Chipp, if I remember correctly from the pre-migration beta, there's a whitelist of allowed HTML tags and disallowed tags are stripped out or escaped.
So when parsing the content of the message, it would probably be fixable if the system matched the entire tag at that stage before parsing it as HTML, as opposed to matching on the angle brackets.
At the moment, if you write >_< it'll treat the angle brackets as HTML tags, but it's not valid HTML so it's not displayed properly. Before parsing it, the system should match for complete tags so that "<" is escaped to < and "<div " , "<p " etc (including optional spaces to allow for attributes) are parsed as html.
In the meantime, to get >_< and other faces, type > and < respectively.
>_< <- test
hmmmmmm
