It wasn’t until today that I found out about P.L. 115-435 during the finalization of a methodology paper for submission to an appropriate journal (such as Sociological Methodology) for which I hoped I would find a credentialed coauthor. Once the methodology found publication I could then write a paper focusing more directly on intergenerational consequences of the latent factors causing loss of social capital (starting with loss of economically valuable characteristics).
Since my motivation for investing time at this relatively shallow level (ie: direct implications for value) was out of a sense of urgency arising from “influencers” palavering about “abundance” and “optimism” while walking on very thin evidencial ice over the abyss of decades of obvious and egregious sovereign violation of consent, I decided maybe to indulge a conceit arising from my leading involvement in another quixotic legislative reform: P.L. 101-611.
A Criterion of Merit for Federal Evidence Building
Sealed-corpus description length, and what it would take to identify the dynamics of a social system
James A. Bowery · September 2026
Hover any underlined term for its definition; definitions are themselves hoverable, so a term you do not know can be unpacked recursively. The full glossary, with 109 entries tagged by provenance, is a separate page.
Abstract
The Foundations for Evidence-Based Policymaking Act of 2018 (P.L. 115–435) created access and governance for evidence building—an openness presumption for public data, a confidentiality framework for restricted data, and the secure-enclave machinery that became the National Secure Data Service—but it defined no criterion of merit. Nothing in the Act says what makes one body of evidence better than another, so the question is settled by expert judgment exercised inside privileged access, and a member of the public cannot contest a finding without obtaining the same access. This is not an abuse of the Act’s access regime; it is that regime working as written. We propose the missing piece: a single hard number, the total description length of a sealed corpus at instrument precision, scored by a custodian who runs submitted programs and publishes only the resulting bit count. The data never move, so confidentiality is preserved exactly; the program is public, so the data-cleaning algorithm is scored rather than presumed; and replication becomes the act of submitting a rival program and beating the number. We then demonstrate the submission format at small scale on 3,146 United States counties observed at eight time points from 1970 to 2024 on 777 items, reporting the identified dynamics, an intervention operator warranted by measured invariance, the curation defects the exercise exposed, and one instructive failure: the sole component of the fitted system that does not extrapolate is the sole component we supplied from outside it.