Metacrap and Interoperability
The O'Reilly Emerging Technology Conference is this week in the Bay Area. I've attended a couple other O'Reilly conferences and really like them. They push at the edges of what is truly emerging technology in a way that other conferences don't. They are also profoundly geeky, so you dig down into lots of minutia which is often (but not always) both entertaining and very revealing.
The conference just got started today with a set of tutorials. I attended a tutorial on the Web Services Stack which maybe took the minutia thing a little bit too far. But in the course of presenting the details of implementing a web service, Sam Ruby of IBM's Emerging Technologies Group, pointed to a great piece by Cory Doctorow.
Cory's essay is officially titled Metacrap: Putting the torch to seven straw-men of the meta-utopia. It is a basic indictment of the idea that by empowering metadata and better metadata standards we will solve all interoperability woes. According to Cory, the utopian view of metadata falls apart because: 1) people lie; 2) people are lazy; 3) people are stupid; 4) mission: impossible -- know thyself [people are lousy observers of their own behavior]; 5) schemas aren't neutral; 6) metrics influence results; and 7) there's more than one way to describe something. I completely agree with Cory; metadata in and of itself won't solve the problems of interoperability. Which is why the combination of SOAP, .Net, WSDL, WS-Inspections, UDDI, etc. are so useful. They allow metadata to be imperfect yet still create interoperation between disparate systems. This stuff is powerful and getting more so every day.
Coincidentally I'm working on a semi-refutation of Cory's piece. Only semi- because I too agree with most of his arguments, just not his conclusions. The biggest fault is the implication that every single piece of metadata has to be hand-crafted - definitely not. Let the machine do the work whenever possible. Also the either/or angle on explicit (title="something", Semantic Web) and implicit (text/link analysis, Google) metadata. Personally I think we'll really start to see the benefit of both aproaches when they are used *together*.
Have a good conference!
You're dead-on right, and so is Cory. I hate getting cynical and turning into an old fart, but I've learned to never underestimate (or overestimate) human nature. :-)
We technologists and early adopters tend to see the positive nature of our inventions, and overlook the issues of abuse, at least initially.
I wrote commentary on a new self-categorization metadata format proposal called Easy News Topics the other day. I fear that this new type of self-created metadata will soon become awash with spammers and other bad actors, and will become as diluted as the META keyword tag of old.
My commentary is here: http://www.sifry.com/alerts/archives/000282.html