| paper | d4-conf-ijcai-1993-0250 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
None of the named results asserts an algorithm, complexity bound, or hardness result. The suggested population extensions would create new allocation or bargaining questions rather than provide a computational mirror of the paper’s theorems.
fails bit a — no named computational result to mirror
The named results concern domain properties, existence of negotiation sets, and utility division; no computational result is covered by a continuous mirror.
The strongest honest positive case is only a motivation, not a qualifying mirror: this paper has no named computational-complexity result to anchor one. Its Theorem 1 says concave TODs are subadditive; Theorems 4 and 5 assert that the negotiation set is nonempty for pure deals in any TOD and for mixed deals in subadditive TODs; Theorem 6 says an optimal mixed-deal procedure in a subadditive two-agent TOD divides available utility equally. These are structural and game-theoretic claims, not results that a problem is in P, NP-hard, or otherwise computationally classified. The authors do mention low computational complexity as a design criterion, but do not establish a complexity bound. They state that proofs of Theorem 1 and the other theorems are in their 1992 technical report. [The paper](https://www.ijcai.org/Proceedings/93-1/Papers/059.pdf)
A large population of agents with recurring, identical task-and-cost profiles could make a high-multiplicity extension of a task-oriented domain plausible. But turning the paper’s two-agent negotiation results into a population-level allocation problem would change the model, and the paper’s theorems do not anchor its computational analysis. So I can’t responsibly state a continuous problem as a mirror of one of its named computational results. That is the weak point—and it prevents a positive case under your anchor requirement.
The proponent names Theorems 1, 4, 5, and 6, but none is a computational result. Theorem 1 relates concavity to subadditivity; Theorems 4 and 5 establish that certain negotiation sets are nonempty; Theorem 6 characterizes the utility split under optimal negotiation between two agents. They specify no computational problem or complexity bound. The paper’s mention of low computational complexity as a design criterion does not supply one. That leaves no named computational result here for a continuous-complexity mirror to investigate.
The best population version would group agents with recurring task sets and costs, then treat their proportions as masses. That regime is plausible, but it does not rescue these anchors. Extending Theorem 1 to aggregate task demand would study a property of the domain’s cost function, not the computational complexity of a population problem. A mass-based version of Theorems 4 and 5 would ask whether an efficient, individually rational allocation exists among many agents. The paper proves existence for two-agent deal models; it gives no computational target for selecting such an allocation in a population. Those could motivate new work, but they would be newly formulated problems rather than mirrors of the paper’s results.
Theorem 6 is the closest candidate for a stronger mirror: take two large classes of agents, aggregate each class’s utility, and ask whether the optimal negotiation divides the available utility equally. But that either retains two bargaining parties, leaving the population continuum outside the theorem’s central object, or requires a new multi-agent bargaining model whose allocation and utility notions the paper does not define. In either version, the equal-split result is an outcome claim, not a computational classification.
So the objection is not that high multiplicity is implausible; recurring task-and-cost profiles make it quite sensible. The gap is that no choice of population scenario turns these particular results into computational claims. A computational study of high-multiplicity negotiation might be worthwhile on its own, but it would be a new research question, not a continuous mirror of this paper.
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.