Ranked Robin (RCV-RR) — Honest Limits¶
One line: Ranked Robin (RCV-RR) is a strong Condorcet method — it reads every rank, allows equal ranks, passes the Equal Vote, and has no center squeeze — but it inherits the two structural limits of ranked, Condorcet counting: it can hit a cycle with no clear winner, and it sees only order, not strength, so it can crown a bland compromise.
→ Companion critical pages (parity across methods): STAR's limits · Approval's limits · RCV-IRV fails the Equal Vote. What RCV-RR is: Ranked Robin. Curriculum: 301.5.
Also called — and why it matters here. These limits belong to the method, not to the brand, so they follow it under every name it travels under. The algorithm is Copeland's method (Copeland//Borda once the margins tiebreak is attached), and after the man who described it in 1299 it is also the Llull method — Wikipedia's article opens "The Copeland or Llull method." The family it sits in answers to Condorcet methods, round-robin voting, paired comparison, and tournament voting. Two campaigns brand it further: Consensus Choice (Better Choices for Democracy — a genuinely different cycle rule, so not a straight synonym) and Instant Round-Robin Voting (IRRV), the banner of the Condorcet Canada Initiative, though that term is used loosely enough — sometimes for the whole family, sometimes for Schulze — to be worth pinning down before you rely on it. If you arrived here from any of those words, §1–§5 are still about your method. Which word means what: the naming decoder · which one we should lead with: what to call this method.
1. Condorcet cycles (no clear winner)¶
Sometimes there is no Condorcet winner: A beats B, B beats C, C beats A — a rock-paper-scissors "cycle." Then RCV-RR must fall back to a resolution rule (Copeland / margins), and any such rule is harder to explain and can feel arbitrary to a voter ("why did we break it that way?"). How much it matters: cycles are rare with real electorates, but they are the honest soft spot of every Condorcet method. See cycle resolution.
2. No preference strength — the "milquetoast" winner¶
A ranked ballot captures order only, not how much. RCV-RR elects the candidate who wins the most head-to-head matchups — which can be a broadly-acceptable compromise everyone ranks second and nobody is excited about, over a candidate a majority passionately prefers. That's sometimes the right, consensus-seeking answer; but the method literally cannot see intensity, so it can't tell "warm second choice" from "grudging tolerance." (This is the mirror image of STAR's design, which does read strength via the 0–5 score.)
3. Fails Later-No-Harm; open to burial¶
Like STAR and Approval, RCV-RR fails Later-No-Harm — ranking a later choice can help them. More specifically it is open to burial: insincerely ranking a strong rival last can flip a pairwise matchup in your favorite's favor. Honest ranking is usually best, but the incentive exists (Gibbard again: no method is strategy-proof).
Two things the topic hub adds, both worth knowing before conceding the point in a debate: no Condorcet method can avoid this — later-no-harm and Condorcet compliance are incompatible, so this is a family property, not a Copeland quirk — and it isn't ours alone: Borda is worse, STAR fails the same criterion, and only IRV largely escapes, by not reading your lower ranks at all. Worked on four methods side by side: Burial. This method's own worked pair: burial.
4. The count is harder to explain¶
"Add the stars" (STAR/Score) or "most approvals" (Approval) is a one-sentence tally. A pairwise matrix — every candidate against every other — plus cycle-handling is more to show a general audience, even though it's fully summable and auditable. Simplicity of explanation is a real adoption cost.
5. Clone independence — teaming (a narrow, tiebreak-specific failure)¶
RR passes vote-splitting (crowding) clone independence — because it counts head-to-head wins, cloning a candidate can't make that candidate lose. But it can fail teaming (running clones of yourself) in the one situation where there's no Condorcet winner and the winner turns on a margin tiebreak: a faction can field clones to reshape the margins and crowd a rival out of the top tier. Worth knowing, easy to overstate — and note it's implementation-dependent: an engine that breaks a 2-way tie by head-to-head (BetterVoting) resists the same attack that a margin tiebreak (LH / the Equal Vote protocol) succumbs to. Worked example: Clone independence — crowding, teaming, and the tiebreak.
Keep it in perspective¶
RCV-RR's strengths are exactly why it's worth taking seriously: it passes the Equal Vote, reads every rank (nothing exhausts), allows equal ranks, and avoids RCV-IRV's center squeeze and non-monotonicity. Its limits — cycles and blindness to intensity — are the genuine cost of a ranked, consensus-first count. Every method trades something away (Arrow, Gibbard); the fair question is which tradeoffs you want, laid out for all four methods on the companion pages above.