| paper | d4-conf-aaai-2026-1705 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
Theorem 4.8 is a numbered result, but it proves IPDA rather than a computational claim. Weighted-population River evaluation is a direct, well-posed question, yet its apparent polynomial-time procedure is not a numbered computational result in this paper, so no anchor passes bit (a).
fails bit a — no named computational result to mirror
The direct mirror covers evaluating River on a distribution over ranking types; it does not turn the paper's axiomatic results into computational results.
The strongest honest positive case is a direct continuous version of River itself, but this paper has no numbered result asserting a computational complexity classification. Theorems 4.2, 4.6, 4.8, 4.10, and 4.14 establish axiomatic properties; Theorem 4.4 gives counterexamples to those properties for other rules. None is an eligible computational anchor. The paper does mention computational results from other work, including polynomial-time computation of River under parallel-universes tiebreaking, but that result is neither proved nor numbered here, and the paper explicitly sets that variant aside.
A plausible mirror would be Continuous River Evaluation. Its instance is a finite alternative set \(A\), a list of \(\tau\) strict rankings with rational masses \(\mu_t\geq 0\) summing to \(1\), and a specified tiebreaker. Type \(t\) is a complete ranking; its mass is the population fraction holding that ranking. Define the pairwise margin by
\[
\Delta_\mu(a,b)=\sum_t \mu_t\bigl(\mathbf{1}[a\succ_t b]-\mathbf{1}[b\succ_t a]\bigr).
\]
Run River on these weighted majority margins, processing weak majority edges in decreasing margin order and applying the cycle and branching conditions. The solution is the resulting tree and its unique source, the winner. A natural instance regime is a large, repeated preference-aggregation exercise—such as standardized feedback from many participants—with many agents but relatively few distinct rankings.
This is close to the paper’s actual object: River is defined through pairwise margins, and its authors present simple computation as one of the method’s advantages. From the definition, computing the margins and sorting the \(O(m^2)\) edges takes polynomial time in \(m,\tau,L\), where \(L\) is the mass encoding length. That is an inference from the procedure, not a result proved in the paper. It would make this a tractable evaluation problem, but it does not show discrete hardness dissolving in the continuum.
Theorem 4.8, proved in the paper, is a tempting thematic anchor but not a qualifying one: it says River satisfies independence of Pareto-dominated alternatives with a Pareto-consistent tiebreaker. Its continuous counterpart would say that if every positive-mass ranking puts \(y\) above \(x\), removing \(x\) leaves the River winner unchanged. The theorem’s margin-based argument makes that extension plausible. Still, it is an axiomatic question, not the computational question required by the programme.
That is the weak point in this case. The authors would likely recognize Continuous River Evaluation as the weighted-population version of their method, but the paper’s numbered contributions concern axioms, not the complexity of evaluating the rule. So this is a plausible population mirror of River, but not a strong anchor-based case for the paper as a ChoCo computational-complexity target.
I would vote against a continuous mirror of this paper’s *contribution*. The key point is that it has no numbered result about the complexity of a computational problem. Theorem 3.2 and Theorems 4.2, 4.6, 4.8, 4.10, and 4.14 establish axiomatic properties; Theorem 4.4 gives counterexamples to other rules’ properties. The paper says River is easy to compute, but does not make its running time a result. The separate polynomial-time result for River with parallel-universes tiebreaking is attributed to other work, and that variant is explicitly outside this paper’s scope.
The proponent is right that a continuous River is easy to define, and that large populations with repeated rankings are a sensible regime. Rankings are complete types for this rule, and their masses give meaningful fractions of a population. That makes the direct mirror plausible, but its evaluation problem appears to have no computational question left to resolve: compute the \(O(m^2)\) pairwise margins from the \(\tau\) types, sort the weak majority edges, and apply River’s cycle and branching checks. This takes polynomial time in \(m,\tau,L\), where \(L\) is the mass encoding length. The paper’s own procedure already supplies the algorithm; the continuous input simply lets it compute the margins without listing every voter.
There is also an exact scaling correspondence. For rational \(\mu\), choose a common denominator \(D\) and replicate each type \(D\mu_t\) times. The discrete margins are \(D\Delta_\mu(a,b)\), so their order and ties—and hence River’s output, with the same tiebreaker—are unchanged. That correspondence is not a collision objection: high multiplicity is a legitimate regime. It does, however, reinforce that weighted River evaluation adds no apparent algorithmic structure to the procedure in this paper.
Theorem 4.8 does not rescue the case. Its continuous counterpart is a sensible statement about deleting a Pareto-dominated alternative, but it remains an axiom about how the winner changes when the *agenda* changes. The same is true of the paper’s other main results. Recasting profiles as distributions does not turn those claims into computational questions, and the programme explicitly sets axiomatic continuization aside.
One could define a new problem, such as the least mass that must change rankings to alter River’s winner. That may be worthwhile research, but it needs an intervention model, costs, and a target that the paper does not study. It would be a problem built on River, rather than a continuous mirror of a computational result in this paper. For the same reason, the externally cited PUT computation does not create an anchor here.
So the negative case is strong against this paper as a ChoCo complexity target, but weaker as a claim that no useful continuous River project exists. I cannot rule out a worthwhile population-robustness or bribery problem built around River; it would be a new direction, not a mirror of the paper’s results.
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.