| paper | d4-conf-ecai-2025-0135 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
Proposition 8 has computational character because it guarantees a property of CP-majority’s output, even without a runtime bound. But the proponent gives no concrete mass-based problem, and the proposed route requires replacing individual comparisons with equal-mass transfers and specifying a measure over background coalitions. No continuous question for the anchor survives as stated, so the grade is red on bit b.
fails bit b-changes — the continuum limit changes the question
No anchor survives. The proponent singled out Proposition 8; a mass-based extension would address only its CP-majority guarantee, leaving the other propositions and experiments outside the mirror.
I can’t make a valid positive case under the required anchor rule. The paper has numbered propositions, but none asserts a complexity result such as membership in P, NP-hardness, or fixed-parameter tractability. Its experiments compute recovery methods, but the paper gives no named runtime or complexity classification to mirror.
The closest result is Proposition 8, proved in this paper: if a total order over items is extended by any of the five lifting rules, then for every compatible partial preorder over coalitions, CP-majority never reverses a strict pairwise preference in the original order; it may return a tie. That is a recovery guarantee, not a computational-complexity result, so it cannot serve as the requested anchor.
A continuous version over masses of item types would also change the paper’s domain: the paper ranks subsets of a finite set of items, rather than studying a population of interchangeable agents. Without a qualifying named result, I can’t state a continuous problem as a defensible mirror of this paper’s computational result.
The strongest case against is that this paper has no named computational result to carry over. Proposition 1 is an impossibility claim about recovering coalition preferences; Propositions 2–12 give recovery and precision guarantees; the remaining evidence is experimental. None states a complexity classification or analyzes an algorithm’s running time.
Proposition 8 is the closest proposed anchor, but it says that CP-majority preserves strict pairwise preferences under the stated assumptions, while allowing ties. That is a mathematical guarantee, not a computational result. One could ask how quickly CP-majority can be computed, but with an explicitly supplied partial ranking it amounts to counting the relevant comparisons. To make that a harder question, one would need to specify a compressed input or a different computational task; that would be a new problem, not a continuous mirror of Proposition 8.
The strongest revised mirror would treat \(X\) as a large population of fungible objects, group identical objects into types, and give the types masses. One could then define fractional coalitions and ask whether a mass-based version of CP-majority preserves the ranking of types. That might be a worthwhile new question, but it needs new definitions: in a nonatomic population, adding one object changes no mass, so the paper’s individual marginal comparisons vanish. Replacing them with equal-mass transfers and a measure over background coalitions could avoid that, but changes the proposition’s central operation rather than continuizing it.
So the negative case is strong against a paper-faithful mirror: there is no computational result here for ChoCo’s complexity programme to mirror. It is weaker as a claim that no adjacent continuous model could ever be worthwhile. A sufficiently reworked type-level model might be interesting; the paper itself does not supply an anchor for it.
The adversarial triple: the proponent anchors on up to three named results; the opponent sees that case and must defeat every anchor; the judge decides which case convinced it. These are the pipeline’s own outputs, generated by tools/triple_run.py — no human edited them. The paper’s own text is not reproduced here beyond the quoted statement above.