Categorical Integrity in Non-AI Domains: The citation-index forensic and a channel contrast

The journal impact factor supplies a non-AI forensic for recursive categorical evaluation. A scalar becomes categorical through quartiles, "high-impact" designations, and career gates; its judgments then feed the next population through editorial optimization, venue selection, and institutional reward. The channel is documented in coercive citation, citation stacking, and journal-level composition engineering. The evidence separates three claims: journal means were already poor article-level proxies by 1997; the journal–article citation relation weakened after 1990; and journal rank is associated with retraction rates in the opposite direction. Only the second is a trajectory, and none identifies recursive use by itself. DORA is read as a channel-weight intervention rather than dismissed as governance: where institutions remove journal metrics from hiring and funding, the recursive channel should weaken even if the citation substrate remains. Its uneven adoption leaves a comparative test. Standardized testing supplies a within-domain contrast. High-stakes state scores rose against lower-stakes NAEP measures across overlapping jurisdictions, but the contrast bundles verdict use with item exposure and format-specific preparation. Peer review marks the intent-dominant boundary. Together the cases vary channel weight and anchor structure while preserving the distinctions needed to test, rather than merely export, the mechanism.