A few weeks ago, I came across an accepted manuscript in the Journal of Public Economics entitled The Old Boy (and Girl) Network: Social Network Formation on University Campuses (subscription req’d). Excerpted abstract follows:
This paper documents the structure and composition of social networks on university campuses and investigates the processes that lead to their formation. Using administrative data and information from Facebook.com, we document the factors that are the strongest predictors of whether two students are friends. Race is strongly related to social ties, even after controlling for a variety of measures of socioeconomic background, ability, and college activities.
In the paper, the authors profile ten Facebook networks in Texas, and use a regression model to predict “friendship” between two individuals in a network. Although standard criticisms of using regression to get at the essential nature of friendship apply, I found the data interesting (and the R-sq’s were appropriately small).
What jumped out at me was the paper’s methodology. As stated by the authors, Facebook provided data from ten schools to the researchers. Furthermore, the researchers were able to correlate the profiles with GPA and registration data, indicating that Facebook shipped personally identifiable information to the researchers. From a research standpoint, this is fine with me – as long as ethical standards are upheld, I’m fine with personally identifiable data being used in research.
All of this made me think back to a post I wrote a few months ago. In it, I documented Facebook saying “we certainly aren’t selling your information to ANYONE. That’s yours“. When I went back to check on this page, I found it missing (though the Internet Archive has versions of the page containing the text). I wonder if this means a policy has changed, or simply if this page was swept up in a redesign. Perhaps I’m the only person who cares.
With selling on the mind, it looks like Microsoft may make a significant investment in Facebook. The New York Times reports 300M to 500M for a 5% stake in the company. While I’ve always seen synergies between Facebook and the major consumer players (G, Y and M), these numbers smack of irrational exuberance. In fact, they make me think back to 1998, when comics genius Todd McFarlane spent 3 million dollars on Mark McGwire’s 70th home run ball. At the time, McFarlane couldn’t imagine such a record could be broken, that his investment was future-proofed. As it turns out, McFarlane was dead wrong; Barry Bonds broke the single season home run record in 2001, smacking 73 home runs. That ball ended us selling for a more rational $517,000. Similarly, while it may be hard to imagine a post-Facebook future (not that Facebook is going away, but that other things will get interesting), Microsoft buying in at 500M is like Todd McFarlane paying 30M for McGwire’s 70th.
Update: Steve Ballmer seems to agree. In an interview with the Times Online, Ballmer is quoted as saying “I think these things [social networks] are going to have some legs, and yet there’s a faddishness, a faddish nature about anything that basically appeals to younger people.” He also said “here can’t be any more deep technology in Facebook than what dozens of people could write in a couple of years. That’s for sure.” Sure, Facebook is overvalued – but what is the point in burning bridges like this?
Fred, thank you for the find, really interesting.
Please !!! Correct the typo — it’s 3M, not 30M in the last sentence.
Not a typo, I actually meant 30M ;)
The big question here for me is lack of transparency. More than just the disappearance of the previous data privacy page, does Facebook have a publicly stated (and/or available) policy on disbursal of data for research purposes? Certainly that’d be nice, especially if they’re more open to said research – but if it’s done on an ad hoc and permission-granted-by-request basis, then I think it’s worth both users and researchers asking why user data is made available for some purposes and not others.
JKD, AFAIK Facebook has no such formal policy. There were stirrings inside the company of a data mining group that would work with external researchers, but to the best of my knowledge that initiative was killed (it did lead to the publication of the Golder paper).
In my last conversation with Facebook, they told me they have cut all ties with researchers, and no researchers are allowed to conduct any automated collection (previously they had exempted me and a few other institutions).
To the best of my knowledge, Facebook is not working with any researchers, but you do wonder how many papers like this are still in submission/rewrite/etc. This was pretty surprising to the FB research community, but there may be more to come.