| paper | d4-journals-mss-2020-0056 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
Theorem 2 and Theorem 3 are axiomatic characterizations, while the propositions and Theorem 1 give structural formulas without an algorithm or complexity claim. The proposed smooth type-mass Shapley limit adds an anonymous mass-based game model and cannot supply the missing named computational anchor, so the paper fails bit (a).
fails bit a — no named computational result to mirror
No computational result is covered; the nearest results, Theorems 2 and 3, are axiomatic characterizations, and the remaining named results are structural.
The strongest honest positive idea is a high-multiplicity limit of the paper’s Intersection rule, but this paper contains no named computational result to anchor that case. So there are zero qualifying anchors, and no per-anchor complexity direction to report.
The nearest named results are Theorem 2, proved here, and Theorem 3, also proved here. Theorem 2 says that the Intersection rule \(I(v,\Omega)=\operatorname{Sh}(v^\Omega)\) is the unique allocation rule satisfying efficiency, additivity, equal treatment of equal agents and the null-agent axiom on games unchanged by restriction, and equality of payoffs when every feasible coalition has worth zero. Theorem 3 says that, for an efficient and additive allocation rule \(f\) on ordinary TU-games, the unique extension satisfying efficiency, additivity, equality, \(f\)-consistency, and invariance under irrelevant changes to the intersection-closed system is \(\Phi_f(v,\Omega)=f(v^\Omega)\); Corollary 2 gives the Shapley-value case. These are axiomatic characterizations, not complexity or algorithm results. Propositions 1–3 and Theorem 1 give structural formulas, but do not analyze an algorithm or its complexity.
A plausible follow-on question would be to compute a continuum limit of the Intersection allocation. Consider a large cooperative organization with finitely many roles, such as workers with a small number of skills. A type records everything relevant to coalition worth and feasibility; agents of that type are interchangeable in every coalition. Let \(\mu_t\) be the population share of type \(t\), let \(x\) be a vector of coalition masses, and let \(W(x)=V(c_\Omega(x))\) be the worth after applying a closure operator for feasible coalitions. Under smoothness assumptions, the candidate output is each type’s per-unit allocation
\[
q_t=\int_0^1 \frac{\partial W}{\partial x_t}(\lambda\mu)\,d\lambda,
\]
so type \(t\) receives total mass allocation \(\mu_t q_t\). This is a concrete continuous analogue of computing the Shapley allocation of the restricted game, and it covers the Intersection rule rather than the paper’s broader family \(\Phi_f\).
The weak point is substantial: the paper allows arbitrary worth functions and intersection-closed systems over named agents. They need not be representable using type masses, and the derivative formula is an added continuum construction, not a result of this paper. The authors might recognize it as a limit question about their rule, but this manuscript supplies no computational anchor or basis for predicting whether that question is tractable.
The proponent identifies the decisive gap: there is no qualifying computational result here to mirror. Theorem 2 and Theorem 3 characterize allocation rules by axioms; Propositions 1–3 and Theorem 1 give structural identities. None states the complexity of a computational problem or specifies an algorithmic input model. So there are zero anchors, and neither theorem supports a continuous complexity claim.
The strongest rescue is the proposed high-multiplicity Shapley limit. With finitely many interchangeable types, an anonymous worth function \(W(x)\) over type masses, and suitable smoothness, the formula
\[
q_t=\int_0^1 \frac{\partial W}{\partial x_t}(\lambda\mu)\,d\lambda
\]
is a coherent continuous allocation. But it needs substantial structure absent from the paper. Its arbitrary coalition worths and intersection-closed systems need not be determined by type counts: pairwise synergies, permission links, or other relations can distinguish agents with the same individual attributes. Preserving those distinctions can make the number of types grow with the population; suppressing them changes the game. In a symmetric, succinctly represented model, the formula may define a worthwhile new computational question, but its complexity depends on that model’s representation and regularity assumptions, not on a result established here.
That rescue is plausible, so the negative case is not airtight as a claim about every possible follow-on project. It is strong as an assessment of this paper: its contribution is axiomatic, and the proposed continuum allocation would be a new model and research question rather than a continuous mirror of a named computational result.
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.