election_title: "The mayor's race (RCV-IRV, full rounds) — Cora comes from third and wins"
scenario_description: >-
  100 ballots, four candidates, one ordinary-looking mayoral race. Counted by FULL
  RCV-IRV: Dean is eliminated first and his ballots lift Cora from 20 to 36, past
  Blake; Blake is eliminated next and his ballots lift Cora to 67. Cora wins — and
  Cora is also the Condorcet winner (beats Ada 67-33, Blake 69-31, Dean 84-16). The
  paper's "single-elimination RCV" never holds those rounds: it keeps only the top two
  (Ada and Blake) and elects Blake. This is the case Kissel calls "quite rare" — the
  streamlined model eliminating a candidate who would win the fuller count — and it
  costs the Condorcet winner. Companion: …_rr.yaml, …_star.yaml, and the contingent /
  supplementary counts run by contingent_vote_report.py.
bv_test_id: BV2277
bv_election_id: tqfdbg
bv_results_url: https://bettervoting.com/tqfdbg/results
voting_method: RCV_IRV
num_winners: 1
# Round 1: Ada 33, Blake 31, Cora 20, Dean 16 — no majority; Dean eliminated -> Cora 36.
# Round 2: Blake lowest (31) -> Blake>Cora, so Cora 67, Ada 33. Cora wins.
# Batch-eliminating to the top two instead (the paper's model) drops Cora in step one.
ballots: |-
  33:Ada>Cora>Blake>Dean     # Ada's voters — Cora is their second choice
  31:Blake>Cora>Ada>Dean     # Blake's voters — Cora is their second choice too
  20:Cora>Blake>Ada>Dean     # the moderates, leaning Blake
  16:Dean>Cora>Blake>Ada     # Dean's voters — Cora again
expected_winners:
  - Cora

# file: bv2277_tqfdbg_mayor_irv.yaml