This document discusses challenges with building HTTP libraries in Python due to differences in how libraries handle I/O operations. Specifically, libraries either integrate I/O directly or use concurrency primitives, limiting code reuse. The author argues this wasted effort could be avoided by separating I/O handling from other concerns. As an example, the hyper-h2 library builds a toolbox around a common HTTP/2 implementation without integrating I/O. In the conclusion, the author asks for help and invites questions.