| paper | d4-conf-atal-2015-0208 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | medium |
| authors would recognise it | yes |
Proposition 1
statement extracted from the paper’s text layer
Given a finite candidate set \(X\), a listed set of ranking types \(S\subseteq L(X)\), and a formula \(\varphi\) with outcome atoms, rational mass-threshold atoms, Boolean connectives, and modalities \(\Diamond_\alpha\), determine whether \(\varphi\) holds for every distribution \(\mu\in\Delta(S)\) and every resolute rule \(F:\Delta(S)\to X\). At \((F,\mu)\), \(\mathsf{Mass}_{t\ge q}\) is true exactly when \(\mu_t\ge q\), and \(\Diamond_\alpha\psi\) holds when some transfer \(\mu'=\mu-\nu+\nu'\), with \(0\le\nu\le\mu\) and \(\sum_t\nu_t=\sum_t\nu'_t=\alpha\), makes \(\psi\) true at \((F,\mu')\).
A society is a distribution \(\mu\) over a finite list of complete ranking types; a resolute social choice rule maps each distribution to one winner. Modal actions transfer a specified mass of voters between types, and validity ranges over all societies and all such rules.
The mirror replaces named-agent control atoms and exact coalitions with mass thresholds and divisible mass moves, creating a validity problem over an uncountable simplex for which the paper’s finite enumeration supplies no algorithmic foothold.
fatal: False
The mirror covers Proposition 1’s validity question; it leaves the completeness result and the formal proof of Arrow’s theorem, including Theorem 10, outside its scope.
The strongest case I can make is narrow: Proposition 1 is the paper’s only named result that directly asserts a computational status. It says that, for fixed finite \(N\) and \(X\), determining whether a formula of \(L[N,X]\) is valid is decidable. The proof is given here: enumerate the finitely many resolute SCFs and check the formula at every profile. This is a genuine decision result, but it gives no running-time bound. Theorem 3 states Arrow’s impossibility theorem, and Theorem 10 gives its syntactic derivation; neither is a complexity result, so I would not use them as anchors.
A plausible high-multiplicity setting is a large electorate in a recurring voting environment where, for a fixed agenda, voters use only a modest number \(\tau\) of complete rankings. There may be millions of voters but relatively few ranking types. A type is the ranking itself, and its mass is the fraction of the electorate with that ranking. This retains the paper’s subject—social choice functions and coalitions’ ability to change ballots—while replacing named agents with divisible mass.
My lead mirror is Mass-Modal Validity. An instance consists of a finite candidate set \(X\), a listed finite set \(S\subseteq L(X)\) of allowed ranking types, and a formula \(\varphi\). The language has outcome atoms \(x\in X\), mass atoms \(\mathsf{Mass}_{t\ge q}\) for \(t\in S\) and rational \(q\in[0,1]\), Boolean connectives, and modalities \(\Diamond_\alpha\) for rational \(\alpha\in[0,1]\). A society is a distribution \(\mu\in\Delta(S)\); a resolute rule is a function \(F:\Delta(S)\to X\), with \(F(\mu)=x\) making outcome atom \(x\) true. The mass atom is true when \(\mu_t\ge q\). Define \(\Diamond_\alpha\psi\) to hold at \((F,\mu)\) if a coalition of mass \(\alpha\) can change its ballots to produce some \(\mu'=\mu-\nu+\nu'\), where \(0\le\nu\le\mu\) and both \(\nu\) and \(\nu'\) have total mass \(\alpha\), such that \(\psi\) holds at \((F,\mu')\).
The question is: is \(\varphi\) valid, meaning that it holds for every \(\mu\in\Delta(S)\) and every resolute \(F:\Delta(S)\to X\)? A solution is a correct yes/no decision. This is a type-level counterpart to the paper’s validity question: it preserves outcome propositions and coalition control, while mass predicates replace propositions about individual ballots. The rule depends on the society’s type distribution, as appropriate when the type description includes everything the problem uses about an agent.
I would tentatively expect a decidability result, perhaps a tractable one for fixed \(|S|\): each mass change and mass predicate is described by linear constraints over a finite-dimensional simplex. A finite formula therefore suggests a route through polyhedral reasoning or linear arithmetic. That is a research hypothesis, not something Proposition 1 establishes; in particular, I would not claim a polynomial bound when \(m\), \(\tau\), and formula length all vary. The natural follow-up questions are the complexity in those parameters, whether arbitrary rules \(F\) create additional difficulty through consistency across repeated visits to the same society, and whether a pointwise version with a supplied rational \(\mu\) is easier.
This mirror covers Proposition 1’s logical validity problem, not the Arrow result or the paper’s completeness theorem. Its weakest point is substantial: Proposition 1 concerns a finite-agent logic, and Mass-Modal Validity requires a new type-level language and anonymous, distribution-based rules. The paper’s discussion of the fixed-\(N\) limitation and its focus on SCFs and coalition control make the extension recognizable, but a referee could reasonably judge it an adjacent formal-methods question rather than the continuous computational-social-choice problem the paper itself studies.
Proposition 1 is the only anchor, and the case against it is a mismatch in what is being computed. Its decidability proof relies on fixed finite \(N\) and \(X\): there are finitely many profiles and finitely many SCFs, so validity can be checked by enumeration. The paper’s main contribution is a syntactic proof of Arrow’s theorem; it gives no complexity bound or algorithmic result about changing an election.
The proposed mass-modal validity problem is coherent for a large anonymous electorate with few ranking types. But it changes the underlying logic: \(p^i_{x\succeq y}\) and \(3_C\) refer to named voters and exact coalitions, whereas mass predicates and \(\Diamond_\alpha\) retain only a distribution and the amount of mass that may move. Profiles with the same histogram become indistinguishable, and the class of rules becomes distribution-based. That is a sensible high-multiplicity restriction, but it is not a continuous version of Proposition 1’s full model class.
The stronger anonymous version still asks a new question about validity over an uncountable state space and arbitrary functions \(F:\Delta(S)\to X\). To make it an algorithmic problem about a supplied rule, one would need to add a finite representation for \(F\); to keep arbitrary \(F\), the question remains one about the new logic’s symbolic validity. Either could be interesting, but the mass thresholds and transfers introduce linear constraints that Proposition 1 does not contain, so any result would chiefly concern the added formalism, not the paper’s enumeration result.
I would not argue that a continuous population is implausible here, or that the new validity question is worthless. The paper itself points toward future work on automated reasoning and verification, which makes this extension especially recognizable. That leaves the negative case weak: the best argument is that this is a neighboring formal-methods project rather than a strong mirror of a computational result in the paper. I do not see a solid basis for the universal claim that no worthwhile continuous mirror exists.
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.