election_title: "Block Voting (3 seats): the majority ceiling — a unanimous candidate holds only 33%"
scenario_description: |-
  Why "the winner got less than half the votes" cannot be read as vote splitting
  in a MULTI-SEAT choose-one race — the arithmetic forbids it.

  30 voters, 3 seats, 5 candidates. Each voter marks 3 names (Block Voting), so
  90 votes are cast in total. Alice is on EVERY single ballot — literal
  unanimity, the most consensus a candidate can possibly have.

  And Alice's share of votes cast is 30 / 90 = 33.3%.

  With k marks per voter the total is k x voters, so no candidate can exceed
  1/k of votes cast no matter how popular: 33% at 3 seats, 25% at 4, 9% at 11.
  A screening rule that flags every race whose leader is under 50% therefore
  flags 100% of multi-seat block-vote races — including unanimous ones. That is
  a denominator artifact, not evidence of a split.

  This does NOT clear Block Voting. Its real defect is the opposite one: a
  bare majority sweeps every seat (see mmp_block_voting.yaml, 60/40 -> 3-0).
  The point is narrower and cuts both ways — the sub-majority test is blind
  here, missing the actual pathology while flagging consensus as a failure.
voting_method: Plurality
num_winners: 3
lot_numbers: [Alice, Bruno, Cleo, Dev, Esme]
ballots: |-
  Alice,Bruno,Cleo,Dev,Esme
  12: 1,1,1,0,0
  10: 1,1,0,1,0
  8:  1,0,1,0,1
expected_winners:
  - Alice
  - Bruno
  - Cleo

# file: mmp_majority_ceiling.yaml