Evaluating Protocol-Like Annotation Schemes in the Book of Revelation: An Exploratory Blocked Benchmark of Domain, Discourse Mode, Authority, and Token Context

This exploratory study evaluates whether a protocol-like annotation scheme for the Book of Revelation captures held-out textual structure beyond conventional linguistic and discourse baselines. The corpus comprises 82 event-expression complexes drawn from seven passage sets. Passage-level blocked evaluation compares frequency, lexical, discourse-cue, first-order sequence, authority-graph, and token-context models. Domain and discourse-mode annotations exhibit learnable structure, but the tested protocol-specific sequence and authority models do not outperform simpler conventional controls. The study is a single-analyst feasibility benchmark and does not establish historical covert communication. This record contains the preprint PDF and a reproducibility supplement including the frozen discovery corpus, annotation documentation, executable code, unit-level predictions, metric tables, machine-readable results, data and model cards, licences, and integrity manifests. This manuscript has not undergone peer review.