In any case it doesn't make much sense to change just one separate docstring. If you want to avoid misleading and support consistent wording, you should examine all occurrences of the word "string" in the documentation, docstrings, error messages and comments -- does it mean Unicode string, bytes-like object that supports the buffer protocol (including memoryview), bytes-like object that supports str-like interface (including bytes, bytearray, but excluding memoryview), either Unicode or bytes string? I tried to do this but abandoned the work on half-way. This is too large work.

> BTW, what do you think about the second TypeError in ``ord``. Shouldn't it be ValueError?

It could be ValueError. But for compatibility it should stay TypeError. This is not wrong if we consider strings of size 1 as separate type.