Seeing The Results

A ranked-choice count doesn't produce a number. It produces a story: who led at the start, who was eliminated, where their supporters' votes went next, and how that changed the standing. The winner is the last line of that story, and often the least interesting one.

So RCV Lab doesn't have "the results screen." It has several views of the same count, each answering a different question. Every one below is live — the real component, rendering a real election.

The round-by-round table

The canonical record. Every candidate, every round, the vote totals and what transferred between them, plus the threshold for winning and the number of ballots that became inactive. If you only look at one view, this is the one that contains everything.

It's also the view designed to be checked. The same numbers appear in the downloadable results, and it follows the layout used by official RCV tabulations, so anyone familiar with those can read it without instruction.

Where each candidate's support came from

The bar chart answers a question the table makes you work for: a candidate gained 400 votes this round — from whom? Each bar is broken into segments showing which eliminated candidate each block of votes arrived from.

This is the view that makes coalitions visible. When one candidate's supporters break heavily toward a particular rival, you see it as a thick band rather than by comparing two columns of numbers.

One round at a time

The pie shows a single round as a whole — every candidate's share of the active vote at that moment — and animates the transition to the next one. As a candidate is eliminated, watch their slice break apart and travel to the candidates their supporters ranked next.

It's the best view for explaining ranked choice to someone who has never seen it, because it makes the central mechanic visible: votes move, and nobody's ballot is thrown away while it still has a preference left.

The whole flow at once

The Sankey diagram is the story compressed into one picture. Each candidate is a band, each round a step, and the ribbons between them are votes moving. You can trace a single candidate's supporters from the first round to the last.

What the ballots say beyond the winner

The statistics view steps back from the count and describes the electorate: how deeply people ranked, how many used every position available, how many ballots became inactive and when.

It also runs the head-to-head comparison — every candidate against every other, one pair at a time, asking which one more voters preferred. If somebody beats everyone, they're a Condorcet winner and the view says so. Sometimes nobody does, because the preferences form a cycle: A beats B, B beats C, and C beats A. That's a genuine property of the electorate rather than a flaw in the count, and it's worth knowing when it happens.

The election above was built to demonstrate exactly that: its candidates are the rival methods for breaking Condorcet cycles, and they form a cycle among themselves.

Comparing two rule sets

Every view can be shown in Split View — the same ballots counted two different ways, side by side, with the rounds stepping together. Change the quota, switch from WIGM to Meek, turn equal ranking on, and watch what actually differs.

This is the honest way to answer "would a different rule have changed the outcome," and usually the answer is no, which is itself worth being able to show. When it is yes, you can see precisely which round diverged. The Analysis page has a library of real elections to try it on, and the methods guide explains what each rule does.

Two more views, when the ballots are available

If you tabulated from ballots rather than from a published summary, two extra views open up. Initial rankings shows the first-choice picture before any elimination — the raw shape of opinion. Ballot trace follows individual ballots through the rounds, which is the view that answers "what actually happened to the people who ranked this candidate first."