When Policy Growth Does Not Become Evaluable Rules: Measuring Policy Computability in Chinese Postdoctoral Governance, 1986-2026

Abstract Policy growth is often taken as evidence that a governance system has become more mature, yet it remains unclear whether more policy text produces rules that can be evaluated, audited and reused in research assessment. This study examines that problem in Chinese postdoctoral governance. We analysed 260 curated policy and governance records issued between 1986-10-27 and 2026-05-29, segmented them into 10,875 clause-level evidence units, and encoded actor, instrument, governance-object and rubric-dimension tags. The framework combines an eight-topic semantic policy twin, a typed clause hypergraph, a Policy Digital Twin Index (PDTI) and an evidence-constrained rubric compiler. Three findings are central. First, policy volume and rule computability were only partly aligned: document length was strongly associated with PDTI (Spearman rho = 0.652) but was unrelated to the share of rubric-ready clauses (rho = 0.002, p = 0.970). Second, mean PDTI increased from 4.452 in the founding stage to 6.368 after 2016, while clause-level rubric readiness remained limited (0.370). Third, the weakest computable dimension after 2016 was fairness safeguard, whose clause prevalence was only 0.028. A validation model combining semantic topics and computability features recovered the post-2015 quality-governance regime with a five-fold AUC of 0.860. The results show that policy modernisation produces more evaluable mechanisms, but it does not automatically turn public values into auditable criteria.