The Extinction Census — Frozen Prereg and Findings: Prediction 6 with a Partial Verdict

This record deposits, unaltered, the registered occupational census behind the manuscript's extinction-asymmetry prediction, with a deliberately PARTIAL verdict. Registered claims - P1: every extinction-grade (>=70%) decline of an information-channel expert occupation traces to a cheap autonomous instrument, never to an aggregation venue; P2: no aggregation event extinguishes its expert channel. Frame: OES national employment 1999/2019/2023, 625 matched occupations, 35 information-channel rows, classifications committed before outcomes (prior-knowledge flags per row); frozen prereg sha256 994d802fe30595ca1a73c591a31d539ed1cf4cb094f897366efafb6b07f097ad, to which the deposited file hashes exactly. Verdict. P2 SURVIVES substantively: every mapped aggregation-exposed expert occupation grew or held (interpreters/translators +278.0% through the machine-translation era; analysts +127.7%; advisors +240.4%; actuaries +102.8%); the sole aggregation-classified contraction (survey researchers, -62.8%) attributes to documented drivers other than aggregation (respondent feedstock collapse, execution automation, a disclosed code reclassification). P1 is UNTESTED IN-WINDOW: zero extinction-grade declines occurred among information-channel occupations 1999-2023, so the universal had nothing to bite on - stated as such, not claimed as support; the window's informative content is the class-outcome asymmetry (deepest technology-attributable contraction is instrument-classified: auto damage appraisers -55.1%, photo/AI estimating). Falsifier check: no falsifying row. Honesty flags, machine-checked by the included ledger (18 checks, exit 0, including that no deposited outcome row reaches the extinction grade - the P1 statement is derivable from the data): the survey-researcher theory-consistent reading is labelled post-hoc with no confirmatory status; the appraiser payroll-versus-licensee series divergence is disclosed; the travel-agents exclusion was ruled ex ante with the standard commission-cap cause named. What this record does and does not do to the theory: neither limb refutes or degrades any claim - P2's survival is supportive at census strength, and P1's in-window emptiness is unfed rather than tested, booked neither way. The record's contribution going forward is exposure: the falsifier is stated and standing, so a future aggregation-caused extinction of an expert channel refutes the prediction.