During my keynote at XML Prague (the video might make more sense than the slides on their own; there are notes on the slides but Slideshare doesn’t do well with Keynote), I talked about how the advantages of using chimeras created from two formats with different underlying models are seldom outweighed by the disadvantages. RDF/XML gets knocked so frequently it’s not even much fun to do it any more, but I’ve applied the same arguments to JSON-LD in the past. My argument was that RDF, XML, JSON and HTML should each be used individually for their strengths rather than trying to find a middle ground that rarely satisfies anyone.
Leigh Dodds’ post on principled use of RDF/XML makes the point that RDF/XML can be useful when it is used in a regular, principled way. And in fact, I am using RDF/XML extensively in my work on Expert Participation for legislation.gov.uk, though slightly differently from how Leigh describes. What I want to explore in this post is when and how it makes sense to use RDF/XML and how that might translate into usage of JSON versions of RDF. The key point I want to make is that RDF chimera are roads, not destinations, and when you’re choosing a road you have to think about the destination you’re aiming for.