Skip to content

BV2158 — IRV buries the centrist — Ossipoff's 303-voter one-dimensional example

Method: RCV-IRV (Instant Runoff) · 1 seat · Expected winner: D · full count →

▶ Live on BetterVoting: vote · results ↗ (election gr72hd, Test ID BV2158; STAR is race 1, and every race matches the LH tabulation below).

Level 301 · the "core support" defense, demolished. IRV defenders excuse center squeeze by saying the eliminated moderate "lacked core support" (first-choice votes). This example (Mike Ossipoff, via rangevoting.org §12) kills that defense: candidate C has the most first choices of anyone and beats every rival head-to-head — and IRV eliminates C anyway, electing D.

A ranked-ballot case (no scores invented). → fairness test: Reading these fairly — the test for an honest "whoops" · cousin: center squeeze · README.

The ballots (303 voters, a 1-D spectrum A…E)

 50 : A>B>C>D>E
 51 : B>A>C>D>E
100 : C>D>B>E>A
 53 : D>E>C>B>A
 49 : E>D>C>B>A

A realistic "one-dimensional politics" taper — support falls off with distance from each voter's favorite. Source: bv2158_gr72hd_ossipoff_centrist_irv.yaml.

C is the favorite by two measures — and still loses

  • Plurality winner: C has 100 first-choices, more than any other candidate (A 50, B 51, D 53, E 49).
  • Condorcet winner: C beats every rival head-to-head — A 202–101, B 202–101, D 201–102, E 201–102 (≈ 2-to-1 each).
  • IRV winner: D. Here is the elimination, straight from the engine:
Round 1:  C 100 · D 53 · B 51 · A 50 · E 49      → E eliminated, flows to D
Round 2:  D 102 · C 100 · B 51 · A 50            → A eliminated, flows to B
Round 3:  D 102 · B 101 · C 100                  → C eliminated (!), flows to D
Final:    D 202 · B 101                          → D wins

C — the plurality leader and the candidate a 2-to-1 majority prefers over everyone — is knocked out in the last round and the win goes to D, whom most voters rank fourth or fifth. The "core support" alibi fails on its own terms: C had the most core support.

Why it's a fair, not cheap, example

Sincere ballots, a textbook 1-D electorate (the source calls it "highly realistic"), and the failure is structural — it's the same center-squeeze mechanism, just dramatic enough to make the point unmissable.

Reading this fairly - How common: structural — center squeeze is a whole region of 1-D electorates; this is a vivid instance, not a knife-edge. - Sincere or strategic: fully sincere. - What IRV does well: guarantees the final pair has majority support and is simple to explain; the failure is specific to squeezed centrists. - The symmetric whoops: STAR isn't spotless — see Whoops 02, where STAR itself misses the Condorcet winner. The honest tally is in the README balance ledger.