| paper | d4-journals-geb-2017-0044 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
P→P discrete result → its continuous mirror
Answer: \(\mathrm P\); indeed, \(R(\beta,\epsilon)=1\) for every admissible parameter pair. · discrete side: polynomial · adjudication: proved
full proof & adversarial review →
Theorem 1
statement extracted from the paper’s text layer
Given a rational distribution \(\mu\) with finite support in \(V=[0,\beta]\), positive maximum value, and supported masses at least \(\epsilon\), choose an allocation intensity \(a(r;\mu)\) and scaled transfer \(b(r;\mu)\) for every report \(r\in V\). In an \(N\)-agent approximation, a reporter receives the object with probability \(a(r;\mu)/N\) and pays \(b(r;\mu)/N\). Require \(\int a(v;\mu)\,d\mu(v)\le1\), \(\int b(v;\mu)\,d\mu(v)=0\), \(v a(v;\mu)-b(v;\mu)\ge v a(r;\mu)-b(r;\mu)\) for all \(v,r\in V\), and individual rationality for supported values. What is the maximum, over such rules, of the worst-case welfare ratio \(\inf_\mu \frac{\int v a(v;\mu)\,d\mu(v)}{\max\operatorname{supp}(\mu)}\)?
A nonatomic population of bidders whose values are types carrying mass \(\mu\), with allocation intensity \(a\) and scaled transfer \(b\); finite approximations give each agent allocation probability and payment divided by \(N\). If the maximum supported value has mass \(\alpha\) and the second-highest supported value is \(v_2\), a menu can assign \(a=1/\alpha,\ b=(1-\alpha)v_2/\alpha\) to reports above \(v_2\), and \(a=0,\ b=-v_2\) to reports at most \(v_2\). For a singleton support, set \(a=1,\ b=0\). This menu is truthful over the full report interval, balances transfers, and allocates all expected supply to the highest-valued cohort, giving welfare ratio \(1\).
The paper does not prove that its finite \(r\)-optimal allocation and payment rules converge to the proposed mass mechanism; the mirror is therefore an extension of its objective rather than a demonstrated limit theorem.
fatal: False
The mirror covers Theorem 1’s welfare objective and allocation toward the highest-valued agents. It does not continuize Proposition 1’s implementability condition, Theorem 3’s ranking Pareto results, or Theorem 4’s general characterization; individual rationality is treated separately in Theorem 2 and Corollary 2.
The positive case is narrow: this paper has no named theorem classifying a problem as polynomial-time, NP-hard, or otherwise. Its strongest usable anchor is Theorem 1, proved in the paper. It says that, among satisfactory ranking mechanisms—those that are dominant-strategy incentive compatible, budget-balanced, and symmetric—an \(r\)-optimal allocation rule exists. Here \(r\)-optimal means maximizing the probability \(\pi_1\) assigned to the highest-valued agent. The rule has \(\pi_1^*=1-\frac{\ell-1}{\binom{n-2}{\ell-1}+\ell}\), \(\pi_i^*=\frac{1}{\binom{n-2}{\ell-1}+\ell}\) for \(2\le i\le\ell\), and \(\pi_i^*=0\) for \(i>\ell\), where \(\ell\) minimizes \(\frac{j-1}{\binom{n-2}{j-1}+j}\) over even \(j\) from \(2\) to \(n-1\). The rule is unique when \(n\ne8\). The paper reports that the highest-valued agent gets the object with at least \(99\%\) probability for \(n\ge14\). This is an exact mechanism-optimization result, not a complexity-class result.
A plausible high-multiplicity setting is a large pool of bidders competing for one indivisible public concession, where private values come from a small set of standardized classes. Each class fixes everything relevant to the mechanism, and each class has at least an \(\epsilon\) fraction of the population. Thus \(N\) is large, the number of types \(\tau\) is fixed, and every type has \(\Theta(N)\) members. The paper’s small-group examples need not fit this regime; the case rests on large, repeated cohorts.
Here is a precise mirror, Atomless Balanced Ranking Design. Let the possible values be \(V=\{v_1>\cdots>v_\tau\ge0\}\), with \(v_1>0\), and let the society be any rational distribution \(\mu\) on \(V\) with \(\mu_t\ge\epsilon\). Find a rule that, for each such \(\mu\), assigns each type \(t\) a nonnegative allocation intensity \(a_t(\mu)\) and a scaled transfer \(b_t(\mu)\). The object is awarded to type \(t\) with probability \(\mu_t a_t(\mu)\), then uniformly to one member of that cohort; hence \(\sum_t\mu_t a_t(\mu)\le1\). In an \(N\)-agent approximation, a member of type \(t\) gets the object with probability \(a_t/N\) and pays \(b_t/N\). The rule must be symmetric within types, budget-balanced \(\sum_t\mu_t b_t(\mu)=0\), truthful at the population scale \(v_ta_t-b_t\ge v_ta_s-b_s\) for every pair \(t,s\), and individually rational \(v_ta_t-b_t\ge0\). Its objective is to maximize the worst-case welfare ratio \(\inf_\mu \frac{\sum_t\mu_t v_t a_t(\mu)}{v_1}\).
This mirror has a direct efficient solution. For \(\tau\ge2\), write \(\alpha=\mu_1\), set \(a_1=1/\alpha\) and \(a_t=0\) for \(t>1\), and set \(b_1=(1-\alpha)v_2/\alpha\) and \(b_t=-v_2\) for \(t>1\). The object always goes to someone in the top-valued cohort, so welfare is \(v_1\), attaining the upper bound of ratio \(1\). The transfers balance. Lower-valued types get scaled utility \(v_2\); the top type’s truthful utility is at least \(v_2\), and neither can gain by switching reports. For a single type, set \(a_1=1,b_1=0\). Thus the robust problem is tractable in this regime, with a compact rule computable from the type masses.
The connection to Theorem 1 is that its finite-population optimum drives \(\pi_1^*\) toward \(1\); with large cohorts, the object’s chance of going to the top-valued *type* approaches one as well. The mirror covers that theorem’s welfare-optimization question, not the paper’s full characterization of satisfactory mechanisms. It keeps the object indivisible: mass describes how many agents share a type, and the type-level allocation is a lottery over individuals.
The soft spot is substantial. The paper requires exact budget balance and incentives at every finite valuation profile. The mirror makes a single agent negligible, so their report does not change \(\mu\); it also scales per-agent lottery probabilities and transfers with \(N\). That change is what permits exact efficiency in the limit. A formal bridge showing that the proposed mechanism is the appropriate limit of the paper’s finite mechanisms—including incentives and payments—is still needed. The finite value classes are also a narrower report domain than the paper’s interval of valuations. If those changes are acceptable as consequences of a high-multiplicity regime, this is a real positive case; if a named complexity-class anchor is mandatory, this paper supplies none.
The strongest objection is that the paper gives ChoCo no computational result to continuize. Theorem 1 optimizes a finite vector of rank probabilities and gives a closed-form choice of \(\ell\); given \(n\), one can check the even candidates directly. Proposition 1 is a single implementability equation, and Theorem 4 is an axiomatic characterization. The paper poses no complexity problem over society instances, and reports no algorithmic landscape for intervention or robustness. That leaves no computational target for the programme’s central question.
Theorem 1 is still the best possible anchor, but the proposed mirror changes its problem. The paper requires DSIC over the full interval \(V=[0,\beta]\), while the mirror restricts reports to a finite set of values. Its efficient rule relies on that restriction: all lower reports receive the same subsidy, and the incentive comparison only ranges over those finitely many values. A mass distribution concentrated on finitely many values does not, by itself, remove the paper’s requirement to handle other reports. The proposed rule therefore establishes a result for a different report domain.
There is a credible finite high-multiplicity scenario here: many bidders may share standardized values, and grouping them by value is sensible. But the proposed construction works for finite populations on that restricted domain; it does not need the continuum to obtain its result. Its source of tractability is the value-domain restriction. A more faithful mirror could retain the interval and take a large-population limit, with rank allocations represented by mass densities. That could be a worthwhile continuum mechanism-design question. With one object, though, preserving meaningful individual incentives requires specifying how vanishing per-person probabilities and transfers are normalized. Once that model is chosen, the result is a new incentive-and-budget-balance problem, without a computational problem or algorithmic claim supplied by this paper.
So the negative case is strong as a case about ChoCo’s remit, but it does not establish that no worthwhile continuous question exists in any scenario. Large bidder cohorts are plausible, and a continuum mechanism-design model could be valuable in its own field. I would reject this paper as a source of a worthwhile ChoCo mirror because its best candidate changes the domain and supplies a new research question, while the paper itself has no computational result for the programme to carry forward.
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.