I don't think it's safe to assume that everyone is working on geenfield projects, where the performance requirements of the project are known at the outset. Wo

I don't think it's safe to assume that everyone is working on geenfield projects, where the performance requirements of the project are known at the outset. Work that increases the upper bound of Python's performance is valuable because there is so much Python code already in existence, and it could be useful to make that code faster, due to shifting requirements for that code, or new knowledge about those requirements. I think similar reasons justify faster runtimes such as PyPy.