Thoughts on branding, design, writing and life by Kevin Potts. Established 2003.

RDFa, Microformats, Standards: Big Questions

While the discussion of RDFa and microformats has been ongoing through the web development community for years, Google recently announced explicit support for both kinds of structured data. Is there room for both in the future, or do we need a true standard to rally around? I dunno. Maybe you do.

Forgive me if I nerd out for a few moments. I’m a big fan of metadata, analytics, semantics and data aggregation and display, and I’m totally fascinated by the discussion on the web regarding the rise of RDFa and how it conflicts with, competes against, and complements the microformats initiative.

I’ve been using microformats for years, and have standardized this adoption across client work, especially using rel attributes in anchor elements, (in particular, employing tags for site taxonomy) and the hCard standard for contact information. On the flipside, I’ve been following the Dublin Core initiative for longer, but have never found a practical application for applying the rich metadata language it offers, despite my hero Joe Clark futureproofing his entire book with the stuff.

With Google announcing support for forward-thinking data structures this week, it appears as if there might be a payoff for the additional development effort. Is it time to pay closer attention?

An Immediate Conflict?

The most obvious problem is the possible conflict of technology between microformats and RDFa. The former uses simple class and id attributes to describe data, which easily bolt onto contemporary HTML, even the aging 4.01 spec. RDFa employs more complicated XML structures, and requires a custom XHTML 1.1 doctype.

Evan Prodromou wrote a great article outlining the differences between RDFa and microformats, and I am not going to reiterate everything here. But my questions echo his.

  • How quickly will frontend developers adopt new metadata attributes that are more complex than the user-friendly language of microformats, and only work with stricter XHTML 1.1 doctypes? Will implementation be relegated to only the cutting edge, or the techno trainspotters? Does Google’s sponsorship help move the meter on adoption?
  • Can web page information be marked up in a hybrid of the semantic languages? Right now, there is no definitive answer.
  • Is there room for both in the future semantic landscape? The world saw a massive stutter-step in the spread of syndication because of competing RSS standards. Ben Adida doesn’t think there is an issue: “Microformats are useful for expressing a few, common, well-defined vocabularies. RDFa is useful for letting publishers mix and match any vocabularies they choose. Both are useful.”
  • Will developers actually spend time to write their own RDFa vocabularies, or simply re-use Dublin Core, FOAF, and others over and over and not bother with content that currently does not have established definitions? (If so, that defeats the entire purpose and everyone should just use microformats.) On a more pragmatic level, is there one core language that is superior to the others for similar schemas, or is it truly a choose your own adventure?

Yes, there are other big, big questions. These are just the ones that strike me as immediately pressing — as a huge advocate of web standards, I’m looking for a standard.

GOOG is not God

Google’s announcement is going to stir pots, especially in the SEO community. SEO experts have wishy washy consensus on whether semantic HTML influences ranking, especially in the use of our friend h1. (Even though our search engine du jour barely acknowledges semantic HTML, and certainly does not endorse W3C validation.)

We have all seen well-structured semantic websites fall short of competition that boasts a bajillion incoming links. This is by design. While Google plays lip service to structured data, does its use dictate a ranking advantage? Perhaps if all else were equal, structured data would offer a razor-thin ascendancy, but Google will never say that, and we plebeians have no way of knowing how it factors into their black box algorithm.

The Cart, the Horse, and the Order of Things

Finally, hardcore adoption of RDFa feels premature. With XHTML 2 many years away, and XML MIME types still not supported, it seems that the technology is almost stillborn. Who wants to use it when its power cannot be fully leveraged? I understand it’s compatible with XHTML 1.1, but the golden road toward an XML-driven web seems more difficult by the day.

On top of everything, HTML 5 seems to be gaining favoritism over XHTML 2. There is clearly confusion in the web development community, and some web illuminati are already building for the next version of HTML. If HTML 5 is the new standard that’s going to push the web forward, does this leaves RDFa out to dry, since it uses strict XML formatting? Are we all going to come full circle to HTML-friendly microformats? Or am I missing something totally obvious?

These are not rhetorical questions — I am not smart enough to even guess most of the answers.

, , , , ,

commentary + criticism


wrote the following on Friday May 15, 2009

HTML 5 is still a specification in development. Though it has been built on implementation reports and issues, so it is more likely to be compatible with today’s browsers.

For RDFa, there are efforts going on this week to write a profile of RDFa that would be possibly included in HTML 5 specification.

On the validation front, HTML 5 being a moving target it is hard to have a stable checking. There is an experimental instance for validation HTML 5 (coded by Henri Sivonen).