Continued use of the truly awful Plurality voting method is the single largest obstacle (yes, even more serious than ballot access) to any other party being able to effectively compete with the two old, declining parties. But no Ranked-Choice Voting (RCV) method works well either. The states of Alaska and Maine switched to IRV (Instant Runoff Voting) in 2016, but there has been no discernable benefit for Libertarians or anyone else. It behooves us to better understand this problem, “get on the same page” and advocate a solution that actually does work
First, it would be good to understand the terminology. There are some 50 or 60 different RCV methods. What people usually mean when they say “RCV” is Instant-Runoff Voting (IRV) which happens to be the worst of the RCV methods. It is now known that Approval Voting (AV) functions somewhat better than the best possible RCV method, but even AV still isn’t very good. Fortunately, other voting methods are now known which work far better than AV, therefore far, far better than any RCV method.
As a physicist and IT professional, I have researched voting methods for the past 14 years. It is a trickier problem than it appears to be at first. I, too, harbored the belief that one of the RCV methods would be best, but that turned out to be just plain wrong.
Plurality restricts voters to identifying just the one candidate they think is best, which is not sufficient data to enable making good choices. In most elections, Plurality strongly motivates voters to vote insincerely (vote for the lesser evil). So, Plurality collects too little info and very often that info is bogus!
IRV is identical to Plurality for two-candidate elections and achieves only a very small improvement for more than two candidates. It allows voters to also say which candidate they think is second (and possibly third) best. But that additional information does not help much with identifying the “correct” winner. The correct winner is the candidate whose election would result in the greatest satisfaction totaled for all voters. IRV is an iterative eliminations method and, incredibly, it ignores the additional information it does collect when deciding which candidate to eliminate next. IRV can easily blunder and eliminate the candidate which should be the winner! However, allowing a second choice does somewhat alleviate the vote-for-the-lesser-evil motivation.
The single data item most helpful to identifying the correct winner is (no surprise) each voter’s choice for the best candidate. The second most helpful single data item is each voter’s choice of the worst candidate. After the best and worst candidates, very few voters have any additional information to provide, and when they do, it improves decision-making very little. Voting methods which empower voters to both vote for candidates they favor and against candidates they oppose can be about 4.8 times better than Plurality and 3.6 times better than IRV in four-candidate elections!
Libertarians have long recognized the importance of empowering voters to say when they don’t like candidates. That’s why NOTA (None Of The Above) is on every ballot. However, NOTA is a “blunt instrument.” It is a vote against all other candidates on the ballot. The surgical precision of indicating the single worst candidate is far more helpful to identifying the correct winner. There is no longer any need to have NOTA on the ballot since it is built into the voting method.
Any voting method which does not allow voters to register dissatisfaction for any candidate is doomed to sometimes commit the horrible blunder of electing a candidate that the majority of voters oppose! That actually happens. Pew Research polling data in 2016 clearly showed that a majority of voters disliked both Trump and Clinton, but Trump was nevertheless elected. The same thing happened in 2020 when Biden won.
Two radically better voting methods were proposed in 2020. The first is AADV (Approve, Approve, Disapprove Voting), which is the improved version of Approval Voting. The second is BAWV (Best, Alternate, Worst Voting), which is the improved version of IRV. Like IRV, BAWV is an iterative elimination method.
These two methods have almost the same (greatly improved) ability to consistently identify the correct winner in all kinds of elections. Since AADV is very simple (simpler than IRV or BAWV), AADV is the preferred method to advocate.
AADV gives voters the option to Approve of either one or two candidates and to Disapprove of one. Each candidate’s disapprovals subtract from its approvals. The candidate having the highest (positive) net approval is the winner. Think of it as conducting a separate yes/no referendum for each candidate. The candidate that wins its referendum by the largest positive amount is the winner.
This complicated subject has been confusing people for 250 years; we have barely scratched the surface in this short article. A comprehensive course on voting methods can be found here: https://home4liberty.org/courses/ Complete details, including the academic papers and data can be found on the “Voting / Elections” page of my website at: http://royminet.org/voting-elections/
The opinions shared here do not necessarily represent the official position of the Libertarian Party. These editorial articles have been submitted by Libertarians across the country, and featuring these topics does not represent an endorsement of the content therein.
The opinions shared here do not necessarily represent the official position of the Libertarian Party. These editorial articles have been submitted by Libertarians across the country, and featuring these topics does not represent an endorsement of the content therein.
AADV sounds like a winner!
Nice summary!
Hells yes!!! I’ve wanted for years a way to vote AGAINST a candidate. One for and one against and let the R and D votes annihilate each other. The only thing I’d change with AADV would be to add a second disapproval to make it AADDV.
Agreed! Ranked Choice (IRV) seems good at first, but falls apart the closer you look at it. It actually tracks very close to plurality voting in terms of accuracy and results. And it’s worse in terms of transparency, auditability, and security. I support STAR Voting, Approval Voting, Score Voting, and other score-based systems.
A fascinating idea! Perhaps, it should be recommended to Rank the Vote & RepresentUs.
https://rankthevote.us
https://represent.us
I’ll send an e-mail suggestion. One moment…
I got a reply from Rank the Vote, and it included these sources.
https://fairvote.org/resources/electoral-systems/#rcv-vs-approval-voting
https://fairvote.org/our-reforms/ranked-choice-voting-information/#_22-how-does-rcv-compare-to-other-alternative-voting-reforms-does-it-matter-which-election-method-is-used
The info seems interesting.
Update: Another reply e-mail from Rank the Vote. From what I understand, Instant-Runoff Voting (I.R.V.) was chosen for greater familiarity & trust in the system, predictability, and calibration to the people’s behaviors. Also, they find it better captures the incentives-induced behaviors of not just voters, but also the candidates, parties, and their personnel. The later-no-harm principle was found more important than Condorcet criterion, though both were considered with merit.
As for RepresentUs, no response as of yet. That’s what I have for now.
Here’s what Steven Brams, an expert on voting systems and a strong supporter of simple Approval Voting, has to say about AADV:
“A 3-tier voting system has been proposed and analyzed over the last 20 years or so by several people. While more expressive than approval voting, it’s also more manipulative.
“For example, if my preference order is D > R > T (third party)–and D and R are in a neck-and-neck race with T out of the running–I would have an incentive not to vote sincerely but definitely cast a negative vote for R, even though I consider R preferable to T, thereby maximizing the distance between D and R. With approval voting, at worst I wouldn’t approve of R. ”
Let’s stick with the simplest Approval Voting, and stop trying to improve on it by making it less than what it is.
In my opinion the first thing Libertarians need to do is get people on the ballot. A winning party always has someone on the ballot for every spot, a perpetual losing party (like the Democrats in Montana) often doesn’t. I think that Georgia has a good system, whoever is going to win has to get over half the votes. The way that benefits the Libertarians is that in the first round of voting the electorate doesn’t have to worry about “wasting” their vote- and so they might be more willing to consider the Libertarian if they know they’ll get another chance, later. According to Ryan Zinke (in 2014) he said Montana is 1/3 Democrat, 1/3 Republican and 1/3 Libertarian. We ought to be able to get 1/3 of the votes- and maybe win an election (for a change…).
And, if the election is close (like in Georgia’s US Senate race in 2020) it gives our state a second look at the options it has and (maybe) cast the deciding vote. I think we should run an online primary with all voters (all parties) eligible to “vote” in our primary and use the internet to get the electorate interested in our candidates. We’re chronically poor and the two major parties are awash in corruption (and money, as the result) and we need to compete in a different way if we want to win.
Imagine what it might be like for a Montana US House Representative to be a Libertarian and throw his hat in the ring to be Speaker promising that he won’t stonewall propositions from either party, and allow debate on issues which can garner a majority of the votes? It would be revolutionary! In my opinion the first thing we need to do is get rid of unconstitutional laws which are on the books and only selectively enforced.
We need a Libertarian with the balls to do it. I would do it, but not alone.
https://www.washingtonpost.com/politics/2023/11/10/third-party-presidential-candidates-2024/?utm_campaign=wp_the_5_minute_fix&utm_medium=email&utm_source=newsletter&wpisrc=nl_fix
You state that no RCV method works well either but don’t explain why it does not work well. IMO this is much preferred voting method compared to what most states have today. Simplicity is the key, from how you describe AADV, I think the general voting public would be loss at the ballot box trying to figure out how to vote properly. From that perspective I think that RCV would be simpler to understand and can be figured out in just a few seconds at the ballot box vs the time it will take to understand the complexity of AADV.
“Approval Voting (AV) functions somewhat better than the best possible RCV method, but even AV still isn’t very good” one of the things we’ve learned over the last several Libertarian conventions is that this is not a true statement. AV is designed with people with good intentions. Voters don’t have good intentions. While there may be many people you wouldn’t hate being in office, Trump supporters would probably rather have someone nearly identical to him like Biden, rather than someone completely different like Jorgenson, but they aren’t going to check the box indicating they approve of Biden as it makes it less likely that their #1 choice, Trump, wins. Its strategic voting. AV is actually worse than the plurality system we have now. While my example may sound silly, for real life examples look at the minutes of 3 of the last 4 LP conventions. The JC is supposed to be selected by AV, but there never seems to be enough JC candidates that are “approved” by a majority of the delegates. As such we’ve ended up with either a JC that is not filed, has fewer elected than there are seats to be filled, or the delegates vote to suspend the convention rules and effectively turn the AV into a plurality election that you are allowed to vote for more candidates than there are seats to be filled (which few delegates do, again this is about “strategy” that is bad enough in an internal Party election that we all are Libertarian for a body that really SHOULD NOT have to do anything most terms, you can only imagine how much worse it would be when you include crazy people like Republicans and Democrats.) This is only one small reason why more and more Libertarians are thining we should just eliminate the Judicial Committee all together.
“AADV gives voters the option to Approve of either one or two candidates and to Disapprove of one.” You clearly have a much higher confidence in the cognitive ability of the voting public than my experience has led me to. There may be other parts fo the world this word work, India, Japan, I don’t know, but in America voters have a hard enough time voting for the candidate that actually want when it’s the same system they’ve always used, but the ballot looks a little different than it did the last time, think Florida 2000 & I have no reason to think voters have gotten smarter since then. Particularly with life expectancies continually increasing, you have more and more stupid old people trying to figure out how to screw the young folks because “they got there’s” and can’t even remember which bathroom to use, let alone how to secretly select a candidate they don’t like. These are all Libertarian pipe dreams. Most western democracies use first-past-the-poll elections just like the States does, and yet they still end up with multi-party democracies. Maybe we need to stop making up voting systems that no one will ever actually use, and fund raise more, educate more, recruit more candidates, and campaign more.
Roy says he wants to make sure he clearly defines what we are talking about but then says, “8 times better than Plurality and 3.6 times better than IRV,” with not only failing to define how this is determined, he isn’t even defining what “better” is. “The correct winner is the candidate whose election would result in the greatest satisfaction totaled for all voters. ” is not a definition of better.
What do you mean by “Western Democracy”. Irland for example uses STV will England is the only one I know that use FFTP? Who else uses FFTP?
My preferred voting system is star or score then automatic run off, where voters give candidates a score. It is also significantly better than AR and RCV (and thus obviously far better than first past the post).
Good article, I just wish it was a little longer to allow more detail 🙂
An example of different outcomes for the same candidates under RCV and AV would help make the case. What we have here are assertions that one is “better” without any clarity on how so.
Some comments:
1. Perfect electoral systems are like perpetual motion machines; they don’t exist, but that doesn’t stop people from trying to invent them.
2. Re: STAR voting (Score Then Automatic Runoff). It violates majority rule (in that a candidate preferred by a majority of the voters to all other candidates can lose) and violates “later no harm” (in that a sincere score for your second favorite candidate can hurt your most favorite candidate). Also, someone who is no one’s favorite can win, which encourages candidates to be bland and not let people know where they really stand on important issues so they can be everyone’s second favorite. RCV has none of those defects.
I served as a co-chair of the LP’s Alternative Voting Committee in 2021, and a core conclusion was that the LP ought to support any and all electoral reforms that address the spoiler effect, such as RCV, Approval, STAR, and particularly Proportional Representation (the only one with a track record of leading to multi-party systems). For more details, see:
tinyurl.com/lpAltVoteRpt
I doubt the LP nor the electoral reform movement need another experimental voting method, as there are plenty that have already proven to be improvements after real-world testing. What we really need is more cooperation and collaboration, as there is way too much infighting between RCV, Approval, and STAR groups (e.g., Seattle). This risks “splitting the vote” against the status quo of plurality and the duopoly, which would be a worst-case scenario for the LP.
A core conclusion of the committee was that the LP should support any and all voting systems that address the spoiler effect. Most any method that allows voters to voice support for multiple candidates should satisfy this, including AADV. I prefer RCV because I think satisfying Later No Harm (LNH) is crucial for a voting system to encourage endorsements that will be important for the LP to grow its influence. Keep in mind that any method that looks at a voter’s entire ballot from the beginning is unlikely to satisfy it (i.e., there is a tradeoff between LNH and expressivity).
I mostly strongly believe that Minet and any other AADV supporters would better serve the LP by campaigning positively for their method instead of attacking other electoral systems that would similarly help the LP. Since it is impossible for any voting system to be perfect, to dwell on the inevitable imperfections would let perfect become the enemy of the good — namely what would be good for the LP.
The science needed to understand voting systems, is mathematics, specifically game theory. Which as branches of math go, is not a hard one. We libs like the freedom of the marketplace, and know that in that competition the best ideas thrive. The Olympics, the Heisman Trophy, and the Academy awards all use Range voting, AKA score voting. All three of these organizations rest their legitimacy and prestige on the quality of their voting systems. The market has spoken. Range voting is the best. Also the mathematicians of game theory that run simulations, also find that Range Voting, performs the best without being excessively complicated. You can see these results at RangeVoting.org
For some people their objection starts because they give five whole seconds to appraising a strategy, and don’t come up with something clear and easy. So here it is.
1) Order your preferences best to worst. First gets max, least gets min.
2) Then consider how the others spread out between, because you are going to allocate the gaps between them. Typically there will be a place on the list where there is a greatest drop off in quality. Assign from your range an estimate on how large this drop off concerns you. If your range is 0-10, and you figure half your preference power should go to preferring the top chunk to the lower chunk, Put a five between those two names.
3) Now you have two lists, and can repeat the process on both. That first choice, leaves how much you have for the other gaps. I think it would be better to express your preference on the high end rather then the low end. Unless that is where the real remaining difference in quality is. Don’t be afraid to bundle groups you consider close as all the same score, so you can better express the gap elsewhere. When you are happy that your gaps match your own appraisal, you are done.
Another method would be to score your political positions. Using larger ranges for the issue you care the most about. Then score all your choices on each issue, and then normalize the totals to final range, reaching across the whole range. (Subtract the lowest score from all, then divide the highest score by the max of your voting range. That is your normalization number that all scores will be divided by. If you did it right at least one will be max, and at least one will be zero.
Range voting (aka Score and Borda) would certainly be an improvement over plurality in addressing the spoiler effect by allowing voters to voice opinions on multiple candidates. Personally, I don’t see it as being as good as RCV, Approval, STAR, or Condorcet because Range is the easiest of these to game: give the worst possible score to the biggest threat to your favorite (i.e., burying). Whether voters rank or score doesn’t affect this issue because it derives from awarding points to candidates. Indeed, the runoff in STAR (Score Then Automatic Runoff) is there to give voters an incentive to score more candidates so as to have a say in the final two. Mathematician Tomas McIntee has more details here:
https://medium.com/basic-voting-theory/the-borda-count-9d4f15b4d20e
Also, there are actually two sciences important to the understanding electoral systems (and thus even more competitors in the field of ideas). The first science is, as Mr. Hall said, voting theory largely within the field of mathematics. The other is electoral systems within the field of political science under the subfield of comparative politics (e.g., Rein Taagepera, Matthew Shugart, Arend Lijphart, Maurice Duverger). There is also voting theory within political science as well (e.g., Kenneth Arrow, Marquis de Condorcet).
One side focuses on the theory, while the other focuses on real-world empirical data from elections. The latter is also key because people — especially groups of people — often behave in very hard-to-predict ways, which is why psychology is a single field, but there are so many social sciences (e.g., political science, economics, sociology, linguistics, anthropology).
And while it is true that the Olympics and the Heisman Trophy uses Range/Score/Borda, the Academy Awards actually uses RCV.
https://www.vanityfair.com/hollywood/2022/03/awards-insider-who-votes-for-the-oscars