Single-winner reform and why FairVote supports instant runoff voting
FairVote works on a range of electoral reforms, of course, but there is one where we have been a national leader and been making particular progress: instant runoff voting (IRV). There are opponents of the system, however, so here is a line of argument for why we support IRV, using two current election stories from Albuquerque (NM) and New Jersey as concrete examples.
1. As a preface, every voting system for a single winner office presents apparent contradictions and paradoxes. Nobel Prize winner Kenneth Arrow has a theorem proving how every ranked ballot system presents paradoxes, while every non-ranked voting reform proposal violates the later-no-harm criteria (where indicating support for a lesser choice counts against your top choice, which, as explained in my recent article for Minnesota Public Radio, results in tactical voting and unfair results).
2. As a result of this fact, defenders of the status quo can focus on the paradoxes associated with any given alternative system and make them seem outlandish. But the debate instead should be about comparing paradoxes and seeing which ones are more problematic. Perfection is impossible.
3. The plurality voting status quo is highly problematic, with straightforward problems. Plurality winners in multi-candidate races often are clearly the wrong winners. One example was last week's mayoral election in Albuquerque, New Mexico. The incumbent, a Latino Democrat, faced another Latino Democrat and a white Republican. The Latino candidates together won 56% of the vote, but the Republican won. In a traditional runoff or an instant runoff, the Republican almost certainly would have lost.
Another common problem with plurality voting is apparent in the current New Jersey governor's race. Democratic incumbent Jon Corzine lacks majority support based on his record, while Republican challenger Chris Christie knows that the New Jersey electorate at this point is not majority Republican -- both are stuck at about 40% support. As a result, their campaigns are based primarily on how bad their opponent is. Independent Chris Daggett gained attention in a recent debate and is generally recognized as talking more substantively about issues than his opponents. While relatively strong, however, Daggett's poll numbers at 14% remain below the level that he is considered viable. He's in a catch-22, as it's hard to build support when not polling higher, and a New York Times article on Monday suggested his primary role is likely to be that of a "spoiler." No matter what one thinks of the frontrunners, it would be fairer for Daggett not to be tagged as a spoiler, and to be judged on his merits as a candidate for governor. The major party candidates would be more accountable to voters and more likely to have to present substantive proposals.
4. Both the Albuquerque "wrong winner" problem and the New Jersey "spoiler" problem would be well addressed with either instant runoff voting or traditional runoff elections where winners need 50%-plus-one of the initial vote to win -- with the top two candidates from the first round facing off in the second if no candidate were to win a majority. Either approach would have provided a more accurate result in Albuquerque, while Daggett would be considered on his qualifications in New Jersey. Note that the only alternatives to plurality voting in one-winner governmental offices around the world are two-round runoffs or instant runoff elections.
5.Both runoff elections and instant runoff voting share exactly the same paradoxes highlighted by some critics of instant runoff voting, such as "nonmonotoncity." In both runoffs and IRV, there are scenarios where elimination of a strong opponent would help a favorite candidate or second favorite candidate win instead of a least favorite. And theoretically, switching a vote to an insincere preference could lead to that result. But (and this is critically important) it is extremely hard and counter-intuitive to figure out these scenarios before an election in order to game the election. It is even harder to then engage in manipulative strategic voting and practically impossible with instant runoff voting. With a traditional runoff, a manipulative group of voters could support a weak opponent in the first round and change to their true favorite in the second round of voting, whereas with IRV any such attempt at manipulation is likely to backfire because voters can't change their first choice between rounds of counting.
6. Meanwhile, plurality voting is much worse in very well documented and common scenarios. The Albuquerque elections were an example of a likely undemocratic result where quite possibly the candidate who was the least favorite of a majority of voters was elected. Plurality voting will much more frequently lead to such undemocratic outcomes, and it always will risk putting darkhorse candidates into the spoiler role.
7. In the choice between runoffs and instant runoff voting, runoff elections can have large disparities in voter turnout between rounds, nearly double the cost of what candidates spend to win the seat and cost taxpayers much more money. Those realities are why Americans generally have moved away from traditional runoffs for state and federal general elections (they were used for many congressional elections in the 1800s, for example), just as New York City is seriously talking about doing away with its runoff after its low-turnout, expensive, and highly negative primary runoffs last month.
Instant runoff voting addresses the chief flaws of plurality voting and accomplishes it with voters making one trip to the polls. FairVote is not alone in suggesting IRV for this reason. It is used by dozens of major private associations and is laid out as a recommended alternative by Robert's Rules of Order. It is used by more than 50 American colleges and universities and by several nations for high-level office. The British prime minister just committed to a national referendum on adopting IRV for electing the House of Commons.
8. Some opponents of IRV want to keep plurality voting or runoff elections. Others suggest that people instead go to other systems (examples being approval voting and range voting) that have no track record in governmental elections and hardly any meaningful experience in private elections. But just as backers of range voting and approval voting can try to mock instant runoff voting, their preferred systems can be easily mocked -- and with flaws that have far more serious consequences. One readily apparent feature of approval voting and range voting makes them hard to imagine being adopted in the U.S.: the candidate who would finish dead last in a plurality race can win under these systems - for instance, a candidate could win who would have finished last in a race where under our current rules the results were 51% to 40% to 9%. Another is that a candidate with 45% support can easily defeat a candidate with 55% in a two-candidate race. Furthermore , because indicating support for a lesser choice counts directly against your favorite choice (violating the later-no-harm criteria, as referenced earlier), these systems also lead to immediate incentives to vote insincerely, unlike instant runoff voting where theoretical scenarios are too convoluted to affect voter behavior.
9. In short, if you're looking for something perfect, you're not going to find it. If you're looking for something better than plurality voting and more viable than runoff elections or more arcane alternatives, instant runoff voting indeed should be your first choice.
Comments currently closed for Single-winner reform and why FairVote supports instant runoff voting
-
My blog MakeEveryVoteCount.com proposes a constitutional amendment that would include instant runoff voting in Presidential elections. This blog honors opinions, both pro and con. The problem of "spoilers" in Presidential electins is likely to get worse because of the Citizens United decision. The Electoral College system is more and more in need of reform.Posted by David M. Gilliam, 2011-08-10 05:38:42 (2 years ago)
-
The thing is -- you are talking about a hypothetical situation where my extra point for Brownback helps him beat McCain, a fairly low-probability situation. I'm talking about a REAL election, in Burlington, Vt., where dozens of voters who voted Wright first, Montroll second, and Kiss third found their votes WASTED (and IRV is supposed to eliminate wasted votes!) in the sense that if they had voted for Montroll, their SECOND choice, first, they would have gotten a better result.
Posted by Bruce R. Gilson, 2009-11-07 05:56:37 (4 years ago) -
Why would I give Brownback 2 points? Because Huckabee was REALLY the kind of person who I could not vote for, under ANY circumstances. If Brownback had been nominated against Obama, I still might have voted for him in November. If Huckabee had been nominated, I would have voted Libertarian.
I still point out that in Burlington, NONE of the votes by Wright's supporters for Montroll in 2nd place against Kiss helped their cause. They got their WORST rated candidate, JUST BECAUSE of the "later-no-harm:" aspect of IRV.
So it is absolutely clear that "later-no-harm" is not a worthwhile goal. Rather, I want a voting system that recognizes that I might prefer someone that can't win, but lets me help a BETTER choice against a WORSE one. In Burlington, IRV was a disaster. It defeated the candidate who would have been most acceptable to most voters, because it left the two most extreme candidates standing and eliminated the candidate who was "everyone's second choice." And IRV will do that every time.
Posted by Bruce R. Gilson, 2009-11-06 10:25:18 (4 years ago) -
Rob, you said:
"Here's what I would do and what I think most people would do. Suppose I really wanted Ron Paul to win in the Republican field. I would give him five points. And I wouldn't give anyone else any points. Period. Even if I thought Romney was okay and certainly a lot better than McCain, I would have wanted to give Dr. Paul my best shot. Giving Romney points counts directly against Dr. Paul."
This is equivalent to arguing that plurality voting doesn't suffer from strategic voting, since people won't want to hurt their favorite candidate by voting for a lesser-liked candidate.
That is unrealistic, because voter psychology usually equates to, "don't waste your vote on a candidate who can't win".
For example, in the 2000 election, exit polling showed that 90% of voters who said that Nader was their favorite actually VOTED for someone other than Nader.
So if Romney and McCain were the clear front-runners, and your honest sentiment was Paul=5, Romney=2, McCain=1, it would be in your STRATEGIC best interest to vote Paul=5, Romney=5, McCain=1. If you were really dead set on expressing yourself rather than being strategic, you'd want to vote sincerely. But there's NO reason for you to vote Paul=5, Romney=1, McCain=1.
Posted by Clay Shentrup, 2009-11-05 18:42:25 (4 years ago) -
Bruce - Your analysis is dependent on your perception that Giuliani couldn't win -- so it was a "free vote." But what if you thought he could win? It all would depend if you were voting more positively (e.g., wanting Giuliani) or negatively (e.g., really not wanting Brownback or Huckabee). If you were voting positively, I can't believe you would have also given McCain points.
In your example, I also am surprised you would give Brownback 2 points -- what if those 2 points helped Brownback defeat McCain by a point? Your ballot would have been the key that turned his defeat into a victory - how would you have felt at that point?
Posted by Rob Richie, 2009-10-27 02:26:09 (4 years ago) -
I happen to have been in exactly the situation you describe. As a Republican I REALLY wanted Rudy Giuliani. In Maryland, by the time the primary came about, it was clear that Giuliani was out of it; I actually voted for McCain as the best of the lot. But in a score voting version of that primary, I would have voted Giuliani 10, McCain 9, Romney 6 (because I REALLY didn't want a Brownback or Huckabee!) Brownback 2, Huckabee 1. (I'm leavingout some, but let's restrict it to those 5 candidates.)
In an IRV version about that primary, I would have had to vote 1 Giuliani, 2 McCain, 3 Romney, 4 Brownback, 5 Huckabee. This does not really represent the intensity of my feeling that the latter two are REALLY bad, and the first two REALLY good, with Romney in the middle -- if it got down to Romney vs. Huckabee, yes I'd want Romney, but if it got down to Romney vs. McCain it's McCain for sure.
The thing is, in that primary IRV would have worked all right for me but score voting would have worked even better. On the other hand, if I were a 2009 Burlington, Vt. voter who wanted to defeat Kiss, and would have even taken Montroll in preference to him -- and such appear to have been quite numerous -- I would have been appalled at how that election played out, because I would have been happier for "later-no-harm" to fail (as much as you seem to like this principle). "Later-no-harm" simply means I can't help ANYONE as long as my favorite is in the running (even though he doesn't have a real chance!)
The fact is, IRV prevents compromises and favors extremists, as Burlington showed. The two extreme ends of the political spectrum survived to the end, while the true compromise candidate, who would have made almose everyone happy, got eliminated in the early rounds.
Posted by Bruce R. Gilson, 2009-10-26 09:00:31 (4 years ago) -
Dylan and Clay,
In my post I've already said no system is perfect. So pointing out flaws is no surprise -- of course you can find flaws because you can do that with any system.
You might way "no, that's not true of range/score voting." I of course think you have it very wrong -- that you're reflecting skills in math a lot more than human psychology.
Here's what I wish you would consider doing. Talk to some people who actually are involved in electoral politics as it is. Explain range/score voting in enough detail that they really get it. You ideally would do this with someone who works on campaigns and really knows the territory, but it still would be instructive with ordinary voters -let's say voters who were very invested in the presidential primaries last year and backed candidates who weren't quite in the top tier, but wanted to be - a Dennis Kucinich or Joe Biden on the Democratic side, say, or a Ron Paul or Fred Thompson on the Republican side.
Let's say you find a Republican who fits that bill. You then would say "Okay, imagine the GOP field was having a winner-take-all contest in one state early on -- New Hampshire, perhaps. Imagine you are a New Hampshire voter and could give any candidate between zero points and five points. Whoever has the most points wins. How do you think you're doing to vote?"
Here's what I would do and what I think most people would do. Suppose I really wanted Ron Paul to win in the Republican field. I would give him five points. And I wouldn't give anyone else any points. Period. Even if I thought Romney was okay and certainly a lot better than McCain, I would have wanted to give Dr. Paul my best shot. Giving Romney points counts directly against Dr. Paul.
Now some voters who are more interested in defeating a certain candidate might do something different. In kind of an approval voting way, they would likely max out on points to all the candidates but the one or two they really didn't want to get the nomination.
I'll be very interested to hear what you find out.
Given that range voting has no track record in meaningful elections that allows us to measure it, we of course don't have anything to weigh it against - Warren's simulations aren't real. But one thing I also think should be tested more is my guess that the less-informed voter will screw up more often. With IRV, you're casting as smart a ballot as anyone else if you follow the instructions -- favorite first, second favorite second and so on. (Yes, I know Clay thinks otherwise, but there's no evidence that some voters don't do this even if afterwards you can look at the data and say something like "oh, changing your vote order would have helped your second choice defeat your last choice.")
But with range voting, you go in and read the instructions about how to vote (note that ballot instructions will only do this, typically, not explain the system).. Say you're kind of a cranky guy who would be a tough grader if a teacher at school. Nobody deserves a 5 on a scale of 0 to five. you give one person a 2 and the rest zero, then. But someone who knows their stuff says "always max out on your favorite." That person is getting a real edge.
Finally, I also would be very interested if you took the time to call a local representative on your city council, school board or something and explained the voting system and asked if they could see it working in your community. One thing that has been true is that advocates of IRV have found ways to advance the issue - -and these folks in nearly every case have been volunteers. There's nothing stopping you -- see if you can make an affirmative case for range voting against the status quo.
Posted by Rob Richie, 2009-10-24 16:53:16 (4 years ago) -
Rob,
You say that you and Sean "simply disagree about how range/score voting would work in practice."
What do you mean by that? How do you think score voting will behave?
Let's say we're talking about a Nader supporter who tactically voted for Gore back in 2000. If you had given such a voter the chance to use score voting, which do you think would be a more likely ballot for him to cast:
Gore=10, Nader=10, Bush=0
or
Nader=10, Gore=0, Bush=0
Posted by Clay Shentrup, 2009-10-24 11:39:19 (4 years ago) -
Rob - you seem to be ignoring Sean's criticisms of the vote-counting algorithm at the heart of IRV.
I don't think that it should merely be about "getting wins," it should be about "getting it right." Your dedication and hard work are admirable, but I'm afraid that you've put all your energy into implementing a voting system that does not solve the problem that you say you want to solve.
Like you, I want a voting system that gives me the freedom to vote for my favorite candidate (say, Ralph Nader) without the fear that this honest vote will effectively only serve to help elect my least-favorite candidate (say, George W. Bush). Unfortunately, IRV does not solve this problem. In fact, IRV is arguably worse than the status-quo of plurality, because with IRV I can inadvertently cause my least-favorite candidate to win the election simply by showing up to the polls and ranking my favorite candidate 1st! There are two very clear examples of how this can easily happen in IRV here
http://www.youtube.com/watch?v=fOwDyGCaOFM
and here
http://www.youtube.com/watch?v=8Z3zi6hQ9OI
IRV merely gives the illusion of solving the spoiler problem by allowing voters to rank the candidates and adding a level of complexity to the voting and tallying.
With IRV we will often have an even worse outcome than we would have had with plurality because people are led to believe that the spoiler problem is solved and happily rank their favorite third-party or independent candidate 1st without realizing that they might be inadvertently ensuring a win for their least-favorite of the two main party candidates, by knocking out their second-favorite candidate.
If there is a flaw in the arithmetic of the two examples I provided above, I urge you to point it out. Otherwise, I hope you will watch those videos, consider them well, then think long and hard about whether you can still consider IRV to be "fair."
At the very least, you must admit that IRV does not solve the spoiler problem and stop implying that it does. In reality, IRV exacerbates the problem if the 3rd party candidate is strong enough to just barely knock out the main party candidate ideologically closest to him/her in the first round because then they are likely to not be strong enough to defeat the other main party candidate in the next round.
There are other arguments for why IRV should be used, such as cost-savings or solving the problem of low participation in run-off elections, but I think those arguments are flawed as well. What's more, I think that the vast majority of citizens who want IRV only want it because they mistakenly believe that it would allow them to vote for an independent like Nader without inadvertently electing a Bush. The above examples demonstrate very clearly why that belief is wrong. In fact, rather than helping third parties and independent candidates, IRV hurts them because as soon as they become almost as popular as one of the two main party candidates, the familiar fear of a spoiler returns.
Posted by Dylan Hirsch-Shell, 2009-10-23 17:06:47 (4 years ago) -
Sean - then give it a whirl. There are hundreds of thousands of elected officials in the United States and far more people you can talk to about using these system for private elections. Bashing folks like me who are taking another approach isn't helping you get wins nor earn allies.
Posted by Rob Richie, 2009-10-21 21:57:33 (4 years ago) -
I think that if you guys can get support for a system that fails the participation, consistency, monotonicity, and most importantly, the favorite betrayal criterion, I think we can get support for something that fails the majority criterion.
Besides, if the 45 outweighed the 55, then the 55 clearly didn't care that much, so the winner had more support. Policymakers should be able to see that.
Posted by Sean Walker, 2009-10-21 21:51:43 (4 years ago) -
Sean,
We simply disagree about how range/score voting would work in practice. I also would urge you to try to get out there and persuade policymakers at any level to support it -- that will be a learning experience, I suspect. Once that policymakers realizes that an intensely supported candidate with the support of 45% of voters could defeat a more modestly supported candidate with the support of 55% of voters, you're going to have a challenge on your hands.
Posted by Rob Richie, 2009-10-18 20:41:55 (4 years ago) -
I'll say this right off the bat: I support Range/Score voting. As such, I disagree with you over the idea that IRV is the best. In attempting to present my case, however, I plan to stick to facts and examples, to be as objective as I can.
Starting off with 1: This incorrectly implies that Arrow's theorem says anything about SV. Arrow himself did not believe that cardinal systems were voting methods, and admitted that his theorem did not apply to it. As such, it says no such thing about failing the LNH criterion (although it is true that SV and AV fail it). Furthermore, LNH is one of the few criteria that can be said to possibly be bad to pass, as illustrated in the following by Clay Shentrup:
"One particularly troubling example is his insistence that the later-no-harm criterion should be passed, because systems which violate it will be susceptible to "tactical voting and unfair results". His underlying argument is effectively identical to claiming that (to use the 2000 US Presidential election as an example) Nader supporters who preferred Gore to Bush would worry that voting for Gore might cause Gore to defeat Nader. Obviously that makes no sense, since their practical concern was about tipping the balance between the two front-runners, and not "wasting" their vote on a candidate who couldn't win. That same strategy exists in score voting (aka range voting), approval voting, and Instant Runoff Voting. Thus our real concern should be what happens AFTER the voter has done that.
Plurality voting - Nothing. You get one vote and you're done. Maximum support for your favorite front-runner means you forfeit the right to support your sincere favorite (whenever your real favorite isn't a front-runner, of course).
IRV - Nothing. Strategically top-ranking your favorite front-runner means you effectively betray your sincere favorite candidate. An example is the last mayoral election in Burlington, Vermont. A group of voters who preferred Republican>Democrat>Progressive could have gotten the Democrat instead of the Progressive if they had tactically top-ranked the Democrat, thus betraying the Republican. Eventually they'll figure out how to game IRV, once they come to understand the futility of top-ranking a Republican in such a liberal area. A wise and tactical voter with this order of preferences will understand the value of betraying the Republican in order to get a better result.
Score voting (aka range voting) and approval voting - It's not over! You can continue adding support for all candidates you like BETTER than your favorite front-runner. For instance, say one of those Burlington Republicans strategically gives the Democrat a 10 and gives the Progressive a 0, because he realizes the Republican probably won't win, and he wants to make sure he doesn't get a Progressive. He can then go ahead and give the Republican a 10 as well. He has no reason to fear doing that. If it turns out that some candidate was really liked better than the front-runners, then he can actually win, even if he didn't have the connections and special interest money to create the appearance of being "electable". This is HUGE.
Thus the Favorite Betrayal Criterion is what voters ought to really be concerned with, NOT the later-no-harm criterion. In fact, it is more accurate to say that IRV *fails* the later-harm criterion than to say it *passes* the later-no-harm criterion."
2 I agree with no perfect system, but that does not mean no system close to it.
3 I also agree with, Plurality is horrible.
4 I agree that in the Daggett situations, IRV would likely fix the spoiler effect, and might have fixed the problem with Albuquerque, provided that supporters of one Democrat would then support the other second.
5 You say that the "critically important" point is that voters can't use non-monotonicity to game the system. We say that point is irrelevant. This response is also by Clay Shentrup:
"The problem with non-monotonicity isn't susceptibility to strategic voting. Indeed, strategic exploitation of non-monotonicity could in some cases produce a BETTER outcome. The problem with failing any criterion which is based on consistency, is that it amounts to a proof that the voting method is picking the wrong outcome. Say for instance that in one IRV scenario, X is the winner. Then say you make a second scenario that is exactly like that first one, except that X is ranked higher on some of the ballots, yet X loses. If X really was the right winner in the first scenario, then X must still be the right winner in the second scenario, and so you know with mathematical certainty that IRV has failed. Or vice versa. That is, if X is NOT the best candidate in the second scenario, then X must also not be the best candidate in the first scenario."
6 While I agree that Plurality fails horribly in those situations, and IRV would (probably) fix the problems presented. On the other hand, SV also fixes those problems, and also has no additional problems (ex. IRV exhibits spoiler effect in a race with 3 distinct, roughly tied candidates) like them. http://scorevoting.net/IRV1519.html
7 I agree that runoffs are costly and expensive to conduct, and should be eliminated. Unfortunately, IRV requires new machines, and must be calculated centrally (non additive), both of which increase cost. While this cost is lower than a runoff, SV can be done with plurality machines and is additive, so does not incur these costs.
In the second part, you appeal to the popularity of IRV as an example of it's goodness. This makes no sense. If popularity = goodness of a system, then Plurality must be incredible. Systems should be evaluated on how well the perform, since one may be popular simply because it was there first, or got more support despite being worse.
8 This is where you get really, really wrong. True, SV has never been used in a government election. However, it has been used more privately than every other voting system in the world combined. Look at the Olympics, which uses truncated SV. Look at the internet, with Amazon and Youtube and IMDB and Hotornot, anywhere that asks you to rate a picture, video, product, etc. Or look at nature, where every animal other than Humans uses range voting. Bees and ants, two wildly successful insects, use it for deciding locations.
Not that the incredible number of times of use guarantees that it works perfectly, as I stated above, but you can't claim it's untested.
Another claim about LNH is made, which was addressed above.
A claim is made that it is bad that a candidate with 45% of the first vote can beat one with 55%. This is an appeal to the Majority criterion. This is wrong. If a 45 beats a 55, then clearly the voters for 55 supported the 45. http://www.rangevoting.org/MajCrit.html#pizza Basically, SV and AV elect the consensus winners, not majority winners. Additionally, the Majority criterion is not necessarily important to pass. It is good, but mutually exclusive of a far more important criterion, the IIA. IIA is a major cause of spoilers, and few if any ordinal systems pass it, whereas SV and AV both pass it.
Also, although pathologies in IRV probably can't affect voting behavior, it does show that the system could have elected the wrong winner, as above for monotonicity.
9 Once again, I agree that no system is perfect. If we go in looking at criteria, then SV seems to be the best. This is subjective however, so the best way to pick is to use utility.
Warren Smith, creator of the CRV is the only one to have done a comprehensive test of many different systems, including SV. As he was the only to include SV, he was the only one to find a system that beats every other system in every one of more that 700 combination of honesty and strategy, utility, # of voters, # of candidates, and noise. That system was Range/Score voting, which beat IRV in expected utility even when SV were perfectly strategic (thus using AV), while IRV was honest. This is despite the fact that 75% of voters are honest with range voting, the same which cannot be said for IRV.
The results of this testing is the reason that we support SV. We oppose IRV because once you change a system, they won't want to change again, and because many agents of IRV use misleading or incorrect statements to persuade the public, which we find unacceptable. Unfortunately, you, Rob Richie, are also one of those spreading misinformation. A shame, since we both agree that Plurality voting must go.
Finally, once again, although you believe IRV to be the right option to replace Plurality, SV is better in effectively every metric, and should be the real goal of reform.
Posted by Sean Walker, 2009-10-17 16:11:18 (4 years ago) -
Thanks, Louis, appreciate you stopping by the site. I simply disagree about how range and approval voting would work -- and the limited experience with them (like the Dartmouth alumni elections, where approval was repealed by 82% to 18% this year by a vote of the alumni)supports my critique about the damaging reality of violating the later-no-harm criteria.
Meanwhile, I stand by IRV being better than runoffs and IRV for the reasons articulated in the article. Your concern about it I assume would be the same with runoffs?
Posted by Rob Richie, 2009-10-14 17:14:17 (4 years ago) -
One straightforward outcome of approval voting and range voting makes them hard to win in the U.S.: the candidate who would finish dead last in a plurality race can win.
Yes, precisely. Because Plurality means that voters have to use strategy to cast their votes.
This criterion is meaningless: what Range (AKA Score) Voting does better than any other method is to find the candidate who minimizes the variance. In other words, RV most often chooses the candidate closest to centroid of the population.
IRV tends to find the candidate who is most satisfactory to the largest faction.
I have no objection to using Single Transferable Vote for multi-winner legislative elections, in fact I favor any method that can increase the diversity of opinion that comes to bear when proposing new laws.
But for single-winner elections, I think we want a method that best approximates a coalescence of independent, un-hedged opinion. IRV does not do that.
BTW, non-monotonicity is not even the strongest argument against IRV. You're just finding a straw man and appealing to ridicule. An even worse problem is that IRV is unstable in close races, and it doesn't scale to large elections.
Posted by Louis Stern, 2009-10-14 16:08:10 (4 years ago)