Dark Clouds - To the Cloud: Big Data in a Turbulent World - page 149

of a search algorithm that both Facebook friends and the general public
can access. Graph Search determines its results by matching phrases and
objects on a site rather than just key words. By combining the information
on a user's site and the relationship of the user to friends and to objects,
Facebook is able to return results that take into account how users feel
about people and things. The “like” function is especially important in this
respect because it enables Graph Search to produce results such as friends
who like the ilms Life of Pi and Zero Dark Thirty and single women in
Manhattan who were born in France.
There is more to it than these relatively innocuous search possibilities
suggest. Graph Search takes putting together social combinations to a
new height or depth, depending on your point of view. Which employers
are most likely to hire racists (i.e., people who identify their employer and
“like” racist or racist-friendly organizations) or people who like sadomas-
ochism (in an early search the “prize” went to Home Depot). How about
people who like the banned religious group Falun Gong—the government
of China might be interested to know which of its citizens have relatives
in the United States who like the outlawed organization. All this and
more is available on the powerful new search engine, and none of it is
subject to fact-checking (Giridharadas 2013b). Making this all the more
remarkable is that most of the work is done by Facebook members who,
of course, labor without compensation so that companies, governments,
and, yes, friends can do a better job of advertising to them, tracking their
behavior, and keeping in touch. It is little wonder that one organization
that tried out Graph Search in its early days declared “the end of privacy
by obscurity.” Or we can view it as the end of privacy through a busi-
ness model that turns every bit of information posted by members into a
marketable commodity and delivers those same members to advertisers
effectively and eficiently.
With all of these elements in place, there remained one key element.
How would the company determine the effectiveness of advertising on its
own and other sites? The irst step was to partner with the data-mining
company Datalogix, which tracks the connection between ads that users
see on Facebook and their in-store purchases. This provided an important
indicator of just how successful the social-media company could be in
turning ads into actual sales. But this was not enough. Facebook wanted
to determine how ads on its site stacked up against those located on others,