(picture)

December 08, 2004

Implicit or Creepy?

David Weinberger (on a new Corante weblog "Operating Manual for Social Tools", which is a bit of a mouthful but looks like a really interesting read) was musing about the social downsides of "social networks":

...It makes complete sense for artificial social networks to notice who's sending what to whom, who's contacting whom, what the hidden contours of the social map are and how the surges of activity relate to other behaviors on the system. ...and more and subtler dimensions.

Yet, at a certain point it becomes creepy. In fact, there are two points.

First, it can be creepy just to know that all your behavior is being tracked.

Second, it can be creepy if the behavior ever surfaces to human eyes.

Two observations of my own. One: yes, it's definitely weird, in a Heisenberg way: knowledge that you might be observed creates a change in behaviour, and you can feel this happening. Danah Boyd follows up with the example of Cobot on LambdaMOO (a "bot" who observes lots of dimensions of social behaviour, and will tell you all about them when asked). I've noticed something very similar with Audioscrobbler, which builds a social database of the music everyone's playing: it's useful, extensive, and fascinating... and if you could see the rubbish I listen to, it would change what you think of me. Right?

Second observation. We're talking about large-scale groupware systems, and agents in those systems which can infer social behaviour "in the small" as well as in macro trends. There are many of these systems, and in the biggest, the users of these analysis tools are not the "subjects", the participants in the social milieu. Walmart has one of the best. Your government has another whole set of them. If something as fluffy as Audioscrobbler has you looking over your shoulder, there are two appropriate reactions: hide, or perform. Dance, baby, dance.