Skip to content

The STAR Whoops Files 🍿 — the manufactured hit-reel against STAR

⚠️ Deliberately unfair, by design — and it's ours. This is the mud-slinging reel a STAR opponent would compile: every warped, contrived, worst-day case against STAR Voting, curated with zero charity. We wrote it about our own method on purpose, because the companion reel does the same to RCV-IRV, and every voting method has a hit-reel — mud is mud. Two honest promises, same as the other page: every "whoops" below is real and documented in this repo (STAR's limits are conceded openly, not hidden — that's what makes the fair case trustworthy), and the grown-up version is always one click away in STAR's honest limits. Enjoy watching us throw mud at ourselves.

STAR is a very good method that this whole library argues for. So here, for balance and sport, is the case for the prosecution — assembled as unfairly as we assembled the one against IRV.

🥇 The criterion failures (STAR really does fail these)

  • Not Condorcet-compliant. STAR can decline to elect the candidate who beats everyone head-to-head — the score-leader can win the runoff over the pairwise winner. → three notions of winner, and the Edelman / Condorcet-myth and FairVote cases.
  • Fails Later-No-Harm. Scoring your second choice can help defeat your favorite. (STAR's camp calls this a feature, not a bug — but a critic gets to list it, and we're being the critic.) → favorite betrayal 301.
  • Fails the majority-favorite and mutual-majority criteria. In constructed electorates, a majority's first choice — or a majority-preferred bloc — can lose. → honest limits.
  • Fails mono-raise-delete. STAR passes ordinary monotonicity, but the stronger Woodall variant it flunks: raise X and delete the now-lower loser, and X can lose. (Our own monotonicity worked set demonstrates it on STAR — we didn't even make IRV do all the suffering.)

🥈 The strategy seams

  • Coordinated burial can squeeze a centrist. Give the strong compromise a 0 and inflate the wings, and STAR can drop the Condorcet winner out of the runoff — French 2017 & Washington 2010, and the 5-1-0 challenge. (Yes, it takes a conspiracy; the prosecution doesn't mention that.)
  • The Chicken / Burr dilemma. Two allies who must beat a third can still be nudged toward a defection standoff — STAR softens it but doesn't abolish it. → chicken dilemma.
  • Turkey-raising exists as a concept even if STAR resists it — a critic will still put it on the slide.

🥉 The "your ballot" gotchas

  • Equal Support = "your vote didn't count." Score both finalists the same and you don't tip their runoff — which an opponent frames as nullifying your vote. → are equal-score votes discounted?.
  • Runoff reversal. The candidate with the most stars overall can lose the automatic runoff. (It's the runoff doing its job — catching a polarizing score-leader — but "the top-scorer lost!" makes a fine attack line.) → runoff-reversal cases.
  • Scale granularity flips the winner. The same opinions on a 0–5 vs. a wider scale can elect different candidates — so the ballot's resolution is part of the outcome. → scale granularity.
  • Ties and dead rungs. STAR can land on genuine ties that need a tiebreak (lot). → tie-break / dead-rung cases.
  • Participation / no-show. In rare constructions, STAR can hand a voter a worse result for showing up. → participation paradox set.

The obligatory conscience clause (and the honest asymmetry)

Fine — the fair version, because turnabout demands it:

  • STAR's limits are real, and STAR concedes them out loud — that candor is a strength, not a weakness. The whole honest-limits page exists so nobody has to compile this reel in secret.
  • But apply the four-part test and the asymmetry shows: most of the above are contrived, strategic, knife-edge, and never observed in a real STAR election. The IRV reel's headliners — Burlington 2009, Alaska 2022, San Francisco 2020 — are structural, sincere, and actually happened on certified ballots. Both reels are unfair; only one has real bodies.
  • A couple of the "whoops" are features a critic reframes as bugs — failing Later-No-Harm is how STAR rescues a compromise; runoff reversal is the runoff working.
  • And STAR still beats Choose-One by a mile. The honest debate is between good options.

There. We threw mud at ourselves as hard as at anyone else — which is the only way the fair pages earn the right to be believed. For the version you'd actually cite, use STAR's honest limits.