For example, such a similarity measure could show numerically just how much what you see while editing in TeXmacs is closer to LaTeX output than what you see while in editing in LyX.
For LaTeX output, maybe TeXmacs would have "90% WYSIWYG/LaTeX" while LyX would have "50% WYSIWYG/LaTeX".
For HTML output, maybe TeXmacs would have "80% WYSIWYG/HTML" while LyX would have "50% WYSIWYG/HTML".
Do you think this would be an improvement over using the term WYSIWYG?
Also, how might one define the similarity measure for this purpose?