IRV and non-Monotonicity
-
This is commonly stated as a complaint about IRV. For instance:
@jack-waugh said in Symmetrical IRV:
Here's why IRV is worse than FPtP, and I get this argument from @Sass .
In FPtP, your entire vote is always counted.
Election security -- FPtP is tallied by the precincts and summed over their reports. The tallying point for IRV is a single point of failure and is vulnerable to a secret, scaled-up attack.
Monotonicity.The main thing I have to say about this complaint is that it is generally stated as a binary. It doesn't matter how likely a monotonicity failure is to happen, but the simple fact that it can happen is all that matters.
I'm not a fan of that sort of black and white approach.
Since IRV has elected the Condorcet winner 99.77% of the time in practice, I'm going to first say it elects the "correct" candidate in those elections.
It is debated whether or not the Burlington, Vt election (in which the Condorcet winner did not win) was a monotonicity failure. CES says yes, FairVote says no. I think both organizations have their obvious biases, but here is what CES says: "Kiss won because some voters ranked him too low."
To which I say.... ok, so? I mean sure, it sounds counterintuitive. But could voters have said "I'm going to rate Kiss higher so he won't win"? Of course not, they wouldn't have known till after the fact it that such a perverse strategy might have worked.
Regardless, the election was very close any way you look at it. If I look at the histograms below I can't instantly say one candidate or the other out of Montroll, Wright or Kiss should have won. (note that Montroll was the Condorcet winner, but significantly lower first choice votes than either Wright or Kiss)
Based on that..... and given that this is the worst case out of 440 elections.... yeah, not something I'd lose sleep over.
And by the way, since this is in response to @Jack-Waugh saying that IRV is worse than even FPTP, it seems very unlikely that Montroll would have won under FPTP either. We can't know, but given that it was a close 3 way race, it is a good bet that most would have voted for their favorite of those three, so..... probably Wright?
-
@Sass, what do you say?
-
@rob Note that even in Burlington the Condorcet loser received the most first-choice votes. So assuming that voters would have cast their votes for whomever was their first choice on the ranked ballots (which I think is a pretty reasonable assumption given that 1. it was a very close election and 2. limited evidence seems to show that voters by and large rank sincerely), then IRV still did better than FPTP would have.
Perhaps a Condorcet method would have done even better, but to say it's worse than FPTP is very silly to me. Also @Jack-Waugh why do you keep tagging Sass lol, I feel like they probably noticed the first 4 tags.
-
@andy-dienes said in IRV and non-Monotonicity:
Perhaps a Condorcet method would have done even better,
What I wonder is, what if, following this Condorcet failure, FairVote had come in and said "instead of repealing RCV entirely, you can make this tiny amendment that will change it to RCV-C, which picks the Condorcet winner if they exist"?
Luckily Burlington has recently brought back RCV (but not to mayoral elections). But yeah, I hope that FairVote warms up to a Condorcet-complaint variation, at least as an option to those that might be concerned about this sort of situation.
It's really odd the way FairVote is so pro-Condorcet on their site. At least they appear to be, their whole message is "IRV almost always picks the Condorcet winner, and that is good" but they never answer why they don't just put a Condorcet variation out there as an option.
Anyway, I know that is straying from the topic and I need to stop posting for a while and do other things ... but eventually I will get around to a post about FairVote and their odd contradictory message with regard to Condorcet.
-
@rob said in IRV and non-Monotonicity:
they never answer why they don't just put a Condorcet variation out there as an option.
It does seem rather strange. The knee-jerk answer is "IRV has momentum and excitement" ... but like FairVote has all the money, and they can generate momentum and excitement on their own with a few well-placed lobbyists and marketing sprees
-
@andy-dienes said in IRV and non-Monotonicity:
Note that even in Burlington the Condorcet loser received the most first-choice votes. So assuming that voters would have cast their votes for whomever was their first choice on the ranked ballots (which I think is a pretty reasonable assumption given that 1. it was a very close election and 2. limited evidence seems to show that voters by and large rank sincerely), then IRV still did better than FPTP would have.
So what? It still failed in every manner. Failed to elect the consistent majority candidate. Failed to protect against the spoiler effect. Failed to protect voters for "Voting their hopes rather than voting their fears" meaning that, like FPTP, it failed to disincentivize tactical voting. And it always fails with precinct summability, requiring the opaque transporting of individual ballot data from polling places to the central tallying location.
And Kurt Wright was not the Condorcet loser. James Simpson was. And if Simpson hadn't been in the election, then Dan Smith would be the Condorcet loser. So it's really just a false claim to begin with.
Perhaps a Condorcet method would have done even better, but to say it's worse than FPTP is very silly to me.
In the manner of precinct summability (which is a good thing supporting process transparency in elections), FPTP is precinct summable. IRV is not. But Condorcet RCV is precinct summable.
-
@rb-j Obviously I mean the Condorcet loser among the competitive candidates. I don't care about the schlubs in the election.
It still failed in every manner.
Except for electing the Condorcet loser, which FPTP would have.
-
@andy-dienes said in IRV and non-Monotonicity:
Obviously I mean the Condorcet loser among the competitive candidates.
Then say so.
I don't care about the schlubs in the election.
Then you need qualifying language to that effect in a stated claim.
It still failed in every manner.
Except for electing the Condorcet loser, which FPTP would have.
No, FPTP would not have elected the Condorcet loser. FPTP would have elected the Condorcet loser among the top three candidates. Or the candidate that Condorcet would rank coming in third.
James Simpson was the Condorcet loser. And if Simpson was out then Dan Smith, not a schlub but a serious candidate that got some serious support, would be the Condorcet loser.
-
@rb-j We've had this argument before and I don't feel like rehashing it. In my experience we get nowhere. I'm very tired of talking about Burlington. Let's just agree to disagree.
-
@rob 440 elections? More like 4. Most IRV elections in modern US history have only had 2 competitive factions, which any voting method would elect the Condorcet winner in almost every time. In the tiny number of them with 3+ competitive factions, the Condorcet winner was elected in I believe one.
-
@sass I didn't qualify my statement to only be counting those that met the definition of having "competitive factions," which sounds very subjective to me. So, my statement stands -- 440 elections, all have had a Condorcet winner. This is true, despite your feeling that we should be counting something else.
I imagine what you mean by "competitive factions" is something similar to "parties." Where I live, San Francisco, we've had ranked choice for a long time, although until fairly recently you could only rank three. Still, it appears to have worked well to reduce factionalization. Which, to me, is a good thing.
So if there weren't what you call "competitive factions" it seems likely that part of that is because Condorcet methods (and ones that strongly tend to produce elect Condorcet winners) are doing their job prior to the election, but encouraging people to run who have wide appeal.
-
@rob I want competitive elections. That's why I'm qualifying your statement -- because it matters. I would agree that Condorcet methods encourage people to run who have wide appeal even in competitive races because the candidates actually need to beat each other. IRV is not a Condorcet method and only appears to tend to elect Condorcet winners because it is so often used in elections that are not competitive. Part of the reason that IRV isn't used in competitive elections is because it doesn't create them, which we need voting method reform to achieve in the US right now. Under IRV, candidates only need to get a majority (really a plurality) of support and then the minimum number of votes they need will naturally flow to them. What you're describing sounds more to me like competition has decreased in favor of one ruling faction, which sounds like the trend across most of California. It doesn't appear to me based on your recounting that IRV has done anything to prevent monopoly rule in San Fransisco.
-
@sass said in IRV and non-Monotonicity:
and only appears to tend to elect Condorcet winners because it is so often used in elections that are not competitive. Part of the reason that IRV isn't used in competitive elections is because it doesn't create them,
Again, both the assumption and the conclusion are not backed by data. IRV is used in plenty of competitive elections, and it does seem to be more likely to elect Condorcet winners, and saying it does not create more competitive elections is speculation which is probably false, because evidence that exists indicates that IRV does lower the barriers to running, and may even encourage more, and more diverse, candidate entry, as theory suggests.
-
@sass Ok well I don't know what you mean by "competitive elections." The elections RCV-Hare have been used in seem quite competitive. I have often used the 2018 San Francisco mayor election as a good example of a competitive election that RCV-Hare did a great job with.
But you said "3+ competitive factions," which I'm not sure what you mean by that. What qualifies as a faction? In the election above, Mark Leno and Jane Kim campaigned together, and were often "co-endorsed" by newspapers and organizations. To me, that's the sort of thing that should happen. I don't know whether you consider Leno and Kim as a single faction or two factions, or really what you mean. Could you give us a definition of "competitive" or "competitive factions"? For instance, does "competitive" mean "highly polarized, with significant portions of the electorate utterly hating at least one front runner"? Or simply "very close"?
