Graph Search is Facebook’s Bid to Compete with…Everyone

1/15/13Follow @wroush

There’s broad agreement in the tech industry that search should be more personal and more social. Google certainly thinks so—that was the whole point of last year’s “Search Plus Your World” initiative, which saw the search and advertising giant begin to customize its search rankings for individual users to highlight material from people connected to them on Google+, its still-nascent social network.

So it’s no surprise to see Facebook, the world’s largest online social network, getting serious about search—and bringing its own unique data and infrastructure to bear on the problem. That was the upshot of a press event Tuesday on the Facebook campus. Co-founder and CEO Mark Zuckerberg told the standing-room-only crowd that Facebook is starting to rearchitect its pages to make it easier for users to explore the world—from TV to movies to travel to restaurants to old photos—using the experiences and preferences of their friends as a filter and guide.

The new social search feature is called “Graph Search,” a reference to the “social graph” connecting all Facebook members, and it combines Facebook’s vast trove of user data with some state-of-the art natural language processing algorithms to give users the unprecedented power to delve into the specifics of their friends’ lives and likes. Graph Search understands almost any social query you can think of, as long as it’s related to a person, a photo, a place, or an interest.

(A few sample queries that returned interesting results when I tried them: “Best photos of my friends in 2011” or “restaurants in Chicago my friends have visited” or “movies liked by my friends” or even “Indian restaurants in San Francisco liked by my friends who are from India.”)

It’s human nature to give more weight to information that’s been filtered by people we know. For that reason alone, Graph Search will create many new occasions for people to visit Facebook. And as users become aware that the information in their profiles and timelines is the raw material for Graph Search, the kinds of information they upload to Facebook may change, making the site an even richer resource than it already is.

But there’s no mistaking the project’s real purpose: to get one step ahead of Google and every other company working to customize search results based on consumers’ preferences and social connections. Viewed through a business lens, the change puts Facebook on a path to competing directly with every company interested in local search, including Yelp, Foursquare, Yahoo, Apple, AOL, CitySearch, CitySquares, TripAdvisor, the Yellow Pages, and, of course, Google.

And that’s just the beginning. Because Facebook has so many members (1 billion and counting) who share so much data (including 300 million new photos every day), and because all that data is connected in a gargantuan mathematical graph consisting of more than a trillion relationships, the company is in a historically unique position to serve up socially filtered recommendations.

Give Graph Search a query like “My friends who like Homeland” and you can immediately see which people in your social graph love the Showtime series—and, more importantly, what other shows they like. (You can short-circuit those two steps by typing “TV shows liked by my friends who like Homeland.”) In one swoop, in other words, Facebook’s project stands to circumvent more than a decade of work on recommendation systems and collaborative filtering algorithms by Amazon, Netflix, and dozens of other search, content, and e-retailing companies.

Then there’s employment recruiting—Graph Search makes it extremely easy to do a search like “My friends who have worked at Google and graduated after 2005.” And dating—“My friends who are single and live in San Francisco.” There’s no doubt folks at companies like LinkedIn and Match.com were watching today’s Facebook announcement with interest.

So far, beta access to the Graph Search feature is limited to about a thousand people, including many of the journalists and analysts who attended today’s event. Facebook says it intends to extend access very gradually—so don’t expect to see it turn up the next time you visit the site.

But soon enough, all English-speaking users will be able to access the feature (other languages will come later, as will mobile version of Graph Search). And Zuckerberg made it clear that the capabilities the company showed off today are only the tip of the iceberg.

“In the future there are some very obvious things we want to get to,” he said. That includes indexing all content on Facebook—not just photos and preferences and check-ins, but the text of status updates and everything else people upload to the site. It also means making Graph Search part of Open Graph, the protocol that Facebook uses to allow outside developers and services to access and update the information stored in users’ Facebook accounts.

Once that happens, look for Graph Search to start powering all sorts of new experiences, whether from Facebook or its partners, such as socially savvy virtual personal assistants. Imagine a version of Siri, for example, that understands not just what’s on your iPod playlists or your iCal calendar, but everything your friends are doing and have done. (Well, not literally Siri, since Apple is unlikely to pay Facebook to connect Siri to Graph Search. But many other companies are working on Siri-like AI technologies.)

“Graph Search is the sort of product we love to build,” Zuckerberg boasted at today’s event. “It’s a big technology problem, and it’s also a big social problem. So it’s the kind of product Facebook and our culture are uniquely suited to build.”

Whether Facebook’s engineers are uniquely good at solving computational problems with a social component would, of course, be hard to prove. But it’s indisputable that Facebook has more data to play with—specifically, connected data about people and their preferences—than any other company on the planet. Google is striving to catch up using Google+, which reportedly has more than 135 million active users—but it will be a long time before the typical Google+ profile is as rich or deep as the typical Facebook profile.

That puts Facebook in an extremely powerful position. Microsoft has been smart enough to partner with the social giant: its Bing search engine is integrated with Graph Search and provides Web-based results whenever social results aren’t available. Google had the chance to collaborate with Facebook on Graph Search, but Zuckerberg said talks between the two companies “fell apart” over technical differences about how users’ private data would be indexed.

The breakdown of those talks “may be a symptom of a bigger strategic rift” between Google and Faceboook, Zuckerberg said. Gee, ya think?

Wade Roush is a contributing editor at Xconomy. Follow @wroush

By posting a comment, you agree to our terms and conditions.

  • http://www.carlease.uk.com/ CarLease UK

    “Graph search” seemingly will offer individuals instant group opinion based upon their collective experiences. It sounds like an obvious natural progression from asking a friend/s opinion, but on a larger scale and in an instant. This could be the point at which social media begins to be a major influence in any individual’s commercial decisions, “where did you spend your money…”