election_title: "Same ballots, different Condorcet rule — Schulze says Ana, Ranked Pairs says Bruno, Split Cycle says both"

scenario_description: |-
  The second worked profile behind "Cycle Resolution — why Minimax, Ranked Pairs,
  and Schulze exist" (05_Ranked_Robin/01_Learn/cycle_resolution.md): the
  point at which the Condorcet family stops agreeing. 40 members of a mural
  committee rank four designs. Majority preference cycles:

      Ana beats Diego by 12 · Diego beats Bruno by 10 · Bruno beats Ana by 4

  with Chloe hanging off the side (she loses to Ana and Bruno by 18 each, but
  beats Diego by 12 — which is what pulls the whole field into the Smith set).
  There is no Condorcet winner, and Copeland — the count under Ranked Robin —
  ties Ana and Bruno at +1 each.

  Now the refined rules split, on identical ballots:

      Minimax       Ana      (her worst defeat, 4, is the mildest in the field)
      Schulze       Ana
      Ranked Pairs  Bruno    (locks the big margins first, and they favor Bruno)
      Split Cycle   Ana AND Bruno

  Split Cycle is the interesting one, and not because it is indecisive: it
  discards the weakest defeat in every cycle, which erases Bruno>Ana (4) and
  Diego>Bruno (10) alike, leaving BOTH Ana and Bruno undefeated. Its claim is
  that Schulze and Ranked Pairs are not discovering a winner here — they are
  applying a convention to a tie the ballots genuinely do not break. Split Cycle
  hands the tie back rather than resolving it silently, which is why its winner
  set is always a superset of theirs.

  Verified with pref_voting:
    uv run STARVote_LH_tabulation_engine/tools_adam/pref_voting_tabulation_engine/cycle_resolution_report.py \
      method_comparisons/cycle_resolution/cases/cycle_schulze_vs_ranked_pairs_c4_b40.yaml

  LH-only (no BetterVoting election): the LH result is a Copeland tie, broken by
  margin then lot, and BV breaks the same tie at random — so this one cannot be
  frozen on BV. The companion case where every rule agrees:
  cycle_copeland_ties_c4_b21.yaml

paradoxes: [condorcet-cycle]

voting_method: RankedRobin
num_winners: 1

lot_numbers: [Ana, Bruno, Chloe, Diego]

ballots: |-
  7:Ana>Bruno>Chloe>Diego
  8:Bruno>Ana>Chloe>Diego
  14:Diego>Bruno>Ana>Chloe
  11:Chloe>Ana>Diego>Bruno

expected_winners:
  - Ana