The issue I found with microformats when I was evaluating it was is that there was no way to automatically transform a microformatted file into a native data structure without first knowing the schema. Writing a microformat-to-JSON parser is hard because you have no way of knowing which classes are significant and which are just there for styling.
