Towards a More Burkean Approach to Computational Social Choice

Strictly under ChoCo’s anchor rule, this paper has no qualifying computational anchor. Its only numbered formal statement is Definition 1, defining executive capacity; it is neither a theorem nor a computational result. The claims that district division is NP-hard and that primaries can change distortion by \(O(1)\) or \(O(n)\) are prose summaries attributed to other papers, not numbered results proved here. I would not mislabel them as anchors.

The strongest honest positive case is therefore prospective rather than paper-anchored: the paper’s Executive Capacity definition gives a very natural population mirror.

Call it Party-Based Continuous Executive-Capacity Robustness. An instance contains finitely many districts, parties and candidates; a finite set \(T\) of voter types; rational masses \(\mu_t\); and a rational metric space of ideological positions. A type records everything the mechanism uses: district, ideal point, party-relevant attributes, and the induced complete ranking of candidates. Millions of voters may share one type. Candidates have initial positions, and the electoral mechanism is the paper’s suggested party-based mechanism: plurality within each district, with the party winning the largest number of districts declared the overall winner.

For a target party \(p^\star\), an adversary may move mass \(x_{t,t'}\) from type \(t\) to type \(t'\), subject to the ideological displacement being at most \(\varepsilon\), while non-target candidates may also move within \(\varepsilon\). The resulting mass is

\[ \mu'_{t'}=\sum_t x_{t,t'}. \]

The problem is to compute the smallest \(\varepsilon\) for which some admissible mass transfer and candidate relocation makes \(p^\star\) lose. Equivalently: given \(\varepsilon\), decide whether \(p^\star\) remains the winner under every such perturbation. A negative answer includes \(x\) and the candidate relocations as a certificate; an optimization answer returns the critical robustness radius.

This is recognisably the continuous version of Definition 1, not merely a fractional outcome model. The winner remains discrete, parties and districts remain explicit, and continuity is introduced in the population: “a small movement of the electorate” becomes a transfer of a small fraction of each voter type. A plausible regime is a national election with millions of voters but only a few hundred recurring ideological-demographic types per district. Such voters are not individually targetable; the meaningful intervention is changing \(0.5\%\) of a type’s mass, exactly the scale ChoCo is designed to expose.

For a fixed district structure and a finite catalogue of ideological states, I would expect this version to be Class A. At a fixed perturbation radius, the adversary’s mass transfer is a capacitated transportation polytope, and plurality conditions are linear inequalities in the resulting masses. The critical radius can be found by checking the finitely many relevant type and candidate distances and solving the associated LPs. This would turn the paper’s qualitative robustness notion into an actual computational study, with approximation and rounding questions connecting it back to finite elections.

The mirror becomes substantially less obviously tractable when the district partition is itself chosen, compactness constraints are imposed, candidate positions range over a continuous space, or coalition formation is endogenous. Those variants may expose Class B hardness inherited from districting or Class C hardness caused by the continuum. That boundary would itself be valuable: it would show exactly when executive capacity benefits from population continuization and when the party/geographic structure defeats it.

The regime is also plausible for the paper’s sensor-array example: many sensors share calibration, location class and observation profile, so a type is a complete sensor state and mass is the fraction of sensors of that kind. Robustness then asks how much of the array’s population-level signal can drift before the reported winner changes.

The weakest point is decisive: the paper leaves \(f\) deliberately general and does not specify a computational input model, encoding, or optimization question. Choosing finite ideological states and fixed districts could therefore be attacked as making the problem tractable by narrowing it beyond the authors’ intended question. Conversely, retaining the fully general metric-space definition leaves complexity underspecified. The mass-transport formulation is a faithful high-multiplicity interpretation of their robustness idea, but it is not a result contained in the paper.

So my positive case is: the paper contains no qualifying named computational result, but its numbered Definition 1 supports one unusually natural continuous-population research problem, likely tractable in a finite high-multiplicity regime and capable of generating meaningful harder extensions. It covers only the Executive Capacity direction; it does not support anchored mirrors for liquid democracy, sortition, primaries, or districting claims.