| paper | d4-journals-geb-2020-0014 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
Neither named result the proponent discusses asserts an algorithmic or computational claim. A computational question about finding a community-size threshold or evaluating learning would be a new research problem, not a continuous analogue anchored in these results.
fails bit a — no named computational result to mirror
No result is covered by a computational mirror; Theorem 5 and Theorem 3 are analytic results, and the paper's other results likewise do not provide a computational anchor.
I can’t make a compliant positive case from this paper: it contains no named computational result. Its theorems and propositions concern equilibrium existence, learning probabilities, and conditions for truth-telling observation; none establishes a problem’s complexity or gives an algorithmic result.
The closest near miss is Theorem 5, proved in the paper: for sufficiently low observation cost and sufficiently large communities, for every \(\epsilon>0\) there exists an equilibrium with truth-telling observation and limiting probability of a correct action greater than \(1-\epsilon\). That is an analytic existence result, not a computational anchor. Theorem 3, attributed to Song (2015), likewise characterizes truth-telling observation by whether \(K(t)\) tends to infinity and gives a limit learning probability, but makes no complexity claim.
Large cohorts of similar agents could motivate a continuous-population model, but proposing a computational problem from that model would not satisfy your requirement to anchor it in a named computational result of this paper. So there is no honest anchor—and therefore no continuous problem per anchor—to offer here. This does not establish that the model has no continuous mirror; it means the paper supplies no computational result for the requested case to build on.
The proponent names no eligible computational anchor. Theorem 5 is an equilibrium-existence result about learning under costly observation; Theorem 3 characterizes when truth-telling observation occurs and gives a limit learning probability. Neither states the complexity of a computational problem. That is the strongest objection, because the programme’s gap is specifically computational.
The most charitable mirror of Theorem 5 would ask for the least community size—or, after a suitable payoff rescaling, the least population mass—at which an equilibrium can achieve a chosen learning level. For Theorem 3, it might ask whether a given observation policy yields truth-telling or what its limiting learning probability is. Those are conceivable new questions, but they are equilibrium and learning questions, not computational results supplied by the paper. The paper specifies no finite input representation or algorithmic target for them. Turning either theorem into an algorithmic problem would require adding a new problem, rather than continuizing the result.
Even the most natural population limit has little aggregate structure to exploit here. Agents in a community share the same signal and observations, and Proposition 1 establishes unanimous actions in equilibrium. A continuum version of a community therefore still has a common information state and a single binary community action; the action mass is concentrated at one endpoint. Scaling payoffs with community mass is a reasonable co-modification, but it does not produce the kind of rich distribution over types or mass-transfer problem at the centre of ChoCo.
Section 6.2 offers a stronger alternative: give agents independent private signals and represent their signal heterogeneity by a continuum of types. That could support a worthwhile continuous social-learning model. But its central object would still be equilibrium learning in a dynamic Bayesian game, rather than the computational social-choice questions this programme targets. I cannot rule out a useful computational extension of that game. The negative case is narrower: it would be a new research direction, not a continuous computational mirror anchored in this paper’s 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.