It's great to see all of the work going into microformats. While it's good stuff, I honestly find microformats to embody a process way more formal than what excites me personally.

Of course, this formality is justified in that many of the microformats are of quite complex design that reflect relatively elaborate aspirations (e.g., for information retrieval, data interchange, knowledge representation, etc.). And, actually, I'd expect this formality to only increase as microformats are more widely adopted.

Now, what excites me a bit more is kind-of going in the opposite direction—doing things in a way that's less reliant on such a formal process. (Don't get me wrong. I am not suggesting that being less formal is a better way to solve the problems that microformats are addressing—it just may be more exciting, depending on your sense of adventure.)

***

Nanoformats are HTML documents where a pattern of values in HTML id attributes are used to indicate HTML elements that contain data of interest.

Here is an example nanoformats document that encodes sets of data compatible with two hypothetical schemas, book-city and webpage-author. View the source, and note some of the id attributes and the input tags in the form.

Nanoformat documents may optionally include a number of hidden-type HTML input elements whose values are used to indicate: the schemas that describe the data constraints, the schema elements used in the document, aliases for schema element names (to avoid name collisions), and the number of sets of elements (in each schema) that are present in the document. (And, some other relevant stuff.)

So, there are actually a few rules for what I am calling nanoformats. These are a few MUST rules and a few more SHOULD and MAY rules.

I'll get into these rules more in a future post. And, I'll also talk about the different use cases for the rules, like how you need less rules if you are using a nanoformat for specific processing, but more if you want to account for different kinds of discovery and mixing of schemas.

Obviously, there is more to say about this, and it still might not be that interesting compared with microformats, depending on your point of view about such things.

***

So, a couple bits of background for those really interested.

The name nanoformat was coined by Lucas Gonze (in his post, Introducing: the nanoformat) in response to my Moment of Silence microformat amusement. I've decided that nanoformat is actually a nice word to describe the technique I've introduced here.

This technique is something that I've personally been using since around 1998, and I'm sure that variants on it have been used by many other people independently. (Let me know if you suspect some site or article to be an early source of this technique!)

Also, in 2004, I was meeting with Tantek Çelik and Kevin Marks, and we ended up talking about using HTML to encode structured data. In that conversation, I became aware that they both, for a long time already, had been thinking about and experimenting with what's now known as microformats.

Finally, Dan Connolly's XSLT for screen-scraping RDF out of real-world data (which led to GRDDL) shows how this idea of encoding data in HTML has been considered in terms of the Semantic Web.