Alaska's ranked-choice elections — the whole picture¶
People (reasonably) get these tangled: which Alaska election was the famous "RCV broke" one? The special or the general? And didn't RCV work fine most of the time? This page is the map. Short version: Alaska has run many RCV races since 2022, and exactly one misfired — the August 2022 US House special. That's the one this repo models in depth; the rest are the honest backdrop that make it stand out.
How Alaska got RCV¶
Alaska voters passed Ballot Measure 2 (2020) (~50.6%), replacing party primaries + Choose-One with a top-four nonpartisan primary followed by an RCV-IRV general election, for all state and federal offices. It was first used in 2022. A 2024 measure to repeal it failed by ~664 votes (50.1% to keep) — so Alaska kept RCV and used it again in 2024.
"RCV" here means ranked ballots counted by instant runoff (IRV) — the FairVote-style single-winner tabulation. Alaska's are single-winner races.
The elections at a glance¶
| When | Race | Winner | Did RCV-IRV misfire? |
|---|---|---|---|
| Aug 2022 | US House — special (Peltola / Begich / Palin) | Peltola | YES — center squeeze. Begich beat both head-to-head (the Condorcet winner) but had the fewest first choices, so IRV eliminated him first. ← the case this repo models. |
| Nov 2022 | US House — general (Peltola / Palin / Begich / Bye) | Peltola | No. Six months later the electorate had shifted; Peltola was the Condorcet winner this time, and IRV elected her — ▶ runnable model, where all four counts agree. |
| Nov 2022 | US Senate (Murkowski / Tshibaka / Chesbro / …) | Lisa Murkowski | No — arguably RCV's best showing: the broadly-acceptable incumbent won by gathering cross-partisan second choices, an outcome a closed partisan primary might have denied her. |
| Nov 2022 | Governor | Mike Dunleavy | No RCV rounds — won an outright first-round majority, so no runoff was needed. |
| Nov 2022 | ~many state legislative races | (various) | No Condorcet failure reported. |
| Nov 2024 | US House (Begich / Peltola / Howe) | Begich | No — the 2022 "victim" won cleanly this time (51.3%), which also shows the 2022 special was a one-off, not a permanent distortion. |
| Nov 2024 | Ballot Measure 2 (repeal) | Repeal failed | RCV retained (50.1%). |
The honest scorecard — credit where it's due¶
This is the part a fair critic must state out loud: Alaska RCV mostly works.
- The empirical study this repo scales its model from — Graham-Squire & McCune, arXiv:2301.12075 — counted 13 Alaska RCV races in 2022 that actually reached a second (runoff) round. Of those thirteen, exactly one produced a Condorcet failure: the US House special. The other twelve elected a candidate with no Condorcet objection.
- In the 2022 Senate race, RCV is widely credited with a better, more representative result than the old system would have produced (a moderate incumbent surviving a same-party challenge by consolidating broad support).
- In 2024, RCV ran again with no reported Condorcet failure, and Alaskans voted to keep it.
So the failures are rare and specific, not a constant. Across the whole national dataset, Condorcet failures showed up in only two of 182 US RCV elections (Burlington 2009 and the Alaska 2022 special) — see the paper. Browse the ones that worked on Equal Vote's Real RCV tool.
So which one does this repo model, and why?¶
The August 2022 US House special — because the failure is the teachable moment. We reproduce it as a runnable, four-count model (Plurality & IRV → Peltola; Ranked Robin & STAR → Begich, the Condorcet winner IRV cut), with a graduated explainer:
- 101 — plain language · 201 — the vote math · 301 — every pathology + is this fair?
- Prose write-up with real vote totals: Alaska 2022 — RCV-IRV case study
The takeaway — the fair version¶
The center squeeze is a real, structural flaw of IRV: it can eliminate the candidate a majority prefers, and in Alaska's August 2022 special it did, in a high-profile federal race. But it is rare and conditional — it only causes a wrong result when a genuine centrist Condorcet winner has too few first choices, which is not most elections (not even most Alaska ones).
So the honest case for STAR and Ranked Robin is not "IRV always fails." It's: IRV has a known, structural failure mode that has struck real elections — and methods that read the whole ballot avoid it entirely, at no cost to the races IRV already handles well. Alaska is Exhibit A for both halves of that sentence.
Sources: Graham-Squire & McCune, arXiv:2301.12075 (the 182-election database; the 13 Alaska 2022 races; the two Condorcet failures) · Clelland, arXiv:2303.00108 (the special + general under STAR and Approval — STAR robustly elects Begich; Approval is indeterminate — which the repo's special and general cases make runnable) · Alaska Beacon, Nov 2024 (2024 House result + repeal outcome) · Equal Vote Real RCV (per-election visualizations) · Wikipedia: 2020 Alaska Measure 2 (neutral overview of the system).