E064 Pre-Registration: AI-Likelihood Score vs Indexation Rate on a Single Small Corpus (Open Instrument, Confound-Controlled, Directional)

Pre-registration, filed before data analysis. Tests whether an open, locally-computed AI-likelihood signal (GPT-2 perplexity and burstiness) predicts Google indexation on thegeolab.net, controlling for content quality and a content-effort proxy. Deviates from the Originality.ai instrument in the original practitioner claim for cost, reproducibility, and vendor-independence; tests the generalised claim, not Originality.ai specifically. Bivariate N=47, confound-controlled N=45. Directional only: the corpus cannot support a proven causal finding and the result is barred from becoming an audit-rule gate without larger-corpus replication. Raw-data md5 hashes recorded for post-publication verification.