| paper | d4-conf-aaai-2005-0008 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
DCVC is an algorithmic contribution, but the paper states no numbered theorem, lemma, corollary, or proposition asserting a computational result; its comparison table reports empirical results. Since no result meets the required anchor criterion, bit (a) fails and no continuous mirror can green the paper.
fails bit a — no named computational result to mirror
The paper's DCVC algorithm partitions a finite, enumerable set of coalition-value calculations, and the experiments compare that partition with SK; neither is stated as a numbered computational result.
I can’t make a case that meets your anchor requirement: the paper has no numbered Theorem, Lemma, Corollary, or Proposition asserting a computational result.
It does present the named algorithm DCVC and claims that it gives exhaustive, disjoint, equally sized task shares without communication, using memory for at most one coalition at a time. It also reports simulation comparisons with SK. Those are computational contributions, but the paper does not state them as numbered results of the required kind, so I can’t use them as anchors or supply a continuous problem “per anchor” without stretching your rules.
The proponent names no anchor, so no continuous question has been established from a result of the paper. The paper has no numbered theorem, lemma, or corollary asserting a computational result. DCVC is a real algorithmic contribution, but its claims concern how to divide a finite list of coalition-value calculations; the experiments compare that division with SK. They do not classify the complexity of a specified value-computation problem. That is the paper’s strongest candidate for a mirror, so it is worth testing directly.
The best version would group indistinguishable agents by type and represent a coalition by the mass it takes from each type. But for type masses \(\mu_t\), those coalitions range over the whole box \(\prod_t [0,\mu_t]\): there are uncountably many. DCVC’s guarantees—exhaustive, disjoint shares, equal numbers of calculations, and at most one coalition stored—presuppose a finite, enumerable set of coalitions. They do not extend to an exact finite algorithm for calculating every value in that continuum.
There are ways to make the task finite, but each changes what is being mirrored. A rational grid gives finitely many coalitions, yet its resolution reinstates individual units and produces a discrete high-multiplicity problem. Restricting coalitions to whole type classes discards partial participation. Supplying only finitely many coalition queries makes the task ordinary workload assignment; the population distribution then does not determine the computational question. A symbolic or parametric value function might yield a new continuous problem, but the paper leaves coalition-value calculation to the developer and specifies no such function or input representation. Defining one would supply the substance the paper leaves open, rather than continuize a result it establishes.
So the negative case is strong against this paper as written: its finite enumeration problem has no exact counterpart over a nonatomic population, and making one finite requires adding or changing the task. The universal claim is less airtight. A carefully chosen parametric coalition-value model could create a worthwhile continuous computational problem, but that would be a new model, not a surviving mirror of DCVC’s stated result.
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.