| paper | d4-conf-atal-2014-0323 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
Theorem 4.6's proof offers an iterative construction of the equilibrium support, so it is the closest computational anchor. Its population counterpart requires replacing the single pivotal manipulator and ordered traders with positive-mass classes and new rules for their actions, leaving a different game and no precise mirror question.
fails bit b-changes — the continuum limit changes the question
A proposed extension would touch Theorem 4.6's equilibrium support analysis and Theorem 5.2's finite sequential convergence; it leaves the simulation claims in Section 5.3 without a guaranteed computational counterpart.
The strongest honest case is that a continuous population model for prediction-market participants is plausible, but this paper gives no named computational result to anchor it to.
A reasonable regime would be a large expert pool whose members fall into a bounded set of complete types: each type specifies its signal information, whether it is honest or manipulative, and its incentives. The mass \(\mu_t\) would be the fraction of participants of type \(t\). One could then ask how to compute an equilibrium aggregate price, or how much manipulator mass changes an early-stopping price. That is a possible research question about the population, not merely about making prices continuous.
But it is not a mirror of a named computational result here. Theorem 4.6, proved in the paper, characterizes the manipulator’s equilibrium support: at equilibrium, the manipulator mixes among reports corresponding to all signals at least as high as some threshold. Theorem 5.2, also proved here, shows that with a sufficiently large trade limit, myopic honest traders fully reveal their signals and the market price converges to the posterior given them. Neither theorem classifies a computational problem as tractable or hard. The paper’s simulations in Section 5.3 are not a numbered complexity result.
So I cannot give the requested continuous problem “per anchor” without presenting a new problem as though this paper had established a computational result about it. The paper’s analysis is about equilibria and market dynamics; its continuous probabilities and repeated rounds do not themselves continuize the population.
The strongest case against is that this paper has no computational result for ChoCo to continuize. Theorem 4.6 characterizes an equilibrium strategy; Theorem 5.2 proves a convergence property; Section 5.3 reports simulations. None states the complexity of a computational problem or gives an algorithmic result. One could ask for an algorithm to compute an equilibrium or an early stopping price in a new model, but that would start a new project rather than mirror a result in this paper.
Theorem 4.6 is a particularly poor anchor for a population mirror. Its threshold characterization concerns a specific asymmetric game: one manipulator moves before one honest trader, who updates beliefs from that move and their private signal. In a large pool, a single manipulator has zero population mass. Giving manipulators positive mass instead turns the singular manipulator into a coalition or a class of manipulators, changing the strategic problem and its beliefs. The best repair—many exchangeable honest experts, a positive mass of advocates, and an algorithmic question about equilibrium prices—could be worth studying. But its answer would depend on new choices about coalition behavior, ordering, and how trades affect price. The paper supplies no computational result that this model would extend.
Theorem 5.2 has a different mismatch. Its subject is a finite sequence of traders, each of whom eventually reveals a signal. In a continuum, an individual trader has zero mass, and the paper’s notion of stopping after all but one trader has revealed has no direct population meaning. One can replace it with a residual-mass threshold or a cutoff in an ordering, but that defines a new stopping rule and a new dynamic model. If signals are replicated in fixed cohorts, the posterior is unchanged by replication; if instead a large pool supplies fresh signals, the amount and dependence of that information must be specified anew. Either way, the theorem itself is about convergence, not computational complexity.
So the negative case is strong for ChoCo’s stated remit: there is no named computational result here, and the most promising population versions require substantial new modeling choices. It is weaker as the universal claim that *no* worthwhile continuous-population computational project can be built from this topic. A model with many exchangeable experts and a positive mass of manipulators is conceivable. I would reject this paper as a mirror candidate, while acknowledging that such a model could motivate a separate research question.
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.