| paper | d4-journals-geb-2013-0107 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
Proposition 1
statement extracted from the paper’s text layer
Given rational masses μ_t over student types, per-capita hospital capacities κ_h, additive hospital values u_h(t), and a stable flow x with ∑_h x_{th} ≤ μ_t and ∑_t x_{th} ≤ κ_h, does there exist a profile of hospital reports, each an ordered list of types, such that mass-based student-proposing deferred acceptance returns x and no hospital can improve its value by changing its report? Each type records a student’s ranking of hospitals and the priority ranks assigned by every hospital; mass-based acceptance may split a type at a capacity boundary.
A fixed set H of hospitals faces a large population represented by rational masses μ_t over a finite support of complete student types. Hospital capacities scale with population as κ_h, and hospital payoffs from assigned flows are ∑_t u_h(t)x_{th}. Hospitals are the strategic players and submit ordered type lists; students propose according to their type rankings, while hospitals accept the highest-ranked available mass up to capacity.
A small-support type must capture each student’s preferences and every hospital’s priority ranking, so the regime requires repeated complete profiles; a type split across hospitals also cannot be selected by the finite construction’s list of named students.
fatal: False
The mirror covers Proposition 1’s stable-matching-to-equilibrium implementation claim. It leaves aside the equilibrium-outcome examples, the limits of dropping strategies, the filled-position and welfare results, and the open characterization of all equilibrium outcomes.
The strongest positive case is narrow. Proposition 1, proved in this paper, is the closest named anchor: in every many-to-one market, every stable matching is induced by some equilibrium consisting of dropping strategies. Consequently, the stable matchings are a subset of the equilibrium outcomes, and an equilibrium outcome exists. The proposition gives an explicit construction from a stable matching, but it does not state a complexity bound. The paper has no named theorem classifying an equilibrium-computation problem as being in P or as hard; its numbered examples about dropping strategies, filled positions, and welfare are not complexity results.
The lead mirror is *Stable-Flow Equilibrium Implementation*. Consider a large centralized placement market with a modest fixed set \(H\) of large employers or hospital systems and a large cohort of students. A student type \(t\) is a complete ranking of \(H\) and the outside option; \(\mu_t\) is the fraction of students of that type, with \(\sum_t\mu_t=1\). Let \(\tau\) be the number of types with positive mass, and take a regime where the cohort grows while \(\tau\) and \(|H|\) stay modest. Each hospital \(h\) has capacity \(\kappa_h\) per unit of population and a responsive ranking over student types. Its payoff from an assignment flow \(x\) is its true per-capita value \(U_h(x)=\sum_t u_h(t)x_{th}\). Scaling capacity with the cohort is part of this mirror: it keeps the many-to-one problem meaningful as the population grows.
An input consists of these rational type masses, hospital capacities and preferences, and a specified *stable flow* \(x\), where \(x_{th}\) is the mass of type \(t\) assigned to \(h\). The flow must respect student mass and hospital capacity; stability means that no positive mass of students and hospital capacity can mutually benefit by matching under the true preferences. Ask whether there is a profile of hospital reports, each an ordered list of acceptable student types, such that student-proposing deferred acceptance on the masses returns exactly \(x\), and no hospital can improve its true payoff by changing its report. A solution is such a report profile, together with the resulting flow and a certificate that no unilateral report improves any hospital’s payoff. The target flow is the feasibility objective here; the problem is to construct an equilibrium implementing it.
This is a recognizable mirror of Proposition 1: its discrete construction has each hospital list the students in its target match, while the continuous question asks whether target assignments of *mass* can be implemented when hospitals report about types. I would expect the fixed-\(|H|\), finite-type version to be tractable, with the explicit construction suggesting a starting point. That is a conjecture about the mirror, not a result in the paper. It generates useful follow-up questions: does implementation hold for every stable flow, and can the profile be found in time polynomial in \(\tau\) and the input bit length?
The weakest point is that the original construction can name particular students, while a type-level report cannot select some otherwise identical students of a type. A target flow may split a type across hospitals in a way the deferred-acceptance process cannot reproduce. That is a real question for this mirror, rather than a reason to assume the Proposition’s construction carries over. The proposed case covers only Proposition 1’s stable-matching-to-equilibrium claim; it does not turn the paper’s welfare or filled-position examples into complexity anchors.
The strongest objection is that Proposition 1 is not a computational result. It proves that every stable matching can be induced by an equilibrium and gives a witness strategy profile. The paper proves no complexity bound for equilibrium computation, and its question about computing the full set of equilibrium outcomes is explicitly left open. The proposed stable-flow mirror therefore has no computational theorem from the paper to continuize; it turns an existence result into a search problem.
For the proposed flow problem, a natural type-level version of the proposition seems to have the same construction: each hospital reports the types that receive positive mass in its target assignment. If the fluid deferred-acceptance model preserves the relevant stability and deviation arguments, this yields an equilibrium witness directly. That could be a valid continuous existence theorem, but it would not by itself chart a computational landscape. Computing all equilibrium outcomes is a stronger alternative, and the paper’s open question points toward it. That is a coherent algorithmic problem, though it is a new research task rather than a computational result established by this paper.
There is also a real restriction on the high-multiplicity regime. To preserve the paper’s strategic incentives, a student type must capture how every hospital ranks that student, as well as the student’s own preferences. The paper allows strict hospital rankings over named students, so generic instances can have nearly one type per student. Keeping the support small requires repeated preference and priority profiles, such as category-based rankings. That regime is plausible for some cohorts, but it is a substantial specialization of the paper’s model. The proponent’s per-capita capacities make the continuum nondegenerate, and the programme explicitly permits that co-modification; they do not resolve the restriction on repeated types.
I cannot make the universal negative airtight. A large cohort with repeated preference and priority profiles, matched to a small number of large employers, is a credible setting for a mass model. The case against is strongest as a claim about fit: this paper contributes equilibrium structure and counterexamples, while its computational question is open and a continuous version would need an additional modeling step. If the standard is that no scenario could produce worthwhile computational work, the supplied material does not support that conclusion.
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.