Aggregate verification stats across all provocations in a reflection log — compute_provocation_provenance

Mode 1's analog of compute_quote_provenance_stats. Walks reflection_log$provocations, extracts each provocation's $provenance field (a QuoteProvenance object built and verified by the per-category function – see R/provocateur.R::.citation_to_provocation), and feeds them through quote_provenance_summary.

Usage

compute_provocation_provenance_stats(reflection_log)

Arguments

reflection_log: A ResearcherReflectionLog, or NULL.

Value

The list returned by quote_provenance_summary.

Details

Provocations from observational categories (absent_voice, parts of assumption_surfacing) carry NULL provenance because the AI is reasoning ABOUT the data rather than quoting it; those are excluded from the verification stats (the Tier-0 dashboard's domain is verbatim claims).