All the votes are apparently in for the 3rd District Congressional seat (oh, I mean the Dem nominee, but they are the huge favorite to win the general as we all assume). And the race will face a recount win with only 52 votes separating the first and second place finishers. And the two top finishers each got less than 22% of the vote…hardly a strong endorsement for either.
Wow, what a way to run a Democracy! We’ll spend a couple of weeks determining the winner and the ultimate winner will have barely gotten the support of 1 in 5 voters in the primary.
Rank choice voting would address these two flaws in the current system, by quickly breaking this “tie vote” and declaring one candidate to have more than 50% of voters having a strong preference for the winner.
For those not familiar with RCV and how it would break the tie and give voters a “majority winner”, here’s what would have happened:
Voters in the 3rd would have voted for their first choice and then rank by preference their support for the other 9 candidates (or however many they want to rank).
The winning candidate must get to 50% support. So in the real case of the 3rd, the 582 votes that last place finisher Leonard Golder received would be reallocated to the candidate each voter designated as their 2nd choice. This wouldn’t get anyone to 50%, so the 1800 votes that 2nd to last place finisher got, would be reallocated based on those voters 2nd choice. This process would continue until a candidate wins more than 50% of the vote (and it is entirely possible that the third place candidate in this scenario, is the ultimate winner with more than 50%).
So with rank choice voting, there is no need for a recount, instead we get an automatic runoff and the winner will have had more than 50% of the voters having a positive preference for the winner.
But this all makes to much sense and we’ll probably live with a system in which the “winner” gets 22% of the vote. Gee, and why are people alienated from politics?
Counterpoint: this race is a good example of a situation where ranked choice voting wouldn’t work well. The basic problem is you’re asking people not to find a single candidate they can live with, but to rank 9 or 10 candidates. That’s simply beyond most people, including denizens of this site (quick, name them all without googling).
Even if you can remember all the names, you have to learn and rate them on their positions that matter to you. I certainly cannot rank all of the candidates in this race in order of their support for single payer health care, or Israel, or defunding ICE, or impeachment, or anything else. Good people I know in the 3rd, people who pay a lot of attention, testified how their perceptions of candidates’ physical appearance or the sound of their voice were narrowing their field, helping them make their ultimate choices.
Even if you know you love Rufus Gifford and hate Dan Koh, can you really order small attention-share candidates like Das and Golder in between them? And if you’re picking candidates because you like their name or the fact that their Dad was yelling about them in a 4th of July parade, are you feeling a real sense of attachment if one of them happens to win?
I suggest that alienation has more to do with elected officials not being responsive to their community, not the circumstances under which they were elected. If that’s the case, the answer is not to make ballots more intricate, but to nominate and vote for people who will represent their constituents instead of their donors.
I don’t agree, and ranked voting does not work that way. If Hitler is on the ballot you have no obligation to cast any vote for him at all.
You vote for as many as you want, in order; your vote is not invalidated if you only vote for 9 or 4 or even 1. Then the process eliminates the candidate that received the fewest votes, repeated until someone gets a majority.
The advantage is not that you feel an emotional investment in the winner (which you might not if your candidate loses), but that you get to cast a vote that reflects your true conviction and dispense with all that “lesser of 2 evils” strategic voting crap.
That is a better process even if it produces the same result.
Point taken that you don’t have to rank everyone. If you happen to only vote for one candidate who gets knocked out your ballot counts as a blank? So the winner is the first to a majority of the active ballots instead of all ballots cast?
But I don’t understand the argument that people aren’t voting their true conviction, especially in a primary. If someone chooses to vote “strategically” instead of voting their heart, why is that a problem with the process?
If I have to vote for Nixon instead of Gandhi out of fear that Hitler won’t win, that’s a problem. It handicaps Gandhi and boosts Nixon
I think it feeds cynicism about voting, and voter apathy, too. Which might be helpful for Hitler.
I don’t know what “true conviction” means, but the idea is to have your vote more fully reflect your preferences. It adds degrees of preference to your vote.
In winner-take-all elections, you choice is binary: your choice and not your choice. That’s not a full reflection of your preferences.
Say you and a group of 5 friends want go out to eat. You have to decide where to eat. Two people want Italian, one wants Chinese, one wants Mexican, and another wants Moroccan. Italian wins because it gets the greatest number of votes. But 3 people–the majority–don’t want to eat Italian! So you decide to nominate all of the ethnic foods. People rank their preferences from love to tolerate to hate. There’s no majority for food people love. But it turns out that 4 people can tolerate Moroccan food. (The 5th person doesn’t like but goes along for fun and the cocktails). A majority are now happier than they would be if the Italian food people won.
If a voter really liked Ralph Nader, but knew he wouldn’t win the general election. Ranked voting would allow that voter to express his preference for Nader as his first choice. As much as that voter preferred Nader, he preferred Al Gore to George W. Bush. The ranking more closely approximates what the voter would like to happen. It also prevents a minor candidate like Nader from spoiling the chances of a major candidate.
Great explanation Mark, it’s like you’re a teacher or something
I also think both examples do a great job of enlightening the debate, but I think they point out some problematic aspects of rcv too. In your food example, the ultimate outcome of rcv is that 4 people instead of 3 wind up eating something they didn’t originally prefer. From what others have said here, this appears to be a feature rather than a bug of rcv – any time rcv changes the outcome compared to a first past the post system you wind up with a situation where fewer people get their first choice. I am not yet convinced that’s a better reflection of the will of the electorate than first past the post.
In your Nader example, the question I have is the inequality of voting. Nader voters, Gore voters, and Bush voters all go into the booth. Then, after counting the votes, the Nader voters are given a second vote. Why do they get 2 votes? If the Gore voters got two votes, their candidate would easily win. Same with the Bush voters. But only the Nader voters get a 2nd try. To try and tie it back to the point of the original post, would fewer people feel politically alienated if 2nd round votes from Nader supporters delivered the Presidency to Gore instead of Bush?
Thanks for the discussion – I feel like I’m learning a lot here.
That is not how RCV works. Everyone has a single vote like they do under the current system. If a voters first choice fails to gain a majority, than that vote goes to their second choice and so forth.
So in your Nader example, the Nader voters first vote is then transferred to Gore. The Gore voters first vote would not transfer since Gore would get a majority. In both cases voters were able to vote their conscience for their first choice while also ensuring their vote ultimately mattered in electing a candidate with majority support.
Everybody get a 2nd vote, Nader, Gore and Bush voters. But realistically, in this case, only the 2nd choice votes of the candidate finishing last (Nader) get reallocated.
Elections are not for the purpose of expressing a preference, they are for the purpose of choosing one winner from among the candidates. Seeding the candidates into brackets and running a series of eliminations would be statistically preferable.
I’m not sure what you’re trying to say here.
If preferences weren’t part of an election, it wouldn’t matter who was chosen.
RCV does all of that Pablo. Every voter is given an opportunity to express their preference and choose one winner with majority support among the candidates. RCV even seeds into brackets in a statistically preferable way without the costs, fairness, and turnout problems from holding multiple first past the post runoff elections.
What part of “however many they want to rank” don’t you understand? You base your objection to something that just isn’t true. Voters to make no preferences or just two or nine preferences.
And what happens when “they” only want to rank one?
The situation we have now. which is what you are trying to protect against.
In the absurd case: If nine people vote for one, each, of ten candidates and the tenth person ranks all ten of the candidates, then the tenth person is the spoiler and essentially gets to decide.
If you want to protect against that case you must add a limit to weight against single-votes: something like winner must get 50% of votes and some percentage ( say, just for the sake of the argument, 50%) of the ballots must be ranked. If the number of ranked ballots falls below that, then what? It’s a mathematical quagmire.
PS: I think this will be the default mode, at least in the first couple of elections if adopted, and possibly the crucial fail if education and sufficient time to break in the system aren’t allowed for…
It doesn’t seem to have been the default mode when they tried it for the first time in Maine earlier this year.
I’ve not heard, either way, regarding the amount of Maine’s ranked balloting . Do you have more information?
I don’t have the “blank” rate from the ME primary. And Ranking only played a role in the Dem Primary. There the candidate that got the most votes in the first round, but not the majority, did indeed get the majority in the end.
From the Bangor Daily News:
You could choose a single candidate if you so desired. There MIGHT even be instances when you want to vote for any other candidate. It works in contests with multiple candidates and multiple seats. I’m not actually sure about IRV.
Join our cause! We have fun meetings and often meet afterward for beer and coffee.
Beer and coffee is a mixed drink I have yet to try. Maybe a nice porter….
Hopefully, not in the same cup.
No no no no . You can’t make this statement because you don’t know how many 2nd place votes Leonard Golder would have received under the hypothetically ranked vote. You simply can’t take an unranked election and speculate what would have happened if it was ranked. It is that different.
Consider the hypothetical instance where 1000 people vote for candidates X, Y, Z and W, Candidate X finisher gets ‘only’ 250 first place votes yet 750 second choice votes… that is to say the remainder of ballots places him/her second. Does she/he get eliminated and all those other votes distributed? A clear 100% of the electorate ranked candidate X first or second but if we eliminate in the manner you choose, hypothetically, the winner might end up being candidate Z, who mostly received 3rd place votes and you’ve defeated your own ranking system.
Yes yes yes. Your hypothetical makes me wonder if you understand RCV (at least how Maine has it organized). If Candidate X gets 250 votes, it is impossible for them to be eliminated in the “first round”. If Y got 499 votes and W and Z split the remaining 251 votes. Given your scenario of candidate X getting ALL the 2nd place ranked votes, then they’ll win the election 501 votes to 499 votes. In your scenario, there is no hypothetical way that a candidate with less votes would surpass candidate X. BUT it certainly possible that a 3rd or 2nd place finisher ultimately defeats a first round “winner” with less than 50% of the vote.
But you do bring up an interesting “navel gazing” question. What if there was a 4 way tie between the four candidates with 250 votes each. But we’d have a similar problem if that occurred without rank choice voting.
The nonono was in reference to the attempt at comparison between the apples of first-past-the-post and oranges of RCV. You simply cannot say such-and-such that occurred via the one system is relevant to what would have occurred in the instance of the other.
I understand it. I just didn’t use a big enough differential. Sloppy of me.: Good catch on your part. The hypothetical still holds in any case where candidate X didn’t get enough votes to survive the first round but got way more second place votes than anyone else.
As I tried to get at in another comment there may be multiple non-unique numerical equalities (i.e. ‘ties’) depending upon how the candidates are eliminated: in a first-past-the-post system there is no possible way more than one candidate can get greater than 50%. This may not be true with RCV. You can eliminate candidates in different ways and get a different candidate with 50%.
For instance for N candidates you can rank the vote itself: which candidate got the most first place votes to which candidate got the least, which candidate got the most second place votes, to which candidate got the least second place votes and so on, and eliminate any candidate who is the lower half in all of those rankings (did not get the top half of any rank,) an re-apportion the vote..
Or you could do the same ranking and eliminate any candidate not in the top half of at least two ranks (meaning the voter deliberately ranked them). Or not in the top half of N/2 rankings or greater… re-apportion their votes.
Or you could assign fractional points to the rankings (1 point for a first place, 1/2 for a second place, 1/3 for third place, ,,, 1/n for nth place. Least point total gets eliminated and votes re-apportioned, etc….
Each of these three scenarios will derive a candidate with 50% of the vote but it won’t necessarily be the same candidate with the same 50% because they won’t necessarily eliminate the same candidates per round if you run them against the actual votes. That’s what I mean when I say there may be non-unique majorities and that the 50% is not ‘found’ but ‘constructed.’ Mathematically speaking they are perfectly valid and perfectly equal expressions of the voters intent, but they are different… I think this is a little more problematic than simple ‘navel gazing’
All that matters in the first round is first ranked votes. If Golder got the least amount of those he’s done, regardless of how many second ranked votes he got.
I think that the predicate assumption here is that there is a unique, and knowable, 50% of support to be found in the data (number and ranking of votes). But what if it is not unique? Suppose there are two, or more, different ways, using the same data, to get two, or more, different results, each supposedly representing a support of 50% of the vote?
If the 50% is not ‘found’ but rather ‘constructed’ by the method of eliminating votes, then (as noted in a previous comment) some candidate X who got 250 first place votes and 750 second place votes — and thus eliminated under the rules explicated above — would have a decent claim of calling the election unfair.
That’s not how ranked choice voting works. People have argued a similar criticism for multi seat elections like with the STV/PR election in Cambridge. The order does matter there, but it’s usually only contentious during vacancies when they do a new count with the vacators votes transferred out. That won’t be the case with RCV which is for a single member seat.
Not really.
This is an advertisement for a Top Two primary.
We are now headed to a general election with a choice between the winner of the recount and Rick Green, who was uncontested for the Republican nomination. Ranked choice might bring us a different Democratic nominee, but the net result is the same, the Democrat versus the sacrificial GOP lamb.
The unofficial results now show Trahan with a 52 vote lead over Koh. Trahan is a Charlie Baker Democrat, and it would be interesting to see what would have happened if she was on the same primary ballot as Rick Green. Who knows? Maybe the GOP leaning independents would have voted for Green, and another Democrat would have emerged at the top.
Still, the idea of RCV has some appeal. Having a top two primary that uses RCV to narrow the field down to two finalists would be an improvement over the current situation, but with the weakness of the Republican party, we have far too many uncontested or lightly contested general elections. Top two gives folks a real choice in November.
Given the composition of the district, Top Two could have a benefit. For example, L’Italien and Matias both have a base in Lawrence, though Matias beat second place L’Italien by 5,897 votes in that city. L’Italien didn’t win a single city or town, but she had quite a few second place finishes. It would be interesting to see what would have happened to L’Italien in a RCV scheme.
RCV helps with the occasional wide open seat, but does nothing with all the one party uncontested general elections. Let’s work a little compromise. Let’s come up with a RCV Top Two primary, where we use RCV to eliminate the lowest candidates and sorts things out until we eliminate all but two, and let the top two move forward to a November general election.
And let’s move the primary back to June.
A top two primary is a separate discussion. But to avoid the potential mess that almost occurred in CA this year, RCV would be necessary for a top two to work. As far as stimulating more general election contests, public financing of campaigns is sure to attract a challenger for every incumbent (and possibly reinvigorate the MA GOP).
Trahan’s donation to Baker is blown out of proportion, plus the person who dropped the dime doesn’t really know party rules FWIW. I can see a top two run-off primary, but if I were a Republican I would not want my one candidate shut out of a spot on the general ballot.
Instant runoff voting would take care of the top two primary without prolonging the election, which takes time, money, and leads to decreased interest in the election.
One potential hybrid between the two systems would be to hold a top four non partisan first past the post primary and then a general with RCV. In that scenario in the CD-3 it is quite possible the Republican would still advance while allowing the top three performing Democrats the opportunity to compete with them in the fall in an RCV manner so the Democrats do not dilute their vote.
My two biggest concerns with top two are that it would make a single party state more so and in the case of CA, create scenarios where a wide Democratic field could be a negative. Which is anti democratic on its face if we believe voter choice matters.
Single party state? Isn’t that what we have when we have folks who would be Republicans elsewhere running, getting elected, and governing as Democrats?
Right now we have a bunch of DINOs run on the Democratic line, and once elected are difficult to dislodge. Republicans won’t run against them because they like the DINOs, and Democrats have the actions of a collection of DeLeo-Baker Democrats running the statehouse.
While we do need to do something to address the occasional plethora of candidates for an open seat, our biggest need is a mechanism that puts credible candidates on the November ballot.
As was the case in the special election for the Fourth Middlesex senate seat, you had a spirited Democratic primary with three Democrats, which eliminated two credible candidates, while the primary winner went on to face a Green Party candidate in the general election. RCV alone wouldn’t have provided the best two choices in the general election. Top Two wouldn’t have excluded a Republican, because nobody ran as a Republican.
IRV would solve all of those issues without the disadvantage of limiting peoples choices. Top two exacerbated those issues of progressive vote splitting when tried in CA swing districts, those issues would certainly still be at play in Democratic primaries in this state as well.
In the case of the Fourth Middlesex special, presumably a progressive district, the most progressive candidate with the majority of support would win. Under a top two, there could be a scenario where more progressive challengers helps a DINO advance to the final runoff where they consolidate unenrolled and Republican support. In a very left wing district in first past the post, the presence of a Green party candidate could be a spoiler. In an RCV election, they are not.
So long as it is a single seat election, IRV does not expand voters choices: They still only get one representative in the end. If the goal is to simply make the voters feel like they participated more in the election, then I suppose any of these schemes are as good as any other.
If you truly want to increase voters choice you must give up the notion of a single representative and embrace proportional representation in all its gory details. You may respond with the notion that this might dilute the representative power of the formerly single representative — and this is true, it will — but that dilution might actually be a more true, that is to say a more accurate and representative, picture of the electorate.
I mean, if we were really living in an ideal world we would have multi member districts and PR non partisan elections like Germany and New Zealand. We would have a much stronger multi party democracy as well.
I have also grown to favor parliamentary democracy over the Presidential republic model for quite some time. That said, we are one of the only major democracies they entrusts party nomination to the voters. As imperfect as we are (caucuses, primary order, open vs. closed) we are still leaps and bounds ahead of most parliamentary countries where the MPs elect their party leader.
@ leaps and bounds ahead: I must say that I think this assumes a measuring stick that may be incorrect.
It appears to me that our nomination process may well be gravely broken. I am not ready to accept, as a matter of dogma and faith, that entrusting party nomination to the voters necessarily results in better nominees.
One of the complaints in the 2016 nomination was that our nominee was deeply flawed (a complaint I reject, but one that was and is still frequently repeated). Certainly the GOP nomination of 2016 is a case study of a broken process.
I think “leaps and bounds ahead” of those other alternatives may reflect faith as much as reality.
I suggest that the political system of the US is very gravely broken. I wonder if an assumption that “the voters” are always correct may be part of that problem — particularly when so many of the voters are so ill-informed (or worse).
IRV alone might result in a different outcome in a multi-candidate Democratic primary,. Unless you open it up as a totally open primary, where the two best candidates move on to November, you will still have meaningless or uncontested races in November.
Uncontested races is a separate (and very important) issue. Based on some research I’m currently doing, public financing of campaigns is a sure way to recruit opponents (because they won’t have the daunting task of raising money agains an already well financed incumbent). And as I alluded to above, the MA GOP would probably be opposed to this for idealogical reasons, when in fact it would be the best thing for them…attracting many candidates to run again t Dems.
My biggest barrier to candidate recruitment was lack of money, fear of being a spoiler, and fear of being subjected to negative campaigns. All of those issues are mitigated by an RCV election. No more fear of being a spoiler, no need for a negative campaign, and while the lack of money is still a separate issue money is no longer the end all be all of messaging.
I think it needs to be said that no voting system is without its drawbacks. The advantages and disadvantages should be weighed against winner-take-all voting.
Fair vote lists a bunch of advantages. These are a few:
I confess to not being very good with numbers, but here’s an apparent paradox for IRV:
I think it needs to be said that no voting system is without its drawbacks.
Mathematician Kenneth Arrow demonstrated this in 1951. It does apply to all voting systems, even the Condorcet system, although it is a special and unlikely case.
The second part of your post, a vs B vs C, is an informal statement of Arrow’s theorem. It applies to winner-take-all pluraity=rules systems as well. So it is not a legitimate criticism of IRV versus other systems—not “a paradox for IRV” versus the system in place on the 3rd, because it applies to that system as well.
Not exactly. Arrow says that it applies to any and all elections with three or more distinct choices and says, in effect, that for such elections, the aggregate choice of the individuals cannot be proven to represent to choice of the group as a whole. That is to say there is a mathematical divergence between the aggregate of individual votes and the resulting vote of the group such that the link between them can not be definitively traced.
The predicate assumption of IRV (and which is not all the assumption of first-past-the-post) is exactly that: the aggregate of individual choices is the choice of the group. Arrow says that can’t be proven.
I do not mind being corrected, but would say that FPTP has a possible result that is exactly analogous, the multiple-way tie.
Whatever method is used to resolve that tie will produce a result that is arbitrarily unfair to the preferences of the other voters.
Though I’m too math illiterate to understand Amartya Sen, but he has addressed Arrow’s argument.
I can;t do the math either, but it looks to me that Sen accepts Arrow and seeks to extend it to other spheres.
I think he resolves some of the apparent contradictions in Arrow. A problem, however, is that people need to understand how the system works.
Is paywalled, but even the Globe is for IRV on this
And we beat them to the punch.
Today’s MASSterList mentioned this diary.
BFD