| paper | d4-conf-ecai-2014-0241 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | unclear |
Propositions 7 and 9 give usable characterizations for computing robust winners, so requiring a complexity classification is too strict. But the proponent states no continuous problem for either anchor; the opponent’s nearest suggestion changes the elicited domain to distributions over ranks without connecting that new target to an anchored result. No anchor therefore passes the continuous-analogue test.
fails bit b-changes — the continuum limit changes the question
The nearest possible extension would concern robust-winner computation from elicited comparisons; it would not directly mirror the paper’s structural results about classes of voting rules or its experimental comparisons of query strategies.
I can’t make an anchored positive case under your criteria: the paper contains no numbered result asserting a computational complexity classification such as polynomial-time solvability or hardness.
The closest results are structural. For example, Propositions 7 and 9 show that the robust rule induced by elicited constraints equals the rule based on the minimal compatible preorder, whether the compatible models are preorders or weak orders. They do not analyze the complexity of a problem on continuous populations. The experimental section compares query-selection heuristics, but it gives no numbered complexity result to anchor a mirror.
So there is no qualifying anchor, and therefore no continuous problem to state *per anchor*. The paper’s focus is eliciting a committee’s ordering over rank-vectors; a continuous-population formulation might be a research direction, but claiming it as a mirror of a named computational result here would overstate the case.
The strongest case against is that the paper gives no named computational-complexity result to mirror. Propositions 7 and 9 characterize the robust rule induced by elicited constraints; they do not classify the complexity of computing it. The other propositions are structural as well. Section 4 reports experiments with query-selection heuristics, but no theorem about their complexity. The proponent raised no qualifying anchor, so there is no named result that survives as a positive case.
The best broader mirror I can construct would replace each candidate’s \(n\)-voter rank vector with a distribution over ranks in \(\Delta_m\), then ask how to elicit a committee’s ordering over those distributions. That is coherent, but the paper supplies no assumptions that make the continuous version computationally tractable or even finitely approximable: its target is an unrestricted weak order, and its experiments rely on a finite domain and a distribution over finite voting instances. To get useful finite-query guarantees on the continuum, one would need to add structure such as a regularity class, a metric, or a sampling distribution. Those additions would define a new preference-learning problem, rather than mirror a result of this paper.
This is a strong reason to reject a *mirror of the paper’s results*. It does not prove that no worthwhile follow-on project could be built from the same idea; that universal claim is weaker than the evidence allows.
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.