election_title: "Alaska 2022 special (reduced model), Palin deleted from every ballot — RCV-IRV now elects Begich"
paradoxes: [condorcet-winner]
scenario_description: |-
  The counterfactual half of the pair. Take alaska_2022_irv_with_palin.yaml and
  strike PALIN — a candidate who did not win — out of every ballot. Nothing else
  changes: no voter is given a preference they did not express, no ballot is
  reordered, no score is invented. The 23 Palin-only ballots simply have nothing
  left to say and exhaust.
  RCV-IRV now elects BEGICH, 93-84 over Peltola — and those are exactly the
  numbers the full election's own pairwise matrix already reported for
  Begich-vs-Peltola, because with Palin gone the instant runoff IS that
  head-to-head. So the loser Palin, not the ballots, decided the 2022 result.
  That is a failure of Independence of Irrelevant Alternatives on a real federal
  race, and it is the direct check on FairVote's claim that "RCV is highly
  resistant to spoilers because it satisfies both the Independence of Irrelevant
  Alternatives and Independence of Clones criteria". Clone independence is
  genuinely satisfied here (4 ballots rank Peltola between the two Republicans,
  so {Palin, Begich} is not a clone set); IIA is not, and IIA is the criterion
  FairVote's own prose definition describes.
  Reading it fairly: a real Palin withdrawal would have changed campaigns and
  turnout too. This is the STRICT criterion test — hold the ballots fixed,
  remove one candidate — which is what IIA asks and all a certified ballot record
  can honestly support.
  Lesson: ../README.md · claim check: ../../fairvote_comparison_table/README.md
voting_method: RCV_IRV
num_winners: 1
lh_only_reason: "A counterfactual profile — these ballots were never cast in this form. Publishing it as a live BetterVoting election would present a hypothetical as a real contest."
ballots: |-
  50:Peltola>Begich
  36:Begich
  29:Begich
  25:Peltola
  16:Begich>Peltola
  12:Begich
  5:Peltola
  4:Peltola>Begich
expected_winners:
  - Begich
