Finite-Sample Signal Uncertainty and Sharp Partial Identification of Diversification Histories: An Affine Signal-to-Decision Workflow with a Bounded CRABS Comparison

Abstract

Unrefereed Evidence Press candidate. Under an exact fixed-stem, stem-survival, homogeneous time-varying birth-death model, this successor constructs a finite-sample confidence set for the pulled scale and propagates it to conditional affine partial-identification decisions. A separate bounded CRABS benchmark retained a failed prespecified H4 utility gate. Producer-side replay and public Linux CI passed; independent reproduction, external specialist review, formal verification, editorial peer review, biological validation and broad CRABS utility are not established. Original prose and project-created data are CC0-1.0; original code is MIT; third-party exceptions retain their recorded terms.