I noticed that HTML is still not considered machine readable…
If its properly formatted, its machine readable, and I think not letting it be an option is not the right approach, although I’m not 100% on where to go from there.
Perhaps making the submitter ensure the markup is valid and microformats/schema validate as well?
Essentially, I think in a few cases there will be machine readable data that will be marked as not because the data is in HTML.
Also think this plays into the worldwide phenomenon of not caring about markup, which hurts no one more than the end user(s), and is a plague on the web.