In fact, this seems like the perfect example of ranked-choice voting giving people the outcome they preferred.
52.5% of people preferred Begich to Peltola. That majority of voters did not get the outcome they preferred.
Imagine an election with 6 candidates. 5 candidates each get 20% of the 1st round voting, and 1 candidate gets 0% of the 1st round voting, but everybody marks him as their 2nd choice. He is nobody's favorite, but everyone likes him over their alternatives to their favorite.
It seems reasonable to me that a good voting system should select this guy who is broadly approved by everyone. But in IRV, he is eliminated in the first round.
The recent Alaska election was much less egregious version of this same failure mode.
In other countries, like Australia, where they're used to alternative styles of voting, they have guides on how to vote and who to put in what rank to match what a given party thinks is the best rundown of potentials.
Alaska will get there.
"Condorcet failure" is a phrase meaning a failure of the voting system to select the Condorcet winner. If a candidate beats all other candidates in a 1-on-1 election, they are what is referred to as the Condorcet winner [1]. If a Condorcet winner exists, and an election system fails to select that candidate, that is a Condorcet failure.
Whether or not you think Condorcet failures are important or not is a matter of opinion. But they are important to many people (at least on an intuitive level).
[1] https://en.wikipedia.org/wiki/Condorcet_winner_criterion
32% Palin
29% Begich
Begich's votes split to 3: 50% for Palin, 29% for Peltola while 21% were exhausted.
That adds up to: 49.67% for Peltola and 46.5% for Palin.
And you said "52.5% of people preferred Begich to Peltola" ignoring that 40 + 32 percent preferred other candidates over Begich. What this says is that if Begich didn't run, more people would have picked Peltola oved Palin.
You could just as easily say that if Palin hadn't run, more people would have picked Begich over Peltola (the 52.5% I mentioned). This is just tautological based on the IRV method. My claim is that if you reason from first principles, the IRV method produces results that are suboptimal.
Begich beats both Peltola and Palin in head to head elections. It seems to me that if a candidate beats every other candidate in a head to head election, that candidate should probably be the one selected. If they aren't selected, that is a failure of the tabulation algorithm you're using. IRV is just one possible tabulation algorithm. My argument is that there are better tabulation algorithms.
If you disagree that a candidate who beats every other candidate in a head to head election should win, that's the core of our disagreement, and I'd be interested to know what you think this criteria is not a good one.
Begich only got more second choice votes because after picking your first, it comes down to "the least evil", on one hand some people preferred Palin because they'd hate Peltola to win, and on the other hand people picked Peltola because they'd hate for Palin to win.
If there were Republican primaries, Palin would have won over Begich, because in that instance you cannot count the votes that went second after Peltola as they're like Democrats. And then in a general election, we'd have the same scenario play as it did with RCV.
The ultimate question is which candidate is socially optimal. Don't think about specific methods like FPTP, RCV, etc. Just think about specific heuristics.
If you somehow had an algorithm for summing up societal satisfaction, which candidate would maximize it?
Given that 52.5% of people prefer Begich over Peltola, do you agree that more Alaskans would be satisfied if Begich was the winner (again, without thinking about the details of the algorithm that selected him)?
Given the result, it is clear that Alaskans are more satisfied with Peltola than Palin, but it's also clear that Alaskans would be even more satisfied with Begich over Peltola. But due to the particular algorithm used, that better outcome was precluded. There are different tabulation algorithms that result in higher expected satisfaction.
The majority of R voters voted for Palin. Without IRV the election would've been between palin and Peltola would've still won. I also disagree that most people would prefer their second choice over the plurality first choice.
There are non-FPTP, non-IRV election methods that would have correctly picked the Condorcet winner, Begich.
Evidently not. Or at least many of them didn't in this case.
edit: see https://www.elections.alaska.gov/results/22SSPG/RcvDetailedR...