election_title: "Exercise 14 — The transfer machine, fully ranked (BV probe variant)"

scenario_description: |-
  The same nine voters as ex14_two_novels.yaml with every ballot completed
  to a full ranking — 5×(Austen>Bronte>Camus>Dickens), 1×(Bronte>Camus>
  Austen>Dickens), 3×(Camus>Dickens>Bronte>Austen). The added trailing
  rankings are never reached by any transfer, so the quota, the rounds
  and the seats are identical to the exercise — under whichever of the
  two published Droop quotas the count applies: Austen elected in round
  1, her surplus transferring to Bronte, and Camus taking the second
  seat. Seats: Austen + Camus.
  This variant exists as a bug probe: BetterVoting's STV tabulator
  returns a server error on the exercise's truncated ballots (BV2201,
  tk776t) — and, it turns out, on this fully-ranked copy too (BV2202,
  bj8dfc), acquitting truncation. The bug has since been fully DIAGNOSED
  (probes BV2203-2205): any STV count whose eliminations leave ONE
  remaining hopeful who then reaches quota crashes — BV's IRV.ts
  redistributes the winner's surplus over an empty candidate list, and
  [].reduce with no initial value throws. This count is exactly that
  shape (Camus elected last and alone). Full bisection, evidence table,
  and ready-to-file issue: 06_Other/STV/bv_stv_sole_survivor_crash/;
  narrative: ex14_transfer_machine.md.
  Live on BetterVoting (Test ID BV2202): https://bettervoting.com/bj8dfc
  — the vote page works; /results currently errors (the bug above).
voting_method: STV
num_winners: 2
bv_test_id: BV2202
bv_election_id: bj8dfc
bv_results_url: https://bettervoting.com/bj8dfc/results
ballots: |-
  5:Austen>Bronte>Camus>Dickens
  1:Bronte>Camus>Austen>Dickens
  3:Camus>Dickens>Bronte>Austen
expected_winners:
  - Austen
  - Camus

# file: ex14_two_novels_fullranks.yaml