Some explanations can, will, and should be written by code authors alone, or by those authors in partnership with LLMs. Others can, will, and should be conjured dynamically by code readers who ask LLMs for explanations on the fly.