| paper | d4-conf-ecai-2016-0130 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
No numbered result in this paper asserts an algorithm, complexity bound, or other computational claim: Theorems 2–4 classify axiomatic properties, and Theorem 1 is prior work. The proposed mass-weighted Hamming merge is a plausible computational follow-on, but it does not mirror a computational result of this paper.
fails bit a — no named computational result to mirror
No result is covered by a continuous computational mirror; the paper's axiomatic classifications and operator-property results remain untouched.
The paper gives a plausible population to continuize. Imagine a large consultation on a fixed set of policy questions: many participants share the same complete belief type, where a type is a knowledge base identified by its models. If there are \(N\) participants but only \(\tau\) distinct types, let \(\lambda_t\) be the fraction holding type \(t\). A natural weighted version of the paper’s \(\Delta_{d_H,\Sigma}\) operator takes a constraint \(\chi\) and returns all models \(w\models\chi\) minimizing
\[
\sum_{t=1}^{\tau}\lambda_t\,d_H(w,K_t),
\qquad
d_H(w,K_t)=\min_{v\models K_t}d_H(w,v).
\]
This has a clean high-multiplicity bridge: if \(n_t\) people hold type \(t\), setting \(\lambda_t=n_t/N\) gives the same minimizers as merging the discrete profile with \(n_t\) copies of each type. The consultation scenario is credible when repeated, stable positions on a small agenda make \(\tau\ll N\). The authors could recognize this as a mass-weighted extension of their distance-based merging operators, though it asks about computing the merge rather than classifying its axiomatic properties.
But there is no eligible anchor in the paper under your stated rule. Theorem 2 says certain properties follow from the IC postulates; Theorem 3 characterizes several properties through assignment conditions; and Theorem 4 classifies which properties \(\Delta_{d_H}\) and \(\Delta_{d_D}\) satisfy. For example, Theorem 4 says both operators fail Monotonicity and Independence of clones; \(\Delta_{d_H}\) also fails Entity resolution, Condorcet’s criterion, and Reversal symmetry, while \(\Delta_{d_D}\) satisfies those three; and no IC merging operator satisfies SC-Majority. These are results about axioms and operators, not claims that a computational problem is in P, NP-hard, FPT, or otherwise. Theorem 1 is a representation result cited from earlier work, not a computational result of this paper. So none can honestly serve as one of the required anchors.
The weighted merge above is still a precise candidate follow-on question: given the propositional alphabet, consistent knowledge-base types \(K_t\), rational masses \(\lambda_t\), and constraint \(\chi\), compute the set of minimizing interpretations. Its general version is expected to be hard from the agenda alone, even with \(\tau=1\): a one-point belief type at the all-false interpretation and a constraint with a high-cost fallback can make the minimizing interpretation reveal whether an arbitrary SAT instance is satisfiable. That would be a new complexity result about the proposed problem, not one of this paper’s results.
The soft point is decisive for the requested case: the paper studies axiomatic classification, not computational questions about a population distribution. The mass-weighted merge is a plausible computational continuation of its subject, but I cannot present it as a mirror anchored in a named computational result from this paper.
The strongest case against is categorical: this paper supplies no computational result to mirror. Theorem 2 derives axioms from the IC postulates; Theorem 3 characterizes axioms through assignment conditions; Theorem 4 classifies which axioms particular operators satisfy. Those are results about normative properties of merging operators, not about the complexity or algorithms for a computational problem. Theorem 1 is a representation theorem attributed to earlier work. None is an eligible anchor, so there is no paper result for a continuous computational mirror to extend.
The proponent’s mass-weighted Hamming merge is a coherent candidate follow-on. With type masses \(\lambda_t\), it minimizes \(\sum_t\lambda_t d_H(w,K_t)\); setting \(\lambda_t=n_t/N\) gives exactly the minimizers of the finite profile with \(n_t\) copies of each type. But that asks a new question—how to compute the output of these operators from a distribution over belief types. The paper studies how to classify their properties, not how to compute their outputs. A continuous-population version of Condorcet or majority would likewise restate axioms over masses; studying whether an operator satisfies those restatements remains axiomatic work, outside ChoCo’s stated scope.
I would therefore reject a claim that this paper contains a worthwhile continuous *mirror*. I would not claim that no worthwhile computational project could grow from it: the weighted merge is plausible, and its complexity could be worth studying on its own. That is the weak point in any universal negative. The paper offers a promising subject for a new computational question, but no named computational result for that question to mirror.
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.