| paper | d4-conf-aaai-2013-0260 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
Theorem 1 characterizes an optimal reservation policy under the well-mixed assumption, but gives no algorithm, complexity bound, or hardness result for computing the threshold. The typed fluid-market optimization would require defining a new limiting process and computational task, so it is a follow-on project rather than a computational mirror anchored in a named result.
fails bit a — no named computational result to mirror
Theorem 1 is the only numbered result considered; Equation (1) presents an approximate threshold calculation, and the welfare comparisons are experimental.
I cannot identify a qualifying anchor for a continuous computational mirror. The only numbered theorem is Theorem 1, proved in this paper: assuming the same non-retrospective mechanism is used each round and the expected number of unmatched agents of each type stays constant, an agent’s optimal accept-or-reject policy is a reservation threshold. The agent rejects matches below \(t^\star\), accepts those above it, and uses the same threshold in every round; \(t^\star\) equals the discounted expected utility of remaining unmatched into a later round. That is an optimal-policy characterization, not a result classifying the complexity of a problem.
The paper’s approximate threshold calculation and its welfare comparisons are presented as an equation, methodology, and experiments—not as numbered computational results. Treating Theorem 1 as an anchor for tractability or hardness would overstate what it proves.
There is a plausible high-multiplicity setting here: kidney-exchange cohorts with many donor–patient pairs across the paper’s nine node types. That could motivate a fluid model of type masses and a question about computing welfare-maximizing thresholds. But it would be a follow-on question, not a continuous mirror anchored in a named computational result of this paper.
Theorem 1 is the only numbered result that could anchor a mirror, and it characterizes a policy rather than a computational problem. Under the well-mixed assumption, a tagged agent’s optimal response has a reservation-threshold form. The theorem gives no complexity claim or algorithm for finding the threshold. Equation (1) supplies an approximation, and the welfare comparisons are experimental. Neither turns the theorem into a result about exact, approximate, or parameterized computation.
The strongest rescue is a typed kidney-exchange fluid model: replace the nine node-type arrival counts with rational masses, retain type-pair compatibility probabilities and the utility distribution, and ask for algorithms to compute stationary unmatched masses and welfare-maximizing thresholds. That is a sensible high-multiplicity setting—the repeated cohorts and few node types make it especially convincing. But it is a follow-on computational project. To get that question, one must specify a limiting matching process, encode its inputs, and define the optimization task. The paper’s theorem supplies none of those computational claims to mirror. A mass-level model would also need to account for how random compatibility graphs and carry-over affect the market state; type masses alone do not specify that dynamics.
There are two routes, and neither preserves an anchor. Keep Theorem 1’s tagged-agent policy question, and the continuum version remains a threshold characterization, not a computational result. Shift to optimizing aggregate thresholds and welfare in the fluid market, and the computational question is newly formulated rather than a mirror of a named result in this paper.
So the case against a *paper-anchored computational mirror* is strong. The broader claim that no continuous version could be worthwhile is not: this paper offers unusually good evidence that a fluid model is sensible. The honest negative verdict is that the proposed computational work would be a new research direction, not a continuization of one of this paper’s computational results.
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.