As a thought experiment, we might consider that any function that is too complicated to explain to ChatGPT and have it produce a working result might need to be refactored for complexity. Obviously not in every case, and our own ability to translate the requirements into a useful prompt must be considered, but I think it’s worth consideration.