The Post-it election, seven ways — all four candidates win, depending on the method¶
▶ Live on BetterVoting: vote · results ↗ (election v8r66y, Test ID BV2177).
The same 20 voters as the Post-it RCV example (BV2176) — the electorate from Equal Vote's video "Updated: How does RCV work? — With Post-its!" — run through the single-winner methods side by side on BetterVoting, leading with STAR. The headline needs only the core five: all four candidates win, depending only on how the same preferences are counted.
One electorate, five single-winner counts, four winners¶
| # | Race (BV) | Method | BV winner | How |
|---|---|---|---|---|
| 1 | STAR | STAR | Blue | scores 46/38/44/44; 44–44 second-finalist tie → head-to-head Blue 10–3 Pink; runoff Blue 10–9 Purple (a Runoff Reversal) |
| 2 | Ranked Robin | RankedRobin (Copeland) | Green | cycle; Green and Blue tie 2–1; BV's head-to-head rung → Green 7–4 (LH's margin rung says Blue — the ladder divergence) |
| 3 | Approval (approve = any support) | Approval | Pink | approvals 12/10/10/8 — the six Green fans' 3-for-their-last-choice carries Pink |
| 4 | RCV-IRV | IRV | Purple | the video's whiteboard count: 7/6/4/3 → 8/7/4 → 9–8, 3 exhausted |
| 5 | Choose-One | Plurality | Purple | 7 first choices (35%) |
Read it as a ladder of how much of the ballot each method consults: Choose-One sees only first choices (Purple). RCV-IRV sees lower ranks, but only for candidates that survive its elimination order (still Purple — it never checks Blue vs Purple). Ranked Robin checks every head-to-head on the same ranked ballots (Green, on BV's tiebreak). Approval reads breadth of support but not strength (Pink). STAR reads strength and finishes with the head-to-head majority check (Blue). The tabulation, not the ballot, decides.
Completeness footnote: the election also carries BV's two multi-winner methods pinned to 1 seat, purely so every BetterVoting tabulator runs on this ballot set. Neither adds to the single-winner argument: STV at 1 seat is IRV by construction (identical rounds → Purple), and STAR-PR at 1 seat is plain Score voting (its per-seat rule is score-only, no runoff → the score leader Purple, and STAR's Runoff Reversal vanishes — a reminder that the runoff, not the scores, is what makes STAR majoritarian). Details below.
Three conversions, three lessons¶
The seven races use three ballot forms derived from one set of preferences — and each derivation quietly makes a choice:
- Ranks → the video's 0–5 scores (race 1): verified strictly order-consistent, ballot by ballot, with the video's rankings — but the gap sizes are editorial, and they matter. With the video's generous 5/4/3 down-ballot scores, Blue reaches the runoff and wins; re-score the same rankings stingily (5/2/1) and the scoring round is Purple 40, Green 34, Blue 32, Pink 26 — STAR elects Purple, agreeing with IRV. Preference strength is real information that ranks don't carry. Full analysis: is the video fair and balanced?
- Scores → approvals (race 3): the video's scores use only 0/3/4/5, so "approve = any support" is unambiguous up to threshold 3 — and elects Pink (12). But approve only 4s and 5s and Blue wins (10/9/8/5); only 5s and Purple wins (7/6/4/3). Three defensible thresholds, three winners: the conversion is the election.
- Ranks as ranks (races 2 and 4): even with no conversion at all, the ranked-ballot family splits — RCV-IRV elects Purple, Ranked Robin elects Green (BV) or Blue (LH). "RCV" names the ballot; the tabulation still decides.
Footnote details: STV(1) and STAR-PR(1)¶
STV at 1 seat is IRV. The Droop quota at one winner, ⌊20/2⌋+1 = 11, is IRV's majority threshold; the rounds are identical (7/6/4/3 → 8/7/4 → Purple 9–8, LH round-for-round).
STAR-PR at one seat is Score voting, not STAR. Allocated Score elects each seat by score alone — the automatic-runoff step exists only in single-winner STAR. At 1 seat that means plain Score voting: Purple's 46 wins and Blue's 10–9 runoff majority never gets consulted; the Runoff Reversal that defines the STAR race simply vanishes. (The LH engine refuses allocated with seats=1 outright and tells you to use STAR — so this leg is BV-only; the live result confirms Purple, score 46, no tie. BV labels the round's tieBreakType "random", but the one-candidate tie list in the raw result shows nothing was actually drawn — 46 is unique.)
Files¶
| Race | YAML | _tabulated mirror |
|---|---|---|
| STAR / RCV-IRV / Ranked Robin / STV | identical ballots to BV2176 — page · star · page · irv · page · ranked_robin | in postit_rcv_example_tabulated/ |
| Approval | page · bv2177_v8r66y_approval.yaml | txt |
| Choose-One | page · bv2177_v8r66y_plurality.yaml | txt |
| STAR-PR (1 seat) | BV-only (LH's allocated requires ≥ 2 seats); scoring round = the STAR yaml's | — |
Frozen BetterVoting export (Election + Ballots + Results): bv2177_v8r66y_bv_export.json — BV's stored winners match every prediction across all seven races.
Related: the BV2176 case page · the switch, made real (BV2178) · is the video fair and balanced? · up: method_comparisons