Microrevolution in web semantics: formatting markup in a meaningful way

I've always thought that microformats are intriguing. They're not an official specification; they add nothing to the existing web standards stack that already isn't there; and they seem to be one of those rare "bottom-up" innovations that become a part of our toolbox solely through an organically increasing user base. They were the topic of a talk by Karsten Januszewski from Microsoft research.

Microformats solve a simple problems really well, and that is perhaps the secret to their increasing popularity: they add meaning to web pages through conventions for XHTML attribute elements (notably the "class" element). This meaning can then be extracted from the markup by software and "sliced and diced" for further effect.

Microformats are both "extensible" and "not". Extensible because it is relatively easy to propose a new microformat (the microformats wiki hosts both the list of ratified microformats and current proposals). Not extensible because once a new format is accepted, one cannot use it to define one's own specialized or fine-grained version of it. In other words, "subclassing" a microformat is impossible.

It's interesting how the microformat movement is going to deal with its challenges. According to Karsten, these are:

• Meaningless microformats (isn't that easy? Don't ratify them).
• Browser support (this one has never been easy, I guess).
• Search engine support (in spite of half a billion microformatted pages, SEs haven't been quick in processing them. I find this a bit paradoxical. Are microformats too "primitive" for SEs? Sometimes, a thing can be good even if doesn't require intricate algorithmics).
• Tooling issues (before long, some open source IDE will start supporting microformat development. Who's first?)

All in all, a nice talk about about an interesting topic. Microformats are a part of the next web. To what extent will become more clear in the coming years. Much lies in our hands as developers. Microformat development, anyone?