election_title: "Reinforcement — combined 9 voters, counted by Ranked Robin (Cara wins)"

scenario_description: |-
  North (6 voters) + South (3 voters) merged, from Brandt, Dong & Peters,
  "Condorcet-Consistent Choice Among Three Candidates" (2024), Theorem 2
  (P1 + P2). Ada was a winner in BOTH districts (outright in South, a co-winner
  in North's dead-heat cycle) — so reinforcement/consistency says Ada should win
  the merged election. Instead a NEW Condorcet winner appears:

      Cara beats Ada 5–4 · Cara beats Ben 5–4 · Ada beats Ben 7–2

  Cara beats everyone head-to-head, so every Condorcet method — Ranked Robin
  included — elects Cara. Ada, the only candidate who won both parts, loses. That
  is the reinforcement paradox, which the paper proves is unavoidable for EVERY
  Condorcet extension once there are ≥ 8 voters.

  The same 9 ballots counted by STAR: reinf_combined_c3_b9_star.yaml (the
  scoring round leads Ada, but the runoff flips to Cara — STAR's runoff catches
  the same pairwise flip). Additive methods (Score/Approval/Plurality) instead
  keep Ada and show no paradox — see the folder README.

paradoxes: [multiple-districts]

bv_test_id: BV2254
bv_election_id: t4by6x
bv_results_url: https://bettervoting.com/t4by6x/results
# Live results: https://bettervoting.com/t4by6x/results — this election's Ranked
# Robin race (one of two on the combined 9 ballots; the sibling STAR race is
# reinf_combined_c3_b9_star.yaml). Frozen export: reinf_combined_c3_b9_bv_export.json.

voting_method: RankedRobin
num_winners: 1

lot_numbers: [Ada, Ben, Cara]

ballots: |-
  2:Ada>Ben>Cara
  2:Ben>Cara>Ada
  3:Cara>Ada>Ben
  2:Ada>Cara>Ben

expected_winners:
  - Cara