| paper | d4-journals-jet-2021-0139 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
Theorem 2 is a named result, but it characterizes dominance and makes no claim about computing it or its complexity. The proposed continuous mass-transfer feasibility question is coherent and has an explicit LP formulation, but it is newly extracted rather than a computational result of the paper, so bit (a) fails.
fails bit a — no named computational result to mirror
The candidate mirror covers Theorem 2's transfer-dominance equivalence and its Hammond-transfer and bidimensional applications; it leaves the paper's other welfare characterizations and decision-under-risk claims aside.
The strongest positive angle is a continuous-population version of the paper’s transfer-dominance question. But under your anchor rule, there is a decisive limitation: the paper has no named computational result. It proves welfare characterizations, not that a problem is in P, NP-hard, FPT, or similar. So I can offer a concrete candidate mirror, but it cannot honestly count as an anchor-based case.
The closest result is Theorem 2, proved in this paper. For a minimal transfer set \(T\) and equal-size integer distributions \(m,m'\), it says three conditions are equivalent: \(m-m'\) is in the discrete cone generated by \(T\); every social welfare function consistent with those transfers ranks \(m\) at least as high as \(m'\); and every consistent utilitarian welfare function \(W_u(m)=\sum_s m_su_s\) does so. Here consistency means \(t\cdot u\ge 0\) for every \(t\in T\). That is a substantial characterization, but it asserts no computational complexity. Theorem 1 and Corollaries 1–2 are also characterizations, not complexity results.
A plausible, explicitly provisional mirror is Continuous Transfer Dominance. An instance gives a finite outcome set \(S\), an explicit minimal set of transfer directions \(T\subseteq\mathbb Z^S\), and rational population shares \(\mu,\nu\in\mathbb Q_+^S\), each summing to one. Each \(s\in S\) is a complete welfare-outcome type, and \(\mu_s\) and \(\nu_s\) are the fractions of the population with that type. The question is whether there are nonnegative real mass transfers \(\lambda_t\) such that
\[
\nu-\mu=\sum_{t\in T}\lambda_t t.
\]
A solution is such a vector \(\lambda\); the decision objective is whether one exists. This keeps the outcome scale and transfer ethics discrete while allowing any fraction of the population to undergo each transfer.
The natural regime is a large regional or national population classified by a finite welfare scale—for example, health categories crossed with income bands. There may be millions of people but only a modest number \(\tau\) of observed types. That is a credible high-multiplicity setting: members of a type are indistinguishable in the model. The paper’s own count-vector framework is supporting evidence that this regime makes sense.
For an explicit transfer list, this candidate is feasible by linear programming, so I would expect Class A. Its rational instances also have a useful bridge back to the discrete problem: after multiplying \(\nu-\mu\) by a common denominator, the result is an integer vector; the Hilbert-basis property used in Theorem 2 makes cone membership coincide with integer-generation membership. That observation comes from the theorem’s mathematical structure, not from a computational result proved in the paper. Further questions would include what happens when transfer directions are implicit, or when the outcome space grows and separation becomes the bottleneck.
This candidate covers transfer dominance in Theorem 2 and its Hammond-transfer and bidimensional examples. It does not supply a mirror for the paper’s other welfare claims, and it adds no transfer costs that the paper never models.
The weakest point is substantial: there is no eligible computational anchor. Also, Section 9 already considers real probability masses and convex-cone transfers for decision under risk. That section treats probabilities over outcomes rather than fractions of a society, so the population interpretation is distinct, but the formal overlap makes the mirror less novel. The honest positive case is that a continuous population is a plausible and useful regime for the paper’s transfer-dominance question—not that the paper has established a computational result whose complexity might change under continuization.
The strongest objection is the one the proponent already concedes: this paper has no named computational result to mirror. Theorem 2 is its central result, but it is a characterization: it equates transfer dominance with unanimous rankings by two classes of welfare functions. Theorem 1 and the corollaries do the same kind of work. None analyzes the complexity of a computational problem. Under the requested anchor rule, Theorem 2 is therefore not an eligible anchor. The proposed decision problem is a computational question built around the theorem, rather than a continuous mirror of a computational result in the paper.
The proposed question is coherent: a society with finitely many outcome types can be represented by rational population shares. But with an explicit outcome set \(S\) and explicit transfer set \(T\), it is just feasibility of
\[
\sum_{t\in T}\lambda_t t=\nu-\mu,\qquad \lambda_t\ge 0,
\]
a linear program. That gives a direct polynomial-time route in the explicit input size. A more demanding problem could encode outcomes or transfers implicitly and ask for a separation algorithm. Yet then the difficulty depends on a new encoding and representation choice; the paper does not identify one. The result would be an algorithmic study of a chosen transfer system, not a complexity question already posed by this paper.
There is a real limit to this negative case. A large population classified by health, income, or other finite outcome categories is a plausible high-multiplicity regime, and the authors would likely recognize mass transfers as their subject. Section 9 also develops closely related mathematics for probability distributions, though that is not itself a continuous-population computational treatment. So I cannot honestly argue that no sensible continuous scenario exists. The narrower and stronger conclusion is that the paper supplies no eligible computational anchor, while its most natural extracted decision problem is already an explicit LP unless additional structure is introduced.
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.