Microformats turn 9 years old

For example, tagging articles with key words so that they can be sorted or selected would be an example of how this sort of stuff is done in practice.

Or you could use it to produce customizable web pages. You know that a user has to have a certain set of features in a template and that the same feature can't be implemented in two different ways. Not an issue. SPARQL, the semantic web's version of SQL, can give you the options and enforce the constraints.

Maybe you want better cross-referencing in a wiki. Specify in SPARQL how to identify relationships of interest.

I like RDFa and tools like Protege for defining content association because it has oomph. I dislike microformats because it gets really, really ugly when you want to use server-side if/then/else logic. However, they all have value.

And none of it is in marketing. Marketing hates the Semantic Web because it's transparent to users (so there's no metric you can use to prove its value) and allows any external user to add their own value to the site (which puts marketing's jobs at risk).

For example, say that the search engine would otherwise have included a website for my search "Linux flower arranging", but since I didn't include "Britany" it decides not to show the site, instead it registers that the site is owed another hit. Sometime later when another web user searches for Britany, but its algorithm would otherwise not consider the site to be quite relevant enough to be included, it then shows them the site and deducts one from the number of hits the site is owed.

I get it that a certain type of person has come to see the web as just a bunch of advertisers passing user eyeballs around for money, but a lot of us are actually trying to get stuff done here. Microformats are one of several ways to _sidestep_ your natural language processing for the case where we want machines to read our content, which in fact is a lot of the time. Personally my stuff uses RDFa, but microformats are a perfectly sensible choice too.

This isn't about writing "My web page is about Britney Spears and I'm 14 and I like ponies and I want to be a hairdresser" so much as "This product has UPC 5060095860304" and "This facility is at lat 3.24140 lon 1.02880" indeed for most of us it's not about search engines at all, the article mentions them only briefly and you seem to have seized on that.

If you'd followed the links you would see that, of course, Google does read microformats. It doesn't trust a microformat tag that says the page is about Coca-Cola any more than it would trust the human language assertion of same, but it does have the advantage of being unambiguous, which you really seem to have underated. Tools processing microformats can be simpler AND more reliable than doing NLP everywhere.

Not everything is about advertising or the open, unbounded, untrusted web. In fact, although Google would much prefer nobody thought about it too hard, it looks increasingly as though actually very little valuable economic activity is generated by the advertisements. They're basically a distraction, albeit one that is currently making some people very rich. My supermarket, for example, is paying for highly targeted and superficially effective grocery advertisements. "Effective" because the people the adverts target are buying those groceries - a few pennies on advertising seems to bring in a ~£50 purchase which is a good return, "Superficially" because those are the people who already buy those groceries every week from that retailer, so the advert made no difference to actual sales.

From a brief read, microformat's appears to be yet another attempt at serialising data in a human readable form that is still easily parsed by a machine. Granted it's a bit unusual in that it doesn't define a format, but rather re-uses existing formats like HTML, JSON, and XML. The core of it appears to be defining naming conventions, and how they should be applied to those formats.

If I was to evaluate it, it would be on whether the naming convention is worth using as opposed to just rolling my own. That could depend on all sorts of things I guess - like availability of compiling and parsing libraries, how many people are likely to want to parse the data I published, how frequently it changes, and who would be using. The one thing it is unlikely to depend on is what a search engine thought of the resulting HTML. And the one thing it absolutely would not depend on is whether it might be adopted by soe practitioners.