After nine years of colonizing the globe and corralling a billion people, Facebook has found a way to unlock the potential of its massive data collection -- a basic semantic search engine that will let it build smarter services for travel, food, recruitment, dating and other verticals that will generate revenue that could rival Google's.
Connecting people via News Feeds (what's going on in the world) and Timelines (who you are) will be remembered as Facebook's Renaissance era, weaving a social fabric into the Internet with 1 trillion connections. Graph Search is the beginning of the Enlightenment, the next major phase in Facebook's history, in which people gain the "power and tools to take any cut of the graph and make any query they want," as CEO Mark Zuckerberg said during the product launch event at Facebook's Menlo Park, Calif., headquarters earlier this week.
Graph Search is about providing answers, extracted from the data your friends feed into Facebook. It's not Web search, which typically generates a series of links for a query, with the exception of current stock prices, weather and many other standard queries. But Graph Search is limited in scope and usefulness at this stage. It is in a beta phase that will last for many months.
Currently, Graph Search is limited to people, places, photos, and interests, and to the English language only. A mobile version is in the works, as well as inclusion of posts that show up in News Feeds, Open Graph data from apps, and support for more languages. However, the rollout of new Graph Search capabilities will happen "little by little," said Graph Search engineering head Lars Rasmussen in an interview with CNET.
"More than 50 percent of the engineering work is responding to inbound feedback. It's a slow rollout similar to way News Feed launched. We are getting a valuable stream of feedback from users, and then we'll figure out what is right order of prioritization [for adding new features]." Rasmussen knows a lot about building complex products. Prior to Facebook he was the co-founder of Google Maps and Google Wave.
Graph Search can tell you basic things like:
- Friends who live in Palo Alto who like "Game of Thrones"
- Friends of friends who are single males in San Francisco and who are from India
- Product managers who have been founders
- NASA Ames employees who are friends of Facebook employees
- Friends who saw "Zero Dark Thirty"
- TV shows liked by doctors
- Music liked by people who like Mitt Romney
- Restaurants in San Francisco liked by Culinary Institute of America graduates
The results are limited by the range of structural data types and data that your friends share. For example, your range of friends, or friends of friends, may not be able to give great information about the best Chinese dumpling restaurants in New York City compared to Yelp or Open Table. Facebook offers some filters, such as restaurant types, and can rank by ratings and the number of check-ins, but it lacks the more integrated user experience and richer metadata set of Yelp and other services. It's not yet designed to show you the best restaurant based on user reviews.
"It's a little more than finding the best place to have a meal tonight," Rasmussen said. "If you know your friends' preference, you can have a more social experience of place. I could argue that watching a terrible movie with your best friends is better than watching an awesome movie alone."
At the same time, Yelp, LinkedIn, OpenTable, Kayak, FourSquare and other vertical sites should be concerned that down the road Facebook will add more full-service functionality to Graph Search.
When asked if Graph Search would include capabilities such as menus and reservations for restaurants, Zuckerberg said, "As that information comes online over coming years...we definitely want to include it in Graph Search."
Facebook is hoping that Graph Search compels users to increase their engagement, adding more Likes, check-ins, tags and other social signals. And businesses will be asking users to Like them so they can show up more prominently in Graph Search. Of course, Like and check-in are not the strongest signals for affinity with a product or service, compared to ratings and reviews.
"There is now a new incentive for businesses to add to their Pages," Graph Search product manager Tom Stocky said in an interview with CNET. Facebook Pages are where businesses and brands stake out their presence on the service.
Searchblogs' John Battelle speculated that Graph Search is essential to recharging user engagement on Facebook. "If people do not constantly feed Facebook with engagement, its value attenuates over time. As the service slows in overall growth, engagement with its current base becomes critical. New connections are the lifeblood of a service like Facebook. Without a steady stream of meaningful Likes, Friend Requests, declared Interests, and such, the platform would wither," Battelle wrote. "Put another way, Facebook needed a service that layered a fresh blanket of value over its core topography. Graph Search is it."
Facebook is expected to generate significant income from paid, targeted search ads, but not at this time. "This could potentially be a business over time, but for now we've really focused on building out this user experience," Zuckerberg said.
It's likely Zuckerberg is waiting until he can deliver a richer and more varied result, which will require integrating Facebook posts and Open Graph data to Graph Search. With the posts, users could get results such as, "Posts by friends about Jennifer Lawrence at the Golden Globes," "Posts by friends who work for CBS about Homeland at the Golden Globes" or "Friends who like Apple who commented on the iPad Mini on Oct. 23, 2012."
Adding data from the Open Graph is a critical factor in making Graph Search and its natural language interface more useful and revenue-generating. Instead of just "like," users will have many more expressive verbs, such as "watch," "listen," "run," "dance" and "play," for querying the search engine, Stocky said. App developers may have to do some additional work to ensure that more custom verbs or actions in their products are captured in the Graph Search taxonomy. In addition, Facebook is mapping out more entities for categories, such as movie types, and integrating entities from Wikipedia, Freebase and other sources of structured data.
When asked if there were a major breakthrough needed to make Graph Search successful, Rasmussen said, "We are past the point where there is a big binary, where we solve a problem and magic will happen. We have laid the foundation for a new paradigm, now we need to built it out."
He added, "We have years and years of work to do."