| paper | d4-conf-aaai-2020-1418 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
The paper reports experimental findings and deploys impartial ranking mechanisms, but states no numbered theorem, lemma, corollary, or proposition with a computational claim. The mechanisms’ guarantees are attributed to Kahng et al. (2018), so they cannot anchor a mirror of this paper; a continuous hiring model would pose a new computational question.
fails bit a — no named computational result to mirror
The studies report empirical results on strategic behavior, ranking accuracy, and worker experience; the impartiality guarantees are cited from earlier work. No result in this paper supplies an eligible computational mirror.
I can’t make a theorem-anchored positive case from this paper: it contains no numbered Theorem, Lemma, Corollary, or Proposition asserting a computational complexity result. Its main results are empirical findings from three studies. The closest candidate is Table 2’s measured ranking accuracy for Kemeny, NAIVE-BIPARTITE, COMMITTEE, and k-PARTITE; those are experimental comparisons, not complexity results. The paper cites the impartiality guarantees of those mechanisms to Kahng et al. (2018), but does not state them as numbered results of this paper.
A large expert crowdsourcing market does suggest a plausible continuous regime: types could describe applicants by their application materials and relevant assessment characteristics, with mass representing fractions of applicants in each type. But a direct continuum raises a serious modeling problem for the paper’s central concern. An individual applicant has zero mass and so cannot affect their own position; impartiality may become automatic. Giving a positive mass to a type instead makes it a cohort that can act together, which changes the strategic unit and is not the individual manipulation studied here.
So there is no eligible anchor, and therefore no per-anchor continuous problem I can honestly hand to a prover under your criteria. The continuous-population idea may be worth exploring for peer ranking, but this paper provides no named computational result to anchor that mirror.
The proponent identifies the decisive problem: this paper has no named computational result to mirror. Its findings are experimental measurements—the strategic exaggeration in Study 1, ranking accuracy in Table 2, and workers’ reported experiences in Study 3. Replacing sample proportions with population proportions might produce a continuous statistical model, but it would not turn those findings into the paper’s computational questions of complexity or algorithm design. The impartiality algorithms’ theoretical guarantees are cited from earlier work, not established here as computational results.
The strongest possible rescue is a high-multiplicity version of peer-assessed hiring: represent applicants and assessors by types, aggregate pairwise assessments, and study how to produce an accurate, impartial ranking. But the paper’s central guarantee is defined for an individual: changing their reports must not change their own position. In an atomless society, an individual’s report has zero mass, so this condition is automatic. If instead an entire positive-mass type is the strategic unit, the question becomes whether a cohort can manipulate its collective position. That is a different guarantee; the paper itself notes that coalitions can manipulate its mechanisms.
There is a further mismatch with the paper’s hiring task. A continuous ranking can rank masses of application types, but the requester hires an individual for a single job. Selecting one person from an atomless mass is not represented by the distribution. Multiple openings or hiring a positive fraction could make selection meaningful, but then the model needs a new account of how individual impartiality works. That may be a sensible research problem, and the paper’s focus on large markets makes it plausible; it is not a continuous version of a computational result this paper reports.
So the best negative case is strong against a mirror *anchored in this paper*: its empirical outcomes do not supply the required computational target, and the most promising redesigned target changes the object of impartiality. The universal claim that no worthwhile scenario exists is weaker. A multi-opening, cohort-based model could be worth studying, but that would be a new computational question motivated by the paper, not a mirror of one of its results.
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.