Apparently people on Bluesky are writing little self-care books for language models and the language models are ignoring them because they're ridiculous