Thanks for the interesting thread! One thing in life is constant - RDF continues to be compared unfavourably to the fashionable file format of the day (XML, and now JSON, occasionally YAML).

I think the RDF world could also benefit from a similar "where are we now, and how might we become more useful" exercise to this one, but we can also indulge in a little "told you so"-ism in that XML data that was in the RDF/XML pattern, is via the underlying data model, totally compatible with JSON data that's in the RDF/JSON-LD pattern. You can parse out the same graph, put it in a common RDF/SPARQL data store, etc etc. So instead of hearing "hey why pay the RDF tax with RDF/XML, you could use native XML and its APIs and standards", we now hear "hey, why pay the RDF tax with JSON-LD, you could use native JSON tooling...". There is certainly a cost and some friction to having this kind of two-tier view of data representation, but there's something to is as well in terms of having an abstraction of what the document/file was communicating, for cases where the payload is simple-ish factual data rather than a tagged document.

Dan