election_title: "Crowded field, rung 5 — 5 candidates, 65 voters, counted by RCV-IRV"

scenario_description: |-
  RUNG 2 — the same voters, two more candidates, and both choose-one methods lose the
  Condorcet winner.

  Diego still beats all four rivals head-to-head (crowded_field_c5_ranked_robin.yaml
  proves it). But first choices are what these methods count, and Bruno at 6 and Elsa at
  14 now stand between Diego and the voters who previously had nobody closer. Diego drops
  from 34 first choices to 9, is eliminated early, and the seat goes to Elsa.

  Round 1 is also the Choose-One count, and it elects Bruno on 23 of 65 — a candidate
  Diego beats head-to-head. Note what caused all of this: not one voter changing their
  mind, only two more names on the ballot.

  Same 65 voters and the same fixed candidate positions as crowded_field_c5_star.yaml,
  on a ranked ballot — the same ballots as crowded_field_c5_ranked_robin.yaml, counted by
  elimination instead of head-to-head.

  Construction: build_ladder.py in this folder. 65 voters in seven blocs at 0, 4, 8, 12,
  16, 20, 24 (sizes 6, 10, 13, 9, 12, 8, 7); candidates fixed at Ana 1 · Bruno 6 ·
  Clara 9 · Diego 11 · Elsa 14 · Felix 16 · Greta 22; utility = minus distance; scores =
  each bloc's own min-max scaling onto 0–5. Nothing is tuned, and no count at any rung is
  settled by a tie-break.

voting_method: RCV_IRV
num_winners: 1

lot_numbers: [Ana, Bruno, Diego, Elsa, Greta]

ballots: |-
  6:Ana>Bruno>Diego>Elsa>Greta    # bloc at 0
  10:Bruno>Ana>Diego>Elsa>Greta    # bloc at 4
  13:Bruno>Diego>Elsa>Ana>Greta    # bloc at 8
  9:Diego>Elsa>Bruno>Greta>Ana    # bloc at 12
  12:Elsa>Diego>Greta>Bruno>Ana    # bloc at 16
  8:Greta>Elsa>Diego>Bruno>Ana    # bloc at 20
  7:Greta>Elsa>Diego>Bruno>Ana    # bloc at 24

expected_winners:
  - Elsa