| paper | d4-journals-geb-2019-0054 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
A continuous electorate over ordinal preference types gives a sensible maximal-lottery computation problem. But the paper's numbered results establish welfare and participation properties or axiomatic characterizations; none asserts a computational result, so bit (a) fails.
fails bit a — no named computational result to mirror
The proposed extension covers computing maximal lotteries from population masses and the participation implication; it does not turn the paper's axiomatic characterizations into computational results.
The strongest honest case is a natural population version of the paper’s maximal lottery results, but it is not anchored by a computational result in this paper. The theorem, proposition, and corollary statements here concern welfare, participation, and axiomatic characterizations. None asserts that a problem is in P, NP-hard, FPT, or otherwise classifies its computational complexity. The theorem environments in the supplied text also have no printed numbers to quote. So there are no qualifying anchors for the requested per-anchor case.
The plausible continuous model is clear. Let a type \(t\) be a complete ordinal preference over the alternatives, represented by its canonical SSB function \(\phi_t\). An instance gives rational masses \(\mu_t\) over \(\tau\) supported types, with \(\sum_t\mu_t=1\). A solution to Maximal Lottery Computation for a Continuous Electorate is a lottery \(p\in\Delta(A)\) satisfying
\[
\sum_t \mu_t\phi_t(p,q)\ge 0 \qquad\text{for every }q\in\Delta(A).
\]
Because the functions are bilinear, it suffices to impose these inequalities for each pure alternative. This is a linear feasibility problem in the lottery probabilities. For a large electorate whose voters share a relatively small set of ballot types, \(\mu_t\) is the share holding each ballot. A subgroup’s abstention is a vector \(\nu\) with \(0\le\nu_t\le\mu_t\); the continuous participation condition is
\[
\sum_t \nu_t\,\phi_t\bigl(f(\mu),f(\mu-\nu)\bigr)\ge 0.
\]
The paper’s result that welfare-maximizing mechanisms satisfy participation has a direct mass-based analogue by the same argument. Its corollary that mechanisms returning maximal lotteries satisfy ordinal participation is a particularly close motivation for this extension.
That is a sensible population continuization: it makes the society a distribution over preference types, rather than treating the lottery over alternatives as the continuization. The case is weakest as a computational mirror. The paper does not analyze the complexity of computing maximal lotteries, and the LP follows from the model rather than from a named complexity result proved here. I would present this as a plausible follow-on question, not claim that the paper supplies the required computational anchor.
The strongest negative starts with the gap the proponent acknowledges: this paper proves no computational-complexity result. Its welfare-implies-participation theorem, converse characterizations, and maximal-lottery corollaries are statements about properties of mechanisms. They do not define an algorithmic problem or classify its complexity.
The proposed continuous electorate is sensible: let each type be an ordinal preference and give each supported type rational mass \(\mu_t\). But computing a maximal lottery then reduces directly to linear feasibility. Set \(M_{xy}=\sum_t\mu_t\phi_t(x,y)\); find \(p\in\Delta(A)\) such that \(\sum_x p_xM_{xy}\ge 0\) for every alternative \(y\). This has polynomial size in \(m,\tau,L\). The same computation applies to a discrete electorate after aggregating its ballot counts. So the mirror preserves the population interpretation, but it does not reveal a new complexity question in the paper’s results.
The participation extension is just as direct. If \(p\) and \(p'\) are maximal lotteries before and after abstention mass \(\nu\), then
\[
\sum_t\nu_t\phi_t(p,p')
=
\sum_t\mu_t\phi_t(p,p')
+
\sum_t(\mu_t-\nu_t)\phi_t(p',p)
\ge 0.
\]
That is the paper’s proof with sums replaced by masses. The characterizations involving homogeneity, Condorcet consistency, and cancellation likewise remain axiomatic questions about mechanisms. Turning them into computational problems would require adding a representation of mechanisms and a new task, such as verifying a property for all mass distributions; that would be a follow-on project, not a computational mirror anchored by this paper.
This defeats the proponent’s strongest mirror under ChoCo’s computational remit, though it does not show that the continuous axiomatic extension lacks mathematical value. The negative case is strong about this paper’s fit for the programme and weaker as a universal claim that no useful scenario could ever be devised.
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.