election_title: "Kissel's five-way example (RCV-IRV) — the compromise is eliminated third"
scenario_description: >-
  The five-candidate field printed on p.5 of Adam Kissel's "Can Ranked-Choice Voting
  Work? A Conservative Approach" (Cardinal Institute for West Virginia Policy), given
  ballots. The paper's shape: A >30%, B 30%, C 20%, D <19%, E <1%. Here A 306, B 300,
  C 202, D 183, E 9 of 1000. C is the moderate — the SECOND choice of both A's and B's
  voters — and beats every rival head-to-head (the Condorcet winner: 511-489 over A,
  700-300 over B). RCV-IRV never finds that out. Round 1 eliminates E, round 2 D,
  round 3 C, and A wins 609-391. The paper says C could only win "if a large majority
  of the D voters choose C"; that is the wrong diagnosis. C's problem is not D's
  transfers, it is that the A and B voters' second choices are never counted at all.
  Same ballots under Ranked Robin (…_rr.yaml) and STAR (…_star.yaml) elect C.
bv_test_id: BV2278
bv_election_id: 8cdkkc
bv_results_url: https://bettervoting.com/8cdkkc/results
voting_method: RCV_IRV
num_winners: 1
# Round 1: A 306, B 300, C 202, D 183, E 9 — no majority; E eliminated -> D (194... 192).
# Round 2: D lowest (192) -> D's 183 to A (489), E's 9 ride on to C (211).
# Round 3: C lowest (211) -> 111 to A, 91 to B, 9 to A.  A 609, B 391. A wins.
# The Condorcet winner C is eliminated one round before the end.
ballots: |-
  306:A>C>D>B>E     # A-partisans — moderate C is their second choice
  300:B>C>D>A>E     # B-partisans — moderate C is their second choice too
  111:C>A>B>D>E     # moderates leaning A
   91:C>B>A>D>E     # moderates leaning B
  183:D>A>C>B>E     # D's voters lean A
    9:E>D>C>A>B     # the <1% candidate the paper says is eliminated first
expected_winners:
  - A

# file: bv2278_8cdkkc_five_way_irv.yaml