I would be curious to know what percentage of Stack Overflow questions are duplicates, as any question flagged as a duplicate is really a misguided search, as the site is intended to be used.<p>But those duplicates still get flooded with answers and votes and such before they get closed as duplicates. I farmed a lot of points on there myself by answering obvious questioned doomed to be closed that would attract lots of votes.<p>> Another question: where will LLMs get coding Q&A training data in the future?<p>I would venture that 90% of SO questions have answers in the documentation or in the code of the library. Discovery is just a tremendous pain. LLMs solve discovery.