election_title: "Reinforcement — Combined, Cara branch (9 voters; both halves say Cara, the whole says Ben)"

scenario_description: |-
  North (6 voters, the symmetric cycle) merged with the CARA-branch South
  (reinf_south_cara_c3_b3_rr.yaml). Cara is a co-winner of North's three-way tie
  and the outright Condorcet winner of South — a winner in both halves — so
  reinforcement (consistency) demands Cara win the union.

  She doesn't. A NEW Condorcet winner appears in the merged electorate:

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

  so every Condorcet method elects BEN. Rotated one more step from the Ben branch,
  and the mechanism is identical: the two 5–4 pairwise majorities that decide the
  merged election exist in neither half by itself. Head-to-head majorities are not
  additive across electorates, and that non-additivity is precisely what Zwicker's
  Proposition 2.5 turns into an impossibility — no Condorcet extension is
  reinforcing, at any number of candidates from three up.

  Branch companions: reinf_combined_c3_b9_rr.yaml (Ada branch → Cara wins),
  reinf_combined_ben_c3_b9_rr.yaml (Ben branch → Ada wins).

paradoxes: [multiple-districts]

voting_method: RankedRobin
num_winners: 1

lot_numbers: [Ada, Ben, Cara]

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

expected_winners:
  - Ben