I think history is worth mining for future ideas for producitvity software – especially when we finish mining everything LLMs and RAG can do, we might go back to past experiments in information retrieval. We might know the history now that we're reading this thread... but who's to say that a developer in 2030 who's never read HN has?