UTH Search Protocol 1.0: The frozen universe, and how each state in it is searched

UTH Search Protocol 1.0 states, in advance and in public, which states are searched for published national calculations of the cost of unequal access to health, what is searched for each of them, and what counts as a find. It is the companion to UTH Standard 1.0, which judges the calculations the protocol finds. Its purpose is to make an empty row mean something. An unsearched row and a row where nothing exists look identical on a page. Under this protocol a state is never recorded as empty: it is recorded as searched on a stated date, by a named route, in named languages, with the search terms used, and nothing found. Anyone can rerun it, and anyone who reruns it and finds something the register missed has found a defect in the protocol rather than in the register's good faith. The universe is frozen at the 44 member states of the OECD together with the European Economic Area, listed in full. Each is searched across six source classes in every official national language and in English: finance ministry, health ministry, national statistics office, national gender-equality body, the public inquiry series, and the social insurance institution. The first run, 22 October 2026, is reported with the protocol: 23 states hold at least one published national monetised gap claim and 79 claims are recorded. No assessment of any publisher's work appears here; under UTH Standard 1.0 none is published before the consultation closes on 15 December 2026. The protocol was not completed in full for a single state: no row reached all six source classes, eleven rows were degraded by the run's search quota, four states have an incomplete national-language pass, and one state returned a null run in which no source class was examined at all. The status of all 44 states is tabulated. How the run was performed is disclosed as part of the protocol rather than as an appendix: the retrieval instruments used, the per-turn query quota shared across agents, and the servers that refused retrieval. A search whose instrument is undisclosed cannot be rerun. Thirteen defects in the protocol itself are published for version 1.1.