[wnoise]

Although FPTP is bad, it's not at all clear that IRV (the election method that most mean when they say the ballot-marking method of RCV) actually leads to less extremism. It's a chaos monkey in any race that approaches competitive, and changes that one might expect to help A (such as more A > B > C votes) can actually hurt A. That is, its bad results show up precisely when it might produce a different outcome than FPTP. See the recent Alaska house election for an actual case.

If you're tied to a full ranking, there are much better choices than IRV; any of the Condorcet methods, for instance. However, rating systems are far more likely to produce compromise candidates. I like Approval, perhaps with a runoff between the top 2.

EDITTED-TO-ADD: In practice IRV and FPTP both suffer from the "center-squeeze effect": https://electionscience.org/library/the-center-squeeze-effec...

[jrgoff]

Would you mind explaining more about what you mean regarding the recent Alaska election? The information I have found about it made it seem fairly reasonable and straightforward to me, but perhaps I am missing something.

[wnoise]

Here's a decent commentary:

https://www.reddit.com/r/EndFPTP/comments/x9oupk/2022_alaska...

> Two popular suspicions are now confirmed. Nick Begich was the Condorcet winner. Sarah Palin was a spoiler candidate - her presence caused Mary Peltola to be elected, by prematurely eliminating Nick Begich.

Here's another, that got down-voted to hell, because of the audience.

https://www.reddit.com/r/ForwardPartyUSA/comments/xb119e/its...

> If 2913 voters who supported Palin first and Begich second flipped their first and second preferences, they’d have gotten a more preferred result.

(i.e. they would have gotten Begich instead of Peltola, by having Palin eliminated first)

> Even worse, if instead 5825 of those same types of voters just decided not to vote, they’d have also gotten a better result. So merely participating in the election hurt them.

While I do prefer the winner it selected, it's not a great election method. Better than FPTP is a terribly low bar.

[peatmoss]

Condorcet has arguably worse features:

1. Even harder to explain to the average voter than RCV

2. Greater likelihood of nobody winning / needing to conduct new election

I am of the belief that RCV is not great, but that it is significantly better than status quo... and it is possible.

[wnoise]

For 1, meh. Marking is the same and it is much closer to incentivizing honest reporting. For 2, everyone advocating for Condorcet has some favorite cycle-breaker. Mostly it doesn't matter; it's a lot more rare than non-monotonicity in IRV.

But my favorite is approval.

[jrgoff]

I see, that was helpful - thank you!

[jpmoral]

IIUC how it would've worked in Australia is:

First round:

Begic and Peltola garner the most first-place votes, Palin is eliminated.

Voters who ranked Palin first get their second choice promoted to first. Presumably that would be Begic for most voters.

Second round:

Begic first choice votes + Palin > Begic votes vs. Peltola votes

Looking at the data in the reddit vote breakdown thread it looks like Begic would've won with this method.

[wnoise]

I don't know if they're exactly following the Australian STV rules, but it is quite similar.

Begich had the fewest amount of _first_ place votes, so was eliminated first.

However, Begich was ranked ahead of Palin (and Pelota for that matter) on more ballots than vice-versa.

[jpmoral]

> Begich had the fewest amount of _first_ place votes, so was eliminated first.

Ah, I misunderstood that part. Under Australian rules voters who ranked Begich first would have had their second choices promoted to first and added to the respective candidates. In this case ~27k for Palin and ~15.5k for Peltola. Then the totals are compared.

Looking at it that way it's not so clear cut that it was unfair that Peltola was declared the winner. I think reasonable arguments can be made either way.

One thing though is that on Australian ballots all candidates must be ranked.

[wnoise]

> Under Australian rules voters who ranked Begich first would have had their second choices promoted to first and added to the respective candidates. In this case ~27k for Palin and ~15.5k for Peltola. Then the totals are compared.

Exactly what happened. Palin was a spoiler -- had she not run, Begich would ha e won instead.

> One thing though is that on Australian ballots all candidates must be ranked.

That's kind of ridiculous. It's quite unwieldy and leads to many spoiled ballots. And I don't think forcing ordering past the point of caring really changes anything.