| paper | d4-journals-mss-2010-0002 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
other→P discrete result → its continuous mirror
Answer: \(\text{All limits exist and are rational; exact vector evaluation is polynomial-time solvable.}\) · discrete side: other · adjudication: proved
full proof & adversarial review →
Proposition 1
Given finitely many unions and rational type masses μ_{kt} summing to one, and an explicitly encoded rational polynomial worth function \(V\) on coalition type-mass vectors, define each replicated finite game by \(v_n(S)=V(x(S)/n)\), where \(x(S)\) counts coalition members by type. For each type \(t\) in union \(k\), compute exactly the limit of \(n\) times an individual’s two-step Shapley value as common-denominator replications \(n\) grow, and determine the evaluation complexity in the number of unions, number of supported types, and input bit length.
Use a type set partitioned into unions, with mass vector μ_k for union \(k\), total mass μ̄_k, and \(w(S)=V(\sum_{\ell\in S}\mu_\ell)\) as the quotient game on unions. The limiting per-capita allocation to type \(t\) in union \(k\) is \(\Psi_{kt}=\int_0^1 \partial_t V(u\mu_k)\,du + (\operatorname{Sh}_k(w)-V(\mu_k))/\bar\mu_k\), combining the within-union marginal value with the equal per-member share of the quotient-game correction.
The proponent’s Φ integral averages marginal contributions across union orders and therefore gives an Owen-style value; a two-step mirror needs the within-union value and the quotient-game solidarity correction.
fatal: False
The mirror covers Proposition 1’s value-computation formula in a large-population regime; it leaves the axiomatic characterizations in Theorems 4 and 5, and the paper’s other axiomatic results, aside.
The strongest honest positive case is narrow, and it does not pass the requested anchor test cleanly: the paper has no named complexity result claiming that a computational problem is in P, hard, FPT, or similar.
The closest candidate is Proposition 1, credited to Kamijo (2009, Theorem 3), so it is not proved here as a new result. It represents each player’s two-step Shapley value \(\eta_i(B,v)\) as the average of the paper-defined marginal contribution \(\delta_i^\pi(v)\) over admissible orders \(\pi\), in which each union’s members appear consecutively. That is a useful order-based computation formula, and the paper uses calculations of values in examples. But it does not give a complexity classification or an algorithmic analysis. Theorem 4 characterizes \(\eta\) by efficiency, additivity, coalitional symmetry, the null-coalition axiom, coherence, and population solidarity within unions; Theorem 5 characterizes it by efficiency, coalitional balanced contributions, population solidarity within unions, and coherence. Both are axiomatic results, which the programme explicitly places outside scope. Under a strict reading of the anchor rule, then, there is no compliant anchor and no per-anchor mirror to hand to a prover.
If Proposition 1 is admitted as a computational-formula anchor, the best lead is Continuum Owen-Density Evaluation. Consider a large workforce divided into \(q\) unions. A type specifies a worker’s union and all game-relevant attributes; \(\mu_{kt}\) is the fraction of the workforce of type \(t\) in union \(k\). The regime is credible when there are many workers but few distinct types. Represent a TU game by a differentiable rational polynomial \(V\) over vectors of coalition type masses. The problem is: given \(V\), the union partition, and the rational masses \(\mu\), compute the limiting per-capita allocation density for every type, obtained by replicating the population and scaling an individual’s two-step Shapley value by the population size.
For a type \(t\) in union \(k\), the proposed density is
\(\Phi_{kt}=\frac{1}{q!}\sum_{\sigma\in S_q}\int_0^1 \partial_{kt}V\!\left(\sum_{\ell\prec_\sigma k}\mu_\ell+u\mu_k\right)\,du\).
Here \(\sigma\) orders the unions, \(\ell\prec_\sigma k\) means union \(\ell\) precedes \(k\), and \(u\) is the within-union position of a representative worker in the large-population limit. This preserves Proposition 1’s order-based marginal-contribution idea while making the population the continuous object. The result would cover the value-computation aspect only, not the paper’s axiom characterizations. A plausible follow-up is to establish the replication-limit identity and its approximation error for finite populations.
For fixed \(q\), evaluating this expression for polynomial \(V\) looks tractable. When \(q\) is part of the input, the order of unions remains explicit, so I would tentatively expect the hard part to persist in the number of unions; the paper gives no basis for a hardness claim or a firm trichotomy classification.
The soft spot is decisive: Proposition 1 is an order-average formula, not a named computational-complexity result, and the proposed continuum evaluation is our extrapolation. The authors might recognize it as a natural computational extension of their value formula, but it is not a continuous mirror of a computational result they establish.
The paper gives no computational result to mirror. Theorem 4 and Theorem 5 are axiomatic characterizations. Proposition 1 is the closest anchor, but it is an order-average formula credited to Kamijo, not a complexity theorem or an algorithmic analysis. It specifies neither an input representation nor a running-time question. The paper’s examples calculate values; they do not establish how hard that calculation is.
The proposed continuum formula also appears to compute the wrong value. Averaging \(\partial_{kt}V\) along paths where unions enter in random order is an Owen-style marginal allocation. The paper’s central object is the two-step Shapley value, which it distinguishes from Owen’s because Owen’s value can fail the paper’s solidarity condition. A direct continuum translation of the two-step definition would instead combine a within-union marginal term with an equal per-member share of the difference between the union’s quotient-game Shapley payoff and its standalone worth. For union \(k\), with total mass \(\bar\mu_k\), that suggests a density of the form
\[
\Psi_{kt}
=
\int_0^1 \partial_t V(u\mu_k)\,du
+
\frac{\operatorname{Sh}_k(w)-V(\mu_k)}{\bar\mu_k},
\qquad
w(S)=V\!\left(\sum_{\ell\in S}\mu_\ell\right).
\]
That is the repair someone should try before claiming to mirror Proposition 1.
The repaired question still needs a chosen encoding for \(V\) and a complexity measure. With an explicit quotient-game table, the remaining Shapley calculation is on the unions; with a succinct polynomial or circuit, the complexity depends on that encoding. The paper gives no reason to select among these problems. Meanwhile, continuous finite-type coalition models already define mass-based games as limits of finite games and provide formulas for computing continuum coalition values, including coalition-structure extensions of Aumann–Shapley methods. That work establishes that the regime is sensible; it does not settle the complexity of this paper’s solidarity-sensitive value. [Calvo, *The coalitional value in finite-type continuum games*](https://www.uv.es/ecalvo/Emilio%20Calvo_archivos/Research/Research%20in%20Game%20theory_archivos/ContOwen.pdf)
So the proponent’s stated anchor does not survive as written, and a corrected version would be a new computational study of an aggregative cooperative-game model, rather than a complexity question the paper itself raises. The negative case is strongest on that paper-specific point. It is weak as a universal claim: a large production or bargaining population with many interchangeable members and a few unions is a plausible high-multiplicity setting, and the corrected density could be useful there. I cannot honestly rule out every worthwhile continuous mirror in such a scenario.
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.