So here I think we differ fundamentally in how we like to read code. I much prefer being able to quickly figure out things of interest about a function by glancing at its signature, rather than look at documentation, or worse, having to read the implementation of the function and the functions it calls (and so on, recursively).
For example, I much prefer a signature like
def f(a: int) -> str:
over
def f(a):
because it allows me to see, without reading the implementation of the function (or, if it exists, and I'm willing to bet on its reliability, the documentation), that it takes an integer, and gives me a string. And yes, this requires that I write viral type annotations on all my functions when I write them, but for me the bottleneck at my job is not writing the code, it's reading it. So that's a small upfront cost I'm very much willing to pay.
> Do you think we should be annotating functions with `expensive` and/or `networking`? And also annotating all of their callers, recursively?
Yes, absolutely, and yes, absolutely. That's just being upfront and honest about an intrinsic property of those functions. A function calling a function that does network I/O by transitivity also does network I/O. I prefer code that's explicit over code that's implicit.
