Surface Weather Station: Claude-Substrate Baseline Reading is the first partial v1.1-conforming scan in the federated, cross-substrate visibility series.
The scan examined eight of twelve objects through nine searches and produced the batch’s most consequential methodological finding: there is no single public surface. Different models and retrieval backends presented materially different versions of the same corpus.
For the Claude-associated scan, Alexanarch was occluded; Lee Sharks, the Crimson Hexagonal Archive, the Semantic Economy Institute, and SPXI survived mainly through older anchors; Provenance Erasure Rate appeared only through secondary citation; Writable Retrieval Basin and Revelation First disappeared behind generic results or homonyms. Yet parallel Kimi and Gemini readings reported several of these objects as strongly visible.
The difference was not merely evaluative disagreement. The systems had retrieved different evidence. The record names this retrieval-stack divergence and separates two experimental layers:
This distinction changed the instrument. It showed that visibility scores can measure the search backend as much as the corpus, and that exact-match support, indexing coverage, personalization, and training priors belong inside the observation environment.
The scan is partial by design, but it opens the federated weather station: a method for observing not one composition layer, but the plurality of machine-specific worlds through which scholarship now arrives.