| paper | d4-conf-aaai-2015-0581 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
The proposed mass-matching problem is a plausible follow-up, but the paper has no numbered computational result to anchor it. Its hardness claims are background, and its own algorithm comparisons are experimental results without a stated guarantee.
fails bit a — no named computational result to mirror
The candidate question captures maximum stable matching over preference types; it does not mirror the paper’s Adaptive Search model or its experimental and parallel performance results.
There is no eligible anchor in this paper: it has no numbered theorem, lemma, corollary, or proposition asserting a computational result. The closest claim is the background statement that MAX-SMTI—finding a maximum-cardinality weakly stable matching with incomplete lists and ties—is NP-hard, including in restricted cases. The paper attributes that claim to Iwama et al. (1999) and Manlove et al. (2002); it does not prove or number it here. The local-search comparisons and parallel speedups are experimental results, not numbered complexity results. So I cannot honestly supply an anchor-specific mirror as requested.
If that background claim is allowed as an unanchored lead, the strongest population mirror is maximum stable mass matching. There are finite sets of men’s and women’s types, each with a rational mass summing to one on its side. A type specifies a complete weak preference order over the opposite-side types, including which are unacceptable; members of the same type are indifferent among partners of any one type. Choose nonnegative flows \(x_{uv}\) on acceptable type pairs, with each row and column sum bounded by its type’s mass. The objective is to maximize \(\sum_{u,v}x_{uv}\), the matched mass. Call a flow stable if, for every acceptable pair \((u,v)\), at least one side has zero mass that strictly prefers the other type to its current assignment. Formally, if \(d^M_{uv}\) and \(d^W_{uv}\) are those masses on the two sides, require \(d^M_{uv}d^W_{uv}=0\). A solution is a stable flow of maximum total mass.
This is a plausible mirror of MAX-SMTI’s underlying question: mass replaces headcount, and the decision remains which agents can be matched while avoiding blocking pairs. A credible regime would be a large matching market with recurring applicant and vacancy-slot profiles: many agents per profile, but relatively few distinct preference types. The problem as stated also allows large support, however. With one type per person and mass \(1/n\), it contains finite MAX-SMTI instances; that suggests hardness transfers in the unrestricted-support problem, though it says nothing decisive about complexity when the number of types is small. The type-based model makes a substantive assumption—agents sharing a type rank partner classes alike and are indifferent within a class—so it needs that cohort interpretation to be credible.
My weakest point is that this question is not anchored in a numbered result of the paper, and its hardness case relies on support growing with the original instance. The paper’s own contribution is a local-search model and experimental evaluation; neither establishes an algorithm or complexity classification for the mass-flow problem. The useful follow-up questions would be whether the rational mass model matches solutions on replicated finite instances, and how its complexity changes with the number of types.
The negative case is strongest at the level of what this paper actually contributes: it has no numbered theorem, lemma, or corollary establishing a computational result. Its NP-hardness and APX-hardness statements are background claims attributed to earlier work. The paper’s own results are an Adaptive Search model and experimental comparisons, including parallel speedups. Those experiments do not give the complexity result that ChoCo’s proposed mirrors are meant to continuize.
The proponent’s maximum stable mass matching is a plausible follow-up problem: large applicant and vacancy cohorts can have rational masses, preferences over counterpart types, and a maximum matched mass subject to stability. But that is an unanchored lead, not a mirror of a computational result established by this paper. Its small-support version also specializes the paper’s identity-level preference lists to cohort preferences, with indifference among people of a type. That can be a sensible market model, but it leaves open whether it preserves the paper’s computational contribution; merely inheriting MAX-SMTI’s background NP-hardness at unrestricted support does not settle the small-support question.
I cannot make the universal negative claim convincing. The cohort model is a credible continuous-population question, and a worthwhile one for ChoCo even if it is not this paper’s contribution. So the case against is strong only as a verdict about this paper’s anchors; it is weak as a claim that no worthwhile continuous mirror exists in any 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.