Ranked Voting (ie, Instant runoff, alternative vote, transferable vote, preferential voting)

[strawman]

The time might be right to push for a better voting system in the US. It's a republic, so it can't be changed nationally without a constitutional convention, but individual states can decide how to apportion their votes in national elections (ie, president), and can choose how to vote for their representatives locally and nationally, so the power is in the people's hands.

We just have to convince them that it's better than what we currently have.

Instant-runoff has some major infrastructural challenges to implementation it would need to overcome, and if it goes full electronic then we need to be super careful about the integrity of the system.

I'm curious what challenges there are to this voting method?

 

[PatrThom]

Kati was just telling me yesterday about how they're considering adopting this in Maine.

--Patrick

 

[SpecialKO]

I'm curious what challenges there are to this voting method?

For a national election with millions of voters, hand-count or punch-card voting systems are simply logistically incompatible with instant run-off (or any ranked vote variation, reallY). There's just no way you could make it work on a practical level.

So every single voting district using paper ballots as the primary ballot (as opposed to maintaining a paper ballot for an auditable paper trail), or older DRE machines that don't store individual ballots as the primary one, would need to either replace their tabulating system or change their ballot system outright. You could simply have them collect all the results, which could theoretically be tabulated via combinatorics (every single ballot variation of each race is individually counted instead of each vote for a person) at a central location, but then you need to ensure that every district is sending those results in the right way/format.

And this gets massively more difficult if people have the option of only ranking as many candidates as they want.

For districts with e-voting machines or optical scan cards that count with computers, it would be a much easier fix, but you would still need to ensure that every one gets properly updated.

This is all doable. But it would take a ton of work. It is why we've really only seen it done in the US for local elections at the municipal/district level.

And if we used something like Condorcet instead of IRV, conceivably, the only way you could get it to work on a national level would be to force the adoption of a nationally-approved e/internet voting infrastructure, which has some fairly obvious logistical and quite possibly legal issue.

 

[GasBandit]

I've always been a vocal proponent of instant runoff voting, but I never thought I would see it come to be implemented without some sort of governmental collapse - I expected too much pushback from the two major parties who directly benefit from the entrenched "two party system." But now even Democrats are starting to speak for it, it seems a step forward to have to actually worry about the logistics of implementing it.

 

[Sara_2814]

This is a great video for explaining Instant Runoff to people.

 

[Eriol]

That video is only one in a series on various election topics: Politics in the Animal Kingdom

All good if you have the time.

 

[AshburnerX]

I will, of course, point out that all voting systems are flawed once they have 3 or more potential choices. (Arrow's Impossibility Theorem)

You can only decide which flaw is the least terrible. In this case, you run into a couple of flaws with Instant Runoff Voting (IRV), most primary of which is that IRV has situations where the person who wins the election doesn't have a majority of total votes. This happened in the Oakland, California mayoral election of 2010. There is also the problem of the order in which candidates get eliminated and have their votes moved can affect the outcome of the final count and decide the winner. There other issues... it can violate Condorcent Winner Criterion and non-monotonicity (giving first place more first place votes can cause it to LOSE the election), both of which happened in the 2009 Burlington, Vermont mayoral election. They no longer use IRV BECAUSE of that election.

Of course, the single biggest issue with IRV is that the person voting has no idea how much of their list is actually going to be used. If your first choice is eliminated in the final round of voting, none of the other candidates you voted for mattered at all because your ranking of lower candidates will never be examined. You might as well have voted for no one EXCEPT your first choice because none of your other choices mattered and will have no bearing on the election.

This isn't to say that IRV doesn't have some advantages over simple plurality, because it does. But it has a whole host of it's own problems that matter just as much.

 

[Sara_2814]

I will, of course, point out that all voting systems are flawed once they have 3 or more potential choices. (Arrow's Impossibility Theorem)

You can only decide which flaw is the least terrible. In this case, you run into a couple of flaws with Instant Runoff Voting (IRV), most primary of which is that IRV has situations where the person who wins the election doesn't have a majority of total votes. This happened in the Oakland, California mayoral election of 2010. There is also the problem of the order in which candidates get eliminated and have their votes moved can affect the outcome of the final count and decide the winner. There other issues... it can violate Condorcent Winner Criterion and non-monotonicity (giving first place more first place votes can cause it to LOSE the election), both of which happened in the 2009 Burlington, Vermont mayoral election. They no longer use IRV BECAUSE of that election.

Then they're idiots, because that's the entire point of IRV. It gives minority viewpoints a voice. Kurt Wright may have had more citizens who wanted him as their first choice, but Bob Kiss has more of the population who wanted him to win over Wright (in some cases as a second/third choice). If Wright was liked by the majority of the population, then all those 2nd choice votes would have gone to him and he would have won easy-peasy. But if supporters of Andy Montroll, Dan Smith, and James Simpson didn't like Wright, then Wright is not representing the majority of the population. And by the results, is does look like the supporters of Montroll/Smith/Simpson preferred Kiss over Wright, so it worked exactly as it was supposed to that Kiss won. More people preferred Kiss over Wright to be mayor. Rather than scrapping IRV, they should have done an education campaign to explain how IRV gives more people representation they can agree with, or at least tolerate.

Of course, the single biggest issue with IRV is that the person voting has no idea how much of their list is actually going to be used. If your first choice is eliminated in the final round of voting, none of the other candidates you voted for mattered at all because your ranking of lower candidates will never be examined. You might as well have voted for no one EXCEPT your first choice because none of your other choices mattered and will have no bearing on the election.

If your first choice is eliminated in the final round of voting...that's called an election? Somebody has to lose an election. I'm not sure why I'm supposed to feel sorry for someone who supports a candidate with a large support base as their first choice? It essentially turns into First Past the Post for them. Why does their ranking of other candidates matter? Their first choice made it to the final round. If the final round ends up being their First Choice vs. their Third Choice, and they actually wanted Third to win the final round, then they should have put him/her as their first choice. I'm not seeing the problem here. Your first choice should be the person who best represents you, the second choice should be the person who you have some disagreement with but you would be okay with them winning, third choice is the person you'll tolerate, and if you think a candidate is a total doucherocket then you don't even put them in your ranking so there is no possibility of them getting a vote from you.

This isn't to say that IRV doesn't have some advantages over simple plurality, because it does. But it has a whole host of it's own problems that matter just as much.

At this time, we need IRV to break the stranglehold the Democrats and Republicans have on our election process and give other views a chance to be heard. The two major parties actively work together to make ballot access extremely difficult, if not outright impossible, for third parties. It's like the one thing they agree on. IRV removes the "spoiler" effect that people are afraid of. With IRV, they could feel confident in choosing a third party that best represents them as first choice, then putting their Major Party Lesser Evil as second choice. Even if their first choice doesn't win, the votes would still be recorded, and then we'd have an accurate picture of just how many people prefer third party policies over the Democrats/Republicans.

 

[Bubble181]

then we'd have an accurate picture of just how many people prefer third party policies over the Democrats/Republicans.

Reason they don't want it, right there.

 

[Dave]

Also, We need to cut down the time campaigns are allowed to do anything publicly. Give them a month or two of campaigning and that's IT. No more of this perpetual presidential campaigning.

 

[GasBandit]

Also, We need to cut down the time campaigns are allowed to do anything publicly. Give them a month or two of campaigning and that's IT. No more of this perpetual presidential campaigning.

Normally I'd agree with you, but on the other hand, do you think the country would have benefited if Trump had had LESS opportunities to show his ass?

 

[MindDetective]

Normally I'd agree with you, but on the other hand, do you think the country would have benefited if Trump had had LESS opportunities to show his ass?

Not only that but there is a huge first amendment problem to get around in order to implement something like that. I love the idea of it, though.

 

[Dave]

Normally I'd agree with you, but on the other hand, do you think the country would have benefited if Trump had had LESS opportunities to show his ass?

I concede the point. In fact, I change my mind completely.

 

[Gruebeard]

I concede the point. In fact, I change my mind completely.

Here's hoping you don't have cause to change it again when you find out that Trump has enough time to pull out a win.

 

[strawman]

This isn't to say that IRV doesn't have some advantages over simple plurality, because it does. But it has a whole host of it's own problems that matter just as much.

In the first example, they didn't run an IRV. They did a form of ranked voting, but rather than eliminate the most-losing candidate and adjust everyone's rankings to remove that person, they started counting second votes, then third votes. I don't think we can conclude anything about IRV from that voting system.

In the second example it appears to have worked correctly, and while the outcome was surprising to some, the main issue was that it's confusing to a lot of people. Further, it did highlight the problem of non-monotonicity within IRV. Even though it didn't materially affect the outcome of the election, the election did experience it, and it showed a great example of the problem - namely that if 750 people had moved a certain candidate to their #1 position, that candidate would have lost earlier in the runoffs than they did - ie, voting highly for someone may actually hurt them in some situations.

I don't think either situation is a slam dunk against IRV.

The only real problem with IRV is that it's hard to understand and explain, and it's easy to manipulate the numbers to make the outcome seem strange. For instance someone may have 34% of the first votes, but then very few of the second and third votes. If they get cut out near the beginning of the runoffs, then they'll complain that they were essentially tied with their competitors in first choice votes, even though their competitors were more frequent second and third choices. It's too easy to mislead the public.

We still have that with the existing electoral college, though, where you can get the popular vote without winning the college.

I still think IRV is better, but it too can be manipulated. The biggest benefit, though, is diversity.

And if it's sold as a way to increase diversity (ie, more women and minorities elected, not to mention third parties) then we may be able to convince the rest of the population that despite its warts, it's still better than the two party system the current election method enables.

 

[Eriol]

STV > IRV > SMP (FPTP) voting IMO. I like the idea of STV as presented by CGP Grey in that playlist I linked earlier in the thread. Lets more voices in, without going to full Proportional Representation, since that silences local voices and puts too much power into parties IMO.


And the whole point of IRV is so that the person with the most first-choices may not get the nomination. But it DOES need to be run in the "dropout" fashion, and not just "now let's pile on 2nd-choice votes" on top for ALL. That seems really weird to me.

 

[strawman]

I like STV better as well, but due to its increased complexity I suspect it would be harder to get people onboard.

Transparency doesn't just mean doing things in the open, but doing them in a way that's easy to comprehend.[DOUBLEPOST=1476379237,1476379197][/DOUBLEPOST]The tax code and the US budget are two examples of things done in the open but are still obscured by complexity and it's easy to misrepresent what's going on.

 

[SpecialKO]

I might be misunderstanding something, but isn't STV in single-winner races functionally the exact same thing as IRV?

 

[strawman]

I might be misunderstanding something, but isn't STV in single-winner races functionally the exact same thing as IRV?

Generally, yes.

 

[AshburnerX]

If your first choice is eliminated in the final round of voting...that's called an election? Somebody has to lose an election. I'm not sure why I'm supposed to feel sorry for someone who supports a candidate with a large support base as their first choice? It essentially turns into First Past the Post for them. Why does their ranking of other candidates matter? Their first choice made it to the final round. If the final round ends up being their First Choice vs. their Third Choice, and they actually wanted Third to win the final round, then they should have put him/her as their first choice. I'm not seeing the problem here. Your first choice should be the person who best represents you, the second choice should be the person who you have some disagreement with but you would be okay with them winning, third choice is the person you'll tolerate, and if you think a candidate is a total doucherocket then you don't even put them in your ranking so there is no possibility of them getting a vote from you.

Yes, but if the final round of voting is your First Choice vs. your Last Choice, the rest of your list doesn't help support your 2nd/3rd/etc choices ether because of how IRV calculates votes. There could exist situations where doing so would actually help another candidate you'd prefer over the winner win the election instead. And, as I pointed out earlier, there also exist situations where voting for your top choice could actually cost them the election.

This is why I prefer Borda Count myself: all of your votes generally affect the outcome of a election in a way you'd want them to because your full list is always used, unlike IRV where your other choices may or may not be used during the election. Unfortunately, Borda Count's cloning problem (teaming in this case) also introduces situations where the weighted vote is thrown off. It does avoid the non-monotonicity though!

I don't think either situation is a slam dunk against IRV...

...And if it's sold as a way to increase diversity (ie, more women and minorities elected, not to mention third parties) then we may be able to convince the rest of the population that despite its warts, it's still better than the two party system the current election method enables.

To be fair, I wasn't really TRYING to slam dunk IRV ether. It does have legitimate, important advantages over what we have now. I just wanted to point out that it does, like all voting systems, have inherent flaws since no one was really talking about them. That doesn't mean that it's not a better system for what we need it for, but it does mean that we need to be vigilant if we implement such a system.

 

[Sara_2814]

Yes, but if the final round of voting is your First Choice vs. your Last Choice, the rest of your list doesn't help support your 2nd/3rd/etc choices ether because of how IRV calculates votes. There could exist situations where doing so would actually help another candidate you'd prefer over the winner win the election instead. And, as I pointed out earlier, there also exist situations where voting for your top choice could actually cost them the election.

I think we're talking about different systems of voting.

I looked up non-monotonicity examples, which is not the same as the system CGP Grey describes in the video I posted, nor on the Wikipedia page for IRV.

In the non-monotonicity examples, they described candidate B being eliminated in the first round, but then winning in the second round. They also seem to be considering your entire list as a unique type of vote, as if voting A > B > C is different than voting for A > C > B. So A > B > C and A > C > B count separately in determining elimination in a round? This makes no sense to me. (I'm still not entirely sure of what I read there, it's in dense academic-speak. )

In the system CGP Grey (and Wikipedia) describes, only first choices are looked at in the first round. If no candidate has a majority, then the candidate with the least number of first choice votes in eliminated and the ballots that listed them as a first choice now go to their second choice candidates in the second round. Repeat as necessary until a majority is reached. There is no way in this system for you to cause your first choice to lose, because you only have one vote and only one of your choices at a time is considered in each round.

 

[SpecialKO]

In the non-monotonicity examples, they described candidate B being eliminated in the first round, but then winning in the second round. They also seem to be considering your entire list as a unique type of vote, as if voting A > B > C is different than voting for A > C > B. So A > B > C and A > C > B count separately in determining elimination in a round? This makes no sense to me. (I'm still not entirely sure of what I read there, it's in dense academic-speak. )

The RangeVoting link is a bit poorly written (probably because the author notes it was heavily revised later). When they say "second" in that example they have at the top, they mean a second example that is a variant of the first where A-candidate voters who also like C-candidate are persuaded to vote C>A>B instead, not a second round of voting.

The point of the example is that when you compare the two scenarios, C-candidate got more first-round votes in the second scenario, but actually lost the election to B-candidate, despite the fact that A-voters who preferred C-candidate to B-candidate are the only ones who changed their vote preference to rank C-candidate higher. In a single-winner election, this is the same process whether we're considering IRV or STV.

 

[Sara_2814]

The RangeVoting link is a bit poorly written (probably because the author notes it was heavily revised later). When they say "second" in that example they have at the top, they mean a second example that is a variant of the first where A-candidate voters who also like C-candidate are persuaded to vote C>A>B instead, not a second round of voting.

Why is it that even being brought up, because being persuaded to change your vote has nothing to do with IRV specifically. Trust me, I'm a third party voter who has spent twenty years being told I'm "wasting" my vote in a FPTP system. If a person makes the choice to be persuaded to vote for a candidate other than the person who best represents them, that's not the fault of IRV (or any voting system).

The point of the example is that when you compare the two scenarios, C-candidate got more first-round votes in the second scenario, but actually lost the election to B-candidate, despite the fact that A-voters who preferred C-candidate to B-candidate are the only ones who changed their vote preference to rank C-candidate higher. In a single-winner election, this is the same process whether we're considering IRV or STV.

How can A-voters who preferred C be the only ones to change their vote preference? It's an automatic system. The A-voters who preferred B also got their second choice moved--to B--in the second round, and helped B win.

C losing even though they got more first round votes is exactly how IRV is supposed to work, because while C may have gotten more votes than B or C in the first round, for it to go to a second round means C didn't get a majority of votes (if C got 80% of the votes in the first round, there would be no second round). And if A supporters like B more than C, then when their second choice is applied to the remaining candidates in the second round (since their candidate A was eliminated in the first round, as if A never ran at all) they would help B win. What the result shows is that if candidate A never ran, B would have gotten more votes than C in the first round. Which means that overall, out of all voters, B has more supporters than C (some as first choice, some as second choice).

If C wants to defeat B in the next election, then they just learned that they better start paying attention to A supporters to try to get their second choice vote. Which is why third party supporters like IRV!

 

[SpecialKO]

You're conflating the two scenarios with two rounds of ranked voting allocation.

The scenarios aren't the same instance of voting, they are separate instances of near-identical voting (think of the first scenario as polling and the second as the actual vote, if it helps), intended to how how both IRV and STV in single-winner elections a voter's bottom preference can win even because they, and only they, switched the rankings of their first two preferred candidates.

You're absolutely right that candidates need to make their case, and that IRV opens the door to 3rd-parties more. But it comes with an inescapable mathematical fact that you can vote for your preferred candidate, and cause that candidate to lose. Adopting IRV or STV for single-winner races means having to live with that.

If C wants to defeat B in the next election, then they just learned that they better start paying attention to A supporters to try to get their second choice vote. Which is why third party supporters like IRV!

There's really no way to reasonably convince people to vote for your side second as a specific outcome unless you only intend to come in second. C would have been much better off trying to convince the B-voters who put C second to put C first than the A-voters who put them last.[DOUBLEPOST=1476401130,1476400973][/DOUBLEPOST]This is why Ash is saying he likes the Borda Count ranking method. This particular scenario we're talking about cannot happen with that method.

 

[Sara_2814]

You're conflating the two scenarios with two rounds of ranked voting allocation.

The scenarios aren't the same instance of voting, they are separate instances of near-identical voting (think of the first scenario as polling and the second as the actual vote, if it helps), intended to how how both IRV and STV in single-winner elections a voter's bottom preference can win even because they, and only they, switched the rankings of their first two preferred candidates.

You're absolutely right that candidates need to make their case, and that IRV opens the door to 3rd-parties more. But it comes with an inescapable mathematical fact that you can vote for your preferred candidate, and cause that candidate to lose. Adopting IRV or STV for single-winner races means having to live with that.

I'm still not understanding what is meant by "voting for your preferred candidate and causing them to lose". How is that even possible? If your first choice passes a round, then your vote helped them move past that round and remains with them in the next rounds until they are eliminated or win. If your first choice gets eliminated, it's not your single vote that made them lose the round (you tried to help them win the round!) but the fact that they didn't have enough overall supporters and came in last place in the round and were eliminated, so your ballot then goes to your next choice.

There's really no way to reasonably convince people to vote for your side second as a specific outcome unless you only intend to come in second. C would have been much better off trying to convince the B-voters who put C second to put C first than the A-voters who put them last.[DOUBLEPOST=1476401130,1476400973][/DOUBLEPOST]This is why Ash is saying he likes the Borda Count ranking method. This particular scenario we're talking about cannot happen with that method.

Okay, let me work through this statement: "C would have been much better off trying to convince the B-voters who put C second to put C first than the A-voters who put them last."

If we assume B and C are major parties (because they made it to the final round) and A is a third party...

The statement is saying Republicans are better off trying to convince Democrats who put Republicans second to put Republicans first, rather than the Libertarians who put Republicans third in their list.

But Republicans and Democrats are most likely not going to be eliminated from an IRV round (except the final one to select the winner) at this point in time. So it doesn't matter what a Republican or Democrat supporter puts as their second, third, etc. choice, because until third parties become stronger, Democrat and Republican candidates are not going to be eliminated in IRV rounds and voters choosing them as their first choice will always make it to the final round with their first choice. Until things change with our political parties, Libertarians, Greens, and other third parties would be the only candidates eliminated in each of the rounds and have to have their ballots go to second, third, etc. choices.

So it makes sense for Republicans and Democrats to listen to and court third parties for their second choice votes in IRV, as those are the candidates most likely to be eliminated and have to go to the other choices on the ballots. Which is going to increase the say third parties have in the political process and allow them to (hopefully) grow to be real competition for the major parties.

Also, Republicans courting Democrats (and vice versa) doesn't work so well under FPTP, so I don't see it working so well under IRV either. To use a saner election as an example, a Bill Clinton supporter may have been okay with Bob Dole winning, but I doubt you could have convinced them to put Dole first on an IRV ballot.

 

[Covar]

I think we can all agree that B is the wrong choice for our country.

 

[Bubble181]

There's really no way to reasonably convince people to vote for your side second as a specific outcome unless you only intend to come in second.

Sure there is. "Dear Bernie supporters, I understand you vote for him as #1, but please consider placing Hillary on #2 because anything but Trump". Same for a Republican candidate asking Libertarians to put him second because "still better than Clinton who wants to Make Government Great Again!"

Trying to convince voters to put you second often comes down to vilifying the other options, but it can be done.

 

[SpecialKO]

Sure there is. "Dear Bernie supporters, I understand you vote for him as #1, but please consider placing Hillary on #2 because anything but Trump". Same for a Republican candidate asking Libertarians to put him second because "still better than Clinton who wants to Make Government Great Again

That's not actually a winning strategy for the leader, though. That's more likely to get liberal voters in general to put Hillary second, and that is specifically counter to her goals. You can't actually message exclusively to Bernie supporters, even in our social media world. Otherwise, Trump would only be screaming his BS to his base, and be presenting a moderate face to everyone else.

 

[Bubble181]

That's not actually a winning strategy for the leader, though. That's more likely to get liberal voters in general to put Hillary second, and that is specifically counter to her goals. You can't actually message exclusively to Bernie supporters, even in our social media world. Otherwise, Trump would only be screaming his BS to his base, and be presenting a moderate face to everyone else.

Maybe in an America where the mindset is very strictly "us or the other guy". I assure you many politicians across many parties here spend a lot of time demonizing specific other parties here, towards everyone. The Green, Socialist and Social-Democrat parties might all be aiming for roughly the same voter and fighting viciously over their votes, they all agree that anything'd better than the nationalists. Of course, it helps if you know who your main voters are, and who are "side" voters who you might convince, and you're not, like American politicians, trying to somehow convince 50% of the people you're the best choice for them.
The social-democrats mainly target liberal high educated white folks, the socialists target lower class left leaning people, the green aim for immigrants and hippies and people who still believe they care about the environment. There's obviously overlap and they'd all prefer if everybody on the left decided to vote for them, exclusively, but they all have a different profile. HRc isn't "left" or "liberal" by any stretch of the imagination, and here she wouldn't be claiming to be, either. She'd be appealing to more moderate-centrist people, people looking for "no big turn around", etc ,whereas Bernie would be appealing to the young and idealistic who still want to change everything for the better. I'm not saying our system is much better, but in some ways it is (in others, it's worse, most assuredly - we could pretty much reduce our parliament down to one guy from each party and have their votes carry different weights and be done with it for all the power they have left).

 

[SpecialKO]

You guys also have a parliamentary system with multiple-winner elections where seat ownership is determined by party votes first, then individuals within that party, that puts a very different spin on the whole thing.

 

[Bubble181]

You guys also have a parliamentary system with multiple-winner elections where seat ownership is determined by party votes first, then individuals within that party, that puts a very different spin on the whole thing.

Definitely. I'm just saying it's possible, not that it'd be easy or would work anywhere in any system.

 

[SpecialKO]

Oh sure. I've long wondered if the way we do our representatives is better or worse than a parliamentary system, but I think a transition at this point, aside from the logistical challenges, would probably actually worsen our current partisan problem.

I don't think Americans need any more incentive to vote by party line before the person at this point.

 

[MindDetective]

So a couple of people are pissed about who won the Cy Young award. Politics notwithstanding, I first thought they used some kind of ranked choice system like we were discussing here but it is points based. That said, looking at the votes, the same outcome would happen as the votes for Britton and Kluber transferred to Porcello. Porcello would still have won, meaning the person with the most first place votes would have lost, still leaving some very invested people heartbroken. It seems like, from a political standpoint, you could end up with a similar result where (for example) Hillary gets the most #1 votes and Trump gets the most #2 votes that transfer over from the Libertarians, pushing him into the lead and thus winning (popular vote or electoral, your choice). In such a circumstance, would we feel any less torn as a nation? Or is ranked choice really too confusing for people?

 

[@Li3n]

It's a republic, so it can't be changed nationally without a constitutional convention,?

I think you mean it's a constitutional republic. Because being a republic says nothing about requiring changes to the constitution to change voting systems.

And actually even then there no obligation to have the constitution include the type of voting system you use.

Pretty sure we changed some vote to FPTP last year just by passing a law (had to stop the corruption % drop somehow).

 

[Gruebeard]

Or is ranked choice really too confusing for people?

It's only really confusing for people who choose to be ignorant, with understanding made more difficult by a news media that promotes that ignorance for ratings and politicians eager to capitalize on it, since ignorant people are (seemingly) easier to manipulate.