Proportional to what? — representation without parties¶

One line: party-list PR divides seats among parties, so the groups are declared before anyone votes; STAR-PR divides seats among quotas of voters, so the constituencies are emergent — discovered by the ballots, with nobody registering, naming, or even knowing they are in one.

→ the method: STAR-PR · seen side by side with the majoritarian count: Bloc STAR vs Proportional STAR · the harder version of this question: what "proportional" actually means

Level: 101 · for voters

The confusion, stated plainly¶

Proportional representation is almost always explained with parties: a party with a quarter of the votes gets a quarter of the seats. That is easy to picture, and most of the world's PR systems work that way.

Then someone says "Proportional STAR", hands you a ballot with nine individual names and no parties on it at all, and the picture stops working. Proportional to what? There are no parties to be proportional to.

This page is that one question.

The answer: the unit is a quota of voters¶

Forget parties. The thing being divided proportionally is not "party vote share" — it is the voters themselves.

Three seats, thirty voters. Then one seat is worth ten voters: the quota. Proportional STAR fills the council by repeatedly asking:

Who is the favourite of some ten voters who don't have a representative yet?

Elect that candidate, mark those ten voters represented, and ask again with the voters who are left. Three seats, three groups of ten, everybody's ten got someone.

That is the whole idea, and notice what it never needed: no parties, no slates, no registration, no declared factions. Just a seat is worth this many voters.

The groups are discovered, not declared¶

This is the part that feels strange coming from party-list PR, so it is worth saying directly.

In a party-list system, the groups exist before the election. Parties register, publish lists, and voters pick one. The count divides seats among groups that were named in advance.

In STAR-PR the groups exist only in the ballots. A "faction" here is nothing more than a set of voters who happened to score the same candidate highly. It has no name, no membership card, no headquarters. It may be:

an ideological camp, if that is what divides the electorate;
a neighbourhood, if geography is what people actually vote on;
everyone who cares about one specific issue;
something nobody predicted and only shows up in the numbers.

The ballots reveal which of these was real. That is why the method works without knowing in advance what the electorate is divided over — and why the same method serves a city council, a co-op board, and a club committee, none of which have parties.

The worked version¶

Look at the Left / Centre / Right example with this in mind. It is drawn as three camps because that makes it readable — but the count never sees a camp.

Ten voters, three seats, so a quota is 3.33 voters. Proportional STAR elects C1, L1, L2, and each of those results is about voters, not parties:

L1 and L2 win because the six left-leaning voters are worth about two quotas. Two seats. The count did not know they were "the left" — it saw six ballots that scored the same three names highly.
C1 wins on breadth, not on a bloc. The two centre voters are below quota on their own and could not elect anybody. But left and right voters also gave the centre candidates 1s and 2s, and those partial scores accumulate. C1 is nobody's favourite and many people's acceptable choice.
The right camp wins nothing, and that is the system working. Two voters is below the 3.33 quota. They have not earned a seat and no proportional method owes them one.

C1 is the case that makes the point. There is no "centre party" doing anything. C1 is elected by a group of voters who never coordinated, never registered, and are spread across all three camps — a group that exists only as a pattern in the scores.

What "represented" actually means — and a correction worth making¶

The natural way to describe the mechanism is: "if you scored the winner highly, your ballot is reduced; if you scored them 0, it keeps full weight."

That describes Reweighted Range Voting, not Proportional STAR. The distinction is the single most confusable thing on this page, and it is the difference between the two schools:

	How it decides who is "represented"	Your ballot after your favourite wins
Allocated Score (the recommended STAR-PR)	A quota's worth of ballots is spent outright — 5-star supporters first, then 4-star, and so on until the quota is full	Either fully spent or fully untouched. If the quota filled before reaching your score group, you keep 100% weight even though you scored the winner 5
Reweighted Range Voting	Every ballot is shaded in proportion to the support it gave	Reduced — a bit, or a lot, depending how highly you scored them

So under STAR-PR proper, being "represented" is not a matter of degree that everyone shares a little of. It is closer to a seat at a table: a quota of voters is seated with each winner, and everyone else walks away with a full vote still in hand. The only voters who lose a fraction are those in the score group that straddles the quota boundary — fractional surplus handling splits the remaining need evenly among them, precisely so that voters who scored the winner identically are treated identically.

That also answers the sharpest version of the question: if you gave the winner 3 stars and they won, were you represented? Only if the quota was still unfilled when allocation reached the 3-star group. Score groups are consumed from the top down; a big enough winner never reaches you, and your ballot goes into the next round intact.

The questions worth asking next¶

These are the right ones, and the honest answers are mixed:

How steep is the reweighting? Steepness is the entire design dial. Too steep and a cohesive majority loses seats it genuinely earned; too shallow and it sweeps. Allocated Score answers with a quota rather than a curve, which is why it passes the Hare Quota Criterion — a quota-sized group can always force a seat.
Does it punish honesty? The real question, and not fully settled. If scoring a backup highly can get your ballot spent on them, there is a theoretical pull toward bullet voting to preserve weight. It is weaker under Allocated Score than under continuous reweighting — being spent requires the quota to actually reach your score group — but "weaker" is not "absent", and the honest position is that this deserves testing rather than assertion. It is one of the open items on the committee's own list.
What happens to surplus? Nothing is wasted: the group on the quota boundary gets the excess returned and shared evenly, keeping partial influence in later rounds.
Can a human audit it? This is the weakest point and the repo says so elsewhere: STAR-PR is not batch-summable. Precincts cannot be tallied separately and added, because reweighting depends on individual ballots. Central tabulation is required — which is an honest cost of proportionality, not of STAR.

Why a 0–5 ballot matters here¶

A choose-one ballot can only tell you which single candidate each voter picked. That records group membership and nothing else — which is why choose-one systems drift toward needing parties to be meaningful.

A 0–5 ballot records how much each voter likes each candidate, so it can see:

partial support (the 2s that got C1 elected),
overlapping membership — a voter can be part of the group backing L1 and part of the group that finds C1 acceptable,
degree of representation, not just its presence.

Real electorates look like that. People are not cleanly sorted, and a ballot that only records sorting cannot represent people who aren't sorted.

What you give up¶

Being honest about the trade, because this is the cost of dropping parties:

With party lists you can promise a specific proportionality — this party got 25% of the vote and will get 25% of the seats, checkable afterwards by anyone. Non-partisan PR cannot make that promise about any named group. It guarantees representation to cohesive groups of voters large enough to reach a quota, but it cannot promise that any particular characteristic — party, geography, demographic — ends up proportional, because it never measured that characteristic in the first place.

That is not a defect being hidden; it is the direct consequence of letting the ballots decide what the electorate is divided over. The longer version, including what the research does and doesn't support, is what "proportional" actually means.