I’m at risk of getting a reputation for pissing on Web 2.0, but here goes. The problem with “microformats,” which Technorati is pushing pretty hard, is that they seem to be no more than poorly implemented metadata standards.
Take the rel=tag specification, for instance. This is a snippet of code you can add to a hyperlink that tells sites, like Technorati, that the page where the link appears should be categorized a certain way.
In other words, you’re tagging the current page (where the link appears) by linking it to a tag page (which presumably collects many similarly-tagged pages). The meaning of the tag comes from the linked page, and is applied to the linking page.
Confused yet? It gets worse.
One of the stated goals of rel-tag is that its tags should be visible. Unlike the META tags we’ve all been putting on our web pages for years, tags are easily spotted and read by humans. Except the specification undercuts even that, telling us: “The last path component of the URL is the text of the tag.” So this link:
is a “tech” tag, even though it looks like a fish. And this link:
is a tag for “haiku,” not “poetry.” And this:
is not a tag for anything, it’s just a link. So much for visibility.
And encoding content as part of a linked page’s URL? How much more inflexible can you get? This is supposed to be an improvement over META tags?
Here’s an example of the problem. The URL for my weblog’s “reviews” category is this:
https://dylan.tweney.com/category/main/
I can’t change that (old) directory structure unless I want to break all the inbound permalinks to it. But thanks to Technorati’s brain-dead tagging scheme, I can’t categorize it under the tag “review” either — instead, it should show up tagged as “main,” which is completely meaningless. And if I want to add some useful additional tags, like “technology” or “Tweney,” forget about it.
The result: I use Technorati tags when I want to make something show up on Technorati. But that’s it.
A decent tagging scheme would allow you to take any arbitrary text, with or without a hyperlink, and add one or more tags to it. It also wouldn’t commit you to using a single company to sort out all your tags. I’m sorry, but rel-tag just doesn’t cut it. The best I can say for this half-baked standard is that Google hasn’t supported it yet.
You want to make metadata visible? Write a browser plugin that lets you view META tags.
So, back to microformats: What makes them different from metadata? Or for that matter the much-maligned but nevertheless-growing semantic web? It seems clear to me that they’re the same thing–except microformats are less well thought-out. If a field was ever crying out in desperate need of a few good librarians, this is surely it.
[tags]metadata, microformat, microformats, technorati[/tags]
Agree with you on most points. Metadata has to be clear and unambigous, otherwise it leads to more confusion. As far plugin that parses pages, this is exactly what we developed at adaptiveblue. I would be curious to know what you think about our work.
http://www.adaptiveblue.com
Alex
Frankly, yes, microformats are bad metadata. But that’s ok.
I think you’ve misunderstood the goals that have driven the microformats design. The idea is to lightly codify and standardize existing practices rather than invent end-all-be-all metadata solutions. Simplicity is more important that comprehensiveness and purity. If you want comprehensive, extensible metadata, RDF already exists. Microformats, by virtue of being simple to implement without big changes to existing tools (eg, rel-tag support can be added to most sites with a minor template tweak) stand a chance of actually being used by the unwashed masses who aren’t librarians or metadata experts. The hope is that widespread adoption of “good enough” metadata will allow us to collectively achieve some of the aims of the semantic web much faster than sparse adoption of “perfect” metadata. With any luck, it might even bootstrap something a little more rigorous into existence.
Anders– I can appreciate your point that microformats are supposed to be quick and dirty. It’s fine to set a low bar. But rel-tag doesn’t even fulfill the modest design goals it sets for itself! Other than facilitating tagging on Technorati (and that not very well), rel-tag is not even close to “good enough.”
For stimulating discussion on metadata, though, it might be useful. I hold out hope that we could have a metadata standard that’s both easy to use, easy to understand, and lightweight — without being brain-dead.
Alex– looks like an interesting plugin… You’re competing against many many other schemes/services for saving & bookmarking info, though. You’ll need luck as well as skill to get any visibility in the world of del.icio.us, furl, digg, etc.
Dylan,
You are right, it is hard to get visibility, but we are working on it 🙂
There is a big difference and value add in what we do compare to collaborative bookmarking sites. We preserve semantics of the objects, enabling people to recognize what they store as everyday things as oppose to pile of links. So stay tuned and send us feedback.
In terms of microformats, we all know its a tough problem, and eventhough I am putting my soul into blueorganizer building top-down solution, i know that ultimate solution needs to be bottom up.
An idea different from microformats is to store information XML and just to embed the the ids of the objects into the pages. So when user clicks on the page, the browser can do: Find the ID of the current object and then ask the page provider for this object. In my mind this is cleaner solution that microformats.
Alex