I hate WYSIWYG editors (also because WYG may differ from what readers using different systems will get). I seriously can't even remember when the last time I used a MS Word-like word processor for something non-trivial. (I use LaTeX or code HTML by hand (with gedit's plugins is much less of a PITA than it sounds like) or stick to plaintext/MarkDown depending on what I need.)
If you write an article in Word, Writer, Scrivener, Google Docs, or another rich text editor, and then copy+paste that rich text into an online WYSIWYG editor like the one on Less Wrong or WordPress, the HTML generated by LW or WordPress is incredibly messy and does tons of weird stuff to your text.
Because of this, I've taken to composing all my posts in Markdown, which is plain text (like HTML) but easier to read, and can be easily converted to clean HTML.
Ideally, though, authors would be able to compose articles in whatever editor they want, and then paste their rich text into a simple web tool that strips all formatting from the HTML except the formatting they want to keep.
HTML Purifier, TIDY, and HTML Tidy aren't quite what we need. Word2CleanHTML, Word HTML Cleaner and WordOff, along with CKEditor's and TinyMCE's 'Paste from Word' features, kinda work, but not really: they still make mistakes pretty often when I try them.
What I was hoping to find was something like Word2CleanHTML but with three changes: