election_title: "The field empties in round two — and Pareto is what keeps Dev out of the tie"

scenario_description: |-
  A total batch elimination is not a round-one curiosity. Six voters, four
  candidates: Alex, Bella and Colin rotate the top three between them, two voters
  each, and every single ballot ranks Dev dead last.
  Round 1 is ordinary. Alex 2, Bella 2, Colin 2, Dev 0 — Dev is the unique fewest,
  so he is eliminated on his own, and because nobody ranked him first there is
  nothing to transfer. Round 2 is where it stops: Alex 2, Bella 2, Colin 2,
  majority is 4 of 6, nobody has it, and all three remaining candidates are tied
  for fewest first choices. Batch elimination removes the entire remaining field,
  and the count ends with all three tied for the win.
  Dev is not in that tie, and the reason is worth naming: every voter prefers all
  three others to Dev, so PARETO excludes him. That is precisely the role Pareto
  plays in Moulin's forced-tie proposition — anonymity and neutrality force a tie
  among the symmetric candidates, and Pareto is what stops a universally-rejected
  candidate from being swept into it. See
  07_Concepts/topics/ties/ties_are_forced.md.
  Verified with pref_voting (Holliday & Pacuit): instant_runoff = {Alex, Bella,
  Colin}, coombs = {Alex, Bella, Colin}, copeland = {Alex, Bella, Colin}. Coombs
  lands in the same place by the mirror route — Dev carries all 6 last-place votes
  and goes first, after which Alex, Bella and Colin have 2 apiece and tie.
  This engine does not print those rounds. The vendored pyrankvote collapses the
  whole thing into ONE round — it rejects Colin and Dev side by side, even though
  Colin was tied at the top of the remaining field and Dev was alone at the bottom
  — and elects Alex 4-2. So the round-2 structure described above is the METHOD's;
  read it off pref_voting, not off the report below. expected_winners records what
  this engine prints for this row order, not the method's answer.
  Lesson: 07_Concepts/topics/ties/batch_elimination.md

video_script: |-
  Four candidates this time, six voters, and one candidate nobody likes.
  Alex, Bella and Colin rotate the top three spots, two voters each. Dev is last
  on every single ballot.
  Round one is completely normal. Dev has zero first choices, he is clearly the
  weakest, out he goes. Nobody ranked him first, so no votes move.
  Round two is where it falls apart. Alex two, Bella two, Colin two. You need four
  for a majority. Nobody has it, and nobody is behind. Eliminate everyone tied for
  last and there is no field left. All three tie.
  Notice who is NOT in that tie. Dev. And the reason is a principle worth knowing
  by name: every voter, without exception, preferred the other three to Dev.
  That is the Pareto condition, and it is the thing standing between "we could not
  separate the good candidates" and "so let's call it a four-way tie."
  The tie is forced. Dev's exclusion from it is not.

voting_method: RCV_IRV
num_winners: 1

ballots: |-
  Alex>Bella>Colin>Dev
  Alex>Bella>Colin>Dev
  Bella>Colin>Alex>Dev
  Bella>Colin>Alex>Dev
  Colin>Alex>Bella>Dev
  Colin>Alex>Bella>Dev

expected_winners:
- Alex
