| paper | d4-conf-wine-2016-0016 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The named results characterize equilibria and bound their welfare or revenue; none asserts an algorithm, a computational hardness result, or another computational result. The symmetric-bidder theorem suggests an equilibrium asymptotics question, but that is not a continuous computational mirror of a result in this paper.
fails bit a — no named computational result to mirror
No result qualifies for a computational mirror. The closest candidate is the symmetric-bidder revenue theorem; the correlated-equilibrium characterization and other coarse-equilibrium welfare and revenue bounds are also analytical results.
I can’t identify a qualifying positive anchor in this paper. Its named results concern equilibrium characterization and welfare or revenue guarantees; none asserts that a computational problem is in \(P\), NP-hard, FPT, or otherwise classifies its complexity. The strongest nearby result is the theorem stating that, with \(n\) symmetric bidders of value \(v\), every coarse equilibrium has revenue at least \(\left(1-\frac{n}{e^{n-1}}\right)v\), and that this bound is tight. That is an equilibrium-performance theorem, not a computational result, and the supplied text does not print a theorem number for it.
The most plausible population-limit question would take \(n\) identical bidders of value \(v\), let \(D_n\) be a coarse correlated equilibrium of their first-price auction, and ask which winning-price distributions can arise as \(n\to\infty\). The theorem already implies that the expected winning price tends to \(v\), so those price distributions converge to a point mass at \(v\). This is a real large-population limit, but it asks about equilibrium asymptotics, not computational complexity. And each bidder’s chance of winning vanishes in the limit, which makes the continuum game’s incentives a substantial modeling issue.
So the honest positive case is weak: the symmetric-bidder result supplies a natural large-population direction, but not a named computational anchor or a continuous computational problem I can hand to a prover under your criteria. The paper’s distributions over bids are continuity in actions and outcomes; they do not themselves make the population continuous.
The strongest case against a mirror is that this paper supplies no computational result to continuize. Its theorems characterize equilibrium outcomes and bound welfare or revenue; none classifies the complexity of a computational problem or gives an algorithm. The remark that finding Nash equilibria is computationally hard is motivation, not a result of this paper. The correlated-equilibrium characterization and the coarse-equilibrium bounds are important, but they answer analytic questions about outcomes.
The symmetric-bidder theorem is the closest anchor. It gives a tight revenue bound of \(\left(1-\frac{n}{e^{n-1}}\right)v\), which tends to \(v\) as \(n\) grows. Repeated bidders with the same value are a plausible high-multiplicity regime, so that regime itself is no objection. But the theorem gives an asymptotic performance statement, not a computational question. It also establishes a bound on expected revenue; by itself, that does not establish convergence of every equilibrium’s winning-price distribution to a point mass.
The strongest upgraded mirror would take a rational distribution over bidder-value types and ask for the minimum revenue or welfare over coarse equilibria. That could be a worthwhile new problem. It is not a continuization of a computational result in this paper, though, and a distribution over types alone does not preserve the single-item auction’s allocation and incentives. One item goes to an individual, while a continuum of bidders assigns zero mass to any individual winner. Coarse-equilibrium constraints still concern each bidder’s deviation. A continuum model could add an individual-level allocation lottery and define how those deviations work; treating each type as a bloc would instead change the incentive constraints. Either approach introduces modeling choices the paper’s results do not settle. Making the item supply a positive mass would change the auction further.
So the negative case is strong for ChoCo’s paper-specific gate: there is no computational anchor here, and the natural symmetric-population limit is an equilibrium-asymptotics question. The universal claim is weaker. Large repeated-value auctions make a continuous population regime plausible, and a carefully specified computational study of continuum coarse equilibria could be worthwhile. It would be a new research direction prompted by the paper, rather than a mirror of one of its results.
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.