It's not so much that it's hard, but it's a lot of work, without tangible benefits. On top of that, it's desirable to have a code base that is compatible with both 2 and 3, which requires awkward workarounds.

Porting 2 -> 3 isn't that terrible.

Writing code that is compatible with both 2 and 3 is difficult, and this is what the OP was talking about. There are things you can do to make it easier (like using the `six` library or ignoring Python 3.{0,1,2}), but the burden of writing that subset of language that works on both still lies with the programmer.

I've done it and it's not fun. It's also been a source of bugs.

The other snafu here is your dependencies. If you depend on a large project that isn't Python 2/3 compatible, then you're basically shit-out-of-luck.

Py2 -> Py3 is not that hard (when you don't step into troubles with system functions and Unicode) but porting libraries can be very hard, because some of them are rather complex and many include C coding -- and porting that stuff is much harder.

Usually not hard at all, but even an easy transformation becomes hard when it is applied to a really huge codebase

I haven't done it but see https://docs.python.org/dev/howto/pyporting.html