election_title: "Runoff reversal, the jarring case — the near-consensus favorite loses a 51–49 runoff to a polarizing rival"
scenario_description: |-
  A *jarring* runoff reversal — the genuine philosophical drawback of STAR, shown honestly. 100 voters, three candidates: Uma is a near-consensus, high-utility candidate (49 give her a 5, 51 give her a 4 — a 4.5/5 average), Rye is polarizing (51 love him, 49 reject him), Tao is a no-hoper. Uma dominates the SCORE round 449 to Rye's 255 (Tao 49) — almost twice Rye's total — yet the runoff is Uma vs Rye, and Rye wins 51–49: a bare majority prefers him head-to-head. A utilitarian says STAR just elected the candidate who makes the electorate less happy overall; a majoritarian says 51 > 49 is the most legitimate thing an election can report. STAR is a hybrid (cardinal ballot, ordinal finish) and this is exactly where those two values pull apart — STAR chooses majority preference over utility maximization. Concede it plainly; it's a real cost, not a bug. Companion reversal_convincing_c3_b100 is the *convincing* reversal.
  Lesson: 01_STAR/01_Learn/the_count/STAR_second_round_FAQ.md
voting_method: STAR
num_winners: 1
ballots: |-
  Count:Uma,Rye,Tao
  51:4,5,0    # love Rye, but Uma is a strong 4
  49:5,0,1    # love Uma, reject Rye
expected_winners:
  - Rye

# file: reversal_jarring_c3_b100.yaml