If you accept this argument, the next logical step is figuring out what features are being used and how. Standard method is just guessing, but there's: 1. What

If you accept this argument, the next logical step is figuring out what features are being used and how. Standard method is just guessing, but there's: 1. What rustc does with crater (linked from article), where you can actually test real code. On a smaller scale, the author of the article points out cryptography package for Python, which runs tests of its most popular upstream dependencies. This latter is probably feasible for many libraries. 2. Analytics, though this requires work to make it private and anonymous, and for libraries maybe requires system-level infrastructure. (Again, this was suggested by Alex.) 3. Beta testing, surveys, and other forms of user feedback.