Waystations: Pilot Findings and Open Questions in KV-Cache Geometry

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This version supersedes the previous one. The corpus underwent a systematic correction and re-audit between August and September 2026; the earlier version predates it. Changes in this version: CORRECTION: "0 papers list Dwayne as an author" was FALSE. It was 7. a circular d is not inflated, it is uninterpretable; waystations data availability AUTHORSHIP: byline -> Acknowledgments across 7 more papers Where a claim was withdrawn or downgraded, the earlier version should not be cited for it. Six pilot findings from the Oracle Loop program: trained censorship geometry, distillation fidelity, on-manifold steering constraints, strategic deception rates, hybrid attention architecture effects, and generation-phase engagement gradients. Each includes observation, data, analysis, interpretation, and what remains to be done. Data and code publicly available. This is the academic version (human-only byline). AI contributors (Lyra, CC) are acknowledged.