Skip to main content

Facebook tackles Google, LinkedIn, Yelp, and its own huge database with Graph Search

Facebook tackles Google, LinkedIn, Yelp, and its own huge database with Graph Search


Zuckerberg has the data, but will users find what they're looking for?

Share this story

Introducing what he called "the third pillar" of Facebook ecosystem today, Mark Zuckerberg was adamant: "Graph Search is not web search," said Facebook's CEO. The implication is that it offers users something different from other search products and Facebook's current features, and that it's technically much harder.

That difficulty means that Zuckerberg has to manage expectations, qualifying Graph Search as "early," as "challenging," as "beta." Without these hedges, if search doesn't work, Wall Street will punish him. But don't be fooled. With Graph Search, Facebook is plugging a big hole in its own features, targeting vulnerable competitors, and taking a measured shot at Google's title as the dominant online company. Facebook wants to be where we find friends, food, bosses, love, news, and the traces of our multimedia lives.

With Graph Search, Facebook is plugging a big hole in its own feature set

Facebook's index of its own database isn't as big as Google's crawl of the open web and its own services, but it's still enormous: a billion people, sharing 240 billion photos and with 1 trillion connections between them, plus their status updates, shares, likes, tags, comments, and other metadata. A single user searching his or her small slice of that social graph doesn't reduce that burden, but increases it. Zuckerberg says that a full ten percent of Facebook's total CPU capacity is spent computing privacy checks. (It's not clear whether that computational burden, or Graph Search's "Beta" tag, is supposed to reassure or to frighten privacy-concerned users.)

Making dumb phrases smart

Facebook has also devoted substantial effort into making search as natural and convenient as possible. After all, Facebook's not interested in pro-level tools, but something that's as appealing and easy as tagging a photo. Graph Search features a natural language parser — not dissimilar to Siri, but based on text, not voice — that maps plain-language phrases onto results that are meaningful inside Facebook. So "photos of my friends in public parks" or "TV shows liked by software engineers" automatically defines both ranges and keywords. All that takes smart code and a lot of computing power. Also, like Siri, it's potentially a source of frustration if you can't get the genie to do what you want.

Still, most of this is basic functionality, not much different from a desktop search by keyword or file type, to help users find text or images they've uploaded or seen in the cloud. Graph Search's more results-driven use cases are designed to addresse the same space as other popular sites and services. "Friends of current employees is a good place to start for recruiting," said Facebook's Tom Stocky — potentially backing Graph Search into LinkedIn. "Single friends of friends" turns Facebook into a lightweight dating and matchmaking site. "Place Search" gives Facebook a foothold on Yelp's turf in local. On Facebook's blog, Vadim Lavrusik touts Graph Search as a tool for news, offering reporters location-specific, topic-tagged photos and "a Rolodex of 1 billion potential sources," making it a rival for Flickr and Twitter's real-time search. And by integrating Bing, Facebook lets social search fail gracefully back into the open web.

But Facebook's graph is orders of magnitude larger than any of these specialty sites or smaller social search players like Path. Also, crucially, Facebook's graph offers the kind of personalized data that Google has been trying, with mixed success at best, to incorporate into its searches. Facebook has a monopoly on this data, both in the sense that only it and its limited partners have access, and in the sense that nobody else has access to anything quite like it.

Facebook as the new Yahoo

Facebook's hope, and Google's worry, has to be that Graph Search will be sufficiently powerful in enough situations to displace web search as the query of first resort. If users start with Facebook search and turn to Google (or worse still, Bing) only when social search fails — much like most users start with vanilla Google search and then drill down to News or Books or Images today — then Facebook cements its place not just as the top social network but as a primary portal onto the web. Likewise, if recruiters start with Facebook first and LinkedIn second, LinkedIn loses much of its luster. And advertisers, who now have a giant audience supplying real intent data rather than just hanging around with their friends, will flock to Facebook.

Facebook, effectively, becomes the new Yahoo, but something closer to what Yahoo, or AOL before that, was before the rise of Google. It is a social network, but also a news feed, a communications suite, and a limited-by-design mix of search and media. Facebook becomes the service that tames the web and delivers people the information and social connection they need.

Facebook Graph Search images