I don't think the intent of marking up things like height and weight is dubious - isn't that the whole point of the semantic web? That you use semantic markup to explicitly define your information content?
Sure, the intent is good. However, would a hypothetical <weight unit=grams>5</weight> be any better than "5gm"? When I see a number followed by the letters 'gm' I know it's a weight in the units of grams. Google knows that too. Making it rely on the markup alone would actually make google dumber and making it understand both would make it more complex. It's better as 5gm.
The main advantage is to make it easier to extract using an xml-like api. But at what cost? "5gm" is way easier for people to type so it will always dominate in real text.
In my mind, markup is useful for adding context that is hard to extract from the text alone: things like groupings, context, delimitating chapters or articles, etc… I wouldn't use it to declare something to be a weight. You use it to add things that are unspoken otherwise. In the <meter> example, it actually lets you convey previously unknown things such as the min/max range.
Maybe if the whole world cared a lot about the differences between different kinds of weights and it was very hard to get from the context of the text, like <catweight> vs. <dogweight> or something, it would be worth marking up. Markup is a form of metadata.
Yeah, but maybe one should use a proper RDFa attribute, like dbpedia-owl:Person/height and dbpedia-owl:Person/weight. HTML5 elements simply aren't detailed enough.
It's in the RDFa Primer [1]; I mean specifically the @property, @about and @typeof attributes, which are used to augment existing HTML elements with RDF semantics.
