| paper | d4-conf-atal-2023-0234 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The paper defines behavioral models and likelihoods and reports empirical estimation results, but it has no numbered theorem, lemma, corollary, or proposition asserting an algorithmic or computational result. The proposed reasoning-level shares may support a population interpretation, but without a qualifying result there is no computational mirror to anchor.
fails bit a — no named computational result to mirror
No result is mirrored; the paper's behavioral-model definitions, likelihood estimation, and empirical comparisons remain unanchored.
The positive case is limited, but there is a natural population model in the paper. In PQCH, agents have reasoning levels \(0,\ldots,3\), with shares given by a truncated Poisson distribution \(L_\tau\) (Definition 3.5); in QRE+L0, a fraction \(\beta\) are non-strategic. For a large cohort facing the same games and sharing the paper’s common value \(v\), those levels or behavioral classes could be types, and their shares could be masses. The paper’s setting of 24 games and 185 participants suggests the shape of such a regime, though not itself a compelling high-multiplicity instance.
But there is no qualifying anchor in the paper: it contains no numbered Theorem, Lemma, Corollary, or Proposition asserting a complexity or algorithmic result. Definitions 3.5 and 3.6 specify behavioral models; equations (2)–(5) specify likelihoods; Tables 2–4 report empirical estimation performance. None establishes tractability or hardness. In particular, Table 2’s finding that PQCH-QL4 estimates values more accurately is an empirical result, not a computational-complexity result, and using L-BFGS-B does not supply an algorithmic guarantee.
So I cannot honestly state one continuous problem per anchor or classify its complexity. The best positive signal is that the paper already models a distribution over behavioral types, making a continuous-population interpretation plausible. The corresponding weakness is also clear: merely treating \(L_\tau\) or \(\beta\) as population masses would largely restate the paper’s model. A useful mirror would need a distinct computational problem over aggregate population data, and this paper supplies no named computational result to anchor it.
The strongest point against a mirror is the proponent’s concession: there is no named computational result here to mirror. The paper reports empirical estimation performance and uses L-BFGS-B, but it states no complexity problem or algorithmic guarantee. So there are no result anchors for a continuous version to inherit.
The best proposed mirror would estimate \(v\) and behavioral parameters from a large cohort represented by masses of reasoning-level types. That is a coherent high-multiplicity regime; it is not objectionable merely because the paper uses such shares already. But \(L_\tau\) in PQCH and \(\beta\) in QRE+L0 already describe population proportions. Estimating them is part of the paper’s existing likelihood model. With the paper’s fixed \(3\times3\) games and finite behavioral classes, replacing a cohort by masses does not create a new computational problem whose complexity the paper studies. Generalizing the game or behavioral family could create one, but that would be a new research question, not a mirror of a result in this paper.
Counterfactual mechanism optimization is the closest route into computational social choice, but it adds a mechanism and an optimization objective the paper does not study. Its allocation games are hypothetical decompositions of observed payoffs, not outcomes generated by a mechanism being compared. The paper’s data and results therefore do not anchor that extension either.
This case does not show that no useful follow-on could ever be built. A large cohort with repeated behavioral types is plausible, and the population interpretation is sound. The negative case is narrower: the paper offers no computational result for which continuization would give ChoCo a worthwhile 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.