Graph Search alert

Facebook's promotional video for its new function Graph Search shows words being typed into a blue search bar: "my friends who like road trips" appears followed by video clips of young people rocking out to music in the front seat of a car or smiling together in the back of a pickup truck as their hair is being whipped in the wind.

But how does this "social search" actually work, and what will it give users?

"The idea of searching for things your friends like is intuitive," says David Short of London-based digital media consultancy, Broadsight, "because people trust people. Facebook searches will be good for what Facebook is good for. So, if I want to find photos of my friends at school reunion, it would be useful for that."

But Short says Graph Search is also Facebook's way of trying to compete with that other Web giant, Google.

"Facebook got plenty of cash from their IPO, and they're experimenting with different business models. In the very long term, they need to create a sustainable business model so they can make a profit every year," says Short.

A relevant search

Facebook's release of the new feature came after social media monitoring firm Social Bakers published figures indicating Facebook's growth has started to stagnate in certain places - including in the United Kingdom and Ireland, where user growth dropped by a little more than 1 percent in December of last year.

"They have a significant percent of the human race as users - and that's including people who don't actually have access to electricity, let alone computers - so they're clearly reaching market saturation," says Short. "The typical approach in that situation is to increase your revenue per user, and clearly this is an attempt to do that."

The attempt involves copying Google's successful strategy of pay-per-click advertising from search results, with Facebook leveraging its huge social database, which is effectively hidden from Google.

Facebook "likes" act as the links in this scheme. When a Facebook member uses Graph Search it generates results based on everything shared with that user on Facebook - which could result in highly relevant results.

But Short says this limited data set also restricts the utility of Facebook's search function.

"If I'm looking for a plumber in the city I live in, it's not going to be very useful. I'm sure they will have algorithms and things under the hood to make the search more relevant, but it's not a very big data set," says Short.

It's this secret formula behind the search that has raised flags for privacy advocates.

Rosario Imperiali of the European Privacy Association lauds the fact that the search only works with information shared among users.

Big data breach

But he says there is a deeper issue behind tapping these vast and complex data sets - known as big data.

"You search for an info, and obtain in exchange sort of a virtual identity of the user, " says Imperiali. "Big data means to be forced or driven to profiling, people, situations, circumstances, relations. There is a business interest in making this big data, big dollars."

Such profiling allowed by Facebook's Graph Search may even come in conflict with European legislation.

"The directive and draft regulation [permit] profiling under two specific conditions: transparency and self-determination. This means that the user has to be empowered and given the right to choose how his personal data can be used," says Imperiali.

For Imperiali, transparency means the user should be informed in advance about what happens with the data he or she shares. It's a difficult proposition, given that the search is based on "likes" that were expressed long before Facebook rolled out the new feature.

"Sharing, or willing to share, is not the same thing as consent," says Imperiali. "The user must specifically consent to the way these connections are made and the fact that these connections are done. The user may also object to these connections."

The game

The privacy advocate also thinks that users should know and understand the logic behind the search. That doesn't mean Facebook should publish its proprietary search algorithm code, says Imperiali. But he does believe a balance should be struck between intellectual property rights and users' privacy rights.

And it's precisely this knowledge of how the search works which could lead to users, particularly commercial ones, "tricking" the search function for marketing purposes.

Broadsight's David Short describes such "gaming" as a real problem.

"The Google search model has been around for quite a long time, and it is fairly heavily gamed, which to some extent reduces its usefulness. So the time is right for new approaches on search," says Short.

As to whether the strategy will see success - well, for now, Short says Graph Search is just "another experiment," but one that has potential.