Scripting and Tooling becomes more robust wrt Unicode. You write a short script which works on some ASCII data. Later that ASCII suddenly becomes utf8. With Python 2 you could easily run into weird issues, when there is a unicode object instead of the expected str. With Python 3 decode is more explicit and robust.
Scripting and Tooling are the major uses of Python for me, so I like Python 3.
I always work with utf-8 strings that get passed around from native C code to python. So far I have never had an issue. The argument for me for unicode support seems weak. Is there anything else beyond string encode/decode/manipulation? You mentioned tooling... like what?
Scripting and Tooling are the major uses of Python for me, so I like Python 3.