| paper | d4-conf-ijcai-2015-0514 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | no |
Theorem 2 constructively supplies a rule, so its lack of a runtime analysis alone does not disqualify it. But when the mechanism is specified only by a distribution over types, one individual's report does not change the distribution; making cohorts strategic or adding pointwise identity-dependent outcomes changes the incentive model. No proposed continuous question preserves the theorem's strategy-proofness content.
fails bit b-changes — the continuum limit changes the question
A pointwise repair would need the mechanism to use an individual's report or identity beyond the type-mass distribution, which the proposed model does not specify.
fatal: True
The proposed mirror concerns Theorem 2's lexicographic exchange-rule existence result; it leaves Theorem 1 and Proposition 5 without separate mirrors.
I can’t give a ChoCo-eligible positive case from this paper: it has no named result establishing a computational complexity classification. Theorem 1, proved here, says that for every common factor \(\triangleright\), an exchange problem with additive-extension preferences is unsuccessful when there are at least two agents and every agent initially owns at least two objects. “Unsuccessful” means no exchange rule is simultaneously individually rational, strategy-proof, and Pareto efficient. Theorem 2, also proved here, says that with lexicographic preferences the problem is successful if the objects have a partition \(H_1,\ldots,H_m\), each agent owns at most one object in each part, and every object in \(H_k\) is commonly preferred to every object in \(H_l\) for \(l>k\). Its proof constructs a rule by applying TTC to each part.
Those are mechanism-existence and axiomatic results, not results that a computational problem is in P, NP-hard, or in another complexity class. The TTC construction is algorithmic, but the paper does not analyze it as a computational complexity result of the kind your anchor rule requires. Proposition 5 is likewise about mechanism properties, not complexity. So there are no qualifying anchors, and no continuous problem I can honestly present as the required mirror of one.
The closest positive candidate would mirror Theorem 2 in a large task-rescheduling market. Let a type specify an initial bundle of object categories and a complete strict ranking consistent with the hierarchy; let \(\mu_\theta\) be the population fraction of type \(\theta\). An allocation variable \(x_{\theta,B}\) gives the mass of type \(\theta\) receiving integral bundle \(B\). Feasibility requires \(\sum_B x_{\theta,B}=\mu_\theta\) and that each object category’s allocated mass equals its per-capita supply. One could ask whether a direct rule mapping every admissible reported type distribution to such allocations is individually rational, Pareto efficient, and strategy-proof.
That is a concrete population model, but it exposes the weak point: under an anonymous rule determined only by type masses, one atomless agent’s report changes no mass, so individual strategy-proofness becomes vacuous. Requiring truthful reporting by a positive-mass cohort would make incentives meaningful, but that changes the paper’s problem. The model also asks for existence of an axiomatic rule, rather than a computational answer to an underlying problem. I’d treat it as a possible new mechanism-design question inspired by Theorem 2, not as a sound continuous computational mirror of this paper.
The strongest negative is that this paper has no computational result to mirror. Theorem 1 proves that a mechanism with three axiomatic properties does not exist on an additive preference domain; Theorem 2 proves existence on a restricted lexicographic domain; Proposition 5 gives another mechanism-property result. None classifies a computational problem or analyzes its complexity. The TTC construction in Theorem 2 is an algorithm, but the paper’s claim is about the properties of the resulting rule, not its computational complexity.
The best version of the proposed mirror would be a large task-rescheduling market: types specify preferences and endowed bundles, supplies scale with population, and the outcome assigns mass to integral bundles. One could ask whether a direct mechanism on those distributions satisfies individual rationality, Pareto efficiency, and strategy-proofness. But with an outcome specified only by type masses, one person’s report does not change the input distribution, so individual strategy-proofness is vacuous. Making positive-mass cohorts the strategic actors changes the incentive property. Adding a pointwise allocation rule can retain individual incentives, but then the question concerns a richer mechanism over reports and individual assignments, not a computational problem stated or solved in this paper.
One could instead ask for an efficient high-multiplicity implementation of the TTC rule, or for a complexity classification of finding mechanisms with the required properties. Those could be worthwhile projects. They would be new computational questions, though: the first studies compressed implementation, and the second turns the paper’s axiomatic existence question into a computational one. Theorem 1 and Proposition 5 have the same gap. Recasting their mechanism claims as mass-allocation questions does not supply a computational result to carry across.
So I would reject the proposed mirror as a ChoCo mirror of this paper. That is a narrower conclusion than “no continuum exchange problem is worth studying”: a continuous task market may be a sensible research setting. The negative case is strong on the programme’s computational gate, but weak as a universal claim about research value in every possible 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.