Measurement-Layer Audit Protocol for Directionality-Adjacent Markers

How should directionality-adjacent claims be measured across a heterogeneous scientific literature without collapsing distinct phenomena into a single thematic category? This study addresses that question by introducing a conservative measurement-layer audit protocol for two explicitly defined markers — entropy production and irreversibility — and applying it to a fixed 12-study cohort. The completed output is a 24-item study x marker matrix built from accepted records produced through machine extraction, source PDF review, and deterministic export. The resulting matrix shows a nested pattern within the present cohort. Explicit irreversibility is broader than explicit entropy-production measurement, and all entropy-production-positive cases also fall within the irreversibility-positive subset. Across the 12 studies, entropy production is positive in 4 cases and irreversibility is positive in 9 cases, with no entropy-positive / irreversibility-negative cases. A separate post-collection inter-rater reliability follow-up was then conducted on the same 24 study-marker items. Three retained streams — a blinded GPT baseline run, a refined-instrument GPT rerun, and a Claude Opus sensitivity run — each produced 18 matches out of 24 items, or 75 percent agreement. Kappa values ranged from 0.50 to 0.52, which falls in the moderate-agreement range under the Landis and Koch heuristic; however, because the present analysis is based on a small 24-item binary matrix with uneven marker prevalences, these kappa values should be read cautiously rather than as prevalence-invariant estimates. The refined GPT rerun did not change the binary result set, while Claude Opus changed which irreversibility cases disagreed without improving the topline agreement. The main contribution is methodological. The protocol separates marker measurement from thematic inference and makes each accepted record source-traceable, providing a conservative and reproducible footing for later hypothesis testing and theory evaluation. Note: Reproducibility Packet for "A Measurement-Layer Audit Protocol for Directionality-Adjacent Markers available at Zenodo. See external links.