I am disappointed that nobody proposed approval voting, which I support. You can read about it at https://electionscience.org/ which is The Center for Election Science (hereafter "The Center"). By letting voters approve of any combination of candidates, candidates are not better or worse off based on how close they are to other candidates. Take California and Washington, which have one primary for all candidates that advance the top two regardless of party. All other things equal, if one party has more candidates, it is harder for those candidates to finish in the top two, and somebody even wrote about if California's 2024 Senate election could have so many more Democratic candidates than Republicans that Democrats could easily win if you add up the votes for all candidates in each party, but Republicans finish in the top two spots because there are so few of them. I am not saying that will happen, and if there was a real risk of that I hope Democrats would drop out, but it is hypothetically possible, and with approval voting Democrats could not lose because there are too many of them. The Center provided an example with people in New York City who voted in the 2012 presidential election. Barack Obama easily won, and Mitt Romney got way more votes than the third place candidate. Later on, a sample of people were allowed to approve of any amount of candidates. Some candidates who got a tiny percent of votes were approved of by more voters than approved of Romney, but those voters voted for Obama and did not have a way of showing they preferred other candidates over Romney. With approval voting, voters never have to choose between their favorite candidate and their preferred candidate who has a chance to win, because voters can approve of more than one candidate. Approval voting might also reduce how much candidates attack each other. In reality, Candidate Y knows that the only votes for him or her are for voters who did not vote for Candidate X (or anybody else). With approval voting, Candidate Y knows that it is possible for voters to approve of him or her and Candidate X, so Candidate Y might give reasons to vote for him or her, rather than attacking Candidate X. The downside of approval voting is that the approvals will add up to greater than 100 percent, which looks strange, but I still support approval voting. Alternatively, the results could be expressed as the amount of approvals without using percentages. Approval voting is used in St. Louis and Fargo, but not any federal or state elections.
I read your articles, and here are my comments. It sounds like BAV gives each candidate the net of their approvals minus their disappovals, divided by the amount of voters to make it be a percent. I would not object to negative numbers, but other people would. With PV or RCV, the percents add up to 100. With AV the percents add up to 100 if they are done as a percent of approvals (which makes 100 percent impossible unless all candidates except the winner got 0 approvals), or add up to greater than 100 if they are done as a percent of voters. With BAV, the percents could be near 0, and they could even be negative.
If all Democrats disapprove of a Republican, and all Republicans disapprove of a Democrat, a third party or Independent (hereafter "TPI") candidate could win by a tiny amount without many approvals or disapprovals, and that TPI could be somebody who would have been the first choice by under 5 percent of voters if they all voted honesty, meaning that TPIs do not get fewer votes because voting for them is a wasted vote. I think the winner should be the first choice of a significant amount of voters. Making up numbers:
Democrat: 48 percent approve, 48 percent disapprove, and 4 percent do neither, net = 0
Republican: 48 percent approve, 48 percent disapprove, and 4 percent do neither, net = 0
TPI: 3 percent approve, 2 percent disapprove, and 95 percent do neither, net = 1, a candidate who was approved of by 3 percent of voters wins
You could make a rule that the winner must have a certain approval percentage, which is kind of like baseball hitters needing 3.1 plate appearances per team game to qualify for rate statistics, but any minimum approval percentage of arbitrary, and makes it more complicated.
I think BAV would antagonize people against it and AV. I think we should hope for AV to be used in several states, and see if voters can adjust from it to BAV. Changing from PV to BAV in one step is too big a change. Changing to BAV might work now in a country that is already using AV and/or has more than two significant parties, but it would not work in the United States now.
Independents in Congress must caucus with Democrats or Republicans to get on committees. Congress and state legislatures operate with a majority party and a minority party, and they might not know how to handle things if enough other candidates won that there was not a majority party. I like when the outcome is determined by voters, not coalitions formed later. The Speaker of the House must get votes from a majority of all House members who vote. If there is no majority party, the plurality party might not get the speaker if the second place party did a better job at attracting TPI representatives. Could we have majority leader, minority leader, and other titles without a majority party?
I have made a similar journey in thinking but turned away from favoring ranked choice voting more than a decade ago. I've documented my journey since then in a series of articles that in some instances analyze the problems of ranked choice voting while other articles examine alternative voting systems. If this interest you, you can find the journal in a series at OpEdNews called "balanced Voting". Another series there called "ranked voting" consists only of the articles that relate to ranked-choice voting.
Thanks for this useful analysis and history. While your focus was on voting process, I do hope that readers took note of your statement that district and assembly size should also be addressed. You said:
“the most significant conclusions from the research suggest that proportional systems and other structural features—district size and assembly size—that support meaningful multiparty representation are best for minority representation.”
The smaller a district is, the harder it is to gerrymander and the more likely that the assembly will contain a multitude of diverse voices more closely representing the diversity of interests in the community. It is, I think, imperative that we increase the size of the US House of Representatives. My personal preference would be for a Cube-Root Rule that set the size of the House at that number computed by taking the cube-root of the US population as counted in each census.
1. You won't get proportional representation at any scale until you first escape the two-party system, which you can only do through a cardinal system like score voting or approval, voting or star voting.
3. There's no evidence that proportional representation is even better than advanced single winner voting methods like approval voting in the first place. We have very robust models showing that approval, voting and other single winner advances can massively improve election representativeness, under essentially any plausible model of human behavior. But we have no way of comparing proportional representation to single winner methods other than plurality voting. Although we may get a small amount of empirical data from the fact that St. Louis recently adopted approval. Voting and Portland recently adopted proportional single transferable vote, and we can maybe parse out some of those effects from all the other noise over the coming decade or so.
Not very surprising; social choice theorists and economists have known that plurality-with-runoffs is a poor system that tends to elect extreme candidates for about 50 years now. Google "Median voter theorem"; instant-runoff isn't subject to it because it's not a Condorcet system, so it frequently fails to elect candidates who have support from a majority of the electorate.
Closely related to this article is the series here on substack called "Unbiased Elections". The article there on "Ranked Voting" takes a critical deep dive into problems with ranked-choice voting and the other two articles explore an alternative approach to ending the two-party duopoly that has let to the toxic polarization we are now experiencing. More recent articles following up on these topics can be found at OpEdNews.com in the series called "Balanced Voting".l
You are wrong on the concept; the concept leads to majority rule.
But you are right that current RCV elections frequently lead to minority rule. But that is due to a simple, yet critical, tabulation error. It has to do with the 50%+1 standard that is implemented. In all too many cases, that 50%+1 standard is formulated on the basis of the number of votes in the so-called "final" round. By way of example, suppose, in that "final" round, say, the fifth round, 80 votes are counted. Of those, Candidate A receives 50%+1 of that number or 41 votes. That candidate is then crowned the winner.
But wait; there's a problem. In the FIRST round, there were 100 votes cast, NOT 80. So an actual majority of all the votes cast would be 51 votes, NOT 41 votes. So, by that latter standard, Candidate A didn't win votes from a majority AT ALL. Instead, they only won the votes of 41% of all the votes cast.
The solution is obvious: The 50%+1 standard should be formulated on the basis of the number of votes cast in the FIRST round and NOT in the so-called "final" round. Using that formulation as the 50%+1 standard, Candidate A is NOT declared the winner in the fifth round. Instead, counting simply proceeds to a sixth round until one of the candidates receives 50%+1, at a minimum, of all the votes cast in the FIRST round. That way the promise of RCV, which is that candidates are elected by a MAJORITY of all the votes cast, is fulfilled, and you never then have the problem of a candidate winning an RCV election with a MINORITY of the votes cast.
You are probably right that many politicians would oppose BAV or for that matter any change that would threaten their current position. But many people, even some successful politicians are concerned about polarization and would prefer there to be more candidates in elections. Some states do promote democracy by allowing for citizen petitions to put issues on the ballot. That is how Maine adopted ranked-choice voting. So voters are not without any power to make changes and sudden changes can be shocking. But we have experienced some shocks in recent elections even without a change in how we vote.
The shock of success by a third-party candidate seems like just what we need to move into a multi-party democracy. And if some voters react by voting against any third-party candidate, others would no doubt react by voting to support third-party candidates and, with BAV, voting against the big-party candidates. A very healthy competition for office with many competitors might be the result.
And as a practical matter adoption of any alternative voting system is apt to begin as experiments with offices that are less powerful, town clerk, or perhaps school boards, with gradual adoption for ever more powerful offices as people grow accustomed to this way of voting. It surely is not going to begin with presidential elections.
Suppose there is an election in which 20.000 voters favored the Republican but opposed the Democrat and suppose that 20,001 favored the Democrat but opposed the Republican. And suppose there is a third candidate, a Green who was supported by 1500 voters but aside from them, no voters had ever heard of. This does not seem a particularly surprising situation.
In a plurality election the Democrat would win, despite being detested by half of the voters.
In an Approval election, the Democrat would win despite being detested by half of the voters.
In a BAV election, the Green candidate would win. Voters might be surprised, but there would fail to be widespread disappointment about their worst fears being realized. And in subsequent elections, voters would be more motivated to learn about minor party candidates.
Even if there would not be widespread disappointment in a minor candidate winning, there would be widespread shock. No state is going to make a voting method that increases the chance of a minor candidate winning. With requirements such as collecting signatures to get on the ballot, one thing Democrats and Republicans agree about is making it harder for anybody else to win. Furthermore, if a minor candidate is on the left like a Green, many Republicans will oppose every Green Party candidate even if they do not know anything about the person other than the person's party. On the opposite side, if a Constitution Party candidate wins, many Democrats will know they are against any candidate from that party.
If BAV was used, candidates could act differently. The Republican would be strongly against the Green. The Republican could tell Republican voters to disapprove of the Green, not just the Democrat. BAV might fool Republicans and Democrats into letting a minor candidate win once, but then Republicans and Democrats would adapt. Minor candidates normally do so poorly that major candidates have no reason to talk about minor candidates or make attack ads, but if major candidates attacked minor candidates, minor candidates would get disapprovals. Let's say voters are 49 percent Democrat, 49 percent Republican, and 2 percent Green. 10 percent of Republicans is 4.9 percent of everybody. If 10 percent of Republicans disapproved of the Green, the Green would get -2.9 percent, and would have 2.45 times as many disapprovals as approvals.
I will be reading and re-reading this in greater depth in the coming days, here are two quick response, which I reserve the opportunity to change my mind. First, I live in Maine and we've had Ranked Choice Voting since 2018 (I believe). Frankly I love RCV. What I have not loved is the resulting election of Rep Jared Golden, who has succeed with RCV on the second or third round instant run-off in three of his four elections. What I detest about Golden is his performative "hippie punching" against the party on whose ticket he runs and that he clearly does not support - among other things. [He's basically become a middle of the party Right Wing Republican, and his Congressional votes demonstrate that.] My short answer to Lee Drutman is: Ranked Choice Voting has not really been tried yet so it is too early to determine what or how this system performs by his criteria of performance. Alaska and Maine, and major cities will need considerable time to show how this system changes our political culture. Neither principal party, but especially the Republicans, have not embraced it (yet?) and do not gear their campaigns or political strategies toward competing in this system. So it is too early to tell. As a voter, however, it place much greater power and opportunity in the voter's hands to direct the outcome of elections in ways that one can embrace and/or accept. I do think a full parliamentary party system would be preferable, and what we partially have (post Newt Gingrich & Karl Rove) in some respects More later, Ted Fletcher
There is an important tabulation error in many RCV elections that I've observed which, if corrected, could materially strengthen the beneficial effect of RCV.
Let me start by pointing to something you said in the above analysis:
"I’ve also been convinced by new analyses that demonstrate that when the electorate is divided, RCV will reflect that division. In a geographically polarized and deeply divided electorate, moderates are unlikely to win more than a few single-winner elections."
There is an important technical reason why this happens more than one would wish in an RCV election. That is because, in many cases, the "majority" standard in the "final" round is 50%+1 of the votes cast in that "final" round.
But that is the wrong standard. Let me illustrate, by way of a very simple example.
100 people vote in an election. No one wins a majority in the first round.
In the second round, as the last-place candidate is eliminated and that candidate's second-choice votes are accordingly re-apportioned, the total number of votes cast in that second round are, say, 91, because of the voters who didn't even bother to cast a vote beyond their first choice. And no one wins a majority in that second round.
And so on. In the "final" round, say, the fifth round, only 80 votes are actually counted and tabulated as more votes are discarded because they didn't name any second- or third-choices, one candidate then emerges with 41 votes, which is 50%+1 of the votes cast in that "final" round, and that candidate is accordingly crowned the "winner."
This is fundamentally undemocratic and unfair. You've ended up with almost exactly the same situation as happens in the traditional first-past-the-post election. And sure, under those circumstances, candidates with support a mile deep but only an inch wide (exaggeration) can win, even in a so-called RCV election.
There is a simple correction that would serve to minimize the number of times an extremist or minority candidate would win an RCV election, and that is to CHANGE the 50%+1 standard by basing it on the number of votes cast in the FIRST round, and NOT on the number of votes cast in the putative "final" round.
To illustrate, and using the example I created above, if one changes the 50%+1 standard to accord with the number of votes in the FIRST round, one ends up with a standard of 50 votes +1, or 51 votes. Once one has established that as the standard, then that fifth-round "winner" in our previous tabulation NO LONGER WINS, because he only received 41 votes. Therefore, one goes on an ADDITIONAL round, eliminating the last-place finisher in the fifth round, apportioning those votes to the next choices ranked, and continuing the counting through a SIXTH round.
The standard to win then becomes higher, and the chances of a minority extremist then becoming elected becomes correspondingly even smaller. Moreover this revision crosses that very important line of being more truly democratic, because the eventual winner of this version of RCV has been approved by a true majority of ALL the votes cast in the election, not just the votes of those in the flawed "final" round of the previous tabulation system.
In fact, this would correct a problem that actually happened in Alaska, where Peltola won the RCV election but with a MINORITY of the total votes cast in the first round. In fact, there were some skeletal claims that if the majority standard in that election had been 50%+1 of the votes cast in the FIRST round, then Peltola might not have even won; instead, her Republican opponent would have won. Regardless of whether or not that is accurate, it is definitely accurate to say that the winner would have been supported by a true majority of all the votes cast in the election.
This improvement could also materially speed up the tabulation process, as follows. You would do a preliminary census to determine which candidates received a total number of votes, whether first-choice, second-choice or nth-choice, of at least 50%+1 of the votes cast in the FIRST round. ANY CANDIDATE THAT DIDN'T MEASURE UP TO THAT STANDARD WOULD IMMEDIATELY BE ELIMINATED, thus ensuring that no candidate who received only a minority of votes could win the election. By eliminating so many candidates up front, the subsequent traditional RCV tabulation would proceed much more quickly, since there would be fewer candidates whose votes needed to be tabulated.
This is not just theoretical. I'm a volunteer political activist, and I've conducted many RCV surveys, along precisely the lines described above, at informal gatherings of fellow activists which I've attended. And my system has worked like a charm. It sounds like a minor technical tweak, but the impact on the results is dramatic and, I submit, ultimately beneficial.
I'm writing from Australia, where we have preferential voting ("ranked choice voting"). I see your point about it being useful in primaries - but we don't have any primaries.
Other than the obvious discouragement of extremism, the other clear benefit of ranked choice is to make viable third party candidacies that simply would not otherwise exist for fear of "wasting" your vote. In this system, your vote can't be wasted, because it just goes to the next preferred candidate. The eventual result is a broader suite of options that more fully represent the electorate (see for example the 'teals', the greens, and conventional independents). I think you might not be able to see that yet so because few of your jurisdictions have had it in place for a general election, and it's not feasible for other options to develop at scale.
While Australian RCV is combined with mandatory voting and independent drawing of electoral boundaries - two clear differences which make it more difficult to isolate the effect of RCV alone - the incentive to be extreme in an RCV system just isn't there like it is in a first-past-the-post system.
Finally, it's a shame that study you cited about Australia is behind a paywall, but to me its conclusions sound like absolute poppycock. While there may be some truth in our having had more extreme politics latterly, this seems a misguided recent import from the US (thinking it would work here) into a jurisdiction where it doesn't fit, and has been severely punished by the electorate since. Extremism is on the nose here, and I doubt your study would have taken these results into account - considering extremism's fashionability is a recent phenomenon. I expect it would have simply said it exists here without exploring how it has been punished.
> the other clear benefit of ranked choice is to make viable third party candidacies that simply would not otherwise exist for fear of "wasting" your vote. In this system, your vote can't be wasted, because it just goes to the next preferred candidate.
This often-repeated claim is not actually true, though. RCV eliminates candidates based only on a tally of first-choice rankings, which can only go to one candidate. So ranking your true favorite as #1 takes a #1 vote away from someone else, and that is the cause of vote-splitting. If you want ranked ballots, you need to count ALL of the rankings of all voters, not just their first choices.
RCV does not do a good job of making third parties viable, either. Australia's House is essentially still a two-party system despite large amounts of support for third parties as evidenced by the STV elections in the Senate. RCV suffers from center-squeeze effect and is biased against moderates and in favor of a polarized two-party system.
I appreciate the engagement, but that's not right.
> "RCV eliminates candidates based only on a tally of first-choice rankings, which can only go to one candidate"
RCV only eliminates the FIRST candidate based on the tally of first-choice rankings, the one with the fewest. After that, preferences are redistributed from the ballots of eliminated candidates, before other candidates are eliminated.
Conceptually, it works like a series of instant runoffs conducted on a single ballot. The candidate with the fewest votes is eliminated. The people who voted for that candidate are then invited to choose from the remaining candidates. That continues until only two candidates remain, and one achieves a majority.
In this system, it doesn't matter if you give your number #1 vote to a smaller or newer party - your vote still counts towards your preferred major party later. The only way your #1 vote "costs" you influence is if your preferred candidate is so successful they surpass others and stay in the count - in which case, they might actually win.
> "RCV does not do a good job of making third parties viable, either. Australia's House is essentially still a two-party system despite large amounts of support for third parties as evidenced by the STV elections in the Senate. RCV suffers from center-squeeze effect and is biased against moderates and in favor of a polarized two-party system."
Australia's House is essentially still a two-party system because people have voted for those two parties (and their predecessors) for 100 years, and those parties command enormous institutional resources. But the success of independents and minor parties, particularly in recent years, from the Greens to the teals, clearly shows that RCV makes such candidacies viable where first-past-the-post systems suppress them.
It's also worth noting that the House system demands concentrated, district-level support, while the Senate (with proportional representation) rewards broader support. That structural difference explains the divergence between the two, not any inherent RCV bias. When minor parties or independents build a strong local base, RCV allows them to win - the Greens in Melbourne being the textbook example.
> RCV only eliminates the FIRST candidate based on the tally of first-choice rankings, the one with the fewest.
Yes.
> After that, preferences are redistributed from the ballots of eliminated candidates, before other candidates are eliminated.
Yes, and candidates at lower rankings are promoted to fill the holes, and in subsequent rounds, candidates are again eliminated based on a tally of only first-choice rankings.
Each round eliminates candidates based on a FPTP tally, which suffers from vote-splitting, just like FPTP.
> Conceptually, it works like a series of instant runoffs conducted on a single ballot. The candidate with the fewest votes is eliminated.
Yes, the candidate with the fewest first-choice rankings is eliminated. But first-choice rankings suffer from vote-splitting, so this is not a democratic way to measure support. It has the same problems as the FPTP system it's supposed to fix.
> That continues until only two candidates remain, and one achieves a majority.
Not a "majority" in any meaningful sense. The candidates who were eliminated may have been preferred by a supermajority over the eventual winner, but were eliminated prematurely due to vote-splitting.
> In this system, it doesn't matter if you give your number #1 vote to a smaller or newer party - your vote still counts towards your preferred major party later.
That's what they claim in the marketing, but it's not actually true. Your honest ranking may cause your preferred major party to be eliminated first, and then your ballot *cannot* transfer to them. Your honest vote has spoiled the election and helped the Greater Evil win, and you should have dishonestly ranked the Lesser Evil first to prevent this scenario, just like under FPTP.
> But the success of independents and minor parties, particularly in recent years, from the Greens to the teals, clearly shows that RCV makes such candidacies viable where first-past-the-post systems suppress them.
Have you looked at the amount of support for third parties on the ballots vs the number of seats they actually win?
As a final observation, in practice RCV actually favours moderates because they are more likely to receive the preferences of multiple camps as others are eliminated. By contrast, more extreme candidates tend to struggle to accumulate those later preferences once their own core support is exhausted.
No, IRV suffers from the center-squeeze effect and is biased *against* moderates. The more "moderate" (representative) candidates there are on the ballot, the more they split the first-choice rankings between them and the more likely they are to be eliminated.
All of the best representatives can be eliminated in turn, leaving only the two least-representative extremists in the final round.
"Thus the single vote method may return the worst of all the candidates ; and although Ware’s method cannot return the worst, it may return the next worst."
The Wikipedia article on Center Squeeze lists a few of the more obvious examples. The same problem affects any plurality-elimination system: Top-Two Runoff, Contingent Vote, Supplementary Vote, Instant-Runoff Voting, etc.
Love this post and your blog, I just wish you (and several others I follow via RSS or email) were using another newsletter platform not ran by problematic owners, like Ghost or Beehiiv.
Also, I’ve long been favorable to party-list PR while seeing RCV as the first of many stepping stones to party-list making an entry here. I wish there was a FairVote-like advocacy group for party-list PR.
But I remember commenting to Steven Hill on that one post he made contrasting RCV to party-list, in which I proposed “spare vote” or preferential party-list PR as a bridge between both which resolves the issue of votes for parties below the percentage threshold for legislative seats, and while this hasn’t yet been implemented anywhere afaik, I still stand by this idea: https://onthethreshold.nz/https://en.wikipedia.org/wiki/Spare_vote
Way way too complicated for the general public. Fancy schmancy nomenclature and yards of explanatory paragraphs distance many/most people. You are creating dead air space. Sadly.
I am disappointed that nobody proposed approval voting, which I support. You can read about it at https://electionscience.org/ which is The Center for Election Science (hereafter "The Center"). By letting voters approve of any combination of candidates, candidates are not better or worse off based on how close they are to other candidates. Take California and Washington, which have one primary for all candidates that advance the top two regardless of party. All other things equal, if one party has more candidates, it is harder for those candidates to finish in the top two, and somebody even wrote about if California's 2024 Senate election could have so many more Democratic candidates than Republicans that Democrats could easily win if you add up the votes for all candidates in each party, but Republicans finish in the top two spots because there are so few of them. I am not saying that will happen, and if there was a real risk of that I hope Democrats would drop out, but it is hypothetically possible, and with approval voting Democrats could not lose because there are too many of them. The Center provided an example with people in New York City who voted in the 2012 presidential election. Barack Obama easily won, and Mitt Romney got way more votes than the third place candidate. Later on, a sample of people were allowed to approve of any amount of candidates. Some candidates who got a tiny percent of votes were approved of by more voters than approved of Romney, but those voters voted for Obama and did not have a way of showing they preferred other candidates over Romney. With approval voting, voters never have to choose between their favorite candidate and their preferred candidate who has a chance to win, because voters can approve of more than one candidate. Approval voting might also reduce how much candidates attack each other. In reality, Candidate Y knows that the only votes for him or her are for voters who did not vote for Candidate X (or anybody else). With approval voting, Candidate Y knows that it is possible for voters to approve of him or her and Candidate X, so Candidate Y might give reasons to vote for him or her, rather than attacking Candidate X. The downside of approval voting is that the approvals will add up to greater than 100 percent, which looks strange, but I still support approval voting. Alternatively, the results could be expressed as the amount of approvals without using percentages. Approval voting is used in St. Louis and Fargo, but not any federal or state elections.
Eric - you may be interested by this debate between Lee and Aaron Hamlin of CES about Ranked Choice v. Approval voting. https://youtu.be/aLBzV0vka98?si=s_iCvAaSlWRUU1L1
Aaron absolutely put Lee through a wood chipper in that debate.
Thank you. I will watch it later.
Approval voting is clearly a much better voting system than ranked-choice voting. I made this point in an article you might want to read: https://www.opednews.com/articles/IRV-Used-to-Rank-Voting-Sy-Ranked-Choice_Ranked-Choice-Voting_Ranked-Voting_Voting-250601-800.html
Still, balanced approval voting is a significant improvement over approval voting. https://www.opednews.com/articles/What-is-so-Special-about-B-Approval_Balanced-Voting_Voting_Voting-Machines-241208-232.html. Balance makes the ballot slightly more cluttered, but it makes voting easier because it allows voters to more accurately express how they feel about the candidates. And it also opens the door much wider for more candidates to compete.
I read your articles, and here are my comments. It sounds like BAV gives each candidate the net of their approvals minus their disappovals, divided by the amount of voters to make it be a percent. I would not object to negative numbers, but other people would. With PV or RCV, the percents add up to 100. With AV the percents add up to 100 if they are done as a percent of approvals (which makes 100 percent impossible unless all candidates except the winner got 0 approvals), or add up to greater than 100 if they are done as a percent of voters. With BAV, the percents could be near 0, and they could even be negative.
If all Democrats disapprove of a Republican, and all Republicans disapprove of a Democrat, a third party or Independent (hereafter "TPI") candidate could win by a tiny amount without many approvals or disapprovals, and that TPI could be somebody who would have been the first choice by under 5 percent of voters if they all voted honesty, meaning that TPIs do not get fewer votes because voting for them is a wasted vote. I think the winner should be the first choice of a significant amount of voters. Making up numbers:
Democrat: 48 percent approve, 48 percent disapprove, and 4 percent do neither, net = 0
Republican: 48 percent approve, 48 percent disapprove, and 4 percent do neither, net = 0
TPI: 3 percent approve, 2 percent disapprove, and 95 percent do neither, net = 1, a candidate who was approved of by 3 percent of voters wins
You could make a rule that the winner must have a certain approval percentage, which is kind of like baseball hitters needing 3.1 plate appearances per team game to qualify for rate statistics, but any minimum approval percentage of arbitrary, and makes it more complicated.
I think BAV would antagonize people against it and AV. I think we should hope for AV to be used in several states, and see if voters can adjust from it to BAV. Changing from PV to BAV in one step is too big a change. Changing to BAV might work now in a country that is already using AV and/or has more than two significant parties, but it would not work in the United States now.
Independents in Congress must caucus with Democrats or Republicans to get on committees. Congress and state legislatures operate with a majority party and a minority party, and they might not know how to handle things if enough other candidates won that there was not a majority party. I like when the outcome is determined by voters, not coalitions formed later. The Speaker of the House must get votes from a majority of all House members who vote. If there is no majority party, the plurality party might not get the speaker if the second place party did a better job at attracting TPI representatives. Could we have majority leader, minority leader, and other titles without a majority party?
The reply I wrote to this somehow did not get positioned correctly. It appears instead as :
https://leedrutman.substack.com/p/how-i-updated-my-views-on-ranked/comment/229454441
Ranked-choice voting has a lot of things wrong with it (see https://www.opednews.com/articles/What-Could-be-Wrong-with-R-by-Paul-Cohen-Election_Ranked-Choice_Ranked-Choice-Voting_Voting-191216-42.html ) and it is unfortunate that so many have joined the bandwagon to support it. But a much simpler voting system can make the two-party system unstable, encouraging more political parties ( https://www.opednews.com/populum/page.php?f=Escaping-Duopoly-Approval_Democrats_Duopoly_Republican-230216-505.html ).
I have made a similar journey in thinking but turned away from favoring ranked choice voting more than a decade ago. I've documented my journey since then in a series of articles that in some instances analyze the problems of ranked choice voting while other articles examine alternative voting systems. If this interest you, you can find the journal in a series at OpEdNews called "balanced Voting". Another series there called "ranked voting" consists only of the articles that relate to ranked-choice voting.
Thanks for this useful analysis and history. While your focus was on voting process, I do hope that readers took note of your statement that district and assembly size should also be addressed. You said:
“the most significant conclusions from the research suggest that proportional systems and other structural features—district size and assembly size—that support meaningful multiparty representation are best for minority representation.”
The smaller a district is, the harder it is to gerrymander and the more likely that the assembly will contain a multitude of diverse voices more closely representing the diversity of interests in the community. It is, I think, imperative that we increase the size of the US House of Representatives. My personal preference would be for a Cube-Root Rule that set the size of the House at that number computed by taking the cube-root of the US population as counted in each census.
This is great, thanks for writing it!!
1. You won't get proportional representation at any scale until you first escape the two-party system, which you can only do through a cardinal system like score voting or approval, voting or star voting.
ScoreVoting.net/PropRep
https://asitoughttobemagazine.com/2010/07/18/score-voting/
2. Fusion voting accomplishes absolutely nothing.
ScoreVoting.net/Fusion
3. There's no evidence that proportional representation is even better than advanced single winner voting methods like approval voting in the first place. We have very robust models showing that approval, voting and other single winner advances can massively improve election representativeness, under essentially any plausible model of human behavior. But we have no way of comparing proportional representation to single winner methods other than plurality voting. Although we may get a small amount of empirical data from the fact that St. Louis recently adopted approval. Voting and Portland recently adopted proportional single transferable vote, and we can maybe parse out some of those effects from all the other noise over the coming decade or so.
https://clayshentrup.medium.com/the-proportional-representation-fallacy-553846a383b3
Not very surprising; social choice theorists and economists have known that plurality-with-runoffs is a poor system that tends to elect extreme candidates for about 50 years now. Google "Median voter theorem"; instant-runoff isn't subject to it because it's not a Condorcet system, so it frequently fails to elect candidates who have support from a majority of the electorate.
Closely related to this article is the series here on substack called "Unbiased Elections". The article there on "Ranked Voting" takes a critical deep dive into problems with ranked-choice voting and the other two articles explore an alternative approach to ending the two-party duopoly that has let to the toxic polarization we are now experiencing. More recent articles following up on these topics can be found at OpEdNews.com in the series called "Balanced Voting".l
RCV is a SCAM. Minority not majority rule.
You are wrong on the concept; the concept leads to majority rule.
But you are right that current RCV elections frequently lead to minority rule. But that is due to a simple, yet critical, tabulation error. It has to do with the 50%+1 standard that is implemented. In all too many cases, that 50%+1 standard is formulated on the basis of the number of votes in the so-called "final" round. By way of example, suppose, in that "final" round, say, the fifth round, 80 votes are counted. Of those, Candidate A receives 50%+1 of that number or 41 votes. That candidate is then crowned the winner.
But wait; there's a problem. In the FIRST round, there were 100 votes cast, NOT 80. So an actual majority of all the votes cast would be 51 votes, NOT 41 votes. So, by that latter standard, Candidate A didn't win votes from a majority AT ALL. Instead, they only won the votes of 41% of all the votes cast.
The solution is obvious: The 50%+1 standard should be formulated on the basis of the number of votes cast in the FIRST round and NOT in the so-called "final" round. Using that formulation as the 50%+1 standard, Candidate A is NOT declared the winner in the fifth round. Instead, counting simply proceeds to a sixth round until one of the candidates receives 50%+1, at a minimum, of all the votes cast in the FIRST round. That way the promise of RCV, which is that candidates are elected by a MAJORITY of all the votes cast, is fulfilled, and you never then have the problem of a candidate winning an RCV election with a MINORITY of the votes cast.
You are probably right that many politicians would oppose BAV or for that matter any change that would threaten their current position. But many people, even some successful politicians are concerned about polarization and would prefer there to be more candidates in elections. Some states do promote democracy by allowing for citizen petitions to put issues on the ballot. That is how Maine adopted ranked-choice voting. So voters are not without any power to make changes and sudden changes can be shocking. But we have experienced some shocks in recent elections even without a change in how we vote.
The shock of success by a third-party candidate seems like just what we need to move into a multi-party democracy. And if some voters react by voting against any third-party candidate, others would no doubt react by voting to support third-party candidates and, with BAV, voting against the big-party candidates. A very healthy competition for office with many competitors might be the result.
And as a practical matter adoption of any alternative voting system is apt to begin as experiments with offices that are less powerful, town clerk, or perhaps school boards, with gradual adoption for ever more powerful offices as people grow accustomed to this way of voting. It surely is not going to begin with presidential elections.
https://www.opednews.com/articles/Familiarity-Breeds-Contemp-Articles-And-Comments_Balanced-Voting_Election-Reform-Activism_Forecast-Election-250907-673.html
Suppose there is an election in which 20.000 voters favored the Republican but opposed the Democrat and suppose that 20,001 favored the Democrat but opposed the Republican. And suppose there is a third candidate, a Green who was supported by 1500 voters but aside from them, no voters had ever heard of. This does not seem a particularly surprising situation.
In a plurality election the Democrat would win, despite being detested by half of the voters.
In an Approval election, the Democrat would win despite being detested by half of the voters.
In a BAV election, the Green candidate would win. Voters might be surprised, but there would fail to be widespread disappointment about their worst fears being realized. And in subsequent elections, voters would be more motivated to learn about minor party candidates.
Even if there would not be widespread disappointment in a minor candidate winning, there would be widespread shock. No state is going to make a voting method that increases the chance of a minor candidate winning. With requirements such as collecting signatures to get on the ballot, one thing Democrats and Republicans agree about is making it harder for anybody else to win. Furthermore, if a minor candidate is on the left like a Green, many Republicans will oppose every Green Party candidate even if they do not know anything about the person other than the person's party. On the opposite side, if a Constitution Party candidate wins, many Democrats will know they are against any candidate from that party.
If BAV was used, candidates could act differently. The Republican would be strongly against the Green. The Republican could tell Republican voters to disapprove of the Green, not just the Democrat. BAV might fool Republicans and Democrats into letting a minor candidate win once, but then Republicans and Democrats would adapt. Minor candidates normally do so poorly that major candidates have no reason to talk about minor candidates or make attack ads, but if major candidates attacked minor candidates, minor candidates would get disapprovals. Let's say voters are 49 percent Democrat, 49 percent Republican, and 2 percent Green. 10 percent of Republicans is 4.9 percent of everybody. If 10 percent of Republicans disapproved of the Green, the Green would get -2.9 percent, and would have 2.45 times as many disapprovals as approvals.
I will be reading and re-reading this in greater depth in the coming days, here are two quick response, which I reserve the opportunity to change my mind. First, I live in Maine and we've had Ranked Choice Voting since 2018 (I believe). Frankly I love RCV. What I have not loved is the resulting election of Rep Jared Golden, who has succeed with RCV on the second or third round instant run-off in three of his four elections. What I detest about Golden is his performative "hippie punching" against the party on whose ticket he runs and that he clearly does not support - among other things. [He's basically become a middle of the party Right Wing Republican, and his Congressional votes demonstrate that.] My short answer to Lee Drutman is: Ranked Choice Voting has not really been tried yet so it is too early to determine what or how this system performs by his criteria of performance. Alaska and Maine, and major cities will need considerable time to show how this system changes our political culture. Neither principal party, but especially the Republicans, have not embraced it (yet?) and do not gear their campaigns or political strategies toward competing in this system. So it is too early to tell. As a voter, however, it place much greater power and opportunity in the voter's hands to direct the outcome of elections in ways that one can embrace and/or accept. I do think a full parliamentary party system would be preferable, and what we partially have (post Newt Gingrich & Karl Rove) in some respects More later, Ted Fletcher
There is an important tabulation error in many RCV elections that I've observed which, if corrected, could materially strengthen the beneficial effect of RCV.
Let me start by pointing to something you said in the above analysis:
"I’ve also been convinced by new analyses that demonstrate that when the electorate is divided, RCV will reflect that division. In a geographically polarized and deeply divided electorate, moderates are unlikely to win more than a few single-winner elections."
There is an important technical reason why this happens more than one would wish in an RCV election. That is because, in many cases, the "majority" standard in the "final" round is 50%+1 of the votes cast in that "final" round.
But that is the wrong standard. Let me illustrate, by way of a very simple example.
100 people vote in an election. No one wins a majority in the first round.
In the second round, as the last-place candidate is eliminated and that candidate's second-choice votes are accordingly re-apportioned, the total number of votes cast in that second round are, say, 91, because of the voters who didn't even bother to cast a vote beyond their first choice. And no one wins a majority in that second round.
And so on. In the "final" round, say, the fifth round, only 80 votes are actually counted and tabulated as more votes are discarded because they didn't name any second- or third-choices, one candidate then emerges with 41 votes, which is 50%+1 of the votes cast in that "final" round, and that candidate is accordingly crowned the "winner."
This is fundamentally undemocratic and unfair. You've ended up with almost exactly the same situation as happens in the traditional first-past-the-post election. And sure, under those circumstances, candidates with support a mile deep but only an inch wide (exaggeration) can win, even in a so-called RCV election.
There is a simple correction that would serve to minimize the number of times an extremist or minority candidate would win an RCV election, and that is to CHANGE the 50%+1 standard by basing it on the number of votes cast in the FIRST round, and NOT on the number of votes cast in the putative "final" round.
To illustrate, and using the example I created above, if one changes the 50%+1 standard to accord with the number of votes in the FIRST round, one ends up with a standard of 50 votes +1, or 51 votes. Once one has established that as the standard, then that fifth-round "winner" in our previous tabulation NO LONGER WINS, because he only received 41 votes. Therefore, one goes on an ADDITIONAL round, eliminating the last-place finisher in the fifth round, apportioning those votes to the next choices ranked, and continuing the counting through a SIXTH round.
The standard to win then becomes higher, and the chances of a minority extremist then becoming elected becomes correspondingly even smaller. Moreover this revision crosses that very important line of being more truly democratic, because the eventual winner of this version of RCV has been approved by a true majority of ALL the votes cast in the election, not just the votes of those in the flawed "final" round of the previous tabulation system.
In fact, this would correct a problem that actually happened in Alaska, where Peltola won the RCV election but with a MINORITY of the total votes cast in the first round. In fact, there were some skeletal claims that if the majority standard in that election had been 50%+1 of the votes cast in the FIRST round, then Peltola might not have even won; instead, her Republican opponent would have won. Regardless of whether or not that is accurate, it is definitely accurate to say that the winner would have been supported by a true majority of all the votes cast in the election.
This improvement could also materially speed up the tabulation process, as follows. You would do a preliminary census to determine which candidates received a total number of votes, whether first-choice, second-choice or nth-choice, of at least 50%+1 of the votes cast in the FIRST round. ANY CANDIDATE THAT DIDN'T MEASURE UP TO THAT STANDARD WOULD IMMEDIATELY BE ELIMINATED, thus ensuring that no candidate who received only a minority of votes could win the election. By eliminating so many candidates up front, the subsequent traditional RCV tabulation would proceed much more quickly, since there would be fewer candidates whose votes needed to be tabulated.
This is not just theoretical. I'm a volunteer political activist, and I've conducted many RCV surveys, along precisely the lines described above, at informal gatherings of fellow activists which I've attended. And my system has worked like a charm. It sounds like a minor technical tweak, but the impact on the results is dramatic and, I submit, ultimately beneficial.
Hi Lee. Thanks for your post, and your honesty.
I'm writing from Australia, where we have preferential voting ("ranked choice voting"). I see your point about it being useful in primaries - but we don't have any primaries.
Other than the obvious discouragement of extremism, the other clear benefit of ranked choice is to make viable third party candidacies that simply would not otherwise exist for fear of "wasting" your vote. In this system, your vote can't be wasted, because it just goes to the next preferred candidate. The eventual result is a broader suite of options that more fully represent the electorate (see for example the 'teals', the greens, and conventional independents). I think you might not be able to see that yet so because few of your jurisdictions have had it in place for a general election, and it's not feasible for other options to develop at scale.
While Australian RCV is combined with mandatory voting and independent drawing of electoral boundaries - two clear differences which make it more difficult to isolate the effect of RCV alone - the incentive to be extreme in an RCV system just isn't there like it is in a first-past-the-post system.
Finally, it's a shame that study you cited about Australia is behind a paywall, but to me its conclusions sound like absolute poppycock. While there may be some truth in our having had more extreme politics latterly, this seems a misguided recent import from the US (thinking it would work here) into a jurisdiction where it doesn't fit, and has been severely punished by the electorate since. Extremism is on the nose here, and I doubt your study would have taken these results into account - considering extremism's fashionability is a recent phenomenon. I expect it would have simply said it exists here without exploring how it has been punished.
> the other clear benefit of ranked choice is to make viable third party candidacies that simply would not otherwise exist for fear of "wasting" your vote. In this system, your vote can't be wasted, because it just goes to the next preferred candidate.
This often-repeated claim is not actually true, though. RCV eliminates candidates based only on a tally of first-choice rankings, which can only go to one candidate. So ranking your true favorite as #1 takes a #1 vote away from someone else, and that is the cause of vote-splitting. If you want ranked ballots, you need to count ALL of the rankings of all voters, not just their first choices.
RCV does not do a good job of making third parties viable, either. Australia's House is essentially still a two-party system despite large amounts of support for third parties as evidenced by the STV elections in the Senate. RCV suffers from center-squeeze effect and is biased against moderates and in favor of a polarized two-party system.
I appreciate the engagement, but that's not right.
> "RCV eliminates candidates based only on a tally of first-choice rankings, which can only go to one candidate"
RCV only eliminates the FIRST candidate based on the tally of first-choice rankings, the one with the fewest. After that, preferences are redistributed from the ballots of eliminated candidates, before other candidates are eliminated.
Conceptually, it works like a series of instant runoffs conducted on a single ballot. The candidate with the fewest votes is eliminated. The people who voted for that candidate are then invited to choose from the remaining candidates. That continues until only two candidates remain, and one achieves a majority.
In this system, it doesn't matter if you give your number #1 vote to a smaller or newer party - your vote still counts towards your preferred major party later. The only way your #1 vote "costs" you influence is if your preferred candidate is so successful they surpass others and stay in the count - in which case, they might actually win.
> "RCV does not do a good job of making third parties viable, either. Australia's House is essentially still a two-party system despite large amounts of support for third parties as evidenced by the STV elections in the Senate. RCV suffers from center-squeeze effect and is biased against moderates and in favor of a polarized two-party system."
Australia's House is essentially still a two-party system because people have voted for those two parties (and their predecessors) for 100 years, and those parties command enormous institutional resources. But the success of independents and minor parties, particularly in recent years, from the Greens to the teals, clearly shows that RCV makes such candidacies viable where first-past-the-post systems suppress them.
It's also worth noting that the House system demands concentrated, district-level support, while the Senate (with proportional representation) rewards broader support. That structural difference explains the divergence between the two, not any inherent RCV bias. When minor parties or independents build a strong local base, RCV allows them to win - the Greens in Melbourne being the textbook example.
> RCV only eliminates the FIRST candidate based on the tally of first-choice rankings, the one with the fewest.
Yes.
> After that, preferences are redistributed from the ballots of eliminated candidates, before other candidates are eliminated.
Yes, and candidates at lower rankings are promoted to fill the holes, and in subsequent rounds, candidates are again eliminated based on a tally of only first-choice rankings.
Each round eliminates candidates based on a FPTP tally, which suffers from vote-splitting, just like FPTP.
> Conceptually, it works like a series of instant runoffs conducted on a single ballot. The candidate with the fewest votes is eliminated.
Yes, the candidate with the fewest first-choice rankings is eliminated. But first-choice rankings suffer from vote-splitting, so this is not a democratic way to measure support. It has the same problems as the FPTP system it's supposed to fix.
> That continues until only two candidates remain, and one achieves a majority.
Not a "majority" in any meaningful sense. The candidates who were eliminated may have been preferred by a supermajority over the eventual winner, but were eliminated prematurely due to vote-splitting.
> In this system, it doesn't matter if you give your number #1 vote to a smaller or newer party - your vote still counts towards your preferred major party later.
That's what they claim in the marketing, but it's not actually true. Your honest ranking may cause your preferred major party to be eliminated first, and then your ballot *cannot* transfer to them. Your honest vote has spoiled the election and helped the Greater Evil win, and you should have dishonestly ranked the Lesser Evil first to prevent this scenario, just like under FPTP.
> But the success of independents and minor parties, particularly in recent years, from the Greens to the teals, clearly shows that RCV makes such candidacies viable where first-past-the-post systems suppress them.
Have you looked at the amount of support for third parties on the ballots vs the number of seats they actually win?
As a final observation, in practice RCV actually favours moderates because they are more likely to receive the preferences of multiple camps as others are eliminated. By contrast, more extreme candidates tend to struggle to accumulate those later preferences once their own core support is exhausted.
No, IRV suffers from the center-squeeze effect and is biased *against* moderates. The more "moderate" (representative) candidates there are on the ballot, the more they split the first-choice rankings between them and the more likely they are to be eliminated.
All of the best representatives can be eliminated in turn, leaving only the two least-representative extremists in the final round.
"Thus the single vote method may return the worst of all the candidates ; and although Ware’s method cannot return the worst, it may return the next worst."
Clearly we disagree. Why not provide an example of an election in which the effects you assert have actually occurred?
The Wikipedia article on Center Squeeze lists a few of the more obvious examples. The same problem affects any plurality-elimination system: Top-Two Runoff, Contingent Vote, Supplementary Vote, Instant-Runoff Voting, etc.
Love this post and your blog, I just wish you (and several others I follow via RSS or email) were using another newsletter platform not ran by problematic owners, like Ghost or Beehiiv.
Also, I’ve long been favorable to party-list PR while seeing RCV as the first of many stepping stones to party-list making an entry here. I wish there was a FairVote-like advocacy group for party-list PR.
But I remember commenting to Steven Hill on that one post he made contrasting RCV to party-list, in which I proposed “spare vote” or preferential party-list PR as a bridge between both which resolves the issue of votes for parties below the percentage threshold for legislative seats, and while this hasn’t yet been implemented anywhere afaik, I still stand by this idea: https://onthethreshold.nz/ https://en.wikipedia.org/wiki/Spare_vote
Way way too complicated for the general public. Fancy schmancy nomenclature and yards of explanatory paragraphs distance many/most people. You are creating dead air space. Sadly.