[EM] December 2024 report: "The future of the instant runoff election reform"

[Kristofer Munsterhjelm]

On 2025-05-20 00:30, Rob Lanphier via Election-Methods wrote:
> Hi folks,
> 
> I thought y'all might be interested in this report from Brookings:
> https://www.brookings.edu/articles/the-future-of-the-instant-runoff-election-reform/ <https://www.brookings.edu/articles/the-future-of-the-instant-runoff-election-reform/>
> 
> I'm curious what y'all think of the report.

Parts of what it's reporting seem quite understandable: that parties 
that are in power don't like having their power challenged by new 
election methods that don't share the flaws that gave them extra power 
to begin with, for instance. It also seems quite understandable that 
parties that were "snubbed" by IRV (e.g. by center squeeze) would feel 
that the method is stacked against them, even if IRV is pretty 
indiscriminate in what direction (left or right) it denies CWs victory.

What surprised me are the arguments *against* IRV: that it violates 
OMOV, that it's complex, and that it's a foreign state invention.

It's true that IRV's outcome can be chaotic (just take a look at Yee 
diagrams). And it's also true that determining why IRV picked the winner 
it did can be hard to do. It can also be difficult to come up with a 
justification for why IRV picked the candidate it did, besides "that's 
the algorithm, and that's the result". But each voter still has one 
ranked vote each, so in that respect it respects OMOV.

As a measure of quality, IRV is complex: the chaos means that a 
candidate may go from unacceptable to acceptable even though there's no 
obvious reason. But I've been led to think that *as an algorithm*, its 
simplicity (tally and eliminate, repeat) is one of its stronger points. 
Maybe that's wrong? Or maybe the voters are reacting to the opacity of 
the outcome, not the process? If it's the latter, then at least 
theoretically, something like Copeland//Plurality or Copeland//TTR could 
actually be at an advantage.

On the other hand, if the voters are feeling there's a violation of OMOV 
because the ranked ballot allows people to vote for multiple candidates, 
then all ranked methods are going to have a problem, and something like 
Approval is going to have an even bigger one, since the ballot format 
makes it look like each voter gets as many votes as candidates they 
approve.[1]

The "it's a California method" objection is reasonable if what it really 
means is "we didn't ask for it, outside sponsors are instead just 
rolling up and ramming it down our throats". But less so if it's a 
not-invented-here objection - after all, states that use FPTP use the 
same FPTP, not their own slightly different method.

(On an aside, I was surprised at the "open primaries" joke, which to me 
seems pretty bad taste, in how it bundles things that have nothing to do 
with each other. But I guess that's the point: that it's shorthand for 
"you already agree with me about other things, now agree with me and 
oppose open primaries".)


Finally, about the indication that IRV may be leading to some 
moderation: I'm reminded of a discussion on Discord about how the 
Australian two-party system might be weakening. While top-two and IRV 
are both subject to center-squeeze[2] and Condorcet methods aren't, both 
TTR and IRV have less of it than Plurality does. The recent Romanian 
election is a good example.

To attempt to treat this more mathematically, in the Left-Center-Right 
election:

51-x: L>C>R
39-x: R>C>L
2x: C>R>L

C wins in FPTP when x > 17, in IRV when x > 13, and in Condorcet when x 
 > 6. So in that respect FPTP's center squeeze is the worst, IRV has 
some (relative to FPTP), and Condorcet has little.

So if new party formation depends on how safe it is to vote for 
candidates closer to the center, then above some threshold of 
center-squeeze, the method so strongly deters new entry that you get 
two-party rule. Below this threshold you would get a tendency for new 
entrants to appear, but they would grow more slowly the closer to the 
threshold the method resides.

It may thus be that IRV does reduce polarization compared to FPTP. And 
it's possible that such reduction of polarization might lead to more 
competition inside parties and growth of third parties, particularly if 
the Australian results are indications of a robust trend. However, even 
if it is, it took Australia a long way to get there - and that's despite 
their other chamber using STV, a proportional multiwinner method.

Is IRV's moderating effect sufficiently strong to stop what's going on 
with US polarization? I don't know. But better methods would produce a 
stronger effect.

(It's also possible that center squeeze isn't the only property that 
matters when it comes to make entry easier for new parties and 
candidates. My main point is more that center-squeeze can be considered 
a sliding scale, not just "this has it" vs "this doesn't have it".)

-km

[1] There are ways to improve on this first impression, but my 
interactions with certain cardinalists have left me disinclined to 
elaborate.

[2] First impressions would suggest that top-two has a stronger center 
squeeze than IRV when there are more than three candidates. But this is 
complicated by the game nature of top-two runoff, where voters can 
change their mind between rounds.