Google Paper Proves Social Web Has Become a Science
Think back a week, a month, a year or almost five years ago to when you started using Facebook or MySpace. It seemed like such a harmless, simple social transaction as you were friended by a friend, colleague or significant other. Folks, it's not so simple anymore because a lot more is at stake. Facebook growth has been rocketlike, gaining some 150 million users and eclipsing MySpace's 125 million users this year. Both sites hope to leverage their user bases for more advertising revenue. These walled gardens have established their brands well enough that they, along with Google, have decided it's time to federate their platforms somewhat, enabling users to move data seamlessly from one Web site to the next. This, of course, brings critical privacy concerns to mind. Accordingly, some Google researchers are getting scientific about this. Monica Chew, Dirk Balfanz and Ben Laurie point out three areas where social networking sites compromise user privacy in a new, thankfully brief white paper, "(Under)mining Privacy in Social Networks." White papers concerning data security on the Web are far from new, but the emergence of one for the so-called social Web spotlights just how serious the social networking phenomenon has become. These researchers cite lack of control over activity streams, unwelcome linkage and deanonymization through merging of social graphs as the three main privacy concerns for social Web users. The activity stream is a collection of events associated with a user, including changes the user made to a profile page, any applications the user runs on a social networking site, shared news items and messages sent to friends. The authors warn that a user may not be aware of all the events that are fed into the activity stream. Second, a user may not be aware of the audience who can see that activity stream. Both issues signal a gap between how a site actually works and how the user perceives it, as well as a lack of adequate communication on the part of providers that wrongfully assume their users know how their technology works. It's okay to set up a Web service and watch it propagate online, but when you start messing with people's data and breaking boundaries without user permission, you're asking for trouble. Duh. Facebook Beacon, CoComment and Google Reader are the great examples the authors cite, all of which caused brouhahas with users. Unwelcome linkage occurs when links on the Internet reveal information about an individual that they had not been intended to reveal. This happens in plenty of places online, but the social Web is great testbed for these issues because people exchange a lot of data on them. Data from one context can leak over into another, sparking some unpleasant reactions. Blog trackbacks can break anonymity barriers. Merging social graphs are an extension of the breakdown of anonymization online. Again, because social networking sites extract a lot of personally identifiable information from people, users can figure out who an otherwise anonymous user is and expose that person. So what are ardent social Web users to do? After all, it's not as if they can bury their heads in the sand. After spending hours nurturing their profiles on social networks, should they suddenly abandon them? That's the worst-case scenario. The authors have a number of recommendations, which ReadWriteWeb's Lidija Davis summarizes:
Notice any themes with these recommendations? Oh wait, they're the same two that I mentioned earlier: user ignorance and poor communication. Educating users about how social Web services operate will clear up a lot of confusion. Service providers can save themselves headaches by granting users greater control over their data online without enabling them to compromise their personal data. That's a fine line to walk ideologically and technically, but it needs to be worked out. In fact, folks at these sites are working on just that. Do you agree or disagree with the positions in the Google paper? Why? What would you recommend? |

Comments (3)
Interesting points on "deanonymization." It seems that there is an unintended blending of anonymity, confidentiality and privacy. Are they the same? I think not. Where is the line? Do our members/readers understand the difference?
Posted by Anne Easterling | January 12, 2009 6:58 PM
Anne:
I see this as the trickiest challenge; rationalizing social aggregation sites in such a way as to ensure data federation without unintentionally exposing user data. I don't have an answer for this anymore than Google, et al. do.
Posted by Clint Boulton
| January 12, 2009 7:40 PM
This paper from Google is very interesting, but seems to be based on old facts.
The example with CityBank is very old and was fixed immediately.
Now, it is clear that social media and the interconnect between the services make user data more rapidly and widely available. One key point here is about the user and the fact that he knows which services he is using and how to control it: if you choose to share you data, then you must understand that it becomes public.
The paper mention that it is critical that the user can keep control on his data, and be able to delete it. coComment offer this in many ways (delete a comment, blacklist a site or a page, anonymize a comment....).
Now, one thing might be missing on the web: private sites should be able to inform the browser of this status (maybe via the site certificate) in order to switch the browser in a "high privacy" mode. This way, the user can ask not to activate any extension/service on such a site and make sure that the data is sent to this site only.
Posted by christophe | January 14, 2009 2:25 AM