| paper | d4-journals-mss-2022-0009 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
The paper's numbered results establish probability, covariance, and structural facts, but none asserts an algorithm, a complexity bound, or hardness. Computing limiting tournament probabilities from a type distribution is a coherent follow-on question, yet it cannot supply the missing computational anchor required for a green grade.
fails bit a — no named computational result to mirror
The proposed extension concerns asymptotic tournament and qualitative-margin probabilities, especially the probability statement in Theorem 4.2. It does not classify the paper's covariance and eigenspace results, conjecture about linearity, finite-voter probabilities, or Split Cycle and Minimax consequences.
The strongest honest positive case is a follow-on question, not a mirror anchored in this paper’s named computational results: the paper contains no named theorem, lemma, corollary, or proposition classifying a computational problem as tractable or hard.
The closest anchor is Theorem 4.2 (also stated as Theorem 1.5): for every tournament \(T\) and total ordering \(\prec\) of its edges, there are \(N\) and \(p>0\) such that, for every \(n\ge N\), an election of \(n\) voters choosing rankings uniformly at random has probability at least \(p\) of producing qualitative margin graph \((T,\prec)\). That is a probability-existence result, not a computational result. Theorem 5.1 gives the covariance matrix’s cycle-space and cut-space eigenvalues; it is also not a complexity result. Section 9 describes a faster Gaussian simulation method, but it is not a numbered result.
A plausible future question would take rankings as types and an arbitrary distribution \(\mu\) over them, then ask how to compute the limiting probability of a specified tournament under the corresponding Gaussian margin distribution. But that is not a computational problem this paper’s named results solve or classify. Its continuous object is the limiting distribution of margin outcomes; it does not make the society itself an input distribution over types. Under the programme’s stated anchor requirement, I can’t honestly present that future question as a continuous mirror of a named computational result in this paper.
The strongest case against a mirror is that this paper gives no named result about the complexity of a computational problem. Theorem 4.2 says qualitative margin graphs occur with positive probability under impartial culture; Theorem 5.1 describes the covariance matrix; Theorem 6.1 decomposes margin vectors. The results about Split Cycle and Minimax concern their asymptotic behavior. Section 9 gives a faster simulation method, but it does not classify a problem’s complexity. There is therefore no computational result here for a continuous mirror to inherit.
The proponent’s proposed question—compute the limiting probability of a tournament from an input distribution \(\mu\) over rankings—is the best repair, but it changes what is being studied. In the paper, randomness comes from drawing finitely many ballots. If \(\mu\) is instead the exact mass of each type in a continuous society, its pairwise margins are fixed; under impartial culture they are all zero. The tournament probabilities and the covariance in the paper come from finite-sample fluctuations around those masses. Keeping those probabilities means retaining a sampling model, rather than asking about the deterministic continuous society itself.
Allowing arbitrary \(\mu\) does not remove that tension. If its expected pairwise margins are strict, the sampled election’s majority graph converges to the graph determined by those margins, so the limiting probability is trivial. The interesting cases require ties in expected margins, as under impartial culture, or a scaling of the society toward ties as the population grows. Computing probabilities in those cases could be a worthwhile problem, but it is a new Gaussian probability-computation question. Its central dimension is the number of candidate pairs, not the number of voter types; adding bribery, control, or robustness would make it a different problem again.
That is why I would reject this as a continuous mirror of the paper’s results: the named results are probabilistic and analytic, and the plausible computational extension either keeps the finite sampling that generates their randomness or adds a new computational task. But the universal negative case is weak at precisely that boundary. The paper explicitly points to other ballot distributions as future work, and computing outcome probabilities from a type distribution is coherent and may interest its authors. I can defeat the claim that this paper already supplies a computational anchor; I cannot honestly rule out that extension being worth ChoCo’s time.
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.