| paper | d4-conf-atal-2021-0041 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The proponent raised no anchor, and the paper’s numbered characterizations and axiomatic results do not assert an algorithmic or complexity claim. The chain-editing hardness is cited prior work, while the interleaving runtime appears only in prose; neither meets the required anchor test.
fails bit a — no named computational result to mirror
No continuous mirror is anchored to a named computational result; the MLE, axiomatic, and interleaving characterizations have no qualifying computational claim.
I find no eligible anchor under your rule. The paper’s numbered Theorems, Lemmas, and Propositions state likelihood, chain-structure, and axiomatic results; none asserts a complexity classification. Chain-editing NP-hardness appears in prose and a footnote citing earlier work, while the \(O(N^3)\) runtime for the cardinality-based interleaving operator appears in prose rather than a numbered result.
So I can’t give a compliant continuous-mirror case anchored to this paper. The cited hardness result is relevant background, but it is not a numbered result of the paper.
The proponent’s anchor screen is right. The paper has no numbered theorem, lemma, or corollary establishing a complexity result. Its main theorem characterizes chain-minimal operators as maximum-likelihood operators; the other prominent theorems characterize chain-definability and interleaving, or encode tie-breaking with weighted Hamming distance. Those are useful mathematical results, but they do not chart the computational landscape ChoCo is meant to chart. Chain-editing hardness is cited as prior work, and the \(O(N^3)\) bound for cardinality-based interleaving appears in prose.
The strongest candidate for a mirror is the MLE characterization. One could give the two sides finite sets of types with masses \(\mu_i,\nu_j\), record matchup frequencies \(p_{ij}\), and seek a chain relation \(q\) minimizing
\[
\sum_{i,j}\mu_i\nu_j\bigl(p_{ij}(1-q_{ij})+(1-p_{ij})q_{ij}\bigr).
\]
That is a coherent continuous-population question. But it is a new weighted editing problem: the paper’s result concerns a finite, individually indexed matchup matrix with unit Hamming costs. The paper proves no algorithmic or complexity claim about the weighted version. The election interpretation has a similar gap: a voter distribution determines pairwise-majority edges, but a margin-weighted editing objective would again change the rule.
So the strongest negative case is narrow: no eligible named computational result survives the required anchor test. I would not claim that no worthwhile continuous problem can be constructed from the paper. The weighted version above is the real weakness in a universal “no mirror” claim; whether it merits ChoCo’s time cannot be settled from this paper’s results alone.
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.