HTML certainly needs better, proper, fragment identifier referencing capability. Remember: unless there is already an id in the right place in the document, it is impossible.
Dretblog proposes some decent options. And it seems it would be a relatively simple project. There may be a tendency to immediate concern for robustness: surely a simple easy scheme would break too easily. But think of basic URLs – they have no robustness, technically, at all. Totally disregarding robustness is well founded, in fact it is right. Not trying too hard makes the web work.
A tangential benefit is a perhaps better way than RDFa to bridge the GGG with the WWW, ie. RDF to HTML pages.
There are two obvious basic approaches:
But another way would be: adjust an intermediary to join them together better: improved HTML URL fragment identifiers. Then external, separate, RDF could address details of web pages. This seems much the best engineeringly, because it is so minimal.
Whereas, RDFa is rather dubious. Its HTML base makes it familiar and acceptable. But it still can't make statements about web-page details without changing the page itself. It is antithetical to RDF's essentially ‘addititve’, always open, graph principle. RDFa is also much more complex than better fragment identifiers: it adds more than it reuses, designwise (is that a tendency of W3C?).
There is not really much of a problem having all three tactics available, and maybe that is best overall. But RDF + Fresnel, with better HTML fragments, would be most powerful.
References: