Fred StutzmanThoughts about information, social networks, and privacy

For the past few days, I’ve been thinking about the information products and byproducts of social computing. Products may be thought of as things we create with intent; our Facebook profile, our home page. Byproducts, respectively, are the things we create with limited intent; our attention data, the traces we leave in server logs, the software products that appropriate our agency.

From a volume standpoint, the amount of data byproducts we produce significantly outweigh our pure data products. Maybe we’ve got 15 profiles on social networks, but Google’s got gigs of our email, search logs, and click streams. Following Irwin Altman’s notion of privacy as boundaries, its easy to see how we delineate between these two data sets, even though they’re identical at the binary level: one we see, and one we don’t.

At SGFoo, I participated in a number of discussion around data byproducts and the social graph. Leveraging your explicit connections (a data product) and attention or network data (byproducts), service providers could expose all sorts of novel information to you. I tend to agree; the jumble of connections and intentions and algorithms can likely tell me all sorts of new and interesting things.

In a post danah boyd wrote a few days ago, she cautioned against where such objectively computational approaches lead us, that the negative effects of such systems may outweigh the perceived gain. I tend to agree; the leaders of the social computing space possess an alarming antipathy towards privacy, especially when weighed against the benefits of derived, latent knowledge. Of course, this is the ideology of Google or competitors; in the graph, we’re all just documents with linkages, our behaviors subject to Map Reduce. The privacy advocate stands in the way of progress, the natural state of industry.

Drawing back to the initial distinction I posed, the product and byproduct, I wonder if there isn’t a self-regulation implicit in the system. Perhaps norms other cultural processes will make taboo the “reveal” implicit in surfacing computed data byproducts. It’s creepy when a computer tries to figure you out, it’s creepier when a computer tries to figure you and your friends out, and perhaps the creepiness of all of this makes leveraging such knowledge in social processes taboo. We may be able to compute it, but we may not actually want the information because the objective boundary is crossed.

In 1996 Sherry Turkle proposed that we were looking for the subjective computer, one that became a place of identity reflection and expansion. At the time, it was alarming to think of a computer to which we bared our souls. Of course, 1996 was a different time for computers: we weren’t hyperconnected, massive data stores like Google were nascent, the notion of sharing one’s real identity online was anything but pervasive. These conditions established a sense of mastery over what one was sharing; the computer could become your second self because, well, you didn’t have to worry about a creepy Facebook app sharing your deep political opinions with your friends without your knowledge.

Do we still seek the subjective computer? I’d argue that, in 2008, the subjective computer seeks us. Since Turkle wrote Life on the Screen, we’ve placed much emphasis on using objective measures to uncover subjective knowledge. Rather than the computer being the device you pour your heart out to, it has become an intelligent proxy. At the same time, there no longer exists the monolith computer; the computer is simply the networked device, routing you to the best places for disclosure and community.

In 2008, we find ourselves in a unique situation where the things we say, and the things we don’t say become central parts of our computer disclosure. It’s no longer simply about our blog post, it’s about who we’ve looked at or talked to. Our machines have frameworks for computing both the intentful and ephemeral things we disclose, our data products and byproducts.

Where does this leave us? When we reached out to the subjective computer, it was a powerful tool that one could master and appropriate for specific purposes. Social interaction, identity play – these were affordances of the device. Now computers master us, leveraging our data to fit us into modeled interactions, exercising tremendous power through selective disclosure, and offering us freedom through a participation process that is essentially repressive.

As I alluded earlier, it is unlikely that we’ll ever become comfortable with the spaces of complete disclosure. There’s always going to be a difference between our shared and mined data, and there will always be social rules standing in the way of leveraging data a person or system has collected about another. This is not to say that the boundaries won’t be tested, or that they aren’t already stretched to frightening levels. Beacon didn’t work because we were uncomfortable with the removal of boundaries, and I’d argue that we’re going to continue to feel this way in similar situations.

It is now time to push back against the devices and networks that seek to master us. It is time to return to places where we exert control, where our data isn’t an asset, and where our mastery over the device sets us free. Horribly naive? Perhaps, but I also might be right. The arms race of analytics may fail simply because we’re not comfortable with the “reveal”. The true loss here, however, is the sense of freedom we once had when the subjective computer was our agent. As we now live in fear of the computer, we’ve lost the ability to seek freedom in it; I think one day we’ll want that back.