| paper | d4-conf-atal-2014-0133 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
The proponent identifies no named computational result to anchor a computational mirror. The proposed mass-splitting question also changes the paper’s single-agent manipulation into coalition manipulation, a limitation the proponent acknowledges.
fails bit a — no named computational result to mirror
The proposal targets Proposition 1’s incompatibility claim; it does not establish a computational mirror of that result or cover the mechanism constructions, distance bounds, or facility-location illustration.
I don’t see a qualifying computational anchor in this paper. Its named results concern mechanism properties and existence, not whether a computational problem is in P, hard, FPT, or similar. In particular, Proposition 1, proved here, says that when the preference domain is symmetric and competitive, false-name-proofness and individual rationality are incompatible for the utilitarian welfare maximizer. Proposition 2, also proved here, gives a Groves transfer rule that is false-name-proof for every preference domain. Both are substantial formal results, but neither classifies an algorithmic problem. So there is no paper result that meets the requested anchor test.
The strongest adjacent mirror I can make is Mass False-Name-Proof Welfare Mechanism Existence, based on Proposition 1. Take a finite outcome set \(X\) and a finite, symmetric, competitive domain \(D\subseteq\mathbb{Q}^{X}\) of complete utility functions. A society is a rational distribution \(\mu\) over \(D\), with support size \(\tau\); \(\mu_u\) is the fraction of the population with utility type \(u\). The outcome must maximize \(\sum_{u\in D}\mu_u u(x)\). A type includes every preference and value the mechanism uses.
The question is whether there exists an anonymous per-unit transfer rule \(\pi(u,\mu)\) that makes this welfare-maximizing mechanism both individually rational and false-name-proof against mass splitting. Precisely, if a bloc of mass \(\beta\) and true type \(u\) replaces its reports \(\beta\delta_u\) with \(\beta\rho\), for any distribution \(\rho\) over reported types, write \(\mu'=\mu-\beta\delta_u+\beta\rho\). The rule must give the bloc at least as much aggregate utility when truthful:
\[
\beta\bigl[u(F(\mu))+\pi(u,\mu)\bigr]
\ge
\beta u(F(\mu'))+\beta\int_D\pi(v,\mu')\,d\rho(v).
\]
It must also give truthful participation at least the bloc’s utility after opting out. A solution is such a transfer rule, defined across all rational societies in the domain.
A plausible regime is a large open-source development community with millions of accounts but relatively few recurring preference and workload profiles, so \(N\) is large while \(\tau\ll N\). The mass-splitting question could describe an organization controlling a block of accounts and distributing that block among aliases. Proposition 1 makes impossibility the natural conjecture for this mirror; further questions include whether the proof survives fractional splits and what continuous counterparts the paper’s individual-rationality distance bounds, Propositions 3 and 5, permit.
The weakness is central: the paper lets one agent *add* arbitrarily many fake identities, while this mirror fixes the bloc’s total mass and lets it repartition that mass. That change makes the continuous society well-defined, but it changes the manipulation the authors study. Also, this is a mechanism-existence question, not a complexity classification, so it does not fit the programme’s A/B/C trichotomy. I would offer it as a plausible research extension, not as a strong continuous-computational mirror of a qualifying result in this paper.
The strongest negative is that this paper has no named computational result to mirror. Proposition 1 is an impossibility theorem about mechanism properties; Proposition 2 constructs a transfer rule; Propositions 3 and 5 bound its distance from individual rationality. The facility-location section illustrates those mechanisms. None classifies an algorithmic problem or gives a complexity or algorithmic result. The paper’s open question is also about mechanism design. That leaves no result for ChoCo’s computational trichotomy to chart.
The proposed mirror of Proposition 1 makes the population continuous by letting a bloc of mass \(\beta\) split its reports. That is a coherent coalition-manipulation question, but it changes the paper’s central manipulation. In the paper, one real agent can add arbitrarily many identities, each with full weight in the welfare sum. In a normalized, atomless society, one agent has measure zero, so splitting their reports has no effect. Giving a positive-mass bloc control over reports makes the attacker a coalition; fixing the bloc’s total mass prevents it from adding the unbounded identity weight the proposition addresses. Allowing unlimited new mass instead requires a different account of population size and transfer scaling. Adding ownership relationships to the type space could preserve who controls which identities, but then the model must track that control structure, not just the distribution of preferences. These are substantive changes to the manipulation being studied.
Proposition 2 does have a natural mass analogue: replace sums of utility gaps by integrals over types. But the proposition gives an explicit payment formula; it makes no claim about the cost of computing it. Computing the selected outcome requires a representation of \(X\) and an algorithm for maximizing welfare over it, neither of which this paper specifies. Choosing a combinatorial outcome space might produce a worthwhile computational problem, but that would attach a new algorithmic question to the model rather than mirror Proposition 2. Facility location makes the population story especially natural: a distribution of resident locations has a median. Yet an individual’s zero mass cannot move that median, while a positive-mass attacker again changes false-name manipulation into bloc manipulation.
The distance bounds in Propositions 3 and 5 face the same issue. They compare an agent’s outcome with the outcome after that agent is removed. Removing a zero-mass agent leaves the continuous society unchanged. Removing a positive mass gives a block-level participation question, and the original bounds’ dependence on the number of agents has no canonical normalized counterpart. Such bounds could be reformulated, but the paper offers no computational result for them to anchor.
I cannot rule out every new algorithmic problem one could build around these mechanisms. That is the negative case’s honest weakness. But the submitted mass-splitting question remains a mechanism-existence problem, and the stronger variants require changing the paper’s manipulation model or adding a computational objective it never studies. Under ChoCo’s stated scope, none is a worthwhile continuous mirror of 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.