| paper | d4-journals-geb-2004-0081 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | no |
Theorem 4 proves convergence in a model that already treats the population as a continuum; it does not give a running-time guarantee for computing the limit. The proposed limit-computation task asks for the conclusion the theorem already supplies, rather than mirroring a computational result in the paper.
fails bit a — no named computational result to mirror
The proposed computation concerns Theorem 4's two-state convergence conclusion; it leaves the other convergence, stationary-point, and biased-sampling results aside.
The strongest positive lead is Theorem 4, proved in this paper. It says that with proportional sampling, \(N>1\), informative signals, the minimal informativeness condition, and payoff-determined initial popularity (two states), the system converges from any initial position to the efficient point \((1,0)\); average efficiency \(U\) is strictly increasing along every trajectory. The theorem gives a clean computational target: the long-run adoption shares in a large population.
A plausible regime is a large market of farmers choosing between two technologies. Agents share the same prior and payoff laws, and are distinguished by their current choice; the masses of the two choice types replace individual counts. There may be millions of farmers but only two types per state, and each newcomer samples a fixed small number \(N\). That is a credible high-multiplicity regime.
The candidate problem would be Two-State Word-of-Mouth Limit Computation. An instance gives rational \(q,\Delta,\underline{\Delta},\gamma,N\), rational initial shares \(x(0)=(x_1(0),x_2(0))\), and finitely encoded signal densities—for example, rational piecewise-constant densities for each state and sample—promised to satisfy Theorem 4’s assumptions. The task is to return the exact limiting shares and limiting average payoff. The solution is \(x^\infty=(1,0)\) and the full-information payoff \(q\Delta+(1-q)(-\underline{\Delta})\). On these promised inputs, the answer is constant and hence tractable. Follow-up questions could remove the two-state assumption, classify stationary corners as in Theorem 2, or study stability under the biased sampling rules of Theorems 6 and 7.
But this is only a candidate computationalization, not a qualifying computational anchor under your criteria. The paper’s numbered results concern monotonicity, convergence, stationary points, and stability; none asserts a complexity classification or gives an algorithm for a computational problem. More decisively, the paper already models a continuum of agents of mass \(1\). The population mirror is therefore already present in the paper: the candidate asks how to compute an outcome of that existing continuous-population model. Its weakest point is the central one—the paper offers a natural computational follow-through, but no genuine new continuization mirror.
Theorem 4 is the strongest candidate, and it fails the programme’s first gate: it proves a convergence result for a deterministic dynamical system. It does not state the complexity of a computational problem or give an algorithm with a running-time guarantee. The paper’s other named results are also analytic: Theorem 2 characterizes stationary corners, while Theorems 6 and 7 give stability and convergence conditions. There is no discrete computational result here to mirror.
The proposed “Two-State Word-of-Mouth Limit Computation” makes the issue clear. On its promised inputs, Theorem 4 already gives the output: \(x^\infty=(1,0)\), with the stated full-information payoff. Returning that answer is a trivial computation of an analytic conclusion, not a complexity question the paper leaves open. One could instead ask for convergence rates, parameter thresholds, or the set of stationary corners under broader assumptions. Those might be worthwhile algorithmic projects, but they would be new questions about this model, not computational mirrors of the paper’s results. Encoding signal densities to make them inputs does not supply the missing discrete problem.
The same applies to the suggested extensions of Theorems 2, 6, and 7. Checking their stated conditions from an encoding could become a decision problem; searching over corners, or synthesizing sampling rules with desired stability properties, could be more substantial. But each requires adding a computational task that the paper neither formulates nor analyzes. The better versions may deserve study on their own merits, yet they do not establish a continuous mirror of a result in this paper.
There is also no population continuization to perform: Section 2 already models a continuum of agents of mass \(1\), tracks population shares, and specifies aggregate evolution directly. That is not an objection based on high-multiplicity work, nor does it mean an analytic treatment already answers a computational question. It means the paper starts on the continuous-population side and supplies no discrete computational result whose high-multiplicity version could fill ChoCo’s gap.
I cannot rule out a worthwhile computational theory of finite encodings of these dynamics. The negative case is narrower: such a theory would be a new research direction, not a continuous mirror of this paper’s computational result, because the paper has no such result.
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.