FairVote’s odd position against Condorcet-compliant RCV

rob

Let me start with a couple of premises/assertions.

Infighting within the voting reform community stands in the way of wider adoption of alternative voting.
If FairVote accepted a Condorcet-compliant version of Ranked Choice as an option, they would make a lot faster progress in replacing “Choose-one” voting, because they would eliminate a large amount of resistance from the voting reform community, without bothering RCV advocates in the least.

Most people agree that if a candidate would beat every other candidate if they ran against them one-on-one, that candidate should be elected. Some people think this should outweigh everything else, while some don’t.

FairVote has a very odd, and seemingly contradictory, position on all of this. My own reaction to their position is: “is that SERIOUSLY the hill you want to die on?” It really makes no sense at all that they are so uncompromising on this issue that would make a trivially tiny difference in actual outcomes of elections (and therefore campaign strategies), but would make a huge difference in reducing the number of people out there arguing against IRV rather than more productively directing their energy against Choose-one.

First, let’s clarify a couple things. When I say “IRV”, I mean the current ranked-ballot system that FairVote advocates, which is also known as Hare. You can call it RCV-Hare. In many contexts, “Ranked Choice” or “RCV” simply refer to his method, but here we have to be careful because Ranked Choice can apply to various Condorcet methods with ranked ballots.

Another method, "Bottom Two runoff" is very similar to Hare in almost all ways. Being a ranked ballot system, it could easily be referred to as Ranked Choice (or even IRV, for that matter.... but here let's assume IRV specifically refers to the Hare version). Bottom Two just has a minor tweak making it Condorcet compliant. FairVote could offer it as an option, and I’d predict it would meet with near zero resistance from outside their organization.

FairVote, though, has refused to do this, and they have been entirely contradictory regarding their reasons. They say, over and over on their site, that IRV is good because it almost always elects the Condorcet candidate (339 out of 340 elections). They say, over and over on their site, that IRV is good because it reduces polarization.

But then, when asked why they aren’t ok with Bottom two runoff, they say “IRV rewards those w/ strong 1st choice support. Bottom-2 rewards those who avoid polarizing stances. We like IRV since it encourages real stances, not just campaigning to avoid the bottom”

Let’s break this down, and show just how contradictory they are.

“IRV rewards those w/ strong 1st choice support.”

Isn’t that precisely what Choose-one, the main system they are against, does? So they are basically saying, “Rewarding first choice is the cause of all problems, and our system fixes those problems. But it still rewards first choice, and that’s good.” What?

“Bottom-2 rewards those who avoid polarizing stances.”

Isn’t that just another way of saying that Bottom-two runoff doesn’t encourage polarization? And yet their site is chock full of arguments that polarization is a bad thing, and IRV reduces polarization. So, again.... what?

“We like IRV since it encourages real stances”

I’m not really sure what qualifies as a “real stance”. Do they mean an extreme stance? A polarized stance? A one-sided stance that appeals to some faction while angering the others? One that fails to find the nuance, fails to find a middle ground?

They really need to clarify this. I have a hard time seeing how they can clarify the meaning of "real stance," without just saying that polarization is a positive (despite all the places they say that polarization is a negative).

To me this sounds like the exact opposite of what they say is positive in just about every other page of their web site.

“not just campaigning to avoid the bottom”

“Just” campaigning to avoid the bottom? “Just”? Are they really suggesting that by changing the method to bottom-2, which would have only changed the outcome in 1 out of 440 elections, that it will change campaign strategy that much, so that candidates will now have a sole priority that is completely different from what it would be with the other method? That's pretty extreme..

The logic behind what they are saying is utterly ridiculous. FairVote is well aware that the one RCV election that didn’t elect the Condorcet winner (Burlington 2009) is the one that has produced by far the most criticism for RCV. While I am fully of the opinion that that election was a very close race, and it wasn’t THAT big an error in the grand scheme of things…. I can’t see any reasonable argument that it picked the “right” winner.

The most charitable interpretation of what the are saying is, we want to reduce the polarization and "favor first choice only" attributes of Choose-one, but EVER SO SLIGHTLY less than we could. Because .... uhhh.... just because.

Finally, the overall irony: FairVote says that RCV encourages compromise, nuance, and give-and-take. Accepting Bottom-Two runoff as a reasonable option would be a perfect way to practice what they preach.

Here are some articles where FairVote argues FOR Condorcet and/or AGAINST polarization:

https://www.fairvote.org/want_to_fix_our_polarized_politics_fix_how_we_vote

https://www.fairvote.org/polarization_key_facts

https://www.fairvote.org/report_finds_lack_of_competition_increased_polarization_in_congressional_elections

https://www.fairvote.org/polarization_under_rcv_in_cambridge

https://www.fairvote.org/every_rcv_election_in_the_bay_area_so_far_has_produced_condorcet_winners

cfrank

@rob said in FairVote’s odd position against Condorcet-compliant RCV:

“IRV rewards those w/ strong 1st choice support.”
Isn’t that precisely what Choose-one, the main system they are against, does? So they are basically saying, “Rewarding first choice is the cause of all problems, and our system fixes those problems. But it still rewards first choice, and that’s good.” What?

Actually, not necessarily, depending on your definition of "1st choice support." Many voters in the Choose-one system will not indicate their "honest" 1st choice, due to the conflicts of interest that can easily be introduced by that form of ballot. The concept of a "real stance" is meaningless, but the influence of conflicting interests on decision-making is quite clear, as most moderate people face this problem every election cycle. You cannot eliminate all conflicting interests, but you can try to diminish the severity of those conflicts. I think it's pretty clear that IRV accomplishes this better than "Choose-one," and to me it doesn't seem very productive to nitpick about the exactitudes of that.

As I've mentioned, I am among those who do not particularly care for the Condorcet criterion as a primary measuring stick. The strongest non-majoritarian argument I've found for it is what I would call the "pragmatic" one, which is to consider the prospect that the majority that pairwise prefers the Condorcet winner to an alternative could--perhaps, in principle, maybe--band together and overthrow the election. Is that actually a rational concern? I don't know, but it seems like there are other practical concerns to take into account that could very well counter that prospect, such as the effort, motivation and organization necessary to accomplish doing that, the social and legal agreements and pressures upholding the legitimacy of the voting system in place, the risks of failure, etc.

I do see it as a huge improvement over FPTP, it’s just that the more I think about it, the more I am personally convinced that it is standing on top of some misguided principles.

rob

@cfrank said in FairVote’s odd position against Condorcet-compliant RCV:

Many voters in the Choose-one system will not indicate their "honest" 1st choice, due to the conflicts of interest that can easily be introduced by that form of ballot.

Of course.

But those conflicts of interest that will cause people to insincerely express that a candidate is first choice, will vary proportionally with whatever degree that the method rewards first choice candidates. You can't have the one without the other.

So lets say Choose-one rewards first choice votes 100%, IRV maybe rewards them 5%. A Condorcet method that resolves cycles with plurality (what @rb-j calls "straight-ahead Condorcet") might reward first choice votes maybe 0.5% or less.

What we would expect is that the degree people are incentivized to insincerely rank their first choice varies exactly the same. The degree you reward first choice, is the exact degree you incentivize insincerely expressing first choice.

(to be clear, I personally don't consider rewarding first choice to be a benefit even if we could count on people voting sincerely. But that is neither here or there...)

@cfrank said in FairVote’s odd position against Condorcet-compliant RCV:

, which is to consider the prospect that the majority that pairwise prefers the Condorcet winner to an alternative could--perhaps, in principle, maybe--band together and overthrow the election

Ok I think we're straying off the topic a bit, since this part of your comment is more about Condorcet vs. various other methods. I'm not really trying to argue here for Condorcet vs. some of these other methods out there, just relative to IRV, since IRV has the bulk of the momentum. And I'm not so much arguing "Condorcet is significantly better than IRV." I'm saying "if you throw a bone to Condorcet fans by endorsing both regular IRV and Condorcet-compliant RCV variants, you'll speed up adoption by quieting a lot of loud anti-RCV voices (such as those still obsessed with the one time IRV failed to elect the Condorcet winner), maybe even getting them on your side".

So, for those primarily interested in the topic of the thread, feel free to stop reading here.

First, let me say that both IRV and Condorcet methods are very much center leaning compared to choose-one, it's mostly just that IRV is more chaotic and can therefore occasionally result in electing a candidate that is fairly obviously "the wrong one".

Although "banding together and trying to overthrow an election," even violently so, is indeed a "thing" now, I'm not sure I'd measure that likelihood purely by looking at the number of people who prefer one candidate or another. You can "prefer" a candidate without wanting to risk everything by taking up arms against the government to put them in office or keep them in office. You need true passion to do that -- worshipping one candidate, despising another, or some combination -- to want to go to that extreme.

Of course this isn't just about the possibility of citizens trying to overthrow elections. This is about government running smoothly in a general sense. When people have that degree of passion is when things go badly. Especially if you consider that passionately loving a candidate tends to inspire passionate hatred toward those running against that candidate. What we're after is reducing destructive polarization.

Often, people criticizing Condorcet-compliance show examples where some percentage of the population absolutely loves a candidate, and note that the Condorcet tabulation effectively ignores the degree of their passion. Everything else being equal, having people love a candidate is good, I suppose. But then again.... what can happen when there is that degree of passion? Isn't that what we've seen recently, where a particular candidate inspires people to extreme degrees, those people don't mind if the candidate blatantly lies, does illegal things to try to stay in office such as bribing foreign leaders, fires those that might investigate him, and otherwise abuses power over and over again? Many of those citizens (including elected representatives) are, as we speak, still trying to justify and/or downplay a violent attempt to stop the democratic process. I try to avoid getting too directly political here, but hopefully recent events are still instructive as to why center leaning methods -- that don't put a lot of weight into whether people are over-the-top worshipful toward a candidate -- have benefits.

I will also add that Condorcet methods seem to me to be the most game-theoretically stable and therefore difficult to game through strategic voting and strategic nomination. It seems readily provable that non-Condorcet compliant methods are going to be more sensitive to "irrelevant" candidates (e.g. adding a candidate can change the outcome even when that candidate doesn't win), which will tend to mean strategic nomination. That's almost built into the definition of Condorcet. If you know the elected candidate was preferred to every other candidate by more than half of the people who expressed a preference, its hard to imagine how an irrelevant candidate could have changed the outcome.

rb-j

I think FairVote's position against Condorcet is along the lines of

"Our current RCV method (Hare) already has momentum that we don't want to damage by conceding it's not the best thing ever since sliced bread."
"We hear enough complaining about how RCV or IRV is too complicated. Condorcet RCV is even more complicated and harder to sell to skeptical voters and policy makers."
"What are they complaining about? IRV and Condorcet elect the same candidate 99.8% of the time. IRV only once did not elect the Condorcet winner. Just once. Likely to not happen again very soon."
"Whenever the Condorcet winner is not elected with IRV it's because the Condorcet candidate did not have enough base support. So the Condorcet candidate deserves to lose."

rob

@rb-j Right. I think #1 and #3 can sort of be turned around and hopefully reframed as "Condorcet RCV is nothing more than a minor tweak, to get that extra 0.2%". It's not conceding, any more than Honda is conceding that the 2022 Civic is bad by making a few changes for the 2023 model.

As far as I'm concerned, they still can not only call bottom two runoff "RCV", they can call it "IRV." Just treat it as the same method, with a minor performance tweak.

The vast majority of those who would weigh in on Ranked Choice give exactly zero fucks as to the difference between Hare and Bottom Two. They both use identical ranked ballots, they both go through an elimination process to eliminate "irrelevant" candidates that could be spoilers, they both are center leaning, and they both remove most of the incentive to vote or nominate strategically compared to choose-one.

As for 4, sure, they can keep arguing that if they want, but I wonder why they bother since it is so few elections it makes a difference. And I'm not sure the concept of "base" is seen as a positive to many these days. It seems to be used more of a pejorative than anything on the news, e.g. "just playing to the base," etc. In other words, it represents polarized politics. It seems like most people who would be interested in ranked choice voting would see it as a negative, anyway.

For #2, a good counterargument is that if you make it Condorcet compliant, it means results can be delivered immediately due to it being precinct summable (via a pairwise matrix). With bottom two runoff, the only case where it is not precinct summable is when there is no Condorcet winner and then it has to actually go through the runoff process. That is far more rare than the case in Hare-IRV where no one has a majority of first choice. For most "regular people," that is what makes it seem complex, having to wait days while they do the runoff process.