Skip to content

Round-robin, Copeland, Condorcet, Ranked Robin — a naming decoder

One family of voting methods, half a dozen names, and two Wikipedia articles that overlap so much they look redundant. If "the Condorcet family" has ever left you unsure whether round-robin voting, Copeland's method, and Ranked Robin are the same thing or different things — you're not confused, the names genuinely blur. This page maps them, links where each is defined, and flags the sources to read critically.

Level: reference · deep dive (a 201/301 aid). The teaching page for the method itself is Ranked Robin; this is just the terminology map.

This page maps what each word means. For the argument about which one we should use — and what each choice costs — see What should we call this method?.

The one-line answer

round-robin voting (the family) ⊃ Copeland's method (the algorithm) ≈ Ranked Robin (a branded Copeland variant).

They're at different levels of generality — a category, a specific method in it, and a brand name for a tweaked version of that method. Say "Condorcet" or "round-robin" when you mean the family; say "Ranked Robin" only for the specific branded method.

The names stack like sports leagues

Round-robin is a tournament format: everyone plays everyone, best record tops the table. Every league that uses the format has its own name and its own tiebreak fine print. These names work exactly that way (the map below links each one — this list is the spoken version):

  • Condorcet — the format, i.e. the family. Any rule that elects the candidate who beats all others head-to-head, whenever one exists. Members include Copeland, Schulze, Ranked Pairs, Minimax.
  • Copeland — one member: rank candidates by head-to-head wins. With the margins tiebreak it is often written Copeland//Borda — but whether "Borda" belongs in that name is contested, including by the person who proposed the tiebreak.
  • Ranked Robin — a campaign name (Equal Vote, 2021) for Condorcet voting on a ranked ballot. Their recommended count as of 2026 is Copeland plus the margins tiebreak — a current default, which they say is under review, not a definition.
  • Consensus Choice — a different campaign's name (Better Choices for Democracy) for the same family, with a different cycle rule ("Most Wins, Smallest Loss"). Same tournament, different fine print.

Spoken aloud in ten seconds: Condorcet is the format, Copeland is the standings rule, Ranked Robin is the league we play under. The members only disagree when nobody goes undefeated — otherwise they all crown the same winner.

The map

Term What level Who uses it Defined at
Condorcet method(s) the family (traditional academic name) social-choice theory Wikipedia
Round-robin voting the family (a newer umbrella name) Wikipedia's recent reframing; reform advocates Wikipedia
Pairwise / paired comparison the family (plain-descriptive) general; both spellings in use Wikipedia
Tournament voting the family (plain-descriptive, the sports metaphor made literal) Wikipedia's round-robin article lists it alongside paired comparison Wikipedia
Instant Round-Robin Voting (IRRV) ambiguous — pin it down before relying on it. electowiki defines a specific method (ranked ballots, approval ballots as tiebreak); the term is also used for Condorcet methods generally, and sometimes for Schulze the Condorcet Canada Initiative flies it as its banner tagline; Wikipedia redirects the term to Round-robin voting electowiki
Copeland's method the algorithm — most head-to-head wins (½ per tie) academic; originated by Ramon Llull ~1299 Wikipedia
Llull method / Llull voting the same algorithm, credited to its 1299 author instead of its 1951 one rare in the wild, but not fringe: Wikipedia's Copeland article leads with "The Copeland or Llull method" Wikipedia
Ranked Robin (RCV-RR) a branded method — Copeland + a defined cycle tiebreak (sum of win margins) Equal Vote Coalition, 2021 — tiebreak proposed by Sass, name coined by Sara Wolk electowiki · equal.vote
~~Consensus Voting~~ unsourced — do not use. This library listed it for months as an Equal Vote alias. It is not on equal.vote, not on electowiki's Ranked Robin page, and not used by Better Choices for Democracy (who say Consensus Choice). Checked 2026-07-29. It is a garbling of Consensus Choice below, whose own page is titled "Consensus Choice Voting" — drop the middle word and you get the phantom. — —
Consensus Choice a sibling brand (Copeland-family, different tiebreak) Better Choices for Democracy betterchoices.vote · FAQ (the cycle rule)
~~RCV-RR~~ this repo's house shorthand — not an external term — —

Other members of the round-robin / Condorcet family — Minimax, Ranked Pairs, Schulze, Kemeny — sit beside Copeland; they differ in how they use the same pairwise results (defeat-dropping or path strength rather than simple win-counting). See the ranked-ballot zoo.

"RR" ≠ "round robin" — the abbreviation collision. In this repo RR always abbreviates Ranked Robin, one branded method. Round robin is the tournament format (everyone plays everyone) and, as "round-robin voting," a name for the whole family. So every Condorcet method is a round-robin count; only one of them is Ranked Robin. Using "RR" for "round robin" silently promotes a brand to the family name — the exact error this page exists to prevent. (And RCV-RR is house shorthand, built to parallel RCV-IRV; say Ranked Robin to people and Copeland to academics and the engine.)

The same collision, one letter over: "IRRV." Instant Round-Robin Voting repeats the Ranked Robin story almost beat for beat — a friendly tournament name, coined for advocacy, that some sources attach to one method and others to the whole family. electowiki gives it a specific definition (ranked ballots, approval ballots as the tiebreak); Wikipedia treats it as a redirect to the family; the Condorcet Canada Initiative uses it as a banner. So IRRV and RR are both names that shrink or swell depending on who is speaking — which is the argument for saying Copeland when you need to be exact, whichever brand your audience arrived with. (electowiki here is a community wiki and reform-adjacent — right tier for what a brand name means, wrong tier for a verdict; condorcet.ca is an advocacy site, cite it for its own usage only.)

The winner has aliases too — and they are not the method's

A separate set of names describes the candidate, not the count, and they are interchangeable in a way the method names above are not:

Condorcet winner = Condorcet candidate = pairwise champion = beats-all winner — the candidate who beats every other head-to-head. All four are the same stipulated definition, and none of them names a method: Minimax, Schulze, Ranked Pairs and Copeland all elect this person whenever one exists. Reach for "beats-all winner" with a general audience — it is the only one of the four that explains itself — and note that it inherits the caution from the "consensus" table below: it is a pairwise-majority fact, not a claim about warmth. → Condorcet winner criterion (electowiki).

Three more aliases are in circulation — and unlike the four above, they are not safe to use interchangeably. Wikipedia and reform sources increasingly gloss the Condorcet winner as a majority winner, a majority-preferred candidate, or a tournament winner. Each names the right candidate and imports a wrong idea:

Alias Names the right candidate, but… Say instead
majority-preferred candidate Scope-ambiguous, and this repo uses it the other way. Preferred by a majority over every rival (the whole field) is the Condorcet winner. Preferred by a majority over one other candidate is a single pairwise result — which is what STAR's automatic runoff delivers, and what the ~15 uses of "the majority-preferred finalist" across this library mean. The two are not the same, and STAR is not Condorcet-compliant. beats-all winner for the field; majority-preferred finalist — keep the noun — for a runoff
majority winner Collides with the ordinary meaning of majority: over half. A Condorcet winner may hold no absolute majority of first choices, and a candidate who does hold one is a different animal — the subject of the majority criterion, which STAR fails and Condorcet methods pass only incidentally. This library's house rule on the word is to say majority only for over-half. beats-all winner; reserve majority winner for >50%
tournament winner "Tournament" already has a technical job here: the C1 tier reads only the tournament graph, and a tournament solution usually returns a set, not a winner. The decoder's own table also files tournament voting as a name for the family. beats-all winner; keep tournament for the graph and the C1 tier

The pattern worth noticing is the same one this page flags for the "round-robin voting" umbrella: the newer names are warmer and less mathematical than the ones they replace, and two of these three trade a precise pairwise fact for the word majority. That is a real gain in accessibility and a real loss in precision — worth making deliberately rather than by drift. Judge the swap on what the word does, not on who prefers it.

One notation footnote, since it changes the arithmetic. Modern Copeland scores a matchup 1 / ½ / 0 (win / tie / loss). Llull's original was 1 / 1 / 0 — a tie counted as a win for both. Same tournament, different standings, so a source using Llull's scoring can crown a different candidate on a ballot set with ties. Worth checking which one an implementation means before comparing results.

Where the name came from — and what its coiner says it means

Equal Vote's own wording has moved. Their current Ranked Robin page now opens by calling it "a modern name for one of the oldest voting methods out there. First described in the literature in 1299" — wording their previous version did not have; that page never mentioned Copeland or 1299 at all.

The coiner says it is a rebrand of the family, not one fixed count. On electowiki's talk page in 2025, Sara Wolk — who coined the term — wrote that she "always intended the name Ranked Robin to be a rebrand of Condorcet," leading with the takeaway "Ranked Robin is a synonym for Condorcet on a ranked ballot." So Copeland-plus-margins is Equal Vote's default recommendation, which she notes is under review — not the whole meaning of the name. That is exactly why this page files Ranked Robin at the species level and Condorcet at the genus, and why a page that says "Ranked Robin is Copeland" is describing today's recommended count, not a definition.

And the objection is as old as the name. That same talk page opens with Markus Schulze objecting in 2021 that the title is misleading, since "round robin" has long covered Condorcet methods generally — the abbreviation collision flagged above, raised at the moment of coining rather than in hindsight.

(Sourcing: these points come from electowiki's Ranked Robin page and its talk page — a community wiki, Equal-Vote-adjacent. That is the right tier for what a branded name means, and the wrong tier for a verdict on the method; see the leans, below.)

Is the margins tiebreak "Borda"? — contested

Sources that call Ranked Robin Copeland//Borda and sources that reject the label are both easy to find, so this page names the dispute instead of picking a winner.

For the label. On the talk page's Black's-method thread, Kristomun derives it in January 2022: a finalist's sum of margins over the other finalists is, on a complete ballot, linearly related to that candidate's rank position — which is what a weighted positional system is, i.e. Borda. On that reading Copeland//Borda is just the technical name, and Ranked Robin sits beside Black's method (Condorcet//Borda) with Copeland swapped in at the first stage. Note the hedging in the original — "I seem to recall," "I think" — this is a wiki contributor reasoning in public, not a settled citation.

Against it. Sass, who proposed the tiebreak, edited the article in June 2024 to say the step after finding the finalists is "NOT Borda" — on the grounds that Borda carries contested rules about explicit ranks, ballot-completion requirements and skipped ranks that Ranked Robin does not inherit. That edit summary grants the arithmetic, calling it mathematically equivalent to tournament-style Borda, while denying the name and describing it as a novel application of the metric.

How to use this. The two sides barely disagree about the arithmetic; they disagree about whether a method name drags its parent's fine print along with it — the same species of question this entire page exists to answer. Safe phrasing: call the tiebreak the sum of win margins, which nobody disputes, and reach for Copeland//Borda only when locating the method in the academic taxonomy — noting that Equal Vote itself does not use that term.

The tier the brands never mention (C1 / C2 / C3)

The table above maps the marketing names. The academic literature sorts the same family a different way — by how much of the pairwise data a rule actually needs — and knowing this one scheme, from Fishburn (1977), stops most cross-talk before it starts:

Class Reads Members
C1 only who beat whom (the tournament; margins discarded) Copeland → Ranked Robin, Smith set, Schwartz set, Top Cycle
C2 the margins too Minimax, Ranked Pairs, Schulze, Kemeny, Split Cycle
C3 must return to the ballots and iterate Dodgson, Young

Two things fall out of it. "Tournament solution" means the C1 tier specifically — not the whole family — so a theorem proved about tournament solutions doesn't automatically apply to Minimax. And the tiers only diverge in a cycle: when a Condorcet winner exists, every rule in every tier elects the same person, which is why so much of the literature argues about a rare case. → the full reading list, with sources and leans.

The word "consensus" carries three different jobs

The table above strikes Consensus Voting as an unsourced phantom. But the word itself shows up in this repo, and across the web, in three senses at different altitudes — and sliding between them is the single most common source of confusion in this family:

Sense What it means Careful
Technical "consensus winner" = the Condorcet winner — beats every other candidate head-to-head A definition, not a description. It's a stipulated label for a pairwise-majority fact.
Brand Consensus Choice = Better Choices for Democracy's brand for the family; Equal Vote's is Ranked Robin. ("Consensus Voting" is neither — see the struck row above.) Two different organizations, two different tiebreaks. Not interchangeable at the byte level.
Ordinary English broad, warm, genuinely-shared agreement — everyone is content Condorcet methods cannot see this. A ranked ballot carries order, not intensity, so the count can't tell a warm second choice from grudging tolerance — see honest limits.

The claim to watch for, common on advocacy sites and forums:

"Condorcet methods are designed to make the best use of limited ranking information to find the consensus."

The first half is a fair steelman — and worth conceding cleanly. It admits ranking is limited information and claims Condorcet extracts the most available from it, which is true: a round-robin reads every rank against every opponent, and nothing exhausts. That's a real advantage over IRV, which reads one rank per ballot per round.

The second half slides from sense 1 to sense 3. "Finds the consensus" is true by definition (whoever wins the pairwise comparisons is called the consensus winner) and unproven in ordinary English (that person may be everyone's tepid second choice). The method doesn't discover consensus; it elects the pairwise-majority winner and the brand supplies the warmer word.

Where the two reform camps part ways is exactly here — and both answers are respectable:

  • Ranked Robin / Condorcet: extract maximum value from limited information. Keep the familiar ranked ballot; count all of it.
  • STAR: don't limit the information. Let the ballot carry degree (0–5), not just order, so intensity is visible at all.

That's a genuine design disagreement about ballots, not a dispute about counting — see scores vs. ranks and preference vs. support.

Looks like Condorcet — but isn't (STAR, 3-2-1)

Two rated methods end with a head-to-head step, which can fool you into filing them under Condorcet. They aren't. Both first narrow the field, then run a pairwise comparison only among the survivors — so the true Condorcet winner can be eliminated before the final:

  • STAR — add 0–5 scores, take the top two by total, then a pairwise automatic runoff. The runoff is pairwise, but the finalists are chosen by score sum, not by beating everyone — so STAR is not Condorcet-compliant. → STAR · three notions of winner.
  • 3-2-1 — Good/OK/Bad, then most-Good → fewest-Bad → pairwise. Same shape: a pairwise final, but the two finalists come from the rating steps, so a Condorcet winner outside them never gets there. Not Condorcet. → 3-2-1 voting.

The tell: a Condorcet method compares every pair across the whole field; these compare one pair among pre-selected finalists. A pairwise step ≠ a Condorcet method.

Why the two Wikipedia articles overlap (the trap you hit)

Round-robin voting and Copeland's method feel redundant for a concrete reason: one is the family, the other is its simplest member — and Copeland is the method that most literally implements "round-robin tournament standings," so the category name and the method sound identical. Two extra wrinkles:

  • "Round-robin voting" as the umbrella name is a recent Wikipedia reframing of what was long just called "Condorcet methods." That renaming has seen editorial back-and-forth (it makes the family sound friendlier and less mathematical), and the churn is part of why the articles feel unsettled.
  • Neither article pins down "Ranked Robin" — it's a brand for a Copeland variant, so its natural home is the Copeland article as a sourced aside, not the category article. (If you're tempted to add it: it rests almost entirely on advocacy sources, so it needs an independent one, and — for a STAR-aligned editor — a conflict-of-interest disclosure and a Talk-page proposal.)

Read the sources knowing where they lean

  • electowiki is the canonical definition for the Ranked Robin name — but it's a community wiki, reform-advocacy-adjacent. Cite it for definitions; lean on academic sources for critical/limits claims.
  • equal.vote is the promoter (STAR / Equal Vote camp) — the source for the brand, not a neutral referee.
  • Wikipedia / academic ("Condorcet method," "Copeland's method") is the neutral ground — but note the "round-robin voting" umbrella framing is itself contested.

(Fuller roster of who-leans-where: advocacy organizations · who's who.)

The clear version, in this repo