Author recognition seems to climb as a smooth gradient with author exposure in LLM's. :)