Your Attention, Please
AttentionTrust is one of those things that provoked a hearty “wuh?” on my part when I first heard about it. Perhaps the initial message wasn’t articulated clearly by its promoters, or maybe I just wasn’t paying, ehm, attention. I finally got it when I saw their presentation at Web 2.0 and immediately realized that their vision is not that far from our own. The elevator pitch is that information about your web surfing patterns is currently stored and managed in a multitude of data silos, mainly large e-commerce sites that use clickstream data to target product offerings and advertising based on user-specific information. If you spent an inordinate amount of time lingering over the Superbitchbag, for example, you might be in the market for other firearm-themed accoutrements. But why donate isolated bits of your surfing profile to third parties when you could capture the whole lot in your browser and use it for your own ends?
A powerful idea, but though it has a number of high-profile backers, it has yet to get any real-world traction. Hence the suggestion by Alex Barnett to retrofit OPML, which has spread like wildfire as a format for representing collections of RSS feed subscriptions, so that it can hold attention data as well. Nick Bradbury agrees with the principle but sees RSS as a better choice because of its <dripping irony>fantastic namespace support</dripping irony>.
I don’t get it. An XML format has value only inasmuch as it represents a contract between multiple parties as to what the specific XML tags actually mean. Experience has shown that establishing this kind of consensus, even for fairly trivial formats, is pretty darn arduous. So newcomers like OPML should be welcomed with open arms: it ain’t rocket science, but it’s a good format for transporting lists of feeds between newsreaders, and for whatever reason it’s achieved widespread adoption. But jamming in a bunch of additional tags, in a foreign namespace, achieves exactly nothing. Folks still have to agree what those tags mean, so we’re right back where we started… except that we’ve taken two distinct XML vocabularies (OPML and attention.xml) and needlessly intertwined them.
Better to stick to attention.xml and address the real issue: why aren’t people adopting it? The answer is that there aren’t yet any compelling applications that would encourage us to do so. The barrier to attention.xml adoption is much higher than for OPML, since it has to be made by every individual user, instead of a handful of feedreader developers. So there had better be some serious mushroom-cloud-layin’ killer apps to motivate us to make the effort. What kinds of apps might fit the bill? The mind boggles. Let’s start with a Memeorandum-like aggregator that chooses which articles to show me based not on lowest common demoninator rankings but on my own surfing patterns; which sites I’ve visited, how long I spent there, whether I left comments and so forth. That’s a better way to drive adoption than to sprinkling on a few more buzzwords.
9 Comments »
Trackback URL RSS feed for comments on this post. TrackBack URI
Leave a comment
Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>
Hey.
Funny you mentioned attention and the aggregator as I’m going to be adding the attention trust toolbar support to TaiRank…
Comment by Kevin Burton — 11/12/2005 @ 10:15 pm
OPML = Attention Data, Attention Engines and Tailrank - podcast with Kevin Burton
Although we met briefly last week, Kevin Burton and I didn’t manage to get enough time to discuss some…
Trackback by Alex Barnett blog — 11/13/2005 @ 12:32 am
You ask
to which I’d reply that often, these 3rd parties add value that cannot be decentralized easily.
My favorite example, as always, is Flickr. The value of of Flickr is not the sharing of JPEG files but the comments, groups, tags and other mechanisms for distribution and socializing.
One point should be made here: I only accept their data gathering because they export much of it as RSS and ATOM feeds, so I can get it out again.
Comment by ingo — 11/13/2005 @ 4:04 pm
Ingo,
You’re right that in many cases the algorithms for managing community-generated data on a centralized server are much simpler than the equivalent decentralized approach. Something like Flickr’s interestingness rating is a particularly sticky case, since it needs aggregate data from a lot of users to do anything useful. (Though I remember seeing an academic paper about a distributed clustering algorithm.)
The point of my post is not that centralized services of this type should disappear, it is that the primary repository for a user’s surfing patterns and suchlike should be their own machine. In the same way that managing certain data centrally enables cool applications like Flickr, I’m convinced that controlling a profile that takes into account a single user’s activities across multiple websites will lead to an explosion of new toys like the personalized Memorandum clone I proposed. Of course, no one is going to stop a site from collecting information about what users are doing on that site.
I should add that you can take a lot of the complexity out of distributed apps, while preserving the advantages, by providing a generic P2P platform to do the heavy lifting. Hmmm, there might be a business idea there.
Comment by Matt — 11/14/2005 @ 7:19 pm
Matt Gertner on OPML, RSS and Attention.xml
Matt Gertner joins the discussion on attention data formats, commenting on the recent posts by Nick Bradbury, Steve Gillmor, and Alex Barnett.
Trackback by AttentionTrust.org — 11/14/2005 @ 7:27 pm
Alex Barnett/Kevin Burton: OPML as attention
Close on the heels of my previous post about attention data formats (though certainly not prompted by it), Alex Barnett and Kevin Burton post a podcast (mp3) about, in part, OPML as attention data.
There’s some good stuff in there, not least th…
Trackback by elliptical . . . — 11/16/2005 @ 3:07 pm
links for 2005-11-15
RSS 1.0 Modules: Taxonomy The taxonomy module is a RSS 1.0 module and a RDF application enabling the identification of topics covered by a RSS channel or item. (tags: RSS ontology rdf taxonomy) Peer Pressure ยป Your Attention, Please…
Trackback by Monkeymagic — 11/18/2005 @ 5:57 pm
I was recently in London meeting creative agencies, and one guy I met had gathered ALL of his personal data collected by any and all means by 3rd parties, and then auctioned it on eBay (it was literally a stack of paper 20 cm thick). Sold it all for 160 pounds - a relatively small sum compared to the effort companies were forced to make to collect the data (which they are compelled to provide on request by law), but still a nice lesson on the value of personal data. Definitely also points to future business models.
Comment by Philip — 11/24/2005 @ 11:42 am
OPML and Attention Data and Tailrank Podcast with Kevin Burton …
Although we met briefly last week, Kevin Burton and I didn't manage to get enough time to discuss…
Trackback by alexbarnett.net blog — 9/23/2006 @ 10:27 pm