So my drive back from Cleveland yesterday was a great sunny-window-down-jamming-Sly-very-loudly-getting-a-sunburn-on-my-left-forearm kind of drive which gave me plenty of time to think about stuff. Alas, I didn’t have any strategy for taking notes or anything, but I got started writing some of this out last night. I ended up lapping myself with another post, so excuse me if there seems to be weird time-shifting going on.
One thing I was thinking about a lot is social bookmarking.
If you’re not familiar with social bookmarking, the “google” of the field is Delicious . Delicious wasn’t the first of its kind, but it stripped the idea down to the simple core, and foregrounded the idea that the “tagging” is what makes the bookmark really valuable. Since then, a multiplicity of variant websites have spawned — most major media websites have “share this” panels to make it easy for you to add any article to bookmark-y sites. In fact, the sheer number of services represented in some of these panels suggests that this is a use case for which future browsers ought to have “baked-in” support, as they mostly do for RSS now. Note that the excellent Mac feedreader NetNewsWire has had “Post to Delicious” support for a couple of years, at least. Anyway, suffice to say that social bookmarking sites are one of the loci of all this tagging stuff, and given that they are abstract enough to let you tag any page on the internet, they fit right in with the things I’ve been writing about recently (including falling prey to the aboutness-of-a-page dilemma.)
More specifically, I was thinking about a rather obscure corner area of social bookmarking known as machine tagging. Machine tagging is a grass-roots semantic web initiative which is similar in spirit to Microformats. Earlier this year, MachineTags.org was established to document and develop conventions for this variant model of tagging, which I didn’t realize until I got home and got online.
The prototypical machine tags represent geographic points. In 2004, Mikel Maron announced an application which mapped geotagged Delicious urls and just this January, Flickr started ramping up machine tag support.
Therefore, it’s no surprise that the first formalized machine tag “namespace” was geo.
Between the Flickr ramp-up post and the still-warm-from-the-oven draft spec for ISBN machine tags, there’s certainly a lot to cause shudders in conservative old school software engineering types. I am definitely not one of those, but as someone whose job is based largely on factoring the surprises out of large web applications, I am sniffing for code smells like someone who suspects the pilot light of his stove has gone out.
After all, look at this list of properties supported by the book machine tag namespace:
- isbn
- title
- subtitle
- series
- author
- editor
- illustrator
- serieseditor
- translator
- introauthor
- edition
- publisher
It is just about impossible for me to get over the idea that specifying any of these after the isbn is any kind of a good idea. I totally respect the successes of the internet-as-hive-mind, but I think most of those are in areas that lend themselves to more fluidity. These are machine tags — so why not design an ecosystem for the software that deals in them and end-run the problems of incomplete and inaccurate tags for values which are invariant properties of “the thing” once you identify it? Assuming that we’re leaving the hard core epistemology to the Academy, what are the building blocks? (If I start titling posts like “the thingness of a thing,” send the guys in the white coats!)
