Attention Profiling and APML

I was chatting about lifestreams and thick profiles with a friend (Matt) and he pointed out the Attention Profiling Mark-up Language (APML). An attention profile makes sense in an attention economy where the way a user spends their time is treated as a scarce commodity to be optimised and exploited. An attention profile captures and describes this commodity for a user, which may be exploited to enhance their web experience (such as a service), or by others (such as retail or advertisement).

APML is an approach toward standardising the description of a users attention profile toward promoting the idea that those who own such collected information (the so-called silos such as Google, Amazon, Ebay, etc, as well as the little guys) make it available to users. The effort is driven by Australian company Faraday Media, that have a broader interest in data portability and providing services related to such information. They describe themselves as "epicentre of the attention ecosystem". The core of the standard appears to be motivated by empowering users by making such information accessible and portable given its inherent value. The standard appears to neglect that this value is exactly what the companies that possess it are designed to capture and exploit. Giving it up may not be desirable for many cases. As with Google scanning books pissing off publishers, this will likely make the owners of the collected attention information uncomfortable (disruptive is good right?).

At this early stage, the standard is quite general, describing concepts, sources, free form data, and applications, with dates of access and scalar user ratings. There are some interesting characters involved in the working group (Digg, Ma.gnolia, popurls, etc), and a growing list of supporting applications most notably, the Faraday Media based particls and engagd. Two other early adopter applications that look interesting are dandelife that converts the lifestream into a personal bibliography, and cluztr for social news (click stream as the attention profile).

Particls is interesting in that it is a desktop application that provides an intelligent (attention mediated) information alert system, learning about the user from the contents of their hard disk (see coverage least year on RWW, TC1, 2). The idea is cool, although appears too technical for the average user case. They need to be bought or license the technology to someone to build a killer user-friendly webapp. Faraday Media are obviously tech guys, because their engagd product is a broader solution to the same problem, providing tools and a service to convert a users lifestream into an APML for use elsewhere (such as in a mashup). If the format takes off, this will be a standard practice across lifestream applications (their vision I suspect).

Faraday seemed to have kicked all of this off in 2006, although APML and their applications seeped into the sphere last year with broader coverage. For example: expectations for the use of APML in personalised advertising by Ross Dawson, a generic overview of the state of APML, Mashable highlighting privacy concerns, the potential for integration in GoogleReader, and another generic overview.

Frankly, user's do not want access to their attention profile, they want the services that do. I see the same arguments as are used with OpenID, in the end, it is better for users and businesses. What is missing are good idea's for using this information! Exploitation of an attention profile into an existing social-based service (social news, etc) is a gimmick at best. Social sites are built around aggregate decisions, not personalised for a user based on a deep profile. Modeling a person based on behaviour is hard and expensive, although collecting attention information is relatively easy. This is an important distinction because all APML provides is a standard way for describing such attention information in aggregate. This is an excellent next-step for the the lifestream applications, although still requires innovators to make use of it.

Frankly, attention information will be used to solve completely different problems to the use of such information in aggregate. First-pass cases naturally include filtering, recommendation, and advertisements all of which are currently performed based on a users group rather than the actual user (an easier problem). I believe that as with the shift to social-powered websites, we will see new paradigms for thinking about personal-powered websites.