election_title: "Preference vs Support — the center SUPPORTED (wings score Blair 4)"

scenario_description: |-
  The "supported" half of a matched pair (companion: the TOLERATED election,
  bv2225_ywx39y_center_tolerated). Same three candidates on a spectrum — Alex (a pole),
  Blair (the center), Cole (the other pole) — and BYTE-IDENTICAL rankings to the
  tolerated election. The ONLY change: the two wings now score the centrist Blair a
  genuine 4 instead of a grudging 1. Because the orders didn't move, the ranked methods
  don't move either: RCV-IRV still center-squeezes Blair (-> Alex); Ranked Robin still
  finds him the Condorcet winner (-> Blair). But STAR reads the support — Blair's real
  4s lift him into the runoff, which he wins — so STAR flips from Alex (tolerated) to
  Blair (supported). Two elections, identical ranks, and only the score-reading method
  can tell that in one Blair is merely tolerated and in the other he is genuinely backed.
  Live results: https://bettervoting.com/82gg36/results

video_script: |-
  Hold the twin election up beside this one. Every ranking is the same; the ranked
  tallies (IRV, Ranked Robin) print the same winners. The only thing that moved is
  Blair's score from the wings — 1 to 4 — and STAR moved with it, from Alex to Blair.
  That's the payoff line: "The rank couldn't see it. The score did." Support is the
  information a ranking throws away.

paradoxes: [condorcet-winner]
voting_method: STAR
num_winners: 1

bv_test_id: BV2226
bv_election_id: 82gg36
bv_results_url: https://bettervoting.com/82gg36/results

ballots: |-
  Count:Alex,Blair,Cole
  15:5,4,0   # Alex > Blair > Cole   (SAME order — Blair now a strong 2nd, scored 4)
  6:1,5,0    # Blair > Alex > Cole   (Blair's base — unchanged in both elections)
  15:0,4,5   # Cole > Blair > Alex   (SAME order — Blair now a strong 2nd, scored 4)

expected_winners:
  - Blair

# file: bv2226_82gg36_center_supported.yaml