Python 3 primer, Part 1: What's new (developerWorks)

It is true that, in order to get bytes out of a string, you need to specify an encoding, but bytes has barely anything to do with strings or characters now. I assume that he was looking at the section on bytes literals, where you can write b'\xc3\x9feta' for his example (ASCII characters for byte values in the 0-127 range), but can't write b'Ùeta' (assuming a UTF-8 source file, that would be putting the desired byte sequence between in quotes in the file on disk, if I did UTF-8 in my head correctly) or b'ɟeta' (using Unicode characters for byte values in the 128-255 range).

They'd probably have made it easier to understand if they'd made bytes literals only permit specifying the contents numerically, like: (not actually allowed): b'xc3 x9f x65 x74 x61' that is, only ignore whitespace, treat everything else except the matching quote like it had a \ in front of it, and disallow non-numeric escapes). It would make it slightly harder to type in binary data with lots of bytes which are (arbitrarily) in the ASCII range, like the GIF file magic number, but I doubt that's very important.