Comment by bjudson

The argument against this approach is that it's not always humans accessing web pages. And even if it is humans, sometimes they need computer assistance to interpret the content. http://en.wikipedia.org/wiki/Semantic_HTML