Don't follow *me*, follow me/tags/interesting-to-you

This post assumes some knowledge of the technologies coming to define the social web, including WebFinger, FOAF and XFN. For an introduction to these technologies, check out my Primer on the Social Web and The Advent of Following.

The Problem

In The Advent of Following, I concluded with a problem: it has become all too easy to follow someone on the Web, but even more difficult to follow just the content you want to. Between FriendFeed and Buzz, following someone and everything they do online is pretty possible and the experience will only get easier and richer. More and more people will sign up and more and more services will be aggregated, either because they publish content in a (possibly de facto) standard format, like Atom, or by brute force (manual integration of some kind). However, already this is making it difficult to sift through the noise and find just the content I am interested in.

Recently, Robert Scoble has been writing about a similar topic. Scoble talks about individual tweets, Flickr images and blog posts as info atoms and the need for curators (bloggers, journalists, etc.) such as himself to create info molecules out of these atoms. Scoble's needs as a blogger are different than mine as a media consumer, but our problems are much the same. Whether a blogger or a media consumer, we need a way to point to a group of related content.

Pointing Towards a Solution

This problem is largely solved on a site-by-site basis. The solution, or a big part of it, is tags. Here are all of the bookmarks I've saved to Delicious having to do with photography. Here are all of the photos I've taken at Mass MoCA. If you want to, today, you can subscribe to feeds of either of those curated groups of content, but this is a site-specific solution. If I post a Mass MoCA image to Smugmug or Facebook, you won't see that.

What we really want is to be able to follow the people we like but automatically filtered to the content we're interested in. If I, as a media producer, am consistent with the tag names I choose from site to site, then this should be possible.

Currently, with WebFinger, you can follow on Buzz (and anything else that can make sense of the FOAF/XFN/WebFinger web (de facto) standards), but you will see everything. The problem is, no one wants to follow everything I do online (even my wife; I'm sure she has no interest in Clojure-related blog posts, for instance). People don't really want to follow, they want to follow As long as I, as content creator, tag content consistently across sites, this could be powerful.

The Wavy-Hands Proposal (where we get a little technical)

These technologies are too new to me for me to suggest anything near fully-baked. However, I think I've got enough of the details down that this isn't too far off. Either way, I'm going to spew out what's been floating around in my head and will just consider it a jumping off point for the ensuing conversation.

Currently, WebFinger uses a site's host-meta file ( to find a URL template that allows one to turn into a URL to an XRD document describing me. If this XRD document points to a FOAF document listing my accounts online, then my Webfinger address,, essentially becomes a pointer to everything I do online (wavy hands flailing here).

Now, just as a host-meta file can support WebFinger by providing a template to turn a WebFinger URI into a URL to a profile page, what if it also provided a template for looking up an account holder's tags? E.g., it could look something like:

<xrd xmlns:hm=""

  <link rel=""
        template="{user}/tags" />

This would provide a standard way for services to look up filtered data about its users. If all publishing services provided this, it could be possible for someone to look me up on Buzz and following anything I tag "Clojure", for instance, whether it be a blog post solving some Project Euler problem, a Flickr image of the class hierarchy of Clojure's code base or a tweet of some Clojure golf fun. They wouldn't have to sift through all of the non-Clojure-related content I produce all over, assuming I take a few seconds to tag everything appropriately.

It may even be possible to extend WebFinger to provide the same kinds of pointers to these curated groups of content as we do people. If points to me, could point to everything I've produced tagged Clojure?

Is there any work like this going on? I haven't seen anything, but I'm new to this space and would be happy to be pointed in the direction of ongoing discussion. If not, is there interest in pursuing something like this? The benefits in terms of media consumption, curation (a la Scoble) and discussion seem huge. We all need a way to sift signal from noise and this could be a big step towards that goal, Web-wide.