RCV-IRV vs. STAR — A Side-by-Side¶
A neutral overview routing to the detailed facet pages. The honest summary: each method has real, documented strengths the other lacks — this is about specific trade-offs, not one being good and the other bad.
Both ask more of voters than Choose-One, and both usually beat plain plurality. They differ in what the ballot captures and how it's counted.
| RCV-IRV (Hare) | STAR | |
|---|---|---|
| Ballot | Rank candidates in order (no ties) | Score each 0–5 (ties allowed) |
| Captures | Order only | Order + strength |
| Count | Eliminate fewest-first-choice, transfer, repeat | Sum scores → automatic runoff of top 2 |
| Equal preferences | ❌ Not allowed | ✅ Allowed |
| Precinct-summable | ❌ No (needs a central count) | ✅ Yes |
| Exhausted ballots | ❌ Possible | ✅ Avoided (blank = 0) |
| Monotonic | ❌ No (more support can hurt) | ✅ Yes |
| Later-No-Harm | ✅ Satisfies | ❌ Fails |
| Majority criterion | ✅ Satisfies | ❌ Can fail |
| Center squeeze | ❌ Vulnerable | ✅ Resists |
| Track record at scale | ✅ Widely used (Maine, Alaska, many cities) | ⚠️ Limited adoption to date |
Read that table both ways. The rows where STAR wins (summability, monotonicity, no exhaustion, center-squeeze resistance) are real; so are the rows where IRV wins (Later-No-Harm, the majority criterion, and a far larger real-world track record). No method escapes Gibbard–Satterthwaite — each has scenarios where honesty is punished.
Where each is genuinely stronger¶
IRV's honest advantages. It satisfies Later-No-Harm (ranking a backup can never hurt your favorite) and the majority criterion (a candidate who is the first choice of an outright majority always wins) — both of which STAR/Score can fail. It also has by far the larger deployment record in real public elections, which matters for legal precedent, vendor support, and voter familiarity.
STAR's honest advantages. It's monotonic, precinct-summable, doesn't exhaust ballots, lets voters express strength and equal support, and resists center squeeze. Its ballot is the familiar five-star format.
Each one's price. IRV's elimination mechanism produces center squeeze, non-monotonicity, exhausted ballots, and no precinct summability. STAR's price is failing Later-No-Harm, the majority criterion, and the Participation criterion (a no-show paradox — adding sincere ballots favoring a candidate can, in rare constructed cases, hurt them; this is the runoff's cost, and is not the same as monotonicity, which STAR does satisfy), plus an incentive toward strategic min/max scoring.
Note on Participation: the no-show paradox is a real but rare failure, and IRV fails Participation too — so it isn't a STAR-vs-IRV differentiator, just an honest caveat that STAR isn't flawless. The runoff that causes it is the same step that guarantees STAR rejects a Condorcet loser. No method escapes Gibbard–Satterthwaite.
The detailed facet pages¶
- Scores vs. ranks · Strict vs. weak ranks
- Tabulation, step by step — same ballots, both counts
- Center squeeze · Monotonicity · Summability
- Exhausted ballots · Residual vote-splitting
- Is RCV "simple"? · Is IRV "just plurality"?
Further reading — note on sources¶
These comparison pages are written by advocacy organizations; read them knowing each leans toward the method it champions, and cross-check claims against the facet pages above.
- Equal Vote Coalition — RCV vs. STAR (pro-STAR)
- STAR Voting — RCV vs. STAR (pro-STAR)
- FairVote (pro-RCV-IRV — the case for ranked choice)
A worked example of this kind of critical reading — a FairVote article on the Condorcet criterion, quoted claim by claim and checked against tabulated elections: FairVote's Condorcet article, claim-checked.