Pipe

The "Content-Type" header has been: "text/html; charset=utf-8" (as well as the equivalent meta tag) since the first day.

As for posting comments, I'm currently being "overly safe" and only allowing keys that can be typed on a US keyboard (minus the ampersand). The idea is that I eventually loosen the rules a bit for most western languages and useful symbols (like euro, pound, and yen.)

The full set is just a huge potential source of abuse with non-printing characters, right-to-left switching, and CJK characters. Within minutes of Soylent News adding it, for example, people where posting pages of braille and other crap.