As I see it, WYSIWYG works great if you create a non-interactive thing, like a document, image or video. As soon as you want to present the same information in several formats, or show headlines of articles in several places you need to abstract those things. None of this is impossible with WYSIWYG, but it is usually much more work than just having some kind of plain-text at the core.
On the other hand, a well-executed WYSIWYG editor with a well-defined abstract document model would be a great thing. The abstract document model is really essential (and probably plain-text based). The minute you let people WYSIWYG edit, they start doing formatting oriented things such as add spacing using actual spaces. It looks the same on their screen, but might break other formats of the same information. If you tell them to stop doing "that kind of thing", then very soon they are essentially doing plain text-editing with a WYSIWYG tool.