Skip to content

Topic: Participation (can showing up to vote ever hurt you?)

Topic hub — a cross-method view. The Participation criterion: adding sincere ballots that prefer X over Y must never change the winner from X to Y. A method that fails it has the no-show paradox — an electorate where some voters get a better result by staying home than by casting an honest ballot. Read forward, the same failure is the Twin paradox: new voters who agree with you exactly arrive, and your group's outcome gets worse.

The one idea to take away: "vote — it can only help your side" is a promise a voting method can keep or break. Pure score-summing methods keep it structurally; every method with an elimination or runoff stage — RCV-IRV, Coombs, plurality-runoff, and STAR — plus every Condorcet-consistent method (Moulin's theorem) can break it in some electorate. The differences are how often and how visibly.

The three-candidate exception. Moulin's theorem needs ≥ 4 candidates. Restricted to exactly three, Brandt, Dong & Peters (2024) prove there is a no-show-immune family of Condorcet extensions — the refinements of maximin (leximin, Nanson), and uniquely so among homogeneous rules. The escape hatch closes again at four candidates. See Condorcet-Consistent Choice Among Three Candidates.

The live demonstration — one electorate, told twice (BV2174 / BV2175)

The centerpiece pair: 8 voters who love April (sincere ranking April > Bruno > Celia) decide whether to show up. Same three candidates, same everyone-else, three races on each election — and the methods split:

Method 8 fans stay home (BV2174 ↗) 8 fans vote (BV2175 ↗) Showing up…
Choose-One Celia April ✅ helped — last choice → favorite
STAR Bruno April ✅ helped — favorite wins (more sincere support moved the result their way)
RCV-IRV Bruno Celia HURT — 2nd choice replaced by LAST choice
Ranked Robin (LH) Celia April ✅ helped (LH-only — see below)

The IRV mechanism, in one sentence: the 8 extra April-first ballots keep April alive past round one, which gets Bruno — the 8 voters' own fallback — eliminated instead, and Bruno's transfers elect Celia, their last choice, 38–24. Their honest votes for their favorite bought their worst outcome; staying home (BV2174) keeps Bruno. Case pages with full engine reports: baseline · show-up.

Two footnotes on the pair, both deliberate:

  • Both electorates are a Condorcet cycle (April > Bruno > Celia > April) — that's the soil no-show paradoxes grow in, and it's why there is no Ranked Robin race on BetterVoting: a Copeland three-way tie resolves by random draw there (not freezable), while the LH engine's margin tiebreak resolves it deterministically (Celia before, April after — see the _tabulated mirrors, and LH vs BV on RR ties).
  • BV2174's STAR count is a Runoff Reversal — Celia tops the scores 136/122/120, Bruno wins the runoff 34–20. One pair, two lessons.

Which methods can punish participation — and where each is treated

Method Participation? Notes Full treatment
Score / Approval ✅ immune a sincere added ballot only adds support where the voter intends — summation can't backfire scoring methods vs ranked
Choose-One (Plurality) ✅ immune more votes for your favorite never hurts your favorite plurality
STAR ❌ rare the runoff stage costs the formal guarantee — live flip: BV2165→BV2166 (Coombs Ex.19), shown honestly STAR's honest limits
Ranked Robin / all Condorcet methods ❌ provably Moulin's theorem: Condorcet consistency ⇒ no-show paradoxes exist; see also Edelman's join-consistency argument Edelman, tabulated
RCV-IRV / plurality-runoff / Coombs ❌ readily the elimination machinery — same root as non-monotonicity no-show paradox

The honest scorecard, spelled out: this page's live pair shows IRV failing while STAR passes — but on other electorates STAR fails too (BV2165/66: two abstainers do better under STAR's runoff as well as Coombs'). No elimination-or-runoff method — and no Condorcet method — carries a participation guarantee. The structural difference: STAR's failures need a finalist to flip and the runoff to reverse (rare, constructed), IRV's need only an elimination-order flip (the everyday center-squeeze machinery). Only pure summation (Score, Approval) and Choose-One are immune outright — and they pay for it elsewhere (what makes a good winner?).

The paradox catalog & the theory

Glossary: no-show paradox.


This is a topic hub (cross-method index). The authoritative write-ups live in the per-method folders and the paradox catalog linked above. See the topics index for the other topic hubs.

file: README.md