There’s been a lot of buzz recently about “leveraging the hive mind” with attention.xml. Basically, attention.xml is an XML spec for publishing your reading habits (how often you read a feed, how much time you spent reading a post, or when you last read a feed). It’s intended as an open foundation for a “cloud of reputational presence and authority [that] can be mined by each group of constituents” as Steve Gillmor says in this article. I think it’s a very cool idea — but having a cloud, although that’s essential, is only the first part to making it rain, as Gillmor readily admits.
The idea is that this pool of information about millions of users’ reading habits can be fed into various embodiments of smarts (like recommendation engines, either built into feed readers or separate services) that munge the attention metadata and generate useful recommendations about what you should read next. Who wouldn’t love that? There’s so much stuff out there to read and the current selection mechanisms are ridiculously crude and inflexible (some feeds are always worth reading, some only occasionally, and keyword search feeds have too much recall and therefore don’t solve the filtering problem well).
So the recommendation engines are where the rain starts to fall from the cloud. I suspect that we’ll see a lot of collaborative filtering and a lot of tag scraping in the first batch of attention mining engines. And then, in another phase, I hope we see engines that move beyond metadata and add content itself into the analysis. When a recommendation engine knows not only that people read blog X more often than blog Y, or that post k from blog Z was significantly more popular than any post from blog Z before or since; when you also know what post Z is about and what blogs X and Y are about and can organize this about-ness and learn from it — without categorizing the world — then you’ve got yourself a recommendation engine. I’m looking foward to that day. Let it rain.
Finally, for those of you who have moved beyond reading altogether, Chris Pirillo talked with Steve Gillmor about attention.xml in this podcast recently.
One interesting scenario that Gillmor held out is this: Because each user’s attention is a form of capital (hey, they’re your habits, why should you share them for nothing), sharing your attention.xml could become a form of payment for access to some content. Agreed, that should be worth something to publishers (as long as your feed reader doesn’t already provide your attention profile to publishes, without your knowledge. Ouch, there’s a though — where’s the source to this feed reader?).