I've never really been that impressed by them, and maybe I'm in the minority about that, but I also will make the effort to use the proper Unicode for many characters (e.g. … ℃ ㎜ π ‽ ½), so I can understand the impulse. Curly quotes and a few other characters happen to be in a range of Windows-1252 encoding that makes them inherently unreliable after coming in contact with Windows.
I'd argue that these encoding issues are not rare, because of this weakness. It doesn't take long to find examples.