> I am happy to bring this to Python-Dev if the people opposed to this change are not on datetime-sig.

I think you should.  As much as I am on your side here, I don't want datetime classes to become special snowflakes in the Python type system.  So far all arguments in favor of retaining the type of datetime subclasses in arithmetic operations apply equally to the other Python types.

BTW, Tim's original datetime.py prototype used type(self) in arithmetic operations, but it was later changed in order to match C.