My sense is that city council is the type of place where you least want proportional representation.
Instead, you want a councillor who knows your area of the city and is responsive to your concerns about noise, construction, parking, traffic, etc. You need someone to complain to when things aren’t going well.
In effect, PR results in no one being directly accountable. Which is probably fine for higher levels of government, but at lower levels you want to know who’s on point.
This is a good thing, actually. Current electeds weight the preferences of local constituents too much and should instead be thinking more of what’s good for the city overall. This is why we have too much zoning, eg.
Plenty of other local problems that are just bad, rather than inconveniences from development and growth. Things like crime hot spots, nuisance businesses, neglected parks, etc.
It's not clear to me that one rep who cares a lot about an issue is more effective at getting it fixed than if all the reps could benefit somewhat by fixing it.
That's the Special Interest Effect in action - when there's an issue that would greatly benefit a small number of people but slightly harm a large number of people, it often happens, even if the net effect is negative. Similarly, policies with wide, small benefits but large, focused harms are unlikely to happen, even if the net effect is positive.
This was exactly my experience when i lived in Detroit. No one represented a particular part of the city, and as you say, no one was responsible --- or responsive.
Note that Matt is endorsing STV, not partisan PR. Under STV, people who particularly care that there is local representation can put all their local candidates first, above all non-local candidates regardless of party. But someone who thinks it is more important to have ethnic representation, or renter representation, or partisan representation, can put the relevant people first. You don’t get the party list effect.
But Matt’s proposed solution is to have “13 at-large members”. If I live in Ward 1, even if I vote for a local candidate, they still aren’t directly responsible for or accountable to Ward 1.
The way Germany and New Zealand do MMP addresses this, but that seems far too complicated for a city council.
That’s interesting! I hadn’t heard of that method. What advantage does it give to use an approval method rather than a ranking method? It seems like that creates a lot more incentive to strategize with one’s vote.
Approval is much more similar to the systems we already have (like where you're voting for multiple at-large seats and can vote for up to N), and the "strategy" in it is very easy to think about. You vote for the person you would've voted for in a FPTP race (your favorite amont the "frontrunners"), plus everyone you like more.
Range systems like STAR (or the proportional variant of that, Reweighted Range) do allow more expressive power, but it turns out that to a large extent, optimal strategy (from the individual perspective) in a Range system is basically to reduce it to approval -- figure out your favorite frontrunner, top-rate them, and then also top-rate anyone you like better. If voters uniformly sacrifice strategy for honest expression of ratings, you should get more socially optimal outcomes, but if only one faction does that, another faction can take advantage to move outcomes in their direction somewhat.
Plus Range is just more expensive to administer and more complicated to explain. Approval is incredibly simple.
“Instead, you want a councillor who knows your area of the city and is responsive to your concerns about noise, construction, parking, traffic, etc. You need someone to complain to when things aren’t going well.”
So you want a council member who will help you be a NIMBY? I am surprised this comment is so popular here.
All of the particular examples brought up by this comment are NIMBY selling points, but there's plenty of other local disamenities that are just bad for everyone. Things like crime hot spots, potholes, nuisance businesses, broken street lights, etc.
Actually I think reducing corruption in politics is a big cause of this. Look up any major iconic building built 1900-1950, there are major corruption issues around all of them. But at least we built them. Bring back corruption!
RCV is good. There’s something deeply demotivating about being forced to vote for the lesser of two evils year after year. On the other hand, eliminating the spoiler effect inherent to FPTP means allowing more parties to succeed and hence better representation.
The spoiler problem is arguably even worse with RCV because voters don’t know about it; they falsely believe that it’s always advantageous to vote and that it’s safe to rank candidates in the order you prefer them.
In Alaska’s RCV election in 2022, if a bunch of Sarah Palin’s supporters had simply stayed home and not voted, a Republican would have won, but instead they voted and so a Democrat won. What’s more, a majority of voters ranked the moderate Republican *higher* than the Democrat yet the Democrat won. RCV does not work the way that people think it does!
Under RCV, the most strategic thing to do is often to vote for the “lesser of two evils” the exact same way you would with FPTP. If instead you rank your real favorite candidate as #1, you may make it more likely that your *least* favorite candidate wins the election. Yikes! https://en.wikipedia.org/wiki/Negative_responsiveness_paradox
(Thank you to the fellow SB commenter who pointed this out a few weeks ago!! I wasn’t fully aware of these issues until then.)
This is all true about what’s possible. But the spoiler effects with IRV are much harder to predict than with FPTP, so they at least are less likely to encourage people to strategically falsify the vote.
And being immune to spoiler effects is logically equivalent to Arrow’s “Independence of Irrelevant Alternatives” condition, which is the one most theorists hold responsible for his impossibility theorem.
"But the spoiler effects with IRV are much harder to predict than with FPTP, so they at least are less likely to encourage people to strategically falsify the vote."
Except it's not actually "falsifying" the vote with FPTP -- you know exactly who you are voting for. With RCV, you as a voter don't really know what happens after the first round because you have no insight into people's second choices.
Bingo! Both FPTP and RCV require you to vote strategically—but FPTP makes it transparent how to do so, whereas it can be much harder to understand the downstream impacts of various voting strategies under RCV.
Under FPTP, what often happens is that one or more lesser candidates will see the polling, realize they have no chance, and eventually drop out and endorse a different candidate. Then the choice at hand starts to become a bit more clear, and people start to coalesce around certain candidates in order to push toward their broader strategic goals.
Think back to when Pete Buttigieg and Amy Klobuchar were doing decently in the polls but probably weren’t going to win, so dropped out and endorsed Joe Biden, and their supporters largely followed suit.
The fact that center squeezes are hard to predict _is not a positive thing_! The FairVote folks say, about instant runoff, that center squeeze scenarios "only" happen in ~1 in 250 races.
There are tens of thousands of distinct elections in America every cycle! They're saying that there will be examples of the center squeeze, all over the country, _every time we run an election_.
Burlington, VT adopted IRV, _immediately_ had a center squeeze for its mayoral race, and the folks on the losing side (backers of both the candidate who got squeezed out, and the people whose second-choice preferences for that candidate _should_ have put them over the top) got together to repeal the whole thing and go back to FPTP.
Mary Peltola's center squeeze win over Nick Begich in Alaska has also led to a ballot initiative to trash their reform, which may well win tomorrow.
Whether strategic voting circumstances being hard to predict is good or bad depends on what you think the problem with them is.
One problem (which doesn’t sound like the one you care about) is that strategic voting circumstances encourage people to write down preferences other than their true ones. If they are hard to identify, then this doesn’t happen very often.
But the problem of a center squeeze is different. This is a problem for a particular normative theory of who should win. If the candidates really are aligned on a single ideological axis, then there’s a good case that the preference of the median voter should carry the day, no matter how few people share that preference. I take it that this is the motivation behind thinking a center squeeze is a bad outcome.
If it’s true that empirically, most elections involve candidates that are best understood as linearly ordered by ideology, then a method that avoids this problem is going to be more valuable.
The problem I see is that a good number of elections aren’t like that - they involve two ideological axes (say, a Clinton/Romney/Trump/Sanders election, which might be satisfied by the median on each axis separately, if that can be identified) or they involve more complex issue spaces, combined with administrative skill ideas.
I don’t think IRV necessarily does well with those things, but I don’t know what methods people are proposing to avoid center squeezes. (FPTP seems like it makes the problem of the center squeezes even worse.)
One good example of how RCV is not perfect is the Alaska Senate election in 2022. Ideally, it would be safe for Democrats to vote for the Democrat first, Murkowski second. But, doing that would be dangerous, as it would risk Murkowski getting eliminated first, which would cause the hard-core Trumper to win (e.g. the "center squeeze"). To allow Murkowski to prevail over her right-wing challenger, many Democrats had to be strategic, realize that the Democrat had no chance anyway, and vote Murkowski first to avoid the center squeeze problem.
To be clear, I do think this is still better than first-past-the-post, where the same need to be strategic exists, but is much worse, but it is an illustration that RCV is not perfect.
Please use the term IRV not RCV when referring to elections using that particular non-Condorcet counting method. To me RCV is a general term referring to all ballot systems, single-winner or proportional, which use ranked ballots. Wikipedia seems to be a bit confusing, as it distinguishes the term “Ranked Voting” (the general term) from RCV (which it sometimes conflates with IRV).
Philosophically I agree! Realistically, in the U.S. in the year 2024, RCV is synonymous with IRV; pushing back against that is probably an uphill battle.
In this comment thread, I used RCV because that’s what Matt and everyone else is using, so sticking to the same term helps with communication.
It appears there's an ongoing wikifight on wikipedia's IRV page regarding this terminology among other things. Going down that rabbit hole, I learned today that 1) IRV is generally called RCV in the US but something else in most other places it's used; 2) Google searches for RCV far outnumber those for IRV; 3) It appears to me that most academic political scientists distinguish the terms as I did; 4) People in general in the US seem to have given up the terminology battle as you suggest; 5) Academic research generally supports Matt's "meh" feelings for RCV/IRV.
The last point is one that I think is significant: Lots of my friends think that RCV/IRV is the “enlightened idea that educated people know is obviously the solution to fix our elections” but have no idea that lots of actual elections researchers who have done the math are not as enthusiastic.
I'd love to see more real world examples of this. The issue for me is that it is extremely easy to see how a spoiler in a FPTP system would work, and while I get the "center squeeze" concept, it's hard to get an intuitive feel for how often that is likely to come up in practice.
Basically, my position is still that Ranked Choice systems are still significantly more likely to reflect the will of the voters, even if there is potential for a spoiler effect. But it's not a very strongly held position.
I'm also with Matt in that it seems like RCV advocates are really passionate about something that seems likely to only help at the margins (though margins are still important!) . Still, even with your input it seems like the kind of marginal benefit with close to no cost and so I'm generally in favor, just not particularly worked up about it.
I think often times what people really want is a system that avoids Ralph Nader/Jill Stein like spoilers but without these weird effects and a clear solution is a French style (Or Georgia/Louisiana) Top two Run off. Get 51% the first time, Congratulations you win! No Majority? Single election FPTP between exactly two candidates, no write ins allowed.
St Louis has a great implementation of this in which the first round uses approval voting, where you can vote for as many candidates as you like to advance to the final round, then the top two vote-getters go to the runoff.
This fixes the vote-splitting problem of the top-two-runoff system used in various U.S. states, in which a party with too many candidates can find themselves locked out of the runoff if the opposition has its votes concentrated around just two candidates.
The issue is exactly what you noted: The spoiler effect exists with both systems, but with FPTP it’s transparent—and can lead lesser candidates to strategically drop out, or voters to strategically consolidate their vote behind a leading candidate—whereas RCV lulls people into a false sense of complacency, thinking there’s no spoiler effect when in fact there absolutely is.
It’s challenging because once people see big tables of numbers, their eyes start to glaze over and they say “I don’t know… can’t I just trust that things will work out magically?” but unfortunately we can’t.
It shouldn't be the case that a less desirable result occurs *because* you voted, which is the case he describes. We know their preferences because they were ranked!
Exactly! Under RCV, you can have a majority of voters rank Candidate 1 higher than Candidate 2 and yet Candidate 2 wins the election. This happened in Alaska in 2022 and also in the Burlington mayoral election in 2009: https://en.wikipedia.org/wiki/2009_Burlington_mayoral_election
I think we should have a strong predisposition towards what is intuitive. It’s intuitive that someone who voted Palin first and Other Republican second would NOT have preferred the Democrat win, yet that’s the result they contributed to. As others have said, this is all extremely complicated for laypeople and so it is likely to make it harder for them to align their preferences with their actions.
In a traditional primary system, the outcome of the Alaska House race would have been exactly the same. Palin would have beat Begich in the Republican primary, then lost to Peltola in the general, with enough Begich voters who couldn't stomach Palin crossing over, just like they chose Peltola as their 2nd choice in the RCV election.
The problem is not RCV, the problem is extremists candidates who appeal only to their hard-code partisans, but are too radioactive to the center to win in the general.
This is the third time Mr. "My t-shirt saying I'm not Hitler is raising a lot of questions already answered by the t-shirt" has been on the ballot. It's making me rethink what a normal election is.
How would you feel if you voted for (e.g.) the Green Party, then they made "evil" concessions in order to make it into government?
I think people don't realize there's an ineliminable tradeoff. Your system is either majoritarian (in which case the majority has to make internal compromises), or minoritarian (which is obviously terrible).
It's obvious once you say it, but you can't give everyone their favorite government.
The flip side is that in a two party FPTP system, the government in red states is run by a majority of the majority, the people who elect the crazies in the republican primaries, even if a majority of the electorate would elect someone else.
Strongly agree that FPTP + populist candidate selection is bad. But I blame the system of candidate selection!
It's probably too late to abolish primaries, but there are other things we can do to fix the problem. Redistrict to have big, diverse districts. Institute a threshold (primaries aren't binding if below n% of the voters in the state don't participate in the primary). Etc.
As I wrote downthread, what's maintaining the US party primary system now is that the state is paying for & running a hugely complex operation. You probably couldn't actually abolish primaries, but a state could just choose to, like...... not pay for them. "Hardworking taxpayers shouldn't bail out political parties", use that kind of rhetoric with voters.
Without the funding & logistics, state-level parties almost certainly couldn't run primaries anymore. It's the best of both worlds, effectively outlawing them without actually doing so
I wasn't arguing against Matt's proposal in the comment you replied to. I was responding to Stew's critique of FPTP.
He said the US's version of FPTP sucks. I agreed, but I said that's because of other features of our system that weaken party discipline - such as primaries. (I could have also mentioned money in politics.)
Repealing FPTP, to my mind, is like bloodletting. Yes, the patient is sick. But this is probably not going to help.
I think the evidence is pretty clear this is not correct. Primary participation at anything lower than the presidential race is really low. Unless you are saying that only the most committed partisans are actually the party, then its clear that the majority of the parties' voters do not vote in many primaries.
1) "Turnout of all eligible voters in 2022 primaries was 21.3%. That compares with 19.9% in 2018, 14.3% in 2014, and 18.3% in 2010."
2) Only 7 states broke 30%.
3) The bottom five states had 12% participation or less.
RCV is better than FPTP, but it's not really "good", by the criterion you specify -- the spoiler effect. All RCV schemes suffer from the spoiler effect except Coombs' Method, which is terrible in other ways. Most RCV schemes make working out spoilers difficult enough that voters may be unable to vote strategically, but they don't ensure voters that they can safely and confidently vote for their most-preferred candidate. They're also complicated for voters to understand, creating confusing ballots and hard-to-follow evaluation processes.
If you're unsure whether the spoiler effect in RCV systems is real, there are many real-world examples. In 2022, Sarah Palin spoiled the election for Nick Begich, giving the victory to Mary Peltola in Alaska's first RCV election. I actually think that was a good outcome, but it wasn't what the voters wanted -- which is causing Alaska voters to demand repeal of RCV to return to FPTP.
But there is a voting system that both completely eliminates the spoiler effect, and is very simple: Approval Voting. Rather than ranking choices, you just vote for all candidates you find acceptable. To determine the winner, officials just count the votes and the candidate with the most votes wins. Everyone can understand it, and it provably eliminates any possibility that voting for your most-preferred candidate could hurt their chances.
The problem with approval voting is that voting for anyone else in addition to you most-prederred candidate can hurt their chances, so it incentivises voters to just pretend it's FPTP, which defeats the entire objective of moving away from FPTP.
No, the objective of moving away from FPTP is to eliminate the need to vote for someone other than your preferred candidate instead of your preferred candidate. That's the sort of strategic voting that creates the need to choose the lesser evil over your true preference. With approval, you may want to strategically cast a vote for the lesser evil in addition to your preferred candidate, but there's never any reason not to vote your true preference, which is the point.
This is different from RCV, where you may actually want to rank the lesser evil above your true preference to avoid giving the election to the greater evil, which perpetuates major-party dominance, and *that* is the point of moving away from FPTP, giving third parties a chance.
Approval voting has the biggest disconnect between academic support and facts on the ground. It's advertised as simple and easy, but nothing could possibly be more confusing or alienating than "come up with your own arbitrary approval threshold". I vote for candidates I don't approve of all the time. Voting theorist need to give this one up, and settle on an RCV method. We all know IRV is suboptimal, but throwing out ranking information is not the solution.
I disagree. I don't think it's at all difficult to come up with your own approval threshold. I can't think of an election I've participated in where I'd have a difficult time at all. Generally, I think I'd end up approving two candidates in each election where I'd vote strategically in a FPTP election: My true preference and the lesser evil major party candidate.
I suppose it could get more difficult in a world where there are many competitive parties, but I think that's a good problem to have, and I think approval is more likely to get us to that many-parties world than any RCV scheme, all of which (except Coombs') fail the no-favorite-betrayal criterion. And I sincerely worry that the complicated winner-determination processes of RCV schemes create too much opportunity for doubt and confusion. Among RCV schemes, my preference is the Schulze method, but it is very complicated.
If complexity isn't seen as a problem, score voting is better than any RCV method. Or perhaps STAR -- it fails more of the theoretical voting system criteria, but simulations suggest that in practice it behaves better. That is, STAR can theoretically exhibit more pathological behaviors, but the structure of the scheme makes such behaviors vanishingly rare.
RCV is the only ballot in which the honest answer is unambiguous. If you have an actual multiparty competitive landscape your proposed algorithm of "actual vote + third party throwaway vote" no longer works. I think worrying about the tabulation complexity is unnecessary. Any method choosing a candidate from the Smith set is basically fine. Going from IRV to Smith/IRV in the future probably wouldn't be too much of a lift. I think concerns about favorite-betrayal are also overblown. The public can just barely grasp strategic voting in FPTP. *Very* few will follow arguments based on anticipated preference cycles
The best I’ve seen is combined approval voting, in which you can rate each candidate as Approve, No opinion, or Disapprove. Each candidate gets a net approval score that’s just the number of Approve votes minus Disapprove votes, and the person with the highest score wins! It’s basically what you see when you see pre-election surveys that sort candidates by net approval, calculating it as what percent of people said they approve of a candidate minus what percent disapprove. It’s basically like the top comment on Reddit; commenters can upvote, downvote, or do nothing, and the comment with the highest upvote-minus-downvote score is the top comment. I think it takes the strengths of traditional approval voting (which is just yes vs. no) but improves upon it by letting you differentiate between candidates you’re okay with and those that you passionately want to block from office.
After that, I like the St Louis system, which uses traditional approval voting in an all-parties primary, then the people with the top two number of votes advance to a two-candidate general election.
There’s also Total Vote Runoff, in which you rank candidates like with traditional RCV, but then candidate popularity is calculated not by looking only at first-choice votes but rather by looking at lower-ranked votes as well, giving them less and less mathematical weight. It eliminates virtually all of the math problems with traditional RCV. You can safely rank candidates in the order that you prefer them! The only issue is that the math becomes even harder for a lay person to understand; that may be fine, though, if it leads to outcomes that seem more intuitively understandable.
Ranked pairs, if you're prepared to bit the bullet and say that any ballot without a full set of preferences is invalid (ie you can't just go 123, you have to number all the way to the bottom)
I'm convinced by whatever they call the system of just letting people vote for whoever they want (e.g., as many candidates as they care for) without ranking. It's easier to understand and doesn't require calculations about who wins that could engender mistrust in the results. But anything but FPTP.
Ed Markey is currently the Dem Senator from MA only because in 1976 he won a House seat with 22% of the vote in a big Dem primary and now we may get to watch him die in office as he suggests he will again run at 80.
He seems like a pretty successful politician that was able to continue getting elected even against a well-funded challengers that was a scion of Americas most famous political family.
Kennedy picked the wrong election to mount a challenge, especially post the "July Crisis"... running against Markey would have much more legs in 2026 than it did in 2020
We instituted it in Maine after the embarrassing governor’s race in 2010, in which two moderately left-of-center candidates split the vote and allowed the odious Paul LePage win. So it seemed like a good response to that situation (ironically, the governor’s race is one of the few situations where RCV doesn’t apply)
There are a few drawbacks. First, it attracts a ton of also-rans, especially for city council. Our own Vermin Supremes are encouraged by the hope that they’re everyone’s second or third or fourth choice, or possibly soothed by the notion that they can’t be blamed for splitting the vote.
The trouble is that it’s really difficult for voters to tell who is a “joke” candidate, and who has a decent shot at winning and competently governing. There aren’t too many resources, even for a political person like me. You’re stuck either researching all 6 candidates to make your own ranking (~ nobody does this) or just voting for the candidate you’ve heard of and ignoring the rest. In which case, what was the point of RCV?
The other problem is that people don’t understand the concept. I think it was Nate Cohn in 2021 who said he talked to a friend of his in finance, a smart guy, who said he was going to rank the viable candidates first, and then some of his favorite marginal candidates second.
Here in Portland, leftists ran 4 or 5 candidates against one right-winger. On the first ballot, the leftists got a large majority of the vote. But so many of their ballots were exhausted, that ultimately the conservative eked out a victory.
RCV makes sense for those of us who spend all day thinking about politics and game theory, but it is frankly confusing and perverse for most voters. It’s better that parties (and the interest groups that comprise them) should hash things out internally, and put up *one* liberal candidate and *one* conservative candidate in a competitive race. It can be done.
The problem with RCV is that the US doesn't use it enough for people to get used to it. If you're in Ireland or Australia, you can say "normal 123 voting" and everyone understands it and uses it effectively because all elections work that way.
But if 95% of your elections are single-choice and 5% are ranked-choice, then only nerds like us spend the time to work out how ranked-choice works.
You're also well in the 99th percentile in political knowledge, like all of us Slow Borers. I think Richard has it right that for most there's a learning curve to be had.
EDIT: I did not expect to get this much pushback, so I'll just say that I agree with all three of you on the bottom line and upvote your comments.
Ranked choice almost always results in increased turnout and I've seen some small sample size studies that indicate there can be some voter confusion, my bet is this is something that goes away after a few election cycles.
Again, not a silver bullet to fix democracy but a worthy reform.
I actually think it's an unlearning curve! The tactical dynamics of FPTP ("splitting the vote", "wasted votes", etc) are so deeply embedded that voters accustomed to it get stuck on that way of thinking and that's how they get it wrong.
I've spoken to voters in Scotland (their local elections are ranked choice with everything else FPTP) who have worried about splitting the vote by marking a second choice. When I ask "how does that work?", they can't explain because they don't even know why "splitting the vote" is a bad thing - they just know it is and should be avoided.
This is why I think it's not a problem in countries that use ranking for all voting - the tactical dynamics aren't embedded in everyone's way of thinking and therefore the learning curve is no steeper than that of FPTP.
TLDR: The problem isn't that new voters will get confused or have a steeper learning curve: it's that old voters have to unlearn what they've known all their lives and learn something new.
Aside: the good news is that people who vote a lot (and so have a lot to unlearn) and people paying a lot of attention to political communication (and so have access to the information necessary to do that unlearning/learning) are very closely correlated.
Low-info voters generally don't have a problem with ranked-choice because they haven't learned all of that traditional FPTP tactical stuff in the first place.
Liberal democracy had a learning curve, too. Europe had a hundred-year-long religious war before they finally started learning to accept their differences.
I just fundamentally reject the idea that we can't ever ask people to adapt.
Oh, it is: it's just people with FPTP brain who "know" they should only vote for viable candidates and should ignore people who have no realistic chance of winning.
That is, they do an only semi-conscious first-pass filter down to a small number of candidates who "have a realistic chance of winning" and that messes up the ranking process - which is how you get "rank the viable candidates first, and then some of his favorite marginal candidates second" as an idea.
But people with "FPTP brain" are actually correct on this one!
You usually cannot safety put your real favorite candidate first under RCV, as that risks helping your least-favorite candidate get elected. Instead you need to look at the polling/guess what other people will do and then put your favorite of the two most likely winners as your #1. (Unless your favorite candidate has no chance of winning, in which case you can safely rank them as #1—but your #2 better be one of the top two viable candidates.) https://youtu.be/JtKAScORevQ?si=wMYo1ME9w7POzx7Z
"Unless your favorite candidate has no chance of winning, in which case you can safely rank them as #1—but your #2 better be one of the top two viable candidates."
Yes, but isn't that the main sales pitch used for promoting RCV? ("You don't have to vote for the lesser of evils -- you can vote for your preferred third-party candidate and then rank the lesser of evils second.")
The trouble is that one of the sales pitches for RCV is “There’s a person who’s your real favorite and then someone who you see as the lesser of two evils. With plurality (FPTP) voting, you have to vote for the lesser of two evils, and you never find out if your preferred candidate had a real chance. With RCV, you can rank your real favorite candidate first, and the outcome might be the same—or maybe it’ll turn out that the person is so popular that they do end up getting elected now that people have a risk-free way of voting for them!”
And so the issue is that the “risk-free” part is incorrect if that person *does* actually have a shot at being a non-trivial candidate, as it could accidentally result in a center squeeze where the less-of-two-evils is kicked out, but your favorite loses too, and your least-favorite wins (as happened for Palin voters in Alaska; she was originally polling behind the moderate Republican but then ended up slightly ahead).
You only help your least favorite candidate if there’s a real risk that your middling candidate gets the fewest first place votes. Just because the situation is possible doesn’t mean it’s likely.
Going back to the 2022 Alaska example: Sarah Palin’s supporters helped get the Democrat elected thanks to rating Sarah Palin #1. If they had instead ranked Sarah Palin last (and their next choice, the moderate Republican first), then the Democrat would have lost.
I don’t suppose you have the numbers at your fingertips but I’d be interested to know:
A) how often do people rank their unrealistic fantasy candidate as #1, the major party candidate whom they like better as #2, and the major party candidate whom they don’t like as #3? How often do they rank them in a different order?
B) how many people vote for their fantasy candidate as #1 and then exhaust their ballots?
People in Ireland may be “comfortable” with RCV, but are they using it in a manner that makes it more effective, or gives them outcomes they like, more often than FPTP?
There aren't "major parties" in that sense in Ireland; it's more of a spectrum from "could provide a PM" (Fine Gael, Fianna Fáil, Sinn Fein) to "unrealistic fantasy" (Renua, Workers', etc) with a whole bunch of parties that can form the minority partner in a government in between (Greens, Labour, Social Democrats, etc). And people do vote all over the shop. There certainly are people who exhaust their ballots after their number one choice is eliminated, but they are relatively few.
The Aontú candidate was eliminated and the SF candidate (the only party they are relatively close to) had already been elected so voters who had voted either just 1 or 12 for those two parties go non-transferable (this is count 5 on the link). There are just 345 non-transferable papers and 1177 transfers, so 77% of voters have a valid preference (be that second, third, fourth or whatever).
For a candidate of a more moderate ilk, the SD candidate eliminated at count 4 only had 83 non-transferables out of 1249 - 93% transfers.
Note that this is multi-winner RCV. If you want examples of single-winner RCV, you need to look at Australian general elections.
You can see it in Australian election results, the major parties pick up ~70% to 80% of the primary vote, i.e., most people ‘vote 1’ for one of the two major parties (labor or the liberal/national coalition).
That seems a little tautological though. If the major parties weren't genuinely widely popular, they wouldn't be the major parties. Most people don't actually have an unrealistic fantasy party they wish were in charge... they think the small parties are kooks and weirdoes organized around some specific ideas too extreme for the major parties. Of course most people vote for them first.
Responding to your point about people not understanding the concept as an Australian.
People will understand the concept when they do it regularly. We have compulsory voting in Australia, and most people understand the concept in our single seat electorates. The process for preferences in our PR elections is more complicated, but simple ‘above the line’ systems mostly solve this problem.
I think most issues that people have with RCV comes from a place of not being able to imagine it working, even though there is extensive evidence that it does work elsewhere.
In saying that, I can’t begin to imagine how much chaos my local Hare-Clark preference system for proportional representation would cause amongst the Stop The Steal types. Understanding the results in a system like that can be challenging for people who want to understand.
I have a somewhat different - and probably somewhat elitist - take. RCV really isn't that complicated. An electorate that can't understand it - and more to the point, a media that can't explain it so that the electorate can understand it - doesn't sound like an electorate that can maintain a stable democracy.
I agree with Matt about proportional representation, but that seems like a lot harder row to hoe - kind of like getting rid of the electoral college.
David, politics is fundamentally zero sum. It's a majority rules enterprise. Every legislature I've ever heard of passes laws with 50%+1. They can all pass a major, controversial, divisive law with 51% in favor and 49% opposed. If you use PR they just form coalition governments, and then the 49% that aren't in the coalition get squashed. I think you're imaging politics is some kind of consensus kumbaya where it takes 75% agreement to pass a law, but it..... just isn't.
Politics is majority rules and that's OK. You can re-arrange the window dressing of the electoral system, but you get the same results any way you shake it up
>>I think you're imaging politics is some kind of consensus kumbaya where it takes 75% agreement to pass a law
You're thinking wrong, then? This is definitely not where I was going with this.
In fact, I'm painfully aware of what you're talking about when you say "politics is fundamentally zero sum". If anything, I actually take that idea MUCH further to its logical conclusion: REALITY is zero-sum. Matter/energy cannot be created nor destroyed. Two particles cannot occupy the same space. Only one chimp can eat any given piece of banana.
So you're ABSOLUTELY correct that politics is fundamentally zero sum. Decisions have to be made by bare majorities! Anything else is the illusion of supermajority, which is actually just minoritarianism by another name (hence my hatred of the filibuster).
However, reality also allows for pockets of positive-sum-ness. The sun's gravity turns it into a fusion engine that emits lotta-yotta-joules of energy per second. The universal arrow of entropy points to higher disorder, but temporary pockets of increased order can perfectly well be established -- this is the foundation of the existence of life itself.
Similarly, although no two humans can eat the same banana, they can cooperate to grow many bananas. This is positive-sumness.
IMO the goal of political systems design should be to set up positive-sum electoral dynamics that mitigate the ambient push of reality towards zero-sum incentives.
I wonder if it's harder for voters to understand outcomes in an RCV system than it is to understand how a candidate with 23.6% of the vote is the "winner" as Matt's post demonstrates.
Wouldn't it be directly proportional if the vote was based on one person, one vote for the whole country instead of filtering through the electoral college?
I appreciate the perspective, but I just can't get on board with a numbered ranking of the candidates being confusing and perverse. Anyone who's ever picked a kickball team on a playground can figure this out.
To continue with the sports metaphor, though, it's not very much like picking kickball teammates. It's much more like picking a competitor in the decathlon.
Do you know the particulars of how the decathlon is scored? I don't.
That's the point of having a top 4 primary first before doing RCV (especially if you require a minimum threshold for support to move on), you filter out the marginal candidates.
I generally think that four or five is about right - I prefer five as fifth place seems to be where a good third party candidate can get to, and having two Dems, two GOP and one third party seems a sensible balance for any semi-swingy area (anything from 65-35 to 35-65 or so)
I’m all for libertarianism, but I don’t think I would bend over backwards for small parties at the district level. That’s better incorporated at the national level through something like a mixed member proportional system.
Under RCV, you should *not* rank your true favorite candidate #1 if they may or may not have a chance, as it risks making them a spoiler who gets your least-favorite candidate elected.
It’s possible that Mr. Finance understood this better than Nate or I do!
But as a practical matter, how would I as a voter decide whom to vote for, being aware of the negative responsiveness paradox? Or should I just pretend it doesn’t exist?
As Matt says, the people who still want to vote like they would for FPTP can still do so: just cast one #1 vote for your favorite candidate and leave the ballot like that. It would help RCV advocates if they pointed that out more often, though.
In my vision for RCV, you must rank exactly four candidates. You cannot rank three, you cannot rank five, you cannot rank one. In order for your ballot to count, it must have exactly four unique candidates ranked.
I don't think that should be that hard too understand. Should be a simple publicity slogan, "Only Ballots with Four Names Ranked Will Be Considered." Then it is up to the voter to have their four names ready when they go to cast.
This is part of my first complaint; if there weren’t RCV, the race would attract fewer fringe lefty candidates, and force more coordination along party lines. People are exhausting their ballots not because they’re Vermin Supreme-or-bust, but because they don’t realize they can rank him first, Angela Davis second, and Normal Leftie 3rd.
On the voter confusion part, a lot of the discussion here seems to be conflating the simple parts voters do (ranking their choices) and the confusing part of how the votes get counted each round (which voters don't have to know the details of in order to vote).
I will never not be furious at the government winning the RCV referendum (Alternative Vote) here in the UK by claiming that voting yes is stealing incubators from babies and ensuring squaddies won't be able to get bulletproof vests. Unbelievably dirty campaigning and it has lead me to supporting RCV on principle everywhere.
One of the problems in some places is that state laws prevent experimentation with PR. So, the political path indeed HAS to go through smaller reforms that give enough power to smaller parties and moderates that they can then advocate for bigger changes to the party system like PR.
Idaho recently became one of them--they mandated the same old terrible geographically defined seats because they were butthurt over Boise electing the same people from the same neighborhoods. PR would still give roughly the same parties the same political power, but the overlords in the legislature just cannot stand anything that even slightly weakens rural power.
The thing that I would want to see before ditching FPTP is some cross-country empirical evidence about the trade-offs between different voting systems. What actually happens when you implement PR? I think the answer is that, often, you end up with weaker political parties and less voter control over the government.
One good thing about FPTP is that you actually get to vote *for a government*. In PR, you may get to vote for your favorite little party, but then they have to cut deals to form a governing coalition, and you as a voter have no control over this process. (As Anthony Downs pointed out in the '50s.) It's also not a process that always works. Belgium's record is 652 days without a government. Germany's current coalition is schizophrenic - one party wants green regulation, another wants no regulation. Such coalitions are harder for voters to hold accountable. (Who's fault is it?)
Another benefit is that FPTP (more generally, single-member districting) is more fascist-proof. There was substantial support for fascism in early 20th-century Britain as in Germany, but it was much harder for fascists to get a toehold in the UK.
As for tactical voting, I don't see the problem. We can't get rid of it, as shown by the Gibbard-Satterthwaite Theorem. All we can do is look at the evidence to see how things shake out for each system in the end. I can think of some happy cases for RCV, like Australia - but they're a small society with mandatory voting, so maybe not a good test case for the US. For PR, I worry that advocates are thinking about peak-performance Germany and ignoring the less rosy examples like Belgium, Israel, and the Germany of today, with AfD on the rise and anemic growth.
Always seemed notable to me that most large, wealthy democracies use a majoritarian system and *not* PR. The US, Japan, Britain, France, Canada, Italy, Australia, South Korea, and Taiwan- all majoritarian. PR might work well for a small, ethnically homogenous country like in the Nordics, but once you reach a certain size and scale it would seem that single party control of the legislature is a desirable thing..... In some cases like France and Italy, they've adopted a majoritarian system after disastrous experiments with PR
Those systems still allow for dynamics that prevent calcification into a two-party system by making it viable to become a breakaway faction (mostly; Italy is a REALLY bad outlier).
Most of those systems are “2 major + 1-5 minor” parties. Ours is strictly 2 major parties that never take input from smaller ones and never suffer breakaways.
This is really BAD, because it incentivizes entryism. A faction like the libertarians or wokeists has to make a “long march” through either party’s institutions. A disruptive authoritarian-aspiring outsider like Trump can leverage winner-take-all dynamics to appropriate an entire party for himself.
The same 2 parties win power every single time, with no exceptions ever, in Britain, Canada, Australia, Taiwan, and I think South Korea though I'm less familiar with their system. (And, uh, the same 1 party wins almost every single time in Japan, whatever you want to call that).
The only non-2 party system in the mix there is France, because of Macron/En Marche. (And they use a 2 round system, which I prefer). So if you think having minor parties who can never actually control the government does something, you can make that argument. But almost all of those countries are effectively 2 party systems. There are never any 'breakaway factions' that control the government
In fact, I DO think that their important role isn't in that they "control" the government, but that they discipline their closest allies and the broader system. UKIP dragged the Tories to the right; the LibDems provide a stalking horse to Labour. Small parties don't have to WIN in order to influence the broader system.
Even when that goes in directions I don't like (UKIP), I'd rather have a Tory party that never has an opportunity to rest and attempt to consolidate control over the entire right wing of UK politics. Because THEN they'd be able to force the Labour and LibDems to merge and do the same on the left.
I think some of the same effect is achieved in the US without these politicians being members of small parties. The Green Party exists only to draw a tiny percentage of the presidential vote away from Democrats, and has no other political significance, but the Democrats face left-wing challengers from people registered in their party (e.g. "The Squad") or a quasi-Independent who has run in the Democratic presidential primary (Bernie Sanders).
The insurgents have been even more successful on the Republican side, where Trump has taken over the party, departed drastically from old pro-free-trade attitudes, and (worryingly) made the party more bluntly authoritarian and in thrall to one man with a huge ego and a disdain for the rules.
It's probably an institutional difference -- US parties have little ability to exclude candidates from their party, but parties other than the big two have no real prospect of getting elected to anything larger than a city council or state legislature, so the strategy for less-conventional politics is to form a "wing" of one of the large parties.
I'd be curious, though, how parties in e.g. the UK exclude candidates they don't like. (Do they have much ability to do so? Do candidates rely on the party per se for their campaign funds?)
>I'd be curious, though, how parties in e.g. the UK exclude candidates they don't like
The major 3 parties hold an extensive candidate interview process that involves panel interviews, training and mock speech sessions, they have to do a group presentation to a board, etc. It's fairly rigorous. Yes, the candidates are completely reliant on the party for funds
To file as the candidate of a party, you have to have a certificate from an official of that party saying you are. If the leadership wants you out, then the official doesn't sign that certificate and that's it.
The problem with having consolidated parties with "wings" is that they're still consolidated AF. The competition happens in primaries instead of out in the open. The dissenting view -- and the argument about it -- happens in a primary dominated by extremists whose incentive is to oversell their extreme vision, instead of maybe to make direct appeals to moderate voters with "out there, but reasonable" visions.
I think a case like AOC demonstrates the strength of our system. She entered politics as in effect a member of a small party (like the Greens or DSA) and has since substantially worked hard to become a party stalwart and a supporter of its national candidates and its larger message while obviously being on the lefthand side of its ideological distribution. I think this has strengthened the Democratic party.
Heck, even Rashida Tlaib has called on her supporters to vote the straight Democratic ticket and if anyone has a gripe against the party leadership, it would be her.
See, here's the thing about this discussion... there's a reason why I avoid direct comparisons and individual cases so much, and it's not because I'm just trying to be evasive.
Rather, it's _path_dependency_.
Pretty much everything here is path dependent. Israel's multiparty system went off the rails from the bat because they had an absurdly low threshold, so their parties were able to establish strong brands that endured even after they raised the thresholds.
Likewise... just because AOC and Tlaib moderated over time, doesn't mean we have a healthy system. *AOC* moderated, but The Groups are still a cancer within our party. The primaries are still broken. Neither of the two major parties has any competition besides each other, and the constant internal insurgencies contribute to negative polarization by confirming each side's worst fears of each other.
So while specific examples can help illuminate healthy *dynamics*, I don't like overindexing on any one scenario. The point is that we want to shape better politics *going forward*.
For instance, imagine if we'd had a more open system before Trump's insurgency, and by now had four parties: 20% MAGA, 30% Republicans, 35% Dems, 10% DSA, and 5% Greens. Dems would run by saying "Don't vote for Republicans, they'll have to coalition with MAGA extremists". The Republicans would run away from MAGA and might even make overtures to the Greens. DSA would be carping hard about Palestine, and Dems could position as centrists by clapping back, confident they could force DSA back to the table if Dems won.
This would probably be healthier. Both major parties incentivized to appeal to the center, to reach across the aisle, repudiate their flanks. The flanks forced to make their case to the center, instead of yelling at their friends.
But what I care about is the dynamic, not how you slice and dice the %'s of each party or when in particular you assume the party system was put in place on the scenario's timeline. The scenario doesn't matter, the dynamic does.
Your scenario sounds like a coin flip where Republicans/MAGA have an even chance of winning and the Republican party would be totally dependent on the MAGA party.
Sounds like about what we have now. Except I might be inclined to switch your 30% vs 20% breakdown. Republicans love Trump and MAGA. Paul Ryan's party is dead. I don't think an electoral arrangement would change that.
My point is that we can sit here arguing about specific scenarios all day long, but what actually matters is the dynamics we learn from observing healthier democracies.
I'd say that Belgium's coalition failure is an example of the system working.
When Belgium is gridlocked, the failure is quite wide and open for everyone to see. The bickering becomes front page news.
When America is gridlocked, we call it a fucking Tuesday, and the idiot voters spend three straight decades failing to connect the filibuster (and all of its precious fucking mythology about "compromise") with the dysfunction, let alone any other broader systemic problems. We keep having elections, politicians keep making promises, but nothing happens to break the gridlock, and voters don't know who to punish. A Republican-controlled House grandstands against a Democratic President for 2-4 years, and then the voters vote for fucking divided government all over again because of "compromise".
I agree that America has political dysfunction, but I don't blame FPTP.
My problem with the US - which seems to be what you have in mind, too - is that it's got anti-majoritarian sand in the gears. The filibuster, the presidentialism, the bicameralism - all of this is designed to prevent voters from electing highly effective governments.
I agree that the basic structure is bad; it's like trying to code an AI in DOS.
But to abuse the metaphor, not ONLY do we have that problem; we've also loaded a bunch of shitty first-party software onto it that locks us into their shitty ecosystem.
Compared to the Senate and EC, the filibuster and FPTP are the easiest root kludges to address. If, all of a sudden (again, still within our metaphor), our corporate board decides that we will no longer honor first-party lock-in agreements, then we can install a new OS. But since the original DOS system is so fragile, we're gonna have to boostrap this new OS to, say, Windows 2.1 (IE RCV) in order to grant us the flexibility to start getting this legacy system up to a modern performance standard.
This is why I just can't get too enthused about discussions regarding various RCV or PR reforms. I appreciate there are a lot of polysci minutiae nerds on this site, and I'm usually amongst them on most topics. But on this particular issue it's too muck like rearranging deck chairs on the SS Madisonian. We should employ a Westminster-adjacent model for both the national and state governments. At that point if we want to experiment with PR perhaps via party lists, have at it.
If a crisis ever presents itself and lends to such a wholesale conversion, best believe I’d be happy to endorse it.
But short of that, I think the best option is to do the slow boring work of creating the reforms that will get us on a path towards the wholesale ideal.
Everyone basically gets a proxy vote for various legislative bodies. If enough people select a certain politician as their proxy (say, 10% of the state or district), then that person wields their votes in the chamber. You can change your proxy from time to time or immediately (we'd probably need some sort of hysteresis/emergency recall procedures in place). Perhaps there'd also be a system of conditional proxies -- for instance, I vote for AOC for Senator, but Chuck Schumer keeps my Senate proxy until AOC has enough votes to hit the threshold, upon which time all her votes convert to her and she joins the Senate. Whoever had the most votes on the waiting list for her House seat in turn gets instantly elected, no fuss.
The counterpoint is that the antimajoritarian sand is the only thing saving the country at this point because either major party would wreck the whole thing if they were actually able to get their way with a simple majority.
Agreed. Without being too much of a nationalist homer, this is where I think America's very, very unique political system is kind of brilliant. Bicameralism, staggered election cycles, strong federalism, etc. The only country with anything close to it is Australia, where a majoritarian lower house has to face an equally powerful Senate elected by PR
This is like saying we should be helicopter parents because kids are dumb. They are. But they'll never grow up unless they sometimes have to fend for themselves.
If children never have to feel the consequences of their actions because they're constantly being thwarted by parents, they tend to become more reckless and irresponsible.
If parties never have to feel the consequences of their political agendas because they're constantly being thwarted by the system, the same principle applies.
I feel like this suffers from the same sort of false analogy as comparing the US federal budget to a household budget? Most critically, when the kid screws up and "feels the consequences of their actions," it's generally speaking only the kid who actually suffers from it other than some sympathetic emotional pain from the parent. When a political party screws up, it can (and does) ruin the lives of millions upon millions of people, often including large numbers of people who were opposed to the policy change.
Either party would have course-corrected us a long time ago if we didn't have the filibuster short-circuiting the democratic feedback loop and raising the stakes by stagnating our politics.
That's probably true, which is why I qualified my statement with "at this point . . . ." If we imagined this change happening in 1900 or something, I'd be somewhat more sanguine.
I don't think anyone would describe the US government over the last 25 year period as being fantastic, and yet is there any major parliamentary government you think has done better?
Even Belgium, who you describe in their government failure as the system working, would you say they've better run than the US over that time period?
I think the USG is benefiting from (1) sheer size from historical immigration, (2) historical good decisions and policies like the Fed, Social Security, and the creation of growth engines like Silicon Valley, and (3) still being the lynchpin of the liberal world order institutions like Bretton-Woods.
If we didn't have the first one, we'd look more like the UK - a spent power that can't sustain even a diminished global hegemony, and unable to incorporate new waves of immigrants.
If we didn't have the second one, we'd look like South Africa or Latin American banana republics, and just fundamentally be unable to develop our institutions or grow our economy through innovation.
If we didn't have that last one, our debt load would look more like Japan's and we'd be shitting ourselves.
Most of those things were also achieved during an era when our party system more resembled "four factions within two parties", and thus was dynamic enough to roll with the significant punches landed on it (Depression + WWII, etc), even if it could be sluggish or acrimonious at times. We don't have that anymore. We escaped a catastrophic pandemic with the strongest recovery of the entire developed world, and our populace can't tell the difference between that and a long-mounting cost-disease crisis exacerbated by the same stagnation and gridlock that we've allowed to fester for the last several decades.
Basically, I'm positing that our ancestors created a machine so powerful that it can't help but keep firing on all cylinders under just minimally-competent governance, despite all the shitty things we do to kill off the golden goose -- NIMBYism, polarization and stagnation, etc. But it's not going to last forever, and we need reforms that will stop strangling the damned goose before it dies (and I don't mean that solely in some vaguely libertarian sense; I also mean that we need more dynamism in our party system).
Belgium is a small power roughly the size of Maryland; similarly, it's largely dependent on a confederated government with its much larger neighbors, although neither Maryland's nor Belgium's own contributions are anything to laugh at. So, I think that's a more appropriate comparison: Does Maryland's state government perform as well as Belgium's? Do (US) states similar in size to Belgium ever suffer from ~2 years of gridlock due to things like divided government or state-leg filibusters? I'm pretty sure we could dig up plenty of examples.
I concede that the US has enormous advantages and that we often waste them. But I also think if you went to the past, you would be surprised by how many people felt the same way you do now about our institutions. There is no period of our past where governance wasn't filled with significant levels of corruption or malignant influence.
Your question about states compared to Belgium is really interesting. I think its extremely unlikely we will find any state having that level of dysfunction.
Mississippi has a lot of issues, but is the problem their government's inability to pass legislation due to veto points or some unique structural problem in their state government system?
Germany set the bar at 5%, and look what happened. Perhaps they could have set the bar at 10% or 20%, but then for all we know, voters for small parties would have felt even more alienated than they do under SMD.
PR might work in local elections, but for federal ones I don't think it's compatible with American culture. "Write to your representative" is deeply ingrained here. In PR there would be no single representative for someone to write to, and it'd be much easier for the representative to ignore constituent comments from people outside their base.
So you write to the party you voted for--or even take it to the next level and join that party, and try to influence its actions.
Yeah, I don't particularly like political parties myself, but at some point I had to grow up beyond the hilarious "Don't blame me, I voted for Kodos!" line, and accept that they're essential parts of politics.
Surely you can see how a democracy in which you contact a party representative is not the same as someone directly impacted by your vote? There's an additional abstraction layer that makes them less accountable to it, and party leadership stronger (which you might prefer if you trust them).
I lived in a country with PR and, while the implementation was different than what MY proposed here, the reps there always seemed more interested in intra-party jockying for spots than they do in the US. That's ultimately what determined their reelection chances. Perhaps that experience left me biased against them.
Note that Matt doesn’t endorse party list PR - he endorses STV. Under STV it’s natural for some candidates to campaign on local representation, and for some people in an area that particularly needs representation to vote for locals first regardless of party. If enough people do this, then they can guarantee a local representative.
But some people might feel a stronger need to be represented by a co-religionist, or someone who shares a hobby or whatever, and it’s possible those representatives might be able to play a similar role to what a geographic representative does. (Though there is no guarantee that every person has one representative that represents them.)
This reminds me of the scene in what is perhaps Robert Heinlein's best novel, "The Moon Is a Harsh Mistress" in which the Moon, having won its independence from Earth, is debating what type of representative body to create. One of the revolution's leaders, Professor de la Paz, urges them to think more creatively than simple geographical representation:
He says, "This is the traditional way; therefore it should be suspect, considered guilty until proven innocent. . . Surely where a man lives is the least important thing about him. Constituencies might be formed by dividing people by occupation . . . or by age. . . or even alphabetically. . . You might even consider installing the candidates who receive the *least* number of votes; unpopular men may be just the sort to save you from a new tyranny. . . .Suppose instead of election a man were qualified for office by petition signed by four thousand citizens. He would then represent those four thousand affirmatively, with no disgruntled minority. . . Or a man with eight thousand supporters might have two votes in this body . . . "
It's nice to know that if I ever ran for some thankless local position, at least 200 people would arbitrarily vote for me without any campaigning at all.
RCV is probably better than the status quo, but Approval Voting is even better. It's simpler, you just mark a vote for all the you think are qualified and approve of, and it also encourages strategic voting less.
Approval Voting actually has a much more central role for “strategic voting” than instant runoff. There is no real meaning to “approve” other than “I think everyone I mark would do better than everyone I don’t”. But if the election is Kamala Harris vs Paul Ryan vs RFK Jr, should I “approve” of Paul Ryan because he won’t break the government, or disapprove because he’s awful? That depends on whether I think there are a large number of people voting for just RFK (in which case I want to support any and all of his opponents) or just Harris.
It becomes more complex in a larger election for something like mayor. In Matt’s San Francisco example, should I approve all three moderates and give no signal about preferring London Breed, or just approve her?
The answer to this is to either use combined approval voting (https://en.wikipedia.org/wiki/Combined_approval_voting?wprov=sfti1#), in which you get 3 choices (approve, neutral, disapprove) or to use the St Louis system in which there’s an approval-voting primary to determine the candidates for a two-candidate general election.
You're right that there is an element of strategy in Approval Voting, but that's because no reasonable voting system can eliminate strategy completely. AV still does much better than RCV. Consider an election with three candidates: Lincoln, Carter, and Hitler. I think Lincoln is great, Carter is fine, and Hitler is awful. Under AV, I have a decision to make whether to approve of Carter or not, and some of that will depend on what I make of his chances vs Lincoln. But even if I mark him as disapprove when really I should approve of him, that's not that different from my sincere preference either way will lead to an OK outcome in the election. In RCV, the temptation is to vote Lincoln > Hitler > Carter rather than my sincere ranking of Lincoln > Carter > Hitler, since the only way to strategically hurt Carter's chances is by helping Hitler's. This is of course both much further from my sincere preference and much more likely to lead to a terrible outcome than either of my votes under AV.
I don’t think IRV ever gives me an incentive to vote L>H>C (at least, I can’t think of how that would happen - as long as I put L first, my other preference can’t do anything until L is eliminated, and in that case there are only two candidates so I have every reason to be honest).
However, it can give me the incentive to put C>L>H, if I worry that Carter might be eliminated first, and his voters might split between Lincoln and Hitler, so I want to keep him in to defeat Hitler in the runoff even though I would prefer Lincoln.
Yeah, that's right, I messed up the example. Hopefully it's still clear that the motivation to vote strategically is at least not worse with AV than RCV.
"and in that case there are only two candidates so I have every reason to be honest"
This is part of the issue I see with RCV/IRV vs. FPTP -- RCV/IRV only really seems to improve things if you have three or more candidates (and much better four or more) who, on a Venn diagram of their preferability, would have substantial overlap between at least two of each of the candidates (better three or more) such that voters actually have a preferred ranking of them. However, probably in most US political races outside of local governments, there are usually only going to be two candidates (in which case, ranking doesn't matter at all) or three or four candidates whether third and fourth candidates (most frequently Libertarian and then Green) and I suspect that the latter two are usually viewed as being so toxic by both Democratic and Republican voters that they either aren't going to bother to rank those candidates at all or will rank them below the main opposing party, in which case again ranking doesn't really matter.
That’s how things go with the current voting system, but with some method other than FPTP in place, you would often get things like the California senate race between Schiff, Porter, Lee, and Garvey, where there are multiple mainstream representatives that differ in meaningful ways that aren’t exactly ideological. And even in the current system, the Green and Libertarian voters often have meaningful preferences between Democrats and Republicans.
It is notable that IRV and approval are the two that are most talked about, and IRV is the one that is most organized and implemented. In the theoretical literature, I think Borda count has a lot of defenders, and the Condorcet criterion does too (though there is less about which particular Condorcet method is best, and all are more complicated to explain than either Borda or IRV).
Approval does have some important difficulties though, in that you need to introduce this hard to motivate threshold of approval, rather than just ranking your preferences. But why IRV won out out of all the methods that count ranked choices, it’s harder to say.
Well, I’m staring at my California ballot now and really wish I didn’t have to vote on things like:
* Airport board
* Hospital board
* Local utility board
I have no idea who any of these people are and all their websites and candidate statements just say stuff about how good at leadership and finding compromise they are and how long they’ve lived in the area. Who cares!
I wish I could vote for “this should not be an elected office”.
Ranked-choice is the best of a bad bunch of solutions to the problem of "how do you do a single-winner election". The real problem is single-winner elections.
If I could make one radical reform to the US, it would be to abolish all single-winner elections.
For offices that have to have a single winner in the end (President, Governor, Mayor, etc), the right approach is to elect (proportionally) an electoral college which sits for the full term of the office, and then have the office-holder chosen by a majority in that electoral college. For cities, that can just be the existing city council, but if you want to have a legislative/executive separation, then have one legislative chamber and one executive one (they can then also approve appointments and be the body responsible for monitoring the performance of the executive). I'd move states to have single-chamber legislatures; they're not internally federal and so don't need a Senate, but this would mean three chambers (two legislative and one executive) at the federal level.
To add a bit more detail: A lot of Americans get mixed up with parliamentarism and proportional representation - the US has neither.
Certainly, this is somewhat *like* parliamentarism, but making the electoral college into a real thing would still have separation between the executive and legislative branches. Each party would put forward their candidate for President; there would then be an election for the electoral college, which would be proportional by list in each state. They would then gather in DC and if a party had a majority (unlikely, as this would lead to multi-party dynamics), they'd elect their President. If not, they'd have time from mid-December to inauguration on Jan 20 to negotiate a multi-party coalition agreement. The smaller parties wouldn't get the Presidency, but they would probably get concessions on both policy and personnel - e.g. the "Aid to Ukraine Party" gets to put their Presidential candidate as Secretary of State and a commitment to spend significantly on aid to Ukraine in exchange for their support.
And if the new President broke the deals, then the minority parties would have the option of pulling the plug, coming up with a new coalition and picking a new President mid-term.
But also - most states should elect their house members in multi-member districts.
I live in a parliamentary country that has lots of single-winner elections. The US has single-winner elections for every single member of Congress, just about every state legislator, etc.
FWIW I completely agree with your assessment. The way I think of it is that single-winner is a form of zero-sum dynamic. These dynamics breed likewise zero-sum thinking amongst their political culture, which is one of the primary evils of the world. Thus, systems should mitigate the impact of zero-sum dynamics, and replace them with positive-sum ones where possible.
Speaking of radical reforms, it would be interesting to get a mailbag question about, "if you could rewrite the US Constitution from scratch, what would it be?" My initial thoughts might be to have a parliamentary system with fewer veto points, but I also worry about what will constrain bad actors, inept actors or ideological extremists from wrecking things as a result of winning a single election. I wonder how much of good governance depends on good institutional design vs having the right people in charge?
Institutional design affects the incentives that determine what people are in charge. A really obvious example is that the design of the Republican primaries determined that Trump won in 2016 and that a different design would have eliminated him.
You need to think through the behaviours you want to incentivise.
I think that removing the executive/legislative split in the US would be far more radical in practice than any other change short of ending elected government, and for that reason I'm inclined to not do it. That's why I vaguely proposed an executive parliament separate from the legislature (which I call the Electoral College because it makes it seem familiar even though it's a radically different institution)
RCV is good. It provides incentives for candidates and parties to be less shitty to each other, and engages both voters and parties who are alienated by the 2 party structure. I think the long term benefit to our civic life is much larger than the direct impact on a few marginal elections that would have flipped in hindsight.
No one in Australia believes that candidates are incentivized to be 'less shitty to each other'
>engages both voters and parties who are alienated by the 2 party structure
Australia's used RCV for a century, and the same 2 parties get 90-95% of their lower house every single election. They actually have a *more* entrenched 2 party system than Britain, which uses FPTP!
Matt’s Lisa Murkowski example shows a very common misconception about RCV.
Let’s suppose Alaska has a top-4 RCV election with the following first-choice results:
—MAGA Republican: 35%
—Lisa Murkowski: 32%
—Moderate Dem: 28%
—Progressive Dem: 5%
What happens? The Progressive Dem gets knocked out and their votes go to the Moderate Dem, who now has 33%. Lisa Murkowski then gets knocked out, and if we suppose that her supporters’ second choices are evenly split between Moderate Dem and MAGA Republican then the result is 49% Moderate Dem, 51% MAGA Republican. The MAGA Republican wins! Or if Murkowski’s supporters’ second choices tilted slightly Dem, then the Moderate Dem wins. In neither case does the median voter’s candidate win. 🫠
Vote-splitting and the spoiler effect are major issues in RCV and can lead to unpredictable consequences like above. The system incentives consolidation into two candidates. In fact, one of the Republicans running for Alaska’s House seat smartly dropped out in order to consolidate Republican support and avoid vote-splitting:
RCV does not magically push more centrist candidates to win elections. In fact, one of its well-known problems is what’s called the “center squeeze” where centrist candidates get squeezed out and people at the fringes get elected: https://en.wikipedia.org/wiki/Center_squeeze
This is definitely possible. But it’s much harder to predict, so it doesn’t encourage as much strategic voting, unless the candidates are very clearly linearly ordered (and then, only by supporters of the most extreme candidates). And even when this does happen, the winner has significant support.
My sense is that city council is the type of place where you least want proportional representation.
Instead, you want a councillor who knows your area of the city and is responsive to your concerns about noise, construction, parking, traffic, etc. You need someone to complain to when things aren’t going well.
In effect, PR results in no one being directly accountable. Which is probably fine for higher levels of government, but at lower levels you want to know who’s on point.
This is a good thing, actually. Current electeds weight the preferences of local constituents too much and should instead be thinking more of what’s good for the city overall. This is why we have too much zoning, eg.
Are there any issues other than zoning which local representation might be useful for? I dunno, potholes?
Plenty of other local problems that are just bad, rather than inconveniences from development and growth. Things like crime hot spots, nuisance businesses, neglected parks, etc.
It's not clear to me that one rep who cares a lot about an issue is more effective at getting it fixed than if all the reps could benefit somewhat by fixing it.
That's the Special Interest Effect in action - when there's an issue that would greatly benefit a small number of people but slightly harm a large number of people, it often happens, even if the net effect is negative. Similarly, policies with wide, small benefits but large, focused harms are unlikely to happen, even if the net effect is positive.
This was exactly my experience when i lived in Detroit. No one represented a particular part of the city, and as you say, no one was responsible --- or responsive.
If you're a YIMBY, who wants local government to pay *less* attention to noise, construction, parking and traffic concerns, then it makes sense.
I am not at all sympathetic to the "everything bagel" theory that the only thing that local government does is restrict development of new housing.
True, it's also bad for transportation/street improvements since every parking spot has a designated defender in office
Note that Matt is endorsing STV, not partisan PR. Under STV, people who particularly care that there is local representation can put all their local candidates first, above all non-local candidates regardless of party. But someone who thinks it is more important to have ethnic representation, or renter representation, or partisan representation, can put the relevant people first. You don’t get the party list effect.
But Matt’s proposed solution is to have “13 at-large members”. If I live in Ward 1, even if I vote for a local candidate, they still aren’t directly responsible for or accountable to Ward 1.
The way Germany and New Zealand do MMP addresses this, but that seems far too complicated for a city council.
He _should_ be endorsing Sequential Proportional Approval (i.e. approval ballots with Jefferson tabulation).
https://en.wikipedia.org/wiki/Sequential_proportional_approval_voting
That’s interesting! I hadn’t heard of that method. What advantage does it give to use an approval method rather than a ranking method? It seems like that creates a lot more incentive to strategize with one’s vote.
Approval is much more similar to the systems we already have (like where you're voting for multiple at-large seats and can vote for up to N), and the "strategy" in it is very easy to think about. You vote for the person you would've voted for in a FPTP race (your favorite amont the "frontrunners"), plus everyone you like more.
Range systems like STAR (or the proportional variant of that, Reweighted Range) do allow more expressive power, but it turns out that to a large extent, optimal strategy (from the individual perspective) in a Range system is basically to reduce it to approval -- figure out your favorite frontrunner, top-rate them, and then also top-rate anyone you like better. If voters uniformly sacrifice strategy for honest expression of ratings, you should get more socially optimal outcomes, but if only one faction does that, another faction can take advantage to move outcomes in their direction somewhat.
Plus Range is just more expensive to administer and more complicated to explain. Approval is incredibly simple.
“Instead, you want a councillor who knows your area of the city and is responsive to your concerns about noise, construction, parking, traffic, etc. You need someone to complain to when things aren’t going well.”
So you want a council member who will help you be a NIMBY? I am surprised this comment is so popular here.
All of the particular examples brought up by this comment are NIMBY selling points, but there's plenty of other local disamenities that are just bad for everyone. Things like crime hot spots, potholes, nuisance businesses, broken street lights, etc.
Agreed
Local pork is good, actually.
Except when it results in a universal NIMBY logroll, as it does in every large American city
Actually I think reducing corruption in politics is a big cause of this. Look up any major iconic building built 1900-1950, there are major corruption issues around all of them. But at least we built them. Bring back corruption!
RCV is good. There’s something deeply demotivating about being forced to vote for the lesser of two evils year after year. On the other hand, eliminating the spoiler effect inherent to FPTP means allowing more parties to succeed and hence better representation.
I regret to inform you that the spoiler effect is a structural feature of *both* FPTP and RCV: https://en.wikipedia.org/wiki/Spoiler_effect
https://en.wikipedia.org/wiki/Arrow's_impossibility_theorem
The spoiler problem is arguably even worse with RCV because voters don’t know about it; they falsely believe that it’s always advantageous to vote and that it’s safe to rank candidates in the order you prefer them.
In Alaska’s RCV election in 2022, if a bunch of Sarah Palin’s supporters had simply stayed home and not voted, a Republican would have won, but instead they voted and so a Democrat won. What’s more, a majority of voters ranked the moderate Republican *higher* than the Democrat yet the Democrat won. RCV does not work the way that people think it does!
https://en.wikipedia.org/wiki/2022_Alaska's_at-large_congressional_district_special_election
This was an example of the infamous “center squeeze” problem inherent to RCV: https://en.wikipedia.org/wiki/Center_squeeze
Under RCV, the most strategic thing to do is often to vote for the “lesser of two evils” the exact same way you would with FPTP. If instead you rank your real favorite candidate as #1, you may make it more likely that your *least* favorite candidate wins the election. Yikes! https://en.wikipedia.org/wiki/Negative_responsiveness_paradox
(Thank you to the fellow SB commenter who pointed this out a few weeks ago!! I wasn’t fully aware of these issues until then.)
This is all true about what’s possible. But the spoiler effects with IRV are much harder to predict than with FPTP, so they at least are less likely to encourage people to strategically falsify the vote.
And being immune to spoiler effects is logically equivalent to Arrow’s “Independence of Irrelevant Alternatives” condition, which is the one most theorists hold responsible for his impossibility theorem.
"But the spoiler effects with IRV are much harder to predict than with FPTP, so they at least are less likely to encourage people to strategically falsify the vote."
Except it's not actually "falsifying" the vote with FPTP -- you know exactly who you are voting for. With RCV, you as a voter don't really know what happens after the first round because you have no insight into people's second choices.
Bingo! Both FPTP and RCV require you to vote strategically—but FPTP makes it transparent how to do so, whereas it can be much harder to understand the downstream impacts of various voting strategies under RCV.
Under FPTP, what often happens is that one or more lesser candidates will see the polling, realize they have no chance, and eventually drop out and endorse a different candidate. Then the choice at hand starts to become a bit more clear, and people start to coalesce around certain candidates in order to push toward their broader strategic goals.
Think back to when Pete Buttigieg and Amy Klobuchar were doing decently in the polls but probably weren’t going to win, so dropped out and endorsed Joe Biden, and their supporters largely followed suit.
The fact that center squeezes are hard to predict _is not a positive thing_! The FairVote folks say, about instant runoff, that center squeeze scenarios "only" happen in ~1 in 250 races.
There are tens of thousands of distinct elections in America every cycle! They're saying that there will be examples of the center squeeze, all over the country, _every time we run an election_.
Burlington, VT adopted IRV, _immediately_ had a center squeeze for its mayoral race, and the folks on the losing side (backers of both the candidate who got squeezed out, and the people whose second-choice preferences for that candidate _should_ have put them over the top) got together to repeal the whole thing and go back to FPTP.
Mary Peltola's center squeeze win over Nick Begich in Alaska has also led to a ballot initiative to trash their reform, which may well win tomorrow.
Whether strategic voting circumstances being hard to predict is good or bad depends on what you think the problem with them is.
One problem (which doesn’t sound like the one you care about) is that strategic voting circumstances encourage people to write down preferences other than their true ones. If they are hard to identify, then this doesn’t happen very often.
But the problem of a center squeeze is different. This is a problem for a particular normative theory of who should win. If the candidates really are aligned on a single ideological axis, then there’s a good case that the preference of the median voter should carry the day, no matter how few people share that preference. I take it that this is the motivation behind thinking a center squeeze is a bad outcome.
If it’s true that empirically, most elections involve candidates that are best understood as linearly ordered by ideology, then a method that avoids this problem is going to be more valuable.
The problem I see is that a good number of elections aren’t like that - they involve two ideological axes (say, a Clinton/Romney/Trump/Sanders election, which might be satisfied by the median on each axis separately, if that can be identified) or they involve more complex issue spaces, combined with administrative skill ideas.
I don’t think IRV necessarily does well with those things, but I don’t know what methods people are proposing to avoid center squeezes. (FPTP seems like it makes the problem of the center squeezes even worse.)
Center squeeze events _piss off_ the majority of voters, and discredit election reform.
And again, this isn't some theoretical thing. It _was repealed_ in Burlington, and may be repealed in Alaska, tomorrow.
And as far as thinking about a multi-dimensional issue space, IRV sucks at that, too.
http://zesty.ca/voting/sim/
One good example of how RCV is not perfect is the Alaska Senate election in 2022. Ideally, it would be safe for Democrats to vote for the Democrat first, Murkowski second. But, doing that would be dangerous, as it would risk Murkowski getting eliminated first, which would cause the hard-core Trumper to win (e.g. the "center squeeze"). To allow Murkowski to prevail over her right-wing challenger, many Democrats had to be strategic, realize that the Democrat had no chance anyway, and vote Murkowski first to avoid the center squeeze problem.
To be clear, I do think this is still better than first-past-the-post, where the same need to be strategic exists, but is much worse, but it is an illustration that RCV is not perfect.
Please use the term IRV not RCV when referring to elections using that particular non-Condorcet counting method. To me RCV is a general term referring to all ballot systems, single-winner or proportional, which use ranked ballots. Wikipedia seems to be a bit confusing, as it distinguishes the term “Ranked Voting” (the general term) from RCV (which it sometimes conflates with IRV).
Philosophically I agree! Realistically, in the U.S. in the year 2024, RCV is synonymous with IRV; pushing back against that is probably an uphill battle.
In this comment thread, I used RCV because that’s what Matt and everyone else is using, so sticking to the same term helps with communication.
It appears there's an ongoing wikifight on wikipedia's IRV page regarding this terminology among other things. Going down that rabbit hole, I learned today that 1) IRV is generally called RCV in the US but something else in most other places it's used; 2) Google searches for RCV far outnumber those for IRV; 3) It appears to me that most academic political scientists distinguish the terms as I did; 4) People in general in the US seem to have given up the terminology battle as you suggest; 5) Academic research generally supports Matt's "meh" feelings for RCV/IRV.
Yep, that all sounds right to me!
The last point is one that I think is significant: Lots of my friends think that RCV/IRV is the “enlightened idea that educated people know is obviously the solution to fix our elections” but have no idea that lots of actual elections researchers who have done the math are not as enthusiastic.
Thanks, this is all useful information.
I'd love to see more real world examples of this. The issue for me is that it is extremely easy to see how a spoiler in a FPTP system would work, and while I get the "center squeeze" concept, it's hard to get an intuitive feel for how often that is likely to come up in practice.
Basically, my position is still that Ranked Choice systems are still significantly more likely to reflect the will of the voters, even if there is potential for a spoiler effect. But it's not a very strongly held position.
I'm also with Matt in that it seems like RCV advocates are really passionate about something that seems likely to only help at the margins (though margins are still important!) . Still, even with your input it seems like the kind of marginal benefit with close to no cost and so I'm generally in favor, just not particularly worked up about it.
I think often times what people really want is a system that avoids Ralph Nader/Jill Stein like spoilers but without these weird effects and a clear solution is a French style (Or Georgia/Louisiana) Top two Run off. Get 51% the first time, Congratulations you win! No Majority? Single election FPTP between exactly two candidates, no write ins allowed.
St Louis has a great implementation of this in which the first round uses approval voting, where you can vote for as many candidates as you like to advance to the final round, then the top two vote-getters go to the runoff.
https://en.wikipedia.org/wiki/Approval_voting?wprov=sfti1#United_States
This fixes the vote-splitting problem of the top-two-runoff system used in various U.S. states, in which a party with too many candidates can find themselves locked out of the runoff if the opposition has its votes concentrated around just two candidates.
This is the way.
Get rid of primaries while you're doing it, and baby you got a stew going!
The issue is exactly what you noted: The spoiler effect exists with both systems, but with FPTP it’s transparent—and can lead lesser candidates to strategically drop out, or voters to strategically consolidate their vote behind a leading candidate—whereas RCV lulls people into a false sense of complacency, thinking there’s no spoiler effect when in fact there absolutely is.
Another real-world example is the 2009 Burlington mayor’s race: https://en.wikipedia.org/wiki/2009_Burlington_mayoral_election
It’s challenging because once people see big tables of numbers, their eyes start to glaze over and they say “I don’t know… can’t I just trust that things will work out magically?” but unfortunately we can’t.
It shouldn't be the case that a less desirable result occurs *because* you voted, which is the case he describes. We know their preferences because they were ranked!
Exactly! Under RCV, you can have a majority of voters rank Candidate 1 higher than Candidate 2 and yet Candidate 2 wins the election. This happened in Alaska in 2022 and also in the Burlington mayoral election in 2009: https://en.wikipedia.org/wiki/2009_Burlington_mayoral_election
The thing I remember most from my discrete math class is that no voting system is free from Condorcet paradoxes.
Which part = “that”? 🙂
I think we should have a strong predisposition towards what is intuitive. It’s intuitive that someone who voted Palin first and Other Republican second would NOT have preferred the Democrat win, yet that’s the result they contributed to. As others have said, this is all extremely complicated for laypeople and so it is likely to make it harder for them to align their preferences with their actions.
In a traditional primary system, the outcome of the Alaska House race would have been exactly the same. Palin would have beat Begich in the Republican primary, then lost to Peltola in the general, with enough Begich voters who couldn't stomach Palin crossing over, just like they chose Peltola as their 2nd choice in the RCV election.
The problem is not RCV, the problem is extremists candidates who appeal only to their hard-code partisans, but are too radioactive to the center to win in the general.
Most of the time (not all of the time), "lesser of two evils" thinking is just very lazy and unhelpful.
"This person makes different compromises than I would" is not "evil", and uncompromising thinking is a childish refusal to engage with the real world.
In a normal election I would agree with you, but this time Hitler is on the ballot.
This is the third time Mr. "My t-shirt saying I'm not Hitler is raising a lot of questions already answered by the t-shirt" has been on the ballot. It's making me rethink what a normal election is.
"My t-shirt saying I'm not Hitler is raising a lot of questions already answered by the t-shirt"
I lol'd. Well played.
That's been a lowkey underrated meme on the internet emerging in the past couple of years or so.
How would you feel if you voted for (e.g.) the Green Party, then they made "evil" concessions in order to make it into government?
I think people don't realize there's an ineliminable tradeoff. Your system is either majoritarian (in which case the majority has to make internal compromises), or minoritarian (which is obviously terrible).
It's obvious once you say it, but you can't give everyone their favorite government.
The flip side is that in a two party FPTP system, the government in red states is run by a majority of the majority, the people who elect the crazies in the republican primaries, even if a majority of the electorate would elect someone else.
Strongly agree that FPTP + populist candidate selection is bad. But I blame the system of candidate selection!
It's probably too late to abolish primaries, but there are other things we can do to fix the problem. Redistrict to have big, diverse districts. Institute a threshold (primaries aren't binding if below n% of the voters in the state don't participate in the primary). Etc.
As I wrote downthread, what's maintaining the US party primary system now is that the state is paying for & running a hugely complex operation. You probably couldn't actually abolish primaries, but a state could just choose to, like...... not pay for them. "Hardworking taxpayers shouldn't bail out political parties", use that kind of rhetoric with voters.
Without the funding & logistics, state-level parties almost certainly couldn't run primaries anymore. It's the best of both worlds, effectively outlawing them without actually doing so
I wasn't arguing against Matt's proposal in the comment you replied to. I was responding to Stew's critique of FPTP.
He said the US's version of FPTP sucks. I agreed, but I said that's because of other features of our system that weaken party discipline - such as primaries. (I could have also mentioned money in politics.)
Repealing FPTP, to my mind, is like bloodletting. Yes, the patient is sick. But this is probably not going to help.
Same happens in blue states, but with even worse results.
I think the evidence is pretty clear this is not correct. Primary participation at anything lower than the presidential race is really low. Unless you are saying that only the most committed partisans are actually the party, then its clear that the majority of the parties' voters do not vote in many primaries.
1) "Turnout of all eligible voters in 2022 primaries was 21.3%. That compares with 19.9% in 2018, 14.3% in 2014, and 18.3% in 2010."
2) Only 7 states broke 30%.
3) The bottom five states had 12% participation or less.
I'm thinking of places like Texas, would any kind of ranked voting system there elect Paxton as AG?
RCV is better than FPTP, but it's not really "good", by the criterion you specify -- the spoiler effect. All RCV schemes suffer from the spoiler effect except Coombs' Method, which is terrible in other ways. Most RCV schemes make working out spoilers difficult enough that voters may be unable to vote strategically, but they don't ensure voters that they can safely and confidently vote for their most-preferred candidate. They're also complicated for voters to understand, creating confusing ballots and hard-to-follow evaluation processes.
If you're unsure whether the spoiler effect in RCV systems is real, there are many real-world examples. In 2022, Sarah Palin spoiled the election for Nick Begich, giving the victory to Mary Peltola in Alaska's first RCV election. I actually think that was a good outcome, but it wasn't what the voters wanted -- which is causing Alaska voters to demand repeal of RCV to return to FPTP.
But there is a voting system that both completely eliminates the spoiler effect, and is very simple: Approval Voting. Rather than ranking choices, you just vote for all candidates you find acceptable. To determine the winner, officials just count the votes and the candidate with the most votes wins. Everyone can understand it, and it provably eliminates any possibility that voting for your most-preferred candidate could hurt their chances.
The problem with approval voting is that voting for anyone else in addition to you most-prederred candidate can hurt their chances, so it incentivises voters to just pretend it's FPTP, which defeats the entire objective of moving away from FPTP.
No, the objective of moving away from FPTP is to eliminate the need to vote for someone other than your preferred candidate instead of your preferred candidate. That's the sort of strategic voting that creates the need to choose the lesser evil over your true preference. With approval, you may want to strategically cast a vote for the lesser evil in addition to your preferred candidate, but there's never any reason not to vote your true preference, which is the point.
This is different from RCV, where you may actually want to rank the lesser evil above your true preference to avoid giving the election to the greater evil, which perpetuates major-party dominance, and *that* is the point of moving away from FPTP, giving third parties a chance.
Approval voting has the biggest disconnect between academic support and facts on the ground. It's advertised as simple and easy, but nothing could possibly be more confusing or alienating than "come up with your own arbitrary approval threshold". I vote for candidates I don't approve of all the time. Voting theorist need to give this one up, and settle on an RCV method. We all know IRV is suboptimal, but throwing out ranking information is not the solution.
I disagree. I don't think it's at all difficult to come up with your own approval threshold. I can't think of an election I've participated in where I'd have a difficult time at all. Generally, I think I'd end up approving two candidates in each election where I'd vote strategically in a FPTP election: My true preference and the lesser evil major party candidate.
I suppose it could get more difficult in a world where there are many competitive parties, but I think that's a good problem to have, and I think approval is more likely to get us to that many-parties world than any RCV scheme, all of which (except Coombs') fail the no-favorite-betrayal criterion. And I sincerely worry that the complicated winner-determination processes of RCV schemes create too much opportunity for doubt and confusion. Among RCV schemes, my preference is the Schulze method, but it is very complicated.
If complexity isn't seen as a problem, score voting is better than any RCV method. Or perhaps STAR -- it fails more of the theoretical voting system criteria, but simulations suggest that in practice it behaves better. That is, STAR can theoretically exhibit more pathological behaviors, but the structure of the scheme makes such behaviors vanishingly rare.
RCV is the only ballot in which the honest answer is unambiguous. If you have an actual multiparty competitive landscape your proposed algorithm of "actual vote + third party throwaway vote" no longer works. I think worrying about the tabulation complexity is unnecessary. Any method choosing a candidate from the Smith set is basically fine. Going from IRV to Smith/IRV in the future probably wouldn't be too much of a lift. I think concerns about favorite-betrayal are also overblown. The public can just barely grasp strategic voting in FPTP. *Very* few will follow arguments based on anticipated preference cycles
> I think concerns about favorite-betrayal are also overblown.
Tell that to voters in Alaska who are arguing for return for FPTP because favorite betrayal put a Democrat in power.
RCV is only 'good' when considering it as the only alternative to FPTP.
I think there are a few! 🎉
The best I’ve seen is combined approval voting, in which you can rate each candidate as Approve, No opinion, or Disapprove. Each candidate gets a net approval score that’s just the number of Approve votes minus Disapprove votes, and the person with the highest score wins! It’s basically what you see when you see pre-election surveys that sort candidates by net approval, calculating it as what percent of people said they approve of a candidate minus what percent disapprove. It’s basically like the top comment on Reddit; commenters can upvote, downvote, or do nothing, and the comment with the highest upvote-minus-downvote score is the top comment. I think it takes the strengths of traditional approval voting (which is just yes vs. no) but improves upon it by letting you differentiate between candidates you’re okay with and those that you passionately want to block from office.
https://en.wikipedia.org/wiki/Combined_approval_voting
After that, I like the St Louis system, which uses traditional approval voting in an all-parties primary, then the people with the top two number of votes advance to a two-candidate general election.
https://en.wikipedia.org/wiki/Approval_voting?wprov=sfti1#United_States
There’s also Total Vote Runoff, in which you rank candidates like with traditional RCV, but then candidate popularity is calculated not by looking only at first-choice votes but rather by looking at lower-ranked votes as well, giving them less and less mathematical weight. It eliminates virtually all of the math problems with traditional RCV. You can safely rank candidates in the order that you prefer them! The only issue is that the math becomes even harder for a lay person to understand; that may be fine, though, if it leads to outcomes that seem more intuitively understandable.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4328946
Beyond that, I think even just plain approval voting is better than traditional RCV.
Matt discusses one in this very article.
That's an alternative, but not necessarily better.
It's good to have your representative
It largely depends on what outcome you are looking to achieve. It should be noted that it doesn't preclude having a representative.
If all the districts are at large, I don't see how you have YOUR representative.
There is no one specifically responsible for your district.
I assumed Mathew inadvertently posted an incomplete reply.
For single winner elections you have fewer options. Approval voting, Condorcet-type voting, range voting. In that scenario RCV could be best.
Ranked pairs, if you're prepared to bit the bullet and say that any ballot without a full set of preferences is invalid (ie you can't just go 123, you have to number all the way to the bottom)
I'm convinced by whatever they call the system of just letting people vote for whoever they want (e.g., as many candidates as they care for) without ranking. It's easier to understand and doesn't require calculations about who wins that could engender mistrust in the results. But anything but FPTP.
Minneapolis adopted RCV in 2009 and none of these things happened though.
Ed Markey is currently the Dem Senator from MA only because in 1976 he won a House seat with 22% of the vote in a big Dem primary and now we may get to watch him die in office as he suggests he will again run at 80.
He seems like a pretty successful politician that was able to continue getting elected even against a well-funded challengers that was a scion of Americas most famous political family.
Kennedy picked the wrong election to mount a challenge, especially post the "July Crisis"... running against Markey would have much more legs in 2026 than it did in 2020
Yep. That challenge was unbelievably ill-advised.
I’ve always been an opponent of RCV.
We instituted it in Maine after the embarrassing governor’s race in 2010, in which two moderately left-of-center candidates split the vote and allowed the odious Paul LePage win. So it seemed like a good response to that situation (ironically, the governor’s race is one of the few situations where RCV doesn’t apply)
There are a few drawbacks. First, it attracts a ton of also-rans, especially for city council. Our own Vermin Supremes are encouraged by the hope that they’re everyone’s second or third or fourth choice, or possibly soothed by the notion that they can’t be blamed for splitting the vote.
The trouble is that it’s really difficult for voters to tell who is a “joke” candidate, and who has a decent shot at winning and competently governing. There aren’t too many resources, even for a political person like me. You’re stuck either researching all 6 candidates to make your own ranking (~ nobody does this) or just voting for the candidate you’ve heard of and ignoring the rest. In which case, what was the point of RCV?
The other problem is that people don’t understand the concept. I think it was Nate Cohn in 2021 who said he talked to a friend of his in finance, a smart guy, who said he was going to rank the viable candidates first, and then some of his favorite marginal candidates second.
Here in Portland, leftists ran 4 or 5 candidates against one right-winger. On the first ballot, the leftists got a large majority of the vote. But so many of their ballots were exhausted, that ultimately the conservative eked out a victory.
RCV makes sense for those of us who spend all day thinking about politics and game theory, but it is frankly confusing and perverse for most voters. It’s better that parties (and the interest groups that comprise them) should hash things out internally, and put up *one* liberal candidate and *one* conservative candidate in a competitive race. It can be done.
The problem with RCV is that the US doesn't use it enough for people to get used to it. If you're in Ireland or Australia, you can say "normal 123 voting" and everyone understands it and uses it effectively because all elections work that way.
But if 95% of your elections are single-choice and 5% are ranked-choice, then only nerds like us spend the time to work out how ranked-choice works.
Ranking candidates seems deeply intuitive and I don't ever really buy this argument.
You're also well in the 99th percentile in political knowledge, like all of us Slow Borers. I think Richard has it right that for most there's a learning curve to be had.
EDIT: I did not expect to get this much pushback, so I'll just say that I agree with all three of you on the bottom line and upvote your comments.
Ranked choice almost always results in increased turnout and I've seen some small sample size studies that indicate there can be some voter confusion, my bet is this is something that goes away after a few election cycles.
Again, not a silver bullet to fix democracy but a worthy reform.
I actually think it's an unlearning curve! The tactical dynamics of FPTP ("splitting the vote", "wasted votes", etc) are so deeply embedded that voters accustomed to it get stuck on that way of thinking and that's how they get it wrong.
I've spoken to voters in Scotland (their local elections are ranked choice with everything else FPTP) who have worried about splitting the vote by marking a second choice. When I ask "how does that work?", they can't explain because they don't even know why "splitting the vote" is a bad thing - they just know it is and should be avoided.
This is why I think it's not a problem in countries that use ranking for all voting - the tactical dynamics aren't embedded in everyone's way of thinking and therefore the learning curve is no steeper than that of FPTP.
TLDR: The problem isn't that new voters will get confused or have a steeper learning curve: it's that old voters have to unlearn what they've known all their lives and learn something new.
Aside: the good news is that people who vote a lot (and so have a lot to unlearn) and people paying a lot of attention to political communication (and so have access to the information necessary to do that unlearning/learning) are very closely correlated.
Low-info voters generally don't have a problem with ranked-choice because they haven't learned all of that traditional FPTP tactical stuff in the first place.
Liberal democracy had a learning curve, too. Europe had a hundred-year-long religious war before they finally started learning to accept their differences.
I just fundamentally reject the idea that we can't ever ask people to adapt.
Oh, it is: it's just people with FPTP brain who "know" they should only vote for viable candidates and should ignore people who have no realistic chance of winning.
That is, they do an only semi-conscious first-pass filter down to a small number of candidates who "have a realistic chance of winning" and that messes up the ranking process - which is how you get "rank the viable candidates first, and then some of his favorite marginal candidates second" as an idea.
But people with "FPTP brain" are actually correct on this one!
You usually cannot safety put your real favorite candidate first under RCV, as that risks helping your least-favorite candidate get elected. Instead you need to look at the polling/guess what other people will do and then put your favorite of the two most likely winners as your #1. (Unless your favorite candidate has no chance of winning, in which case you can safely rank them as #1—but your #2 better be one of the top two viable candidates.) https://youtu.be/JtKAScORevQ?si=wMYo1ME9w7POzx7Z
"Unless your favorite candidate has no chance of winning, in which case you can safely rank them as #1—but your #2 better be one of the top two viable candidates."
Yes, but isn't that the main sales pitch used for promoting RCV? ("You don't have to vote for the lesser of evils -- you can vote for your preferred third-party candidate and then rank the lesser of evils second.")
Yes and no.
The trouble is that one of the sales pitches for RCV is “There’s a person who’s your real favorite and then someone who you see as the lesser of two evils. With plurality (FPTP) voting, you have to vote for the lesser of two evils, and you never find out if your preferred candidate had a real chance. With RCV, you can rank your real favorite candidate first, and the outcome might be the same—or maybe it’ll turn out that the person is so popular that they do end up getting elected now that people have a risk-free way of voting for them!”
And so the issue is that the “risk-free” part is incorrect if that person *does* actually have a shot at being a non-trivial candidate, as it could accidentally result in a center squeeze where the less-of-two-evils is kicked out, but your favorite loses too, and your least-favorite wins (as happened for Palin voters in Alaska; she was originally polling behind the moderate Republican but then ended up slightly ahead).
You only help your least favorite candidate if there’s a real risk that your middling candidate gets the fewest first place votes. Just because the situation is possible doesn’t mean it’s likely.
That doesn’t seem right.
Going back to the 2022 Alaska example: Sarah Palin’s supporters helped get the Democrat elected thanks to rating Sarah Palin #1. If they had instead ranked Sarah Palin last (and their next choice, the moderate Republican first), then the Democrat would have lost.
I think people have difficulty with things that are new and over time they get used to them and do better at it.
Sounds similar to the experience with roundabouts.
I don’t suppose you have the numbers at your fingertips but I’d be interested to know:
A) how often do people rank their unrealistic fantasy candidate as #1, the major party candidate whom they like better as #2, and the major party candidate whom they don’t like as #3? How often do they rank them in a different order?
B) how many people vote for their fantasy candidate as #1 and then exhaust their ballots?
People in Ireland may be “comfortable” with RCV, but are they using it in a manner that makes it more effective, or gives them outcomes they like, more often than FPTP?
There aren't "major parties" in that sense in Ireland; it's more of a spectrum from "could provide a PM" (Fine Gael, Fianna Fáil, Sinn Fein) to "unrealistic fantasy" (Renua, Workers', etc) with a whole bunch of parties that can form the minority partner in a government in between (Greens, Labour, Social Democrats, etc). And people do vote all over the shop. There certainly are people who exhaust their ballots after their number one choice is eliminated, but they are relatively few.
To pick one example, 2020 General Election, Dublin West (chosen because it's an unusually left-wing seat) https://www.electionsireland.org/counts.cfm?election=2020&cons=112&ref=
The Aontú candidate was eliminated and the SF candidate (the only party they are relatively close to) had already been elected so voters who had voted either just 1 or 12 for those two parties go non-transferable (this is count 5 on the link). There are just 345 non-transferable papers and 1177 transfers, so 77% of voters have a valid preference (be that second, third, fourth or whatever).
For a candidate of a more moderate ilk, the SD candidate eliminated at count 4 only had 83 non-transferables out of 1249 - 93% transfers.
Note that this is multi-winner RCV. If you want examples of single-winner RCV, you need to look at Australian general elections.
You can see it in Australian election results, the major parties pick up ~70% to 80% of the primary vote, i.e., most people ‘vote 1’ for one of the two major parties (labor or the liberal/national coalition).
That seems a little tautological though. If the major parties weren't genuinely widely popular, they wouldn't be the major parties. Most people don't actually have an unrealistic fantasy party they wish were in charge... they think the small parties are kooks and weirdoes organized around some specific ideas too extreme for the major parties. Of course most people vote for them first.
Responding to your point about people not understanding the concept as an Australian.
People will understand the concept when they do it regularly. We have compulsory voting in Australia, and most people understand the concept in our single seat electorates. The process for preferences in our PR elections is more complicated, but simple ‘above the line’ systems mostly solve this problem.
I think most issues that people have with RCV comes from a place of not being able to imagine it working, even though there is extensive evidence that it does work elsewhere.
In saying that, I can’t begin to imagine how much chaos my local Hare-Clark preference system for proportional representation would cause amongst the Stop The Steal types. Understanding the results in a system like that can be challenging for people who want to understand.
https://antonygreen.com.au/how-a-hare-clark-count-works/
I have a somewhat different - and probably somewhat elitist - take. RCV really isn't that complicated. An electorate that can't understand it - and more to the point, a media that can't explain it so that the electorate can understand it - doesn't sound like an electorate that can maintain a stable democracy.
I agree with Matt about proportional representation, but that seems like a lot harder row to hoe - kind of like getting rid of the electoral college.
FPTP breeds its own zero sum culture that prevents us from seeing how the zero-sumness is hurting us.
David, politics is fundamentally zero sum. It's a majority rules enterprise. Every legislature I've ever heard of passes laws with 50%+1. They can all pass a major, controversial, divisive law with 51% in favor and 49% opposed. If you use PR they just form coalition governments, and then the 49% that aren't in the coalition get squashed. I think you're imaging politics is some kind of consensus kumbaya where it takes 75% agreement to pass a law, but it..... just isn't.
Politics is majority rules and that's OK. You can re-arrange the window dressing of the electoral system, but you get the same results any way you shake it up
>>I think you're imaging politics is some kind of consensus kumbaya where it takes 75% agreement to pass a law
You're thinking wrong, then? This is definitely not where I was going with this.
In fact, I'm painfully aware of what you're talking about when you say "politics is fundamentally zero sum". If anything, I actually take that idea MUCH further to its logical conclusion: REALITY is zero-sum. Matter/energy cannot be created nor destroyed. Two particles cannot occupy the same space. Only one chimp can eat any given piece of banana.
So you're ABSOLUTELY correct that politics is fundamentally zero sum. Decisions have to be made by bare majorities! Anything else is the illusion of supermajority, which is actually just minoritarianism by another name (hence my hatred of the filibuster).
However, reality also allows for pockets of positive-sum-ness. The sun's gravity turns it into a fusion engine that emits lotta-yotta-joules of energy per second. The universal arrow of entropy points to higher disorder, but temporary pockets of increased order can perfectly well be established -- this is the foundation of the existence of life itself.
Similarly, although no two humans can eat the same banana, they can cooperate to grow many bananas. This is positive-sumness.
IMO the goal of political systems design should be to set up positive-sum electoral dynamics that mitigate the ambient push of reality towards zero-sum incentives.
That's what I was referring to as "pockets of positive-sum-ness".
I wonder if it's harder for voters to understand outcomes in an RCV system than it is to understand how a candidate with 23.6% of the vote is the "winner" as Matt's post demonstrates.
The electoral college is the only way to do PR for the Presidency.
Wouldn't it be directly proportional if the vote was based on one person, one vote for the whole country instead of filtering through the electoral college?
Well, yes, but how do you get half a president if you got 50% of the vote?
I appreciate the perspective, but I just can't get on board with a numbered ranking of the candidates being confusing and perverse. Anyone who's ever picked a kickball team on a playground can figure this out.
To continue with the sports metaphor, though, it's not very much like picking kickball teammates. It's much more like picking a competitor in the decathlon.
Do you know the particulars of how the decathlon is scored? I don't.
That's the point of having a top 4 primary first before doing RCV (especially if you require a minimum threshold for support to move on), you filter out the marginal candidates.
I’m fine with referring to it as a two part general election.
I generally think that four or five is about right - I prefer five as fifth place seems to be where a good third party candidate can get to, and having two Dems, two GOP and one third party seems a sensible balance for any semi-swingy area (anything from 65-35 to 35-65 or so)
I’m all for libertarianism, but I don’t think I would bend over backwards for small parties at the district level. That’s better incorporated at the national level through something like a mixed member proportional system.
Nate Cohn’s smart friend in finance was correct!
Under RCV, you should *not* rank your true favorite candidate #1 if they may or may not have a chance, as it risks making them a spoiler who gets your least-favorite candidate elected.
https://youtu.be/JtKAScORevQ?si=wMYo1ME9w7POzx7Z
https://en.wikipedia.org/wiki/Negative_responsiveness_paradox
It’s possible that Mr. Finance understood this better than Nate or I do!
But as a practical matter, how would I as a voter decide whom to vote for, being aware of the negative responsiveness paradox? Or should I just pretend it doesn’t exist?
As Matt says, the people who still want to vote like they would for FPTP can still do so: just cast one #1 vote for your favorite candidate and leave the ballot like that. It would help RCV advocates if they pointed that out more often, though.
In my vision for RCV, you must rank exactly four candidates. You cannot rank three, you cannot rank five, you cannot rank one. In order for your ballot to count, it must have exactly four unique candidates ranked.
I don't think that should be that hard too understand. Should be a simple publicity slogan, "Only Ballots with Four Names Ranked Will Be Considered." Then it is up to the voter to have their four names ready when they go to cast.
Appellate courts have ruled that you can't force voters to rank candidates that they don't want to. So that idea is out the window
How do ballots get exhausted? Sounds like if it wasn't RCV the conservative still would have won
This is part of my first complaint; if there weren’t RCV, the race would attract fewer fringe lefty candidates, and force more coordination along party lines. People are exhausting their ballots not because they’re Vermin Supreme-or-bust, but because they don’t realize they can rank him first, Angela Davis second, and Normal Leftie 3rd.
On the voter confusion part, a lot of the discussion here seems to be conflating the simple parts voters do (ranking their choices) and the confusing part of how the votes get counted each round (which voters don't have to know the details of in order to vote).
An understanding of the latter is required in order to strategically do the former in a way that makes your desired outcomes more likely
I will never not be furious at the government winning the RCV referendum (Alternative Vote) here in the UK by claiming that voting yes is stealing incubators from babies and ensuring squaddies won't be able to get bulletproof vests. Unbelievably dirty campaigning and it has lead me to supporting RCV on principle everywhere.
Wha-???
One of the problems in some places is that state laws prevent experimentation with PR. So, the political path indeed HAS to go through smaller reforms that give enough power to smaller parties and moderates that they can then advocate for bigger changes to the party system like PR.
Idaho recently became one of them--they mandated the same old terrible geographically defined seats because they were butthurt over Boise electing the same people from the same neighborhoods. PR would still give roughly the same parties the same political power, but the overlords in the legislature just cannot stand anything that even slightly weakens rural power.
"with only seven out of 43 states meaningfully contested in the presidential election"
"Dear Mr. President, there are too many states nowadays. Please eliminate seven. I am *not* a crackpot!"
"There are only 49 stars on that flag."
"I'll be deep in the cold, cold ground before I recognize Missourah!"
More about proposals for electoral reforms, please.
Noted!
The thing that I would want to see before ditching FPTP is some cross-country empirical evidence about the trade-offs between different voting systems. What actually happens when you implement PR? I think the answer is that, often, you end up with weaker political parties and less voter control over the government.
One good thing about FPTP is that you actually get to vote *for a government*. In PR, you may get to vote for your favorite little party, but then they have to cut deals to form a governing coalition, and you as a voter have no control over this process. (As Anthony Downs pointed out in the '50s.) It's also not a process that always works. Belgium's record is 652 days without a government. Germany's current coalition is schizophrenic - one party wants green regulation, another wants no regulation. Such coalitions are harder for voters to hold accountable. (Who's fault is it?)
Another benefit is that FPTP (more generally, single-member districting) is more fascist-proof. There was substantial support for fascism in early 20th-century Britain as in Germany, but it was much harder for fascists to get a toehold in the UK.
As for tactical voting, I don't see the problem. We can't get rid of it, as shown by the Gibbard-Satterthwaite Theorem. All we can do is look at the evidence to see how things shake out for each system in the end. I can think of some happy cases for RCV, like Australia - but they're a small society with mandatory voting, so maybe not a good test case for the US. For PR, I worry that advocates are thinking about peak-performance Germany and ignoring the less rosy examples like Belgium, Israel, and the Germany of today, with AfD on the rise and anemic growth.
[Edited to remove some snark.]
Always seemed notable to me that most large, wealthy democracies use a majoritarian system and *not* PR. The US, Japan, Britain, France, Canada, Italy, Australia, South Korea, and Taiwan- all majoritarian. PR might work well for a small, ethnically homogenous country like in the Nordics, but once you reach a certain size and scale it would seem that single party control of the legislature is a desirable thing..... In some cases like France and Italy, they've adopted a majoritarian system after disastrous experiments with PR
Those systems still allow for dynamics that prevent calcification into a two-party system by making it viable to become a breakaway faction (mostly; Italy is a REALLY bad outlier).
Most of those systems are “2 major + 1-5 minor” parties. Ours is strictly 2 major parties that never take input from smaller ones and never suffer breakaways.
This is really BAD, because it incentivizes entryism. A faction like the libertarians or wokeists has to make a “long march” through either party’s institutions. A disruptive authoritarian-aspiring outsider like Trump can leverage winner-take-all dynamics to appropriate an entire party for himself.
The same 2 parties win power every single time, with no exceptions ever, in Britain, Canada, Australia, Taiwan, and I think South Korea though I'm less familiar with their system. (And, uh, the same 1 party wins almost every single time in Japan, whatever you want to call that).
The only non-2 party system in the mix there is France, because of Macron/En Marche. (And they use a 2 round system, which I prefer). So if you think having minor parties who can never actually control the government does something, you can make that argument. But almost all of those countries are effectively 2 party systems. There are never any 'breakaway factions' that control the government
In fact, I DO think that their important role isn't in that they "control" the government, but that they discipline their closest allies and the broader system. UKIP dragged the Tories to the right; the LibDems provide a stalking horse to Labour. Small parties don't have to WIN in order to influence the broader system.
Even when that goes in directions I don't like (UKIP), I'd rather have a Tory party that never has an opportunity to rest and attempt to consolidate control over the entire right wing of UK politics. Because THEN they'd be able to force the Labour and LibDems to merge and do the same on the left.
I think some of the same effect is achieved in the US without these politicians being members of small parties. The Green Party exists only to draw a tiny percentage of the presidential vote away from Democrats, and has no other political significance, but the Democrats face left-wing challengers from people registered in their party (e.g. "The Squad") or a quasi-Independent who has run in the Democratic presidential primary (Bernie Sanders).
The insurgents have been even more successful on the Republican side, where Trump has taken over the party, departed drastically from old pro-free-trade attitudes, and (worryingly) made the party more bluntly authoritarian and in thrall to one man with a huge ego and a disdain for the rules.
It's probably an institutional difference -- US parties have little ability to exclude candidates from their party, but parties other than the big two have no real prospect of getting elected to anything larger than a city council or state legislature, so the strategy for less-conventional politics is to form a "wing" of one of the large parties.
I'd be curious, though, how parties in e.g. the UK exclude candidates they don't like. (Do they have much ability to do so? Do candidates rely on the party per se for their campaign funds?)
>I'd be curious, though, how parties in e.g. the UK exclude candidates they don't like
The major 3 parties hold an extensive candidate interview process that involves panel interviews, training and mock speech sessions, they have to do a group presentation to a board, etc. It's fairly rigorous. Yes, the candidates are completely reliant on the party for funds
To file as the candidate of a party, you have to have a certificate from an official of that party saying you are. If the leadership wants you out, then the official doesn't sign that certificate and that's it.
The problem with having consolidated parties with "wings" is that they're still consolidated AF. The competition happens in primaries instead of out in the open. The dissenting view -- and the argument about it -- happens in a primary dominated by extremists whose incentive is to oversell their extreme vision, instead of maybe to make direct appeals to moderate voters with "out there, but reasonable" visions.
I think a case like AOC demonstrates the strength of our system. She entered politics as in effect a member of a small party (like the Greens or DSA) and has since substantially worked hard to become a party stalwart and a supporter of its national candidates and its larger message while obviously being on the lefthand side of its ideological distribution. I think this has strengthened the Democratic party.
Heck, even Rashida Tlaib has called on her supporters to vote the straight Democratic ticket and if anyone has a gripe against the party leadership, it would be her.
See, here's the thing about this discussion... there's a reason why I avoid direct comparisons and individual cases so much, and it's not because I'm just trying to be evasive.
Rather, it's _path_dependency_.
Pretty much everything here is path dependent. Israel's multiparty system went off the rails from the bat because they had an absurdly low threshold, so their parties were able to establish strong brands that endured even after they raised the thresholds.
Likewise... just because AOC and Tlaib moderated over time, doesn't mean we have a healthy system. *AOC* moderated, but The Groups are still a cancer within our party. The primaries are still broken. Neither of the two major parties has any competition besides each other, and the constant internal insurgencies contribute to negative polarization by confirming each side's worst fears of each other.
So while specific examples can help illuminate healthy *dynamics*, I don't like overindexing on any one scenario. The point is that we want to shape better politics *going forward*.
For instance, imagine if we'd had a more open system before Trump's insurgency, and by now had four parties: 20% MAGA, 30% Republicans, 35% Dems, 10% DSA, and 5% Greens. Dems would run by saying "Don't vote for Republicans, they'll have to coalition with MAGA extremists". The Republicans would run away from MAGA and might even make overtures to the Greens. DSA would be carping hard about Palestine, and Dems could position as centrists by clapping back, confident they could force DSA back to the table if Dems won.
This would probably be healthier. Both major parties incentivized to appeal to the center, to reach across the aisle, repudiate their flanks. The flanks forced to make their case to the center, instead of yelling at their friends.
But what I care about is the dynamic, not how you slice and dice the %'s of each party or when in particular you assume the party system was put in place on the scenario's timeline. The scenario doesn't matter, the dynamic does.
Your scenario sounds like a coin flip where Republicans/MAGA have an even chance of winning and the Republican party would be totally dependent on the MAGA party.
Sounds like about what we have now. Except I might be inclined to switch your 30% vs 20% breakdown. Republicans love Trump and MAGA. Paul Ryan's party is dead. I don't think an electoral arrangement would change that.
My point is that we can sit here arguing about specific scenarios all day long, but what actually matters is the dynamics we learn from observing healthier democracies.
I'd say that Belgium's coalition failure is an example of the system working.
When Belgium is gridlocked, the failure is quite wide and open for everyone to see. The bickering becomes front page news.
When America is gridlocked, we call it a fucking Tuesday, and the idiot voters spend three straight decades failing to connect the filibuster (and all of its precious fucking mythology about "compromise") with the dysfunction, let alone any other broader systemic problems. We keep having elections, politicians keep making promises, but nothing happens to break the gridlock, and voters don't know who to punish. A Republican-controlled House grandstands against a Democratic President for 2-4 years, and then the voters vote for fucking divided government all over again because of "compromise".
I agree that America has political dysfunction, but I don't blame FPTP.
My problem with the US - which seems to be what you have in mind, too - is that it's got anti-majoritarian sand in the gears. The filibuster, the presidentialism, the bicameralism - all of this is designed to prevent voters from electing highly effective governments.
Steven Teles calls the result "kludgeocracy." https://www.nationalaffairs.com/publications/detail/kludgeocracy-in-america
I agree that the basic structure is bad; it's like trying to code an AI in DOS.
But to abuse the metaphor, not ONLY do we have that problem; we've also loaded a bunch of shitty first-party software onto it that locks us into their shitty ecosystem.
Compared to the Senate and EC, the filibuster and FPTP are the easiest root kludges to address. If, all of a sudden (again, still within our metaphor), our corporate board decides that we will no longer honor first-party lock-in agreements, then we can install a new OS. But since the original DOS system is so fragile, we're gonna have to boostrap this new OS to, say, Windows 2.1 (IE RCV) in order to grant us the flexibility to start getting this legacy system up to a modern performance standard.
This is why I just can't get too enthused about discussions regarding various RCV or PR reforms. I appreciate there are a lot of polysci minutiae nerds on this site, and I'm usually amongst them on most topics. But on this particular issue it's too muck like rearranging deck chairs on the SS Madisonian. We should employ a Westminster-adjacent model for both the national and state governments. At that point if we want to experiment with PR perhaps via party lists, have at it.
If a crisis ever presents itself and lends to such a wholesale conversion, best believe I’d be happy to endorse it.
But short of that, I think the best option is to do the slow boring work of creating the reforms that will get us on a path towards the wholesale ideal.
What electoral system is OS/2 Warp in this analogy?
Liquid democracy.
Everyone basically gets a proxy vote for various legislative bodies. If enough people select a certain politician as their proxy (say, 10% of the state or district), then that person wields their votes in the chamber. You can change your proxy from time to time or immediately (we'd probably need some sort of hysteresis/emergency recall procedures in place). Perhaps there'd also be a system of conditional proxies -- for instance, I vote for AOC for Senator, but Chuck Schumer keeps my Senate proxy until AOC has enough votes to hit the threshold, upon which time all her votes convert to her and she joins the Senate. Whoever had the most votes on the waiting list for her House seat in turn gets instantly elected, no fuss.
Only on Slow Boring can you ask a question like this and get a serious answer!
The counterpoint is that the antimajoritarian sand is the only thing saving the country at this point because either major party would wreck the whole thing if they were actually able to get their way with a simple majority.
Agreed. Without being too much of a nationalist homer, this is where I think America's very, very unique political system is kind of brilliant. Bicameralism, staggered election cycles, strong federalism, etc. The only country with anything close to it is Australia, where a majoritarian lower house has to face an equally powerful Senate elected by PR
This is like saying we should be helicopter parents because kids are dumb. They are. But they'll never grow up unless they sometimes have to fend for themselves.
If children never have to feel the consequences of their actions because they're constantly being thwarted by parents, they tend to become more reckless and irresponsible.
If parties never have to feel the consequences of their political agendas because they're constantly being thwarted by the system, the same principle applies.
I feel like this suffers from the same sort of false analogy as comparing the US federal budget to a household budget? Most critically, when the kid screws up and "feels the consequences of their actions," it's generally speaking only the kid who actually suffers from it other than some sympathetic emotional pain from the parent. When a political party screws up, it can (and does) ruin the lives of millions upon millions of people, often including large numbers of people who were opposed to the policy change.
Either party would have course-corrected us a long time ago if we didn't have the filibuster short-circuiting the democratic feedback loop and raising the stakes by stagnating our politics.
That's probably true, which is why I qualified my statement with "at this point . . . ." If we imagined this change happening in 1900 or something, I'd be somewhat more sanguine.
"either major party would wreck the whole thing if they were actually able to get their way with a simple majority."
Sigh.
I can't tell if that sigh is sympathetic or critical?
The two parties are not the same.
We've always been at war with Bothsidesism.
I don't think anyone would describe the US government over the last 25 year period as being fantastic, and yet is there any major parliamentary government you think has done better?
Even Belgium, who you describe in their government failure as the system working, would you say they've better run than the US over that time period?
I think the USG is benefiting from (1) sheer size from historical immigration, (2) historical good decisions and policies like the Fed, Social Security, and the creation of growth engines like Silicon Valley, and (3) still being the lynchpin of the liberal world order institutions like Bretton-Woods.
If we didn't have the first one, we'd look more like the UK - a spent power that can't sustain even a diminished global hegemony, and unable to incorporate new waves of immigrants.
If we didn't have the second one, we'd look like South Africa or Latin American banana republics, and just fundamentally be unable to develop our institutions or grow our economy through innovation.
If we didn't have that last one, our debt load would look more like Japan's and we'd be shitting ourselves.
Most of those things were also achieved during an era when our party system more resembled "four factions within two parties", and thus was dynamic enough to roll with the significant punches landed on it (Depression + WWII, etc), even if it could be sluggish or acrimonious at times. We don't have that anymore. We escaped a catastrophic pandemic with the strongest recovery of the entire developed world, and our populace can't tell the difference between that and a long-mounting cost-disease crisis exacerbated by the same stagnation and gridlock that we've allowed to fester for the last several decades.
Basically, I'm positing that our ancestors created a machine so powerful that it can't help but keep firing on all cylinders under just minimally-competent governance, despite all the shitty things we do to kill off the golden goose -- NIMBYism, polarization and stagnation, etc. But it's not going to last forever, and we need reforms that will stop strangling the damned goose before it dies (and I don't mean that solely in some vaguely libertarian sense; I also mean that we need more dynamism in our party system).
Belgium is a small power roughly the size of Maryland; similarly, it's largely dependent on a confederated government with its much larger neighbors, although neither Maryland's nor Belgium's own contributions are anything to laugh at. So, I think that's a more appropriate comparison: Does Maryland's state government perform as well as Belgium's? Do (US) states similar in size to Belgium ever suffer from ~2 years of gridlock due to things like divided government or state-leg filibusters? I'm pretty sure we could dig up plenty of examples.
I concede that the US has enormous advantages and that we often waste them. But I also think if you went to the past, you would be surprised by how many people felt the same way you do now about our institutions. There is no period of our past where governance wasn't filled with significant levels of corruption or malignant influence.
Your question about states compared to Belgium is really interesting. I think its extremely unlikely we will find any state having that level of dysfunction.
Mississippi.
Can you elaborate?
Mississippi has a lot of issues, but is the problem their government's inability to pass legislation due to veto points or some unique structural problem in their state government system?
Many of these problems with PR are fixed by setting a fairly high minimum vote threshold for a party to get seats in the legislature.
Again, I'd want to see empirical evidence.
Germany set the bar at 5%, and look what happened. Perhaps they could have set the bar at 10% or 20%, but then for all we know, voters for small parties would have felt even more alienated than they do under SMD.
PR might work in local elections, but for federal ones I don't think it's compatible with American culture. "Write to your representative" is deeply ingrained here. In PR there would be no single representative for someone to write to, and it'd be much easier for the representative to ignore constituent comments from people outside their base.
So you write to the party you voted for--or even take it to the next level and join that party, and try to influence its actions.
Yeah, I don't particularly like political parties myself, but at some point I had to grow up beyond the hilarious "Don't blame me, I voted for Kodos!" line, and accept that they're essential parts of politics.
Surely you can see how a democracy in which you contact a party representative is not the same as someone directly impacted by your vote? There's an additional abstraction layer that makes them less accountable to it, and party leadership stronger (which you might prefer if you trust them).
I lived in a country with PR and, while the implementation was different than what MY proposed here, the reps there always seemed more interested in intra-party jockying for spots than they do in the US. That's ultimately what determined their reelection chances. Perhaps that experience left me biased against them.
>>and it'd be much easier for the representative to ignore constituent comments from people outside their base
They already do that.
I'd say YMMV but it ultimately depends on how safe their seat is and how partisan the issue is.
Note that Matt doesn’t endorse party list PR - he endorses STV. Under STV it’s natural for some candidates to campaign on local representation, and for some people in an area that particularly needs representation to vote for locals first regardless of party. If enough people do this, then they can guarantee a local representative.
But some people might feel a stronger need to be represented by a co-religionist, or someone who shares a hobby or whatever, and it’s possible those representatives might be able to play a similar role to what a geographic representative does. (Though there is no guarantee that every person has one representative that represents them.)
This reminds me of the scene in what is perhaps Robert Heinlein's best novel, "The Moon Is a Harsh Mistress" in which the Moon, having won its independence from Earth, is debating what type of representative body to create. One of the revolution's leaders, Professor de la Paz, urges them to think more creatively than simple geographical representation:
He says, "This is the traditional way; therefore it should be suspect, considered guilty until proven innocent. . . Surely where a man lives is the least important thing about him. Constituencies might be formed by dividing people by occupation . . . or by age. . . or even alphabetically. . . You might even consider installing the candidates who receive the *least* number of votes; unpopular men may be just the sort to save you from a new tyranny. . . .Suppose instead of election a man were qualified for office by petition signed by four thousand citizens. He would then represent those four thousand affirmatively, with no disgruntled minority. . . Or a man with eight thousand supporters might have two votes in this body . . . "
It's nice to know that if I ever ran for some thankless local position, at least 200 people would arbitrarily vote for me without any campaigning at all.
Think how much better you’d do if your name were some variant of “Ebonee.”
RCV is probably better than the status quo, but Approval Voting is even better. It's simpler, you just mark a vote for all the you think are qualified and approve of, and it also encourages strategic voting less.
Approval Voting actually has a much more central role for “strategic voting” than instant runoff. There is no real meaning to “approve” other than “I think everyone I mark would do better than everyone I don’t”. But if the election is Kamala Harris vs Paul Ryan vs RFK Jr, should I “approve” of Paul Ryan because he won’t break the government, or disapprove because he’s awful? That depends on whether I think there are a large number of people voting for just RFK (in which case I want to support any and all of his opponents) or just Harris.
It becomes more complex in a larger election for something like mayor. In Matt’s San Francisco example, should I approve all three moderates and give no signal about preferring London Breed, or just approve her?
The answer to this is to either use combined approval voting (https://en.wikipedia.org/wiki/Combined_approval_voting?wprov=sfti1#), in which you get 3 choices (approve, neutral, disapprove) or to use the St Louis system in which there’s an approval-voting primary to determine the candidates for a two-candidate general election.
I like that idea of approval selecting a slate for a runoff conducted by other means!
You're right that there is an element of strategy in Approval Voting, but that's because no reasonable voting system can eliminate strategy completely. AV still does much better than RCV. Consider an election with three candidates: Lincoln, Carter, and Hitler. I think Lincoln is great, Carter is fine, and Hitler is awful. Under AV, I have a decision to make whether to approve of Carter or not, and some of that will depend on what I make of his chances vs Lincoln. But even if I mark him as disapprove when really I should approve of him, that's not that different from my sincere preference either way will lead to an OK outcome in the election. In RCV, the temptation is to vote Lincoln > Hitler > Carter rather than my sincere ranking of Lincoln > Carter > Hitler, since the only way to strategically hurt Carter's chances is by helping Hitler's. This is of course both much further from my sincere preference and much more likely to lead to a terrible outcome than either of my votes under AV.
I don’t think IRV ever gives me an incentive to vote L>H>C (at least, I can’t think of how that would happen - as long as I put L first, my other preference can’t do anything until L is eliminated, and in that case there are only two candidates so I have every reason to be honest).
However, it can give me the incentive to put C>L>H, if I worry that Carter might be eliminated first, and his voters might split between Lincoln and Hitler, so I want to keep him in to defeat Hitler in the runoff even though I would prefer Lincoln.
Yeah, that's right, I messed up the example. Hopefully it's still clear that the motivation to vote strategically is at least not worse with AV than RCV.
"and in that case there are only two candidates so I have every reason to be honest"
This is part of the issue I see with RCV/IRV vs. FPTP -- RCV/IRV only really seems to improve things if you have three or more candidates (and much better four or more) who, on a Venn diagram of their preferability, would have substantial overlap between at least two of each of the candidates (better three or more) such that voters actually have a preferred ranking of them. However, probably in most US political races outside of local governments, there are usually only going to be two candidates (in which case, ranking doesn't matter at all) or three or four candidates whether third and fourth candidates (most frequently Libertarian and then Green) and I suspect that the latter two are usually viewed as being so toxic by both Democratic and Republican voters that they either aren't going to bother to rank those candidates at all or will rank them below the main opposing party, in which case again ranking doesn't really matter.
That’s how things go with the current voting system, but with some method other than FPTP in place, you would often get things like the California senate race between Schiff, Porter, Lee, and Garvey, where there are multiple mainstream representatives that differ in meaningful ways that aren’t exactly ideological. And even in the current system, the Green and Libertarian voters often have meaningful preferences between Democrats and Republicans.
agree, outrageous there isn't more discussion of approval voting here. Maybe people don't know it exists. I didn't until quite recently
I've asked a few RCV orgs why they landed on that instead of Approval Voting and I haven't gotten more than a shrug in reply.
RCV apparently has amazing brand loyalty.
It is notable that IRV and approval are the two that are most talked about, and IRV is the one that is most organized and implemented. In the theoretical literature, I think Borda count has a lot of defenders, and the Condorcet criterion does too (though there is less about which particular Condorcet method is best, and all are more complicated to explain than either Borda or IRV).
Approval does have some important difficulties though, in that you need to introduce this hard to motivate threshold of approval, rather than just ranking your preferences. But why IRV won out out of all the methods that count ranked choices, it’s harder to say.
IRV copes best with partial preference lists. Most of the other systems really struggle with voters who don't have a full set of choices.
STL has Top-2 AV for mayor races.
Well, I’m staring at my California ballot now and really wish I didn’t have to vote on things like:
* Airport board
* Hospital board
* Local utility board
I have no idea who any of these people are and all their websites and candidate statements just say stuff about how good at leadership and finding compromise they are and how long they’ve lived in the area. Who cares!
I wish I could vote for “this should not be an elected office”.
Ranked-choice is the best of a bad bunch of solutions to the problem of "how do you do a single-winner election". The real problem is single-winner elections.
If I could make one radical reform to the US, it would be to abolish all single-winner elections.
For offices that have to have a single winner in the end (President, Governor, Mayor, etc), the right approach is to elect (proportionally) an electoral college which sits for the full term of the office, and then have the office-holder chosen by a majority in that electoral college. For cities, that can just be the existing city council, but if you want to have a legislative/executive separation, then have one legislative chamber and one executive one (they can then also approve appointments and be the body responsible for monitoring the performance of the executive). I'd move states to have single-chamber legislatures; they're not internally federal and so don't need a Senate, but this would mean three chambers (two legislative and one executive) at the federal level.
That’s basically parliamentarism.
To add a bit more detail: A lot of Americans get mixed up with parliamentarism and proportional representation - the US has neither.
Certainly, this is somewhat *like* parliamentarism, but making the electoral college into a real thing would still have separation between the executive and legislative branches. Each party would put forward their candidate for President; there would then be an election for the electoral college, which would be proportional by list in each state. They would then gather in DC and if a party had a majority (unlikely, as this would lead to multi-party dynamics), they'd elect their President. If not, they'd have time from mid-December to inauguration on Jan 20 to negotiate a multi-party coalition agreement. The smaller parties wouldn't get the Presidency, but they would probably get concessions on both policy and personnel - e.g. the "Aid to Ukraine Party" gets to put their Presidential candidate as Secretary of State and a commitment to spend significantly on aid to Ukraine in exchange for their support.
And if the new President broke the deals, then the minority parties would have the option of pulling the plug, coming up with a new coalition and picking a new President mid-term.
But also - most states should elect their house members in multi-member districts.
I live in a parliamentary country that has lots of single-winner elections. The US has single-winner elections for every single member of Congress, just about every state legislator, etc.
FWIW I completely agree with your assessment. The way I think of it is that single-winner is a form of zero-sum dynamic. These dynamics breed likewise zero-sum thinking amongst their political culture, which is one of the primary evils of the world. Thus, systems should mitigate the impact of zero-sum dynamics, and replace them with positive-sum ones where possible.
Speaking of radical reforms, it would be interesting to get a mailbag question about, "if you could rewrite the US Constitution from scratch, what would it be?" My initial thoughts might be to have a parliamentary system with fewer veto points, but I also worry about what will constrain bad actors, inept actors or ideological extremists from wrecking things as a result of winning a single election. I wonder how much of good governance depends on good institutional design vs having the right people in charge?
Institutional design affects the incentives that determine what people are in charge. A really obvious example is that the design of the Republican primaries determined that Trump won in 2016 and that a different design would have eliminated him.
You need to think through the behaviours you want to incentivise.
I think that removing the executive/legislative split in the US would be far more radical in practice than any other change short of ending elected government, and for that reason I'm inclined to not do it. That's why I vaguely proposed an executive parliament separate from the legislature (which I call the Electoral College because it makes it seem familiar even though it's a radically different institution)
RCV is good. It provides incentives for candidates and parties to be less shitty to each other, and engages both voters and parties who are alienated by the 2 party structure. I think the long term benefit to our civic life is much larger than the direct impact on a few marginal elections that would have flipped in hindsight.
No one in Australia believes that candidates are incentivized to be 'less shitty to each other'
>engages both voters and parties who are alienated by the 2 party structure
Australia's used RCV for a century, and the same 2 parties get 90-95% of their lower house every single election. They actually have a *more* entrenched 2 party system than Britain, which uses FPTP!
Do you think candidates would be worse or the 2 party system less entrenched if Australia switched to FPTP?
+1000. I've been trying to make this argument for years.
San Francisco has RCV. Mayoral candidates are nasty to each other and it’s an entrenched one party system.
would it be better if it was a FFTP democratic primary which decided the Mayors race?
California eliminated partisan primaries, except for President, in 2010 with Proposition 14.
Matt’s Lisa Murkowski example shows a very common misconception about RCV.
Let’s suppose Alaska has a top-4 RCV election with the following first-choice results:
—MAGA Republican: 35%
—Lisa Murkowski: 32%
—Moderate Dem: 28%
—Progressive Dem: 5%
What happens? The Progressive Dem gets knocked out and their votes go to the Moderate Dem, who now has 33%. Lisa Murkowski then gets knocked out, and if we suppose that her supporters’ second choices are evenly split between Moderate Dem and MAGA Republican then the result is 49% Moderate Dem, 51% MAGA Republican. The MAGA Republican wins! Or if Murkowski’s supporters’ second choices tilted slightly Dem, then the Moderate Dem wins. In neither case does the median voter’s candidate win. 🫠
Vote-splitting and the spoiler effect are major issues in RCV and can lead to unpredictable consequences like above. The system incentives consolidation into two candidates. In fact, one of the Republicans running for Alaska’s House seat smartly dropped out in order to consolidate Republican support and avoid vote-splitting:
https://alaskabeacon.com/2024/08/27/why-are-some-republican-candidates-quitting-alaskas-general-election-strategy/
RCV does not magically push more centrist candidates to win elections. In fact, one of its well-known problems is what’s called the “center squeeze” where centrist candidates get squeezed out and people at the fringes get elected: https://en.wikipedia.org/wiki/Center_squeeze
Alaska itself infamously had this problem two years ago: https://en.wikipedia.org/wiki/2022_Alaska's_at-large_congressional_district_special_election
This is definitely possible. But it’s much harder to predict, so it doesn’t encourage as much strategic voting, unless the candidates are very clearly linearly ordered (and then, only by supporters of the most extreme candidates). And even when this does happen, the winner has significant support.
Alaska’s problem is due to IRV, they should keep ranked voting but use a better method which will select the Condorcet winner if there is one.
And the center squeeze is an even bigger problem with two-party FPTP elections, it’s not due to ranked voting.