| paper | d4-conf-ijcai-1999-0149 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
The paper’s numbered results prove strategic or Nash outcome equivalence between auction formats. They do not state an algorithmic guarantee, a hardness result, or a computational formulation; the scaled multi-unit model discussed in the cases would be a new problem without an anchored computational result.
fails bit a — no named computational result to mirror
Theorems 1–4 and Corollary 1 establish strategic or Nash outcome equivalences. The possible scaled multi-unit follow-up is not covered by those results.
The positive case is weak under the programme’s anchor rule: the paper has no named result classifying a computational problem as tractable or hard. Theorem 1, proved here, says the \((n-1)\)-round survival auction is strategically equivalent to the Japanese auction. Theorem 2, also proved here, establishes Nash outcome equivalence for the paper’s survival auctions under its symmetry assumptions; Corollary 1 applies that result to the two-round auction. Theorems 3 and 4 give the corresponding claims for multiple goods. These are results about strategic and outcome equivalence. The fixed round counts are meaningful protocol-performance claims, but they do not classify the complexity of an underlying computational problem. So there are no eligible anchors, and no honest per-anchor continuous problem to hand to a prover.
There is a plausible regime for a *future* continuous question: computerized buyers competing for scalable capacity, as in the paper’s resource-allocation motivation. Types could encode each buyer’s valuation-relevant information and preferences; \(\mu_t\) would be the fraction of buyers of type \(t\). The actions would be type-contingent bid strategies in successive survival rounds, and one could ask for the equilibrium allocation and revenue per unit of population. This sounds closer to the paper than an unrelated continuum-agent model because the paper explicitly studies computerized bidders and resource allocation.
But the mechanism creates a serious obstacle to that mirror. In the two-round result, only two bidders survive to the final round. As the population grows, those two bidders have vanishing mass. Keeping a positive mass of finalists would require changing the survival rule, and scaling the resource supply would change the allocation problem. Those co-modifications might define a sensible new model, but Theorem 2 does not establish its equilibrium properties. A question about computing its equilibrium would also be a new computational problem, not a computational result this paper anchors. I would not assign it a tractability or hardness class from this paper.
The strongest case against a ChoCo mirror is that this paper has no computational result to mirror. Theorems 1–4 and Corollary 1 establish strategic or Nash outcome equivalence. The fixed round counts matter for protocol speed, but the paper gives no complexity classification or algorithm for a computational problem. Asking how to compute equilibria in a new continuous auction would be a new question, not a mirror of one of those results.
The best-motivated setting is the paper’s own: computerized bidders competing for scalable capacity. A type could include all the valuation-relevant information, so repeated types are entirely plausible. But the central equivalence results track individual bidders, their ordered dropouts, and, in the two-round auction, the final two bidders. In an atomless population, each bidder has zero mass; eliminating one bidder removes no mass, and allocating one indivisible good to one bidder does not produce a positive-mass outcome. A mass-based version would have to eliminate quantiles or scale up the supply. Those are reasonable new models, but they change what the mechanism reveals and what its outcome records. The paper’s equivalence proofs do not supply a computational question about either model.
The paper’s multi-good extension makes a positive-density version especially plausible: one could scale both bidders and units, then study quantile-based survival rules. That is the strongest candidate for a worthwhile follow-up. But asking for its equilibrium properties would be an analytic auction-theory question; adding an algorithmic task would still require choosing a new computational objective, such as equilibrium computation or mechanism design. Neither is a named result here.
So the negative case is strong for the paper’s results, but weak as a universal claim: a continuous, computational study of scaled multi-unit auctions might be worthwhile in its own right. The case against is that it would not be a continuous mirror of this paper’s computational result—there is no such result to anchor it.
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.