Preregistered measurement protocol reports an empirical design to quantify answer materiality in undeclared price-list updates, highlighting optimal re-check scheduling for knowledge systems.
A system holding knowledge whose changes originate outside it must decide how often to re-check. The scheduling side of that problem has an established theory — refresh under a finite budget as a restless multi-armed bandit solved via Whittle indices, with Age of Information and Age of Incorrect Information as the regulated quantities, and fresh caching as the applied line (Whittle 1988; Kaul et al.; Abolhassani, Tadrous, Eryilmaz and co-authors; Koley and Singh 2024). These policies regulate on a change signal. An earlier record (Bering 2026a, `10.5281/zenodo.22148710`) preregistered a measurement of the **answer-materiality rate** *m* — the proportion of dated change events that overturn at least one answer of a fixed question set — for software releases carrying a materiality level declared at the source under semantic versioning. That class has a property which limits what it can settle: the source already declares. A reader may reasonably answer that one should simply check on major releases. This record preregisters the complementary case: a source that **dates its changes but declares nothing about them**. Published price-list versions carry an effective date and a version identifier; they carry no indication of whether any price changed. Materiality is therefore decidable by comparison of values rather than by judgement, which removes the model from the measurement. The record further preregisters a **coverage sweep**. The rate *m* is not a property of a source alone but of the pair (source, set of facts held): a system holding more facts is exposed to more of the same change stream. Sweeping the number of held facts *N* turns that dependence from a caveat into a measured curve, and tests it against the prediction that would hold if changes were independent across facts. No results are reported. The protocol is fixed before measurement, and the selection of sources was carried out on metadata only.
No takes yet. Share an insight, caveat, or question.
Alexander Bering (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: