| paper | d4-conf-ecai-2024-0213 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
The proposed continuous \(R2R_b\) aggregation problem is coherent, but the paper gives no numbered theorem, lemma, corollary, or proposition asserting a computational result. Its algorithm defines a procedure and its evaluation reports empirical findings; neither supplies the required named computational anchor.
fails bit a — no named computational result to mirror
The proposed mirror covers aggregation only; it leaves the elicitation algorithm and the paper's empirical comparisons outside its scope.
The strongest positive case is a narrow mirror of the paper’s aggregation component. Imagine a large peer-review cohort assessing the same small set of projects under a common rubric. A type \(t\) is a reviewer’s complete grade vector \(s_t:C\to\{1,\dots,5\}\), together with the strict ranking \(\succ_t\) that resolves ties between equal grades. The society is a rational distribution \(\mu\) over the \(\tau\) observed types.
A precise candidate problem is Continuous \(R2R_b\) Aggregation: given \(C\) and the supported types with their masses, output the weak ranking produced by the paper’s median-then-Borda rule. For each project \(c\), set its median grade to
\[
a(c)=\max\{d:\sum_{t:s_t(c)\ge d}\mu_t\ge\tfrac12\}.
\]
Within each equal-median bucket, order projects by their Borda scores, computed from the mass of types preferring one project to another; preserve ties in those scores. The output is the resulting weak order. The decision is the ranking itself, not an optimization over it.
A plausible regime is a large online course or shared review round: many reviewers assess the same handful of projects, while the five-grade rubric and repeated judgment patterns keep the number of distinct complete types relatively small. This continuizes the population, and it covers aggregation only. With an explicit support list, the rule appears directly computable in time polynomial in \(m\), \(\tau\), and the input bit length. That gives a sensible continuous computational question, but a modest one: it does not capture the paper’s main elicitation contribution or establish that the continuous version is easier in a meaningful complexity-theoretic sense.
There is a decisive limitation: the paper contains no numbered theorem, lemma, corollary, proposition, or other named computational result asserting that a problem is hard, tractable, FPT, or in another complexity class. Its algorithm and definitions specify the method; its findings in Sections 5.3–5.5 are empirical comparisons, not named complexity results. So there are no eligible anchors to quote or restate, and this aggregation problem is a possible research question—not a mirror anchored in a computational result of the paper. The weak point in the positive case is also the scope of the paper itself: the reported setting has small sessions, and replacing elicitation of individual judgments with an already-known distribution leaves out much of what \(R2R\) was designed to accomplish.
The strongest objection is the missing anchor. The paper states no theorem, lemma, or corollary about the complexity of a computational problem. Its R2R algorithm and Borda and Copeland definitions specify a method; the reported reductions in communication and ties are empirical findings. So the proposed mass-based ranking problem is a natural formalization of the method, but it does not mirror a complexity result established by the paper.
That objection does not rule out the aggregation question on its merits. Given reviewer types and their masses, the median grades come from cumulative mass, and pairwise Borda totals come from summing masses. A large cohort reviewing a shared set of projects is a plausible high-multiplicity setting. The problem is therefore well-posed; the negative case is that it asks to evaluate the paper’s rule on a distribution, while the paper’s contribution is chiefly about eliciting judgments with less burden and comparing methods empirically. That makes it a weak fit for a programme aimed at mapping computational complexity.
A stronger candidate would ask how much reviewer mass must change its reports to alter the winner. One could define each type by its full grade vector and ranking, then assign a cost to moving mass between types. This connects directly to the paper’s concern about strategic underscoring. But the paper does not specify the allowed report changes or their costs. Choosing them would create a new manipulation or robustness problem, rather than carry over a computational result from the paper. Still, because strategic behavior is part of the paper’s motivation, this is a plausible extension, and it may be one the authors would recognize.
So the absence of a complexity result is a strong reason to reject the proposed aggregation as a *result-anchored* mirror. It does not sustain the broader claim that no worthwhile continuous mirror exists. The aggregation rule already has a plausible population regime, and a mass-based robustness question has a natural connection to the paper’s concerns. The honest negative case is weak at that broader claim.
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.