Still undecided on Question 2 for Ranked Choice Voting or don’t know how it works? Watch Elizabeth Warren and Maura Healey explain how it works and why it yields better campaigns and fairer, majority outcomes.
Please share widely!
Reality-based commentary on politics.
Thanks for posting this, the graphic helps me better understand the counting process when no candidate has a majority of first choice votes.
I don’t mean to belabor the obvious — I’m a programmer and I’m looking for a description of the vote-counting algorithm that has enough detail for me to understand. This video helps me see how the second-choice of voters whose first choice came in last are allocated among the remaining candidates.
So that explains how the second choice of each voter is used. In what situations does the THIRD choice matter?
It sounds as though this algorithm looks only at the second choice of voters for the LEAST popular candidate. I think that means that the choices of supporters of unpopular candidates get more weight than the choices of popular candidates. Is that intentional? Is that good?
The candidates in the 1992 presidential election and their popular vote totals were:
Bill Clinton: 44,908,254 (43.3%)
George H. Bush: 39,102,343 (37.7%))
Ross Perot: 19,743,821 (19.0%)
Total: 103,754,418
I put together a little model of this in a spreadsheet. The vote between the two front runners split 53-47%. If 66% of the voters who listed Mr. Perot as their first choice chose Mr. Bush as their second choice (13,000,000) then Mr. Bush would have won the popular vote 50.2% to 49.8%.
So the second choice of 12.5% of the voters (13,000,000/103,754,418) would have made George H. W. Bush our president in 1992 if the proposed RCV process had been used for that election.
I understand the motivation for RCV, and I think the video from Ms. Warren and Ms. Healey does an excellent job of posing the problem.
I remain unconvinced that the proposed change solves that problem in a way that I agree with.
——–
Nerd note:
I wonder if any of our political scientists can say more about the following different algorithm for using rankings:
This algorithm is motivated by my desire to make the second and third choices of every voter equally important in determining the outcome, regardless of whether or not their first choice wins.
If my ballot in the 1992 election had been this way, I would have listed Mr. Clinton as my first preference and left my second and third preferences blank. I suspect that most voters who chose Mr. Clinton or Mr. Bush as their first preference would do the same as me. So I suspect that the only voters who would have supplied a second or third preference would have been the 19% who chose Mr. Perot as their first preference.
Since I think this would devolve to the same outcome as the proposed RCV process, I’m just offering it as a way of encourage discussion about how these systems work.
Nerd reply: you are describing a voting method called the Borda Count.
Though it may not be obvious, Borda is actually a more radical change from our current method of voting compared to the transferable ballot that is at the heart of IRV. IRV transfers a single vote under some conditions, while Borda entails multiple votes simultaneously.
Borda has some defects that I do not favor. In any case, I doubt it would pass muster with the state constitution.
The “perfect is the enemy of the good” best system is arguably Condorcet, least defective, hardest (by far) to use (as a voter).
Great to hear this video was helpful.
In that 1992 election, yes the later preferences for Clinton and Bush would not have come into play. It’s a lot like a runoff, where if you voted for Clinton or Bush in the first round and they made it to the runoff, presumably you would continue to vote for them in the next round.
I know you were just proposing a hypothetical, but for what it’s worth, I don’t think Bush would have won the 1992 election under RCV. It’s possible, but I believe the data showed Perot pulled from each evenly, and the norm is that challengers split the vote with one another more than they split the vote with an incumbent. Anyway, that’s all hypothetical, so I digress …
Third and later preferences can come into play when you have more than 3 candidates. So in 1992 there was a Libertarian candidate that won a tiny fraction of the vote. Someone who voted for the Libertarian first and Perot second, or vice-versa, would have seen their ballot eventually count towards Bush or Clinton in the final round. The idea is to incorporate as many voters as possible in the final decision — again, a lot like a runoff.
The key problem with any points-based voting methods is that voters quickly realize that voting for their second choice hurts the chances their first choice is elected. That’s been the history of the Borda count and similar methods: voting becomes highly strategic and frustrating. Under RCV, a vote for your second choice never hurts your first choice, because it isn’t counted unless your first choice is out of running. That’s been key to ensuring people use the ranks and use them sincerely.
And voters do use them. In the 2018 Democratic primary for Congress in Maine, the very first to use RCV, 87% of voters ranked at least a second choice. If experience with RCV around the country is any indication, even among Bush or Clinton voter in 1992, many of them would still rank backup choices, because … why not? It doesn’t hurt Bush or Clinton to do that, it’s a way of expressing your complete feelings about the candidates, and it helps you in the odd case that, for whatever reason, they wind up with too few votes to make the next round.
It’s a very theoretical video.
I would expect the Attorney General, who approved this for the ballot, to discuss the details of how this would work in a state that counts ballots on a municipal level. MA-04 has more than 39 municipalities; MA-01 has more than 60.
In Cambridge, the RCV is done in the office of the city’s election commission. How do we coordinate across municipalities? What happens when Franklin loses 3,000 ballots? Remember, we can’t transfer to the second count until we have a full and complete first count.
Also, and this is a huge question that has not been answered, do we have the time to conduct a RCV on a race such as MA-04, with the possibility of recounts at any of the decision points, with a September primary and a federal requirement to mail military ballots 45 days before the general election?
Another question: If we intend to move forward with expanded mail-in voting, what is the deadline for accepting ballots postmarked on election day? Remember, the first ballot transfer and the second count cannot be run until the first count is complete for every precinct in the jurisdiction.
If by “theoretical” you mean “clear and concise,” I agree.
Local elections are not changed; if we want the transferable ballot in Arlington, we will have to do a charter change. Then it would indeed be the Town Clerk’s problem.
Not so statewide and district elections, which will be tallied by the Secretary of State from ballots cast and collected locally.
So in some cases we could indeed be waiting on Franklin to find its ballots.
Would that were the least of our worries.
Plus we need to wait for Franklin either way. There is already central coordination required today: the Secretary cannot certify the results of a plurality election today until all the ballot data is received. He doesn’t certify the media reports — he certifies the actual reports from the municipalities. And once the ballot data is in, the tally itself would only take a minute or two to run.
We’re also talking about Bill Galvin, the same Secretary of Commonwealth that Paul endorsed — the same one that has already called for ranked choice voting to be implemented on multiple occasions. Paul is more concerned about implementation than the one whose job it is. Either Galvin can handle it fine or Paul is indicting his own ability to endorse competent individuals.
Still, nobody is describing how the count will be implemented.
This isn’t a question of the competency of anyone, but it is calling out the failure of the proponents to describe how ballots will be counted. I want to know about how this impacts a situation like Franklin. I want to know how it works in a place like Monroe, MA.
The video describes it pretty well.
As for Franklin, if that is a proxy for a low-level fumble at a town clerk’s office, the world will still have to ‘bate its breath for them. But not more so than it already did in the last go round.
This isn’t hard. You count everyone’s first vote and if no majority you redistribute the ballots allotted to the last place finisher by their second votes. If still no majority you redistribute the ballots of that round’s last place finisher by their next votes. Rinse and repeat until someone does have the majority. I find it easiest to visualize literal piles of ballots sorted by top votes, then taking the shortest pile and putting those ballots on the other piles. The first count needs to include the total number of ballots cast so you know what a majority is.
There is a classical children’s game that is a staple of programming courses — “Tower of Hanoi” (https://www.mathsisfun.com/games/towerofhanoi.html). Three “towers”, three rings of different sizes. At the start of the game, all three are on the left-most tower, stacked largest to smallest. The goal is to move all three to the right-most tower, one at a time, in the minimum number of moves. A larger disk cannot be placed atop a smaller disk. For extra credit, do the same with more than three towers and more than three disks.
I mention this because the problem is deceptively simple at first blush. It is only when actually playing it that its complexity becomes more apparent.
I agree that the contemplated RCV process is not hard. I do not agree that the consequences of this process are obvious or even understood at the moment.
I understand the concept of how it is done. I want the pragmatic response of exactly how this will be accomplished. How will you coordinate the count among three dozen jurisdictions.
Show me the instructions that will be sent to my town clerk, describing how we will conduct a ranked choice ballot count. That’s the part of the argument the RCV people don’t want to discuss, because it’s messy.
“How will you coordinate the count among three dozen jurisdictions.”
Greg already pointed this out. The same way it is done today. Cambridge has never had to wait very long for a result in the 70+ year history of using this method. The issues with vote counting in Maine had to do with the ballot design, the lack of voting machines calibrated for a ranked choice election, and rural precincts. Here, the SoC will print the ballots and they can work in all the machines. Galvin has not only endorsed this transition but testified that it will not be more costly or that different to run.
Also for the record, i am willing to wait longer and pay more for a truly democratic result where the candidate the majority of voters select wins the election.
I think Pablo dislikes this for the same reason Porcupine dislikes this, an erroneous belief that it will hurt their party. For an unenrolled voter like myself and the majority of Massachusetts voters, this system should expand our choices and help us make more informed ones. For Porcupine I argued it would help move her party back to the center and end the far rights influence on her party. For Pablo i would argue this prevents the stuff he always complains about-third party spoilers or the failure of progressive in a primary to consolidate around a candidate.
Nobody has explained how, in a state that averages 1.3 candidates per seat on the November ballot, how RCV will expand choices or encourage more people to run for office.
Greg and I have. It makes it more likely that more people will run in a crowded primary or that voters have multiple options in a general election by removing the spoiler effect and encouraging cooperative primaries. It substantially reduces the need for negative campaigning which turns off a lot of potential candidates who do not want to face that scrutiny. It makes it easier to vote for the person rather than the party or to vote for a first choice candidates even if their support base is initially small. It’s our best shot at transforming the system you decry. Defeating this proposal does not advance your proposal, if angry hint, it entrenches a status quo we both agree does not work.
I agree with you on this.
It seems to me that this system, at least in the short term, reduces the ability of “fringe”candidates, on the left and on the right, to exert leverage– which would put a merciful end to the Green-Rainbow “There is no meaningful difference between Trump and Clinton” BS, at least at the state level. This also makes it more difficult for ideological activists to gain concessions from the center by way of hostage-taking: “Support XYZ or all of the Bernie voters will stay home, and you’ll lose to the Republican.” At the moment, this is a big deal because that is the left’s entire strategy– a reasonable one under the existing system.
That makes the pitch to them a little like the pitch for single-payer to people like me: Give up this thing that you actually have in hand, which is not great, but is actually in-hand, and in return get something that might be better, but might also be a complete fiasco. That pitch doesn’t really budge me on MFA, and I suppose it might not budge them on RCV.
RCV seems likely to amplify the power of the centrist candidates in the short term. I view that as a feature, rather than a bug, but I would imagine that others might disagree.
It also reduces the power of the party apparatus. In Massachusetts, where one party is dead, but still moving, like a zombie, and the other is all things to all people, but corrupt all the way– this is for me a HUGE benefit of the plan.
In the long run, though, it also seems to me that this provides an avenue for a “fringe” policy position to migrate to more general acceptance more quickly than can happen now.
The more impassioned advocates here for single payer/MFA have argued for years and years that single-payer health care enjoys broad and enthusiastic support by a healthy majority of voters, and have complained that this support is not reflected in the actual candidates that are elected. For a very long time, support for single-payer was confined to fringier candidates– like the Green-Rainbow candidates, along with a few major-party candidates from more liberal constituencies.
It might have been that people for whom single-payer was a decisive issue went ahead and voted for that GR candidate, and that others who might also support that view did not, because they did not want to waste a vote on a fringe candidate. This would result in the support for the position being under-counted in the actual election, and reinforces the perception that this is a fringe issue.
RCV would at least have recorded and demonstrated that support– even if it is not immediately reflected in the results for that particular election. Politicians are good at sniffing for votes, and if it becomes clear that single payer– or whatever else– commands significant support among voters, then mainstream candidates will adjust sail accordingly, and the “fringe” position that actually commands voter majorities will move more quickly into the mainstream. RCV is more likely to drive this process because it is data from an ACTUAL election, based on people who ACTUALLY vote.
Who cares if a poll shows that a majority of voters supports XYZ if they don’t actually vote for it? Well, now you can see what is driving people who actually show up on election day.
Of course, it could also work coming from the ideological right. But so what?
217 of 218 ballot positions reserved for a Green-Rainbow candidate are vacant. This brings me back to a key point of the argument. In a state where three quarters of the seats are uncontested, a RCV scheme won’t fix our underlying problems.
People may be more willing to support GR if there is less risk of spoiling, which will strengthen the party in a few cycles and allow them to field more candidates. That said I was never under the impression that lack of candidates was the problem RCV was designed too solve. If anything it is a method that ameliorates potential pitfalls of a race that is already contested multiple ways.
I agree with much of your analysis CMD. I think we should evaluate this change based on whether it increases political competition and expands voters choice. It does both. What it does not do is advance a particular partisan or ideological agenda. I think jungle primaries, whether in red states or blue states tend to create one party states which is bad for small d democracy in the long run. It also wastes National Party money on what is essentially a November primary. Ranked choice is no panacea, but it does do things these other systems does not. Ensure every voter has a meaningful choice and ensure every winner is supported by a majority of the electorate. Two huge gains compared to the status quo.
No. Greg has not explained this. Cambridge is one jurisdiction, and they have physical control over all of the ballots cast in their city.
We have congressional districts that span dozens of separate municipalities, all with their own town clerk or city election commission. Show me the process for counting the ballots across multiple municipalities.
I’ll point out that Cambridge is the one municipality where Green-Rainbow candidates have been elected and that is due to ranked choice voting. The Green Party is a political player in European countries which has ranked choice or proportional terse their on.
As a former third party organizer I can tell you it’s a negative feedback loop. Nobody wants to join a third party or run as a candidate since they don’t want to be a spoiler, even in a state rep race. Requiring people to join the party in order for it to exist was an existential threat when nobody wants to lose their ability as an unenrolled voter to pick their primary ballot in a presidential election. For parties that do not have meaningful primaries, it becomes a self fulfilling prophecy. So they run against an unopposed statewide official to keep their 5% instead of enrolling 1% of the electorate which is much harder.
Ranked choice voting does not solve this problem, but it makes it easier for a third party to get to 5% on the first count and to cooperate with a major party on a ballot ranking ticket. It also helps ensure intraparty competitions are won by a candidate favored by the majority. It also helps non partisan candidates, women, and minority candidates get elected.
Cambridge elections are non-partisan.
You can’t be a spoiler in a two person race. Three quarters of the 218 seats on the November ballot, and Green-Rainbow is running a candidate in only one of these 218 races. Filing to run in a Green-Rainbow or Libertarian primary is a guaranteed ticket to a spot on the November ballot.
It’s not messy. They simply keep counting the votes until one candidate has a majority. Cambridge has done it without issue for 70+ years. They do this at the local level and then give Galvin the results. The same as a regular election, so the Franklin issue happened in first past the post. That does not discredit first post the post, it discredits the Franklin Town Clerk. We will still have to deal with human error and lack of uniformity since so much of the actual vote counting process in this country is done at the local level. Support a muscular FEC and we have a muscular SOC in Galvin. If he favors this reform, and he opposed many others, then he will implement it properly.
Franklin had a problem withA first past the post election, I really don’t see your point in bringing it up to somehow discredit RCV.
Franklin (or the problem town in another election) needs to resolve the issue before we can determine which candidate is in last place. You can’t proceed with the ballot transfer and the second count until you have a full and complete first count.