| paper | d4-journals-mss-2019-0047 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
Neither named result asserts an algorithm, a computational guarantee, or hardness. The proposed population version of balancedness is a coherent axiomatic extension, but it does not supply a computational result to mirror.
fails bit a — no named computational result to mirror
No computational result is covered; Theorem 1, Theorems 2–3 and 5, the lemmas, and the examples concern axiomatic properties or characterizations.
There is no anchor that meets the programme’s computational-result requirement. The paper’s named theorems concern social-choice properties and characterizations; none establishes a complexity classification or algorithm.
The closest result is Theorem 5, proved here: for \(n=3\) and \(m>3\), if a scoring social choice correspondence satisfies balancedness, then it is the Borda correspondence. Theorem 2, also proved here, says that for \(m,n\geq3\), a correspondence satisfying tops-only and balancedness is constant on all non-unanimous profiles. These are substantive results, but they are axiomatic, not computational anchors.
A plausible population version of the balancedness question would take rankings \(r\in L(X)\) as types and \(\mu_r\) as their population fractions. For adjacent, oppositely ordered \(x,y\) in types \(r,s\), transfer equal mass \(\varepsilon\) from each type to the type obtained by swapping \(x,y\). One could ask which scoring systems have the same winner set before and after every such transfer, using scores \(S_\mu(x)=\sum_r\mu_r\,\operatorname{rank}_r(x)\). That is a natural axiomatic extension, but it is not a computational problem anchored by this paper. Theorem 5 also fixes the electorate at three voters, so it does not itself establish a high-multiplicity regime.
The decisive weakness is therefore the missing computational result, not a lack of a plausible continuous population model. Under ChoCo’s scope, this paper supplies no positive continuous mirror to argue for.
The decisive point is that the paper has no named computational result to mirror. Its Theorems 2 and 5 characterize social-choice properties; neither gives an algorithm or a complexity classification. The proposed extension—put ranking types on a probability simplex and ask whether balancedness is preserved by equal-mass transfers—is a coherent continuous axiom. But replacing voter counts by masses does not turn an axiomatic characterization into a computational question.
Theorem 5 is the strongest anchor. Its \(n=3\) restriction does not by itself make a population model implausible: one could scale the three ranking types to large cohorts and define balancedness through transfers of mass between them. But the result would still ask which scoring rules satisfy an invariance condition. For a scoring vector, a paired transfer involving adjacent ranks \(k\) and \(\ell\) changes one alternative’s aggregate score by a quantity proportional to
\[
(s_{k+1}-s_k)-(s_{\ell+1}-s_\ell).
\]
That makes the continuous version a natural way to restate the axiom, and perhaps to prove a broader characterization. It does not supply a hard algorithmic task: for a specified society and transfer, the score change is directly computable, while testing whether all adjacent score gaps agree is elementary. Any deeper characterization remains a theorem about which rules are balanced.
Theorem 2 has the same problem. Extending its domain to distributions and replacing “non-unanimous” by a condition on the support could yield an interesting statement about correspondences on the simplex. Even if the constant-outcome conclusion carries over, it classifies rules; it does not ask how to compute an intervention, optimize over a society, or decide a complexity question.
The best computational salvage would add a problem such as finding a balancedness-violating society, or minimizing the mass of paired swaps needed to change a winner. That would require specifying an input representation, transfer costs, and an adversarial or intervention objective that the paper does not provide. It could become a worthwhile new robustness problem, but it would be built around a newly imposed task. For scoring rules, the score effects are explicit; for arbitrary correspondences, the paper supplies no computational representation from which to derive an algorithmic problem. Either way, it is not a computational mirror of the results proved here.
So the negative case is strong for ChoCo’s stated scope: this paper offers no computational result whose population version could chart the intended complexity landscape. It is weaker as a universal claim that no continuous work inspired by the paper could be worthwhile. A continuous balancedness characterization may be mathematically and socially interesting; it simply belongs to axiomatic social choice rather than this programme.
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.