Topic: The four kinds of insincere vote¶
Level: 201 → 301 · for debaters
Topic hub — a cross-method view. The Equal Vote Coalition's taxonomy names four distinct ways to vote insincerely on a rated or ranked ballot. The names matter because people argue past each other by confusing them: "STAR is vulnerable to strategy" and "IRV is vulnerable to strategy" can be true of the same word and mean opposite things about the ballot, the risk, and the cure.
The one idea to take away: two of the four invert your order — they state a preference you don't have. Two erase it — they state an indifference you don't have. Inversion is only available where the count reads order; erasure is only available where the count reads degree. That single split explains which strategies each method is exposed to, and it is why choose-one is the worst of both worlds: it forces you to erase every distinction below your one mark, and it rewards inverting the top.
| Type | Also called | What the voter does | Which ballots it's possible on | Full page |
|---|---|---|---|---|
| Strong insincerity | Favorite betrayal, compromising, decapitation | Gives someone a higher score/rank than their true favorite — inverts the top | any ballot with more than one level | Favorite betrayal — Voting 301 · the worked pair |
| Weak insincerity | Burial, skipping | Keeps the favorite on top, but puts a less-preferred candidate below an even-less-preferred one — inverts below the top | ranked, and rated | Burial (topic hub) |
| Restrictive sincerity | Tactical minimization, bullet voting, truncation (down-voting) | Lowers support for non-favorites — erases your preferences downward | rated, approval, truncatable ranked | Restrictive sincerity |
| Expansive sincerity | Tactical maximization (up-voting) | Raises support for non-favorites — erases your preferences upward | rated, approval | Expansive sincerity |
Which one am I looking at?¶
Four questions, in order — the first "yes" names it:
- Is someone above my true favorite? → strong insincerity (favorite betrayal).
- Is anyone below someone I like less? → weak insincerity (burial).
- Did I flatten people down to the same low mark? → restrictive sincerity.
- Did I flatten people up to the same high mark? → expansive sincerity.
The second pair looks harmless — you never claimed to prefer anyone you don't — which is exactly why it's the pair most voters actually commit, usually without thinking of it as strategy at all.
What each method is exposed to¶
| Strategy | Choose-One | RCV-IRV | Ranked Robin | Approval | STAR |
|---|---|---|---|---|---|
| Favorite betrayal | structural — the only defence against vote-splitting | possible (center squeeze) | rare | not needed | rare, conceded |
| Burial | n/a (no order) | largely resisted (later-no-harm) | its named risk | n/a (no order) | possible, rarely pays |
| Restrictive (bullet / truncate) | forced — the ballot is a bullet vote | truncation, → exhausted ballots | truncation | half the threshold dilemma | backfires (worked) |
| Expansive (up-vote / hedge) | n/a — one mark | n/a — order is zero-sum | n/a | the other half | backfires (worked) |
Two readings of that grid worth stating plainly:
- A ranked ballot has no volume knob. Raising one candidate necessarily lowers another, so rows 3–4 barely exist there — a ranked-ballot voter who wants to "support harder" has only inversion available. That's the honest case for a rated ballot, and the reason the STAR/Approval strategy conversation is about volume while the RCV-IRV one is about order.
- A rated ballot's exposure is the pair that backfires. Both erasure strategies cost the voter information they might have needed, which is what a ~1:1 help-to-backfire ratio means in practice.
Runnable here¶
| Case | Kind | What it shows |
|---|---|---|
| Honest → betrayal pays | strong | nine voters demote their own favorite and get a better result — STAR's conceded FBC failure, live on BetterVoting |
| Sincere → burial pays | weak | 15 of 42 voters rank the Condorcet winner last, manufacture a cycle, and win the tiebreak |
| Honest → bullet backfire | restrictive | four fans zero out their second choice and elect the candidate they scored 0 |
| Honest → hedged | expansive | four voters hedge to a 5, register Equal Support in the runoff, and lose it |
| One opinion, three approval lines | both erasures | the same nine honest opinions produce three different Approval winners depending only on where the room draws its line |
The framing this repo keeps¶
No method is strategy-proof — Gibbard–Satterthwaite forbids it, so "you can construct a strategy against X" is never news. The questions that separate methods are the ones this taxonomy makes askable: which kind of insincerity does it invite, how many voters must coordinate, how good must their polling be, and what happens to them when it misfires. A strategy that needs a third of the electorate and a three-point knife edge is a heist; one that a lone voter commits by instinct on election day is a Tuesday.
This is a topic hub (cross-method index). Parent essay: strategic voting across the Equal Vote methods. See also the five strategic pathologies · PVSI, the strategic-incentive metric · Gibbard–Satterthwaite · the topics index. Glossary: bullet voting, burial, Equal Support, favorite betrayal.