| paper | d4-journals-jet-2020-0075 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The named results establish existence, generic uniqueness, and equilibrium implementation, but do not state a computational task with an algorithm, complexity bound, or hardness classification. Section 8’s enumeration procedure and the abstract’s “easy to compute” claim do not provide a computational guarantee in a named result. The type-mass coalition proposal requires a new worth model and computational question, so no continuous analogue of a qualifying result survives.
fails bit a — no named computational result to mirror
No named computational result is covered; Theorems 1–3 address existence, uniqueness, and equilibrium implementation.
I can’t make an anchor-compliant positive case: the paper contains no named computational-complexity result.
The closest candidates are Theorem 1, proved here, which says a SCOOP exists for every TU game; Theorem 2, proved here, which says the SCOOP is generically unique for superadditive games with zero-worth singleton coalitions; and Theorem 3, proved here, which says that as the discount factor approaches one, any limit point of stationary subgame-perfect equilibrium outcomes with positive probability of agreement is a SCOOP outcome, and such a sequence of equilibria exists. These are existence, uniqueness, and equilibrium-implementation results. None classifies a computational problem as tractable, hard, fixed-parameter tractable, or otherwise.
The abstract calls the SCOOP “easy to compute,” and Section 8 gives an algorithm. It enumerates subsets of coalitions and solves and checks systems for each candidate. The paper gives no running-time bound for that procedure. The claim and algorithm are useful evidence that computation matters to the authors, but they do not supply the numbered computational anchor your criteria require.
There is a plausible direction for a future mirror: a large labor or producer population with finitely many complete agent types, represented by type masses \(\mu_t\), and coalition worth determined by the composition of the participating mass. The output could be probabilities of forming each coalition composition and per-member surplus shares, with disagreement payoffs derived from consistent counterfactuals. That scenario fits the paper’s coalition-formation question. But it needs a new continuum version of SCOOP and a structured worth function; arbitrary coalition values in the paper do not automatically reduce to a fixed number of types. The closest named anchor, Theorem 2, concerns uniqueness rather than computational difficulty. I would treat this as a research direction, not submit it as a mirror of a named computational result.
The proponent raised no computational anchor to defeat. They are right that Theorems 1–3 concern existence, generic uniqueness, and equilibrium implementation. The abstract’s “easy to compute” claim and Section 8’s algorithm do not give a complexity result: there is no running-time bound or input model. Theorem 2 is the closest named result, but its uniqueness claim is not a computational problem.
The strongest salvage is a large population with finitely many complete agent types, such as workers with different skill profiles, where coalition worth depends on the mass of each type in the coalition. One could define a continuous version of SCOOP over coalition compositions, with a suitable scaling of worth and per-capita payoffs. That is a plausible high-multiplicity regime; I would not dismiss it merely because the paper’s examples use named players.
But that proposal needs structure the paper does not supply. Its characteristic function allows different worths for any two coalitions, even when they have the same type composition. Aggregating agents into masses preserves the game only when worth is invariant to which same-type agents are present. In a general game, preserving those distinctions can require as many types as players, eliminating the multiplicity gain. Restricting to composition-based worth may be sensible, but choosing that family and its representation is a new modeling contribution.
It also leaves the computational question to be built. Section 8 enumerates candidate sets of relevant coalitions, but does not specify an encoding, a complexity measure, or a bound. In a continuous composition model, one must specify how worth is represented and computed, define probabilities and counterfactual beliefs over the new coalition space, and state what output the algorithm must produce. Those choices could yield a worthwhile project, but they do not follow from the paper’s named results. Treating its probabilities over winning coalitions as the continuous object would not help: those probabilities describe uncertainty about the outcome, not a continuous population.
So the strongest honest case against is about the programme’s gate: this paper supplies no named computational result whose continuous-population mirror can be assessed, and its natural large-population extension requires substantial new structure before there is a computational problem to study. That does not prove no worthwhile project could grow from it. The type-composition model is a plausible research direction; it is simply not yet a computational mirror of a result in this paper.
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.