Topic: Burial (sinking a rival you actually like)¶
Topic hub — a cross-method view. Burial is voting a strong rival below candidates you honestly prefer less, to knock them out of a comparison they would otherwise win. Your favorite stays on top — that's what separates burial from favorite betrayal — so it feels like a free move, and the whole question is whether the count lets it pay.
The one idea to take away: burial is not Ranked Robin's private wart. It is the strategy available against any method that reads a rival's position relative to other candidates — Condorcet, Borda, and (in its score form) STAR. What differs is the price: how many voters must coordinate, how much they risk, and — for the Condorcet family — which completion rule the count uses when the burial manufactures a cycle. The method that most nearly escapes it is the one this library criticizes most, RCV-IRV, and it escapes for an unflattering reason: it doesn't read your lower ranks at all until your higher ones are eliminated.
In the Equal Vote taxonomy of insincere votes, burial is weak insincerity — it distorts the order of candidates you don't lead with, while favorite betrayal (strong insincerity) distorts the top of the ballot. Formally, a method is open to burial only if it fails later-no-harm: if a lower ranking can never hurt your favorite, moving a rival down can never help them either.
Which methods it works on — and where each is treated¶
| Method | Buriable? | Why | Full page |
|---|---|---|---|
| Ranked Robin / Copeland | ❌ yes — its named risk | purely ordinal: sink a rival below weaker candidates and a matchup flips, often manufacturing a cycle the record-tiebreak then settles | RR honest limits §3 · the worked pair |
| Margin-based Condorcet (Minimax, Ranked Pairs, Schulze) | ⚠️ harder | they read how much each matchup was won by, so a thin manufactured cycle is resolved on numbers the buriers can't cheaply fake | Alaska 2022, buried · Minimax |
| Borda | ❌ yes, notoriously | every rank is worth points, so burying is plain arithmetic — the classic Dark Horse | The Dark Horse |
| STAR | ⚠️ possible, rarely pays | fails later-no-harm, so scoring a strong second 0 can help your favorite reach the runoff — but the runoff can then hand the win to the candidate you buried | STAR's honest limits · criteria failures · FAQ |
| RCV-IRV (Hare) | ✅ largely no | satisfies later-no-harm — your 3rd choice is never read while your 1st is alive, so there's nothing to gain by mis-ordering below the top. IRV's exposure is compromising and center squeeze instead | Which RCV-IRV? |
| Approval / Score | ✅ n/a (Approval) | Approval has no order to invert; withholding support is bullet voting, a different strategy with a different cure | Approval's honest limits |
The pattern worth naming: later-no-harm and Condorcet compliance can't both be had (a small impossibility result you can feel here). IRV buys burial-resistance by ignoring most of your ballot; Condorcet methods buy the head-to-head winner by reading all of it, and pay for that with burial. STAR sits between, and pays a smaller price than either.
Runnable here — burial worked on four different methods¶
| Case | Method attacked | What it shows |
|---|---|---|
| Burial in Ranked Robin — the sincere/buried pair | Ranked Robin | 15 of 42 voters rank the Condorcet winner last, manufacture a cycle, and win the record tie (sincere · buried) — triple-checked, deterministic on both engines |
| Even Condorcet methods can be buried — Alaska 2022 | Condorcet, on real numbers | rb-j's attack on the actual Alaska special: bury Begich, manufacture a cycle. The attack's success depends entirely on the completion method — margin-based rules shrug it off, a Hare/runoff completion falls for it (buried ballots) |
| Manipulability — the textbook P₃ profile | Copeland and STAR | Zwicker's own illustration of single-voter manipulability is an attack on Copeland; the page runs the same profile against STAR, where two voters bury their 4th choice and elect their favorite, and against a milder no-burial variant |
| FairVote's white-paper burials | STAR (score burial) | the critics' own constructions, reproduced: Washington 2010 — the squeeze works on STAR, but IRV resists it — and French 2017, where every faction scores Macron 0 |
What a burial actually costs¶
Every page above lands on the same three facts, which is why this library treats burial as a real but expensive attack rather than a routine one:
- It needs a large, disciplined bloc sitting inside the victim's majorities. In the Ranked Robin pair that's 15 of 42 voters — over a third of the electorate — coordinating, with decent polling, and willing to rank a candidate they genuinely like dead last.
- It backfires when mis-aimed. Bury into the wrong cycle and you elect the buriers' third choice. In STAR the buried candidate can simply win the runoff you pushed them into.
- It leaves fingerprints. Sincere ballots showed no cycle; the buried ballots do. A cycle appearing in a race whose polling showed a clear head-to-head winner is the anomaly, and the printed round-robin table is where an auditor would catch it.
And the even-handed framing this repo keeps returning to: no method is strategy-proof — Gibbard–Satterthwaite forbids it. The honest comparison isn't "which method has no wart," it's which warts are systematic and which need a heist. Burial is a heist. Center squeeze is a Tuesday.
Closely related: strategic voting (the four-kinds taxonomy) · the five strategic pathologies (where the Dark Horse row hides this mechanism) · center squeeze (IRV's systematic wart, the contrast case) · cycle resolution (what happens after a burial succeeds) · the Smith set. Glossary: burial (weak insincerity).
This is a topic hub (cross-method index). The authoritative write-ups live in the per-method folders linked above. See the topics index for the other topic hubs.