RDF is tricky

I find RDF tricky.

I love the concept of linked data and the Semantic Web. It’s all lovely and simple and clean. And from that simplicity you can implement insanely powerful things.

But… RDF just always seems to make my head hurt.

I’m not quite sure why.

Maybe it just fosters yak shaving more than it should.

Maybe it fools you into thinking that you can describe the world really precisely and unambiguously if you just try hard enough (you really can’t).

Maybe it’s just RDF-XML that gets me down. I’d get all hyped up on the lovely simplicity of RDF graphs to describe things, then find its common form really non-intuitive and frustrating.

I think my main problem is that I want all pages on the web to have RDF metadata embedded in them by default and nobody should ever have to think twice about it.

I want to run SPARQL queries on the web itself. All of it.

Search engines should just be SPARQL end points with triple-caches.

But the problem is the way RDF is currently described and implemented takes a lot of effort to grok. You need to understand the underlying graph structure concept, various encodings, HTTP requests, headers, error codes, redirects and whatever else.

To become ubiquitous, it needs to be as easy to implement as, say, microformats. Or even HTML. It needs to just be an ingrained, routine part of creating web sites.

But even then, I’ve seen (and committed) untold HTML atrocities with the best of intentions. It’s really hard to get even these basic things right.

I’m sure I’ll come back to RDF at some point. But maybe by then it’ll be a moot point, the web having been replaced by a zillion flashy-but-opaque javascript-based web apps.