Namespaces, Microformats, RDFa, HTML, XHTML

We were recently at Yahoo’s Hack Day and one of the presentations I attended was by the SearchMonkey guys.  What piqued my interest here was that the teaser paragraph talked about augmenting search results with the semantic web — e.g. Microformats, XSLT, RDFa, et al. Hmmm, sounds applicable…

After it was finished I spoke with Evan Goer (who co-presented the session along with Paul Tarjan) about what we were doing around this licensing and attribution microformat.  Will SearchMonkey support this microformat? Certainly, just as soon as it’s accepted into microformats.org and achieves a critical mass of popularity, but if it’s RDFa it’s supported already.

Free Lunch

And herein lies a compelling case for RDFa.  If another service implements an RDFa parser anyone publishing in RDFa is immediately visible to that service.  This is the kind of interoperability that microformats lack.  Microformats are little islands of ideas united into a federation by their consistent approach to markup.  But their consistencies don’t yield automatic discovery of unknown microformats.  Each one needs to be implemented and supported independently of the rest.  Put another way, microformats are less of a standard than a bunch of mini-standards.

So we can yell from hilltops about a given microformat all we like but we’re one voice among many hilltops; the Internet is a loud place.  The RDFa standard allows us to automatically describe a format by putting it into a namespace and immediately join the party.  Yay RDFa!

The Catch

There had to be one, didn’t there?  A format aimed at facilitating media reuse on the Web has to travel well.  The Web is still by-and-large HTML.  RDFa is XML.  Doh!  What does that mean?  It means that putting RDFa tags into HTML pages will “break” them.  I quote “break” because it won’t break them in the sense that matters to most people.  Everything will look fine, all the major browsers will parse and display the content without complaint, but technically we would be violating the HTML specification and it will no longer validate correctly.  No namespaces in HTML.  Is this really a problem or am I just being pedantic?  I have to admit, I’m a little uneasy about using a standard that will break people’s validation.  XHTML is fine with RDFa of course and we should all be using it but we’re not there yet.

Another aspect of travelling well however is allowing the automatic interoperability that RDFa provides.  Ben Adida (co-editor and task-force chair for RDFa) has called it “Bridging the clickable and data webs”, and this concept is really at the heart of media reuse.  We need to talk to each other people!  And as we generally can’t agree on what to say, let’s agree on how to say it.

Survivor

Will RDFa’s incompatability with HTML sink it?  Actually, I think it will be the other way around.  More and more the Web is about connecting services and sharing data.  Providing namespaces to describe the many, many disparate data models is critical to enabling that.  OAuth, OpenSocial, Yahoo’s various APIs, it’s all moving that way and we all need help understanding what we’re talking about.  Perhaps I’m wrong with RDFa and HTML specifically but in the larger picture extensible, automatically discoverable (dare I say viral?) concepts will outlive brittle and inflexible ones.  It’s evolution folks and right now my money’s on RDFa.

Microformats and RDFa

While researching Microformats and RDFa I came across a very consice post by Mike Birbeck addressing this issue directly.  Given he’s a co-designer of RDFa it’s unsurprising that he comes down on his home team’s side but I think his points are salient.  Evan Prodromou has weighed in on the microformats vs. RDFa issue too.  He feels that the two formats are quite close and could be bridged with little effort.  What do you think?  If Ben Adida and Tantek Çelik got in a room together do you think they would emerge with a unified format? I’m not so sure.  I agree that they look very similar on the surface but the philosophies behind them are very different.  Mind you, Evan’s post is a little dated now (May 29, 2006) and RDFa has made a lot of progress in the meantime, I wonder if he still feels the same way.

I should mention that I found both of those links on Mike Linksvayer’s “Microformats are Worse” post where by “worse” he means “better”, or at least “simpler”.  Here’s a little worse is better primer.  The simplicity of microformats is compelling but it comes at a cost of centralized control and limited extensibility.  Sure RDFa is a little more complicated but the range of application becomes so much larger as a result.  I think the increased complexity is worth it.

I’m going to see what some of the ideas we’ve been discussing here would look like in an RDFa format.  Watch for a post about that soon.

[Edit]
Here’s another article by Manu Sporny talking about the limitations of Microformats due to their lack of namespaces.  He feels that RDFa is the inevitable successor to Microformats, and I’m starting to agree.