem "fun", But thousands of voters have used it in Fargo and St Louis without any issue. All the evidence says it gets better results, is more resistant to strategic voting, and has a better chance of replacing the status quo.
Approval is also simpler to count. Ranked choice ballots can't even be properly distributed among congressional districts, as a result it takes longer to get results.
The only advantages ranked choice have over approval are entirely theoretical. Another advantage approval has is that it trends towards moderation.
Approval isn't perfect though (personally I prefer STAR, but it requires ballots to change).
> The big advantage of ranked choice is that it is simple to understand and implement
No, its not. That would be a big advantage of the method equivalent to it but without loser elimination and instead counting each rank down on all ballots simultaneously, which is both easier to understand and much easier to implement.
Proportional representation, the federal government enacts laws and levies taxes on everyone and isn't just some far away entity you can ignore. The multi level democracy USA has now where each party sends candidates based on local votes makes sense when the federated entity doesn't have direct power over people, but since USA has such a strong federal government that taxes people directly it needs to better represent its people than now.
Alternatively we can scale back the power of the federal government massively and have it be funded by the states rather than via direct taxes, similar to how EU works. But right now it is the worst of both worlds, taxes and control like a local government but poor representation like a federated one.
Sums up most of the points about it.
I want to highlight the "leads to two party dominance in countries that use it" (australia for example) point. This is exactly what we want to avoid.
I prefer approval or score voting for single member elections as it does away with all these bad voting incentives.
But regardless, I think single member elections are stupid and proportional systems are far better and more democratic. The idea that the thing most representative of you is the piece of land you live on is very outdated. Zweitmandat is the most advanced implementation of proportional representation imo.
I absolutely agree about the Zweitmandat system. For those unfamiliar:
> The Zweitmandat (English: second mandate) is a feature in the variation of mixed-member proportional representation (MMP) used to elect the Landtag of Baden-Württemberg. Unlike most variations of MMP, such as the German federal electoral system, Baden-Württemberg's system does not use party lists. Instead, proportional seats are filled by losing candidates who won the highest proportion of votes.
The main distinction is that Ranked Choice can elect someone which is not the best compromise of everyone's choices, while Schulze and Ranked pairs will always elect the best compromise of everyone's choices. If there is a candidate who is preferred over the other candidates, when compared in turn with each of the others, they guarantee that candidate will win, while Ranked Choice does not.
The downside to Schulze or Ranked Pairs is that saying that someone is your second choice could hurt the chances of your first choice to be elected, because by putting forward a compromising candidate, and since Schulze or Ranked Pairs optimize for that, you lean the tally towards your potential compromises (which are your second, third, etc. choices) in the case where the election are a "close call".
If I understand correctly.
I personally would prefer a system that optimizes for the compromise, as I think it's more important to get the candidate that the least people dislike, than it is to get the candidate that the most people adore.
So if I understand correctly, it means something like:
If you have A, B and C.
100% think B is second best.
40% think A is best.
19% think B is best.
41% think C is best.
With Schulze or Ranked Pairs B will win, but with Ranked Choice C will win.
That's because C is most people's first choice, so they win. But if you asked people to pick between C or B, B would technically win, because 19% would pick B (as it is their best), and 40% that think A is best but B is second best would also pick B, thus B vs C would get 59% votes for B and only 41% votes for C, that means that in a vote for B against C, B would win, but with Ranked Choice B loses, and C is elected.
The fallacy in the argument is that, first choice only, NO candidate has even a slim majority, let alone a strong one.
If you decay the hypothetical 3 party set of options by eliminating the least popular candidate then you're back to first past the post which is the current status quo in the US.
Schulze and similar methods use the ranked lists to evaluate pairs of candidates and eliminate the candidates that are universally the worst first / retain candidates that rank well in isolation. This is more likely to result in a compromise that works best for the most people.
> Ranked choice seems to be the best one I've read about.
IRV (sometimes called “ranked choice”, but its among the worst of the ranked choice single-member methods) can be improved by getting rid of the loser elimination step, and counting down on all ballots (instead of those with an eliminated loser) until some candidate had a total that crosses the winning threshold (this is a version of Bucklin, but historical Bucklin implementations have often done dumb things like limiting the number of preference ranks to a much smaller number than candidates (often 2-3) rather than using fully-ranked (forced preference or unforced preference) ballots.
But, other than things like unique executive offices, single-member elections should be avoided. Legislative elections with small multimember districts (say 5 members) with a system like STV (or a Bucklinish cousin, again without loser elimination) gets decent proportionality, the candidate accountability of single-member districts (but more, because parties are likely to run more candidates than they'll win seats, so there os general election accountability even within preferred parties) and avoids high-stakes districting (eliminating gerrymandering opportunities.)
Even executive offices with a designated successor (e.g., governor and lt. governor) can be made multiwinner (sequential rather than proportional) using a ranked ballots method with a normal single-winner majority threshold: once the first winner is selected, eliminate that candidate and recount the ballots (for some methods, this can simply be continue the count till the next winner crosses the threshold with the same effect) to select the winner of the successor office. This improves candidate accountability when one party is clearly preferred, because a disfavored incumbent can be demoted without abandoning the preferred party.
The Condorcet Method has a huge disadvantage that it is hard to understand.
Being simple to understand is crucial for voting systems and I would argue is the most important factor since an election without public buy-in is worse than worthless.
Understanding is one thing, counting is another. Plurality voting is great because it’s easy to count by hand and audit with volunteer observers. I’ve personally observed the counting of ballots as a volunteer and it gave me a lot of confidence that the election had been conducted fairly according to the law.
While I would really love to see what outcomes could be produced by a Condorcet method I can’t imagine actually implementing one without relying on computerized counting and software. This makes it impossible for volunteers to audit due to the need to audit the hardware and software of the computers doing the counting. If only security experts are capable of auditing such a system then our whole democracy is at their mercy. That’s essentially a technocracy.
Sweden has an extremely complex system for tallying votes and very very few people can explain the exact algorithm that turns a collection of ballots into a list of names of people in Parliament. Even people that follow politics closely only have a vague idea of how it works. Yet most people feel the system is fair and reasonable and there is no real push to change it or make it simpler to understand.
If your business relies on some big pile of spaghetti code, you might be very reluctant to change it, but you also would be unlikely to recommend that design to someone who was starting from scratch (unless they were a competitor).
I'm not recommending the Swedish system per se. Just pointing out that as long as people feel they understand how to vote to nudge the result in their desired direction, and that the outcome (ie. who ends up in parliament) feels reasonable and representative then people probably don't care too much about the details of the voting system
The Condorcet property (not method) is easy enough to understand.
"Imagine this candidate ran in a two person race against each of the other candidates in turn. If he'd win ALL those races individually, he should win when running against all of them at the same time too."
Now, explaining various methods which guarantee this property is not as easy - nor is it easy to understand why you might prefer one such method to another.
But this isn't actually a big deal. In the many, many countries that use proportional representation, it's the outcome ("parties share of seats in parliament should be roughly the same as parties share of votes in the election") which is agreed upon. The actual mathematics to achieve it are somewhat counterintuitive and not many people are aware of them, but that doesn't matter. We can all see it works, and so we would in an election with, say, Ranked Pairs.
Well, I'm not really seeing people complain about the math itself, but instead more about fraud in the votes themselves, or machines/people cheating when tallying.
Just to be clear, the difference would be something like having to say B won because he was the most common second choice and beat every other candidates in a one on one. Versus saying B won because it had the most first choices, yet did not have a majority of first choices.
I think it's easy enough to understand that while someone had more first choices, they didn't have a majority first choices and since someone else had more second choices they took the win.
After watching people freak out when NYC implemented Ranked Choice, I feel like most Americans won’t tolerate anything even slightly more complicated. It’s our only shot.
I’m from NZ which changed from FPP to MMP in 1993.
Two big problems:
1. Half of the politicians that make it into parliament are from the party list (which is chosen by the party) and are not voted for individually. I believe much of the power of democracy is the ability to vote people out, and the party list mostly prevents that.
2. There is a lot of talk of strategic voting. I guess all systems can be gamed, but MMP seems to encourage it.
https://elections.nz/democracy-in-nz/what-is-new-zealands-sy... “Every candidate who wins an electorate gets a seat in Parliament. They are called electorate MPs. The remaining seats are filled from party lists. Every party has a party list, which is a list of candidates ranked in the order the party wants those candidates to be elected to Parliament. Candidates elected from a party list are called list MPs.”
The NZ voting system looks very much like the German system to me. And while that system certainly has its flaws (e.g. when a party has more directly voted members than seats from the overall proportion of votes), it does respect a higher number of cast votes than FPTP. Any vote in FPTP that wasn't cast for the winning local candidate is lost. In MMP, every vote for a party influences the result in every case.
As a german I can tell you that not being able to get rid of certain people because they will always re-enter parliament via party list is annoying as hell.
The intended way to do that is to join a party and try to work towards that from within. Parties are required to have democratic internal processes including voting on the candidate lists. But let's face it - only people with political ambitions of their own will even think about that.
I'll take second worst over worst. The current system has been in place for hundreds of years, I'm not going to hold my breath for the "perfect" voting system.
> It is the second WORST voting system after first past the post
I dunno, I think random ballot is worse.
Its arguably the worst system that uses ranked choice ballots and ever considers more than the first position, though, which is why in should be called its proper name, “instant runoff voting”, and not “ranked choice”.
It is the second WORST voting system after first past the post, which is what we have now.