Wikis (or at least the ones I've seen) describe how the document looks, eg __foo__ might mean underline foo.

I want my markup to describe what the document says and later let my parser/processor implement the look. Then I get a consistent look, I can change it whenever I want and I can implement cool content related features, here's a pseudo code example.

My favorite book is BOOK<Perl Programming>.  I also like BOOK<The Hobbit>.

If I want to, I could render all book titles in bold, then I could later decide I want to render them in italics.I could put them in a list of all books mentioned in my document, I could make each book title include a link to Amazon... etc.. etc...

You can do this with docbook, but it's a huge pain to actually write in. <document type=book link=true>The Hobbit</document> Note, that's not real docbook, but real docbook is that cumbersome.