When Facebookin Menlo Park, Calif., this week, the announcement was no doubt being watched closely in nearby Mountain View.
The data that Google engineers have craved for years -- rich portraits of connections between people, places, and things, all tied to real identities -- was suddenly searchable on Facebook. It's data Google is trying to replicate, using Google+ and other products, but there's no doubt that Facebook's billion-member network has given it the advantage. You can imagine the Google crew listening to Mark Zuckerberg describe the data that underpins Graph Search and wishing they could get their hands on it.
In fact, two of them did. Lars Rasmussen and Tom Stocky, who both worked on search products at Google, defected to Facebook and began working on a new kind of search product. In 2011 they started on what would be introduced Tuesday as Graph Search. From the get-go it was clear that searching your social network was materially different than searching the Web. Like the Web, it could provide answers (who's a good dentist?) and entertainment (show me photos of my friends in Paris). But social networks promise something more, the duo said in this week at CNET headquarters: bringing you closer to friends by helping you share experiences.
It's an approach that differs sharply from that of Google, where the search team is interested in getting users the best possible answers, at lightning speed, regardless of who their friends are. The question is which approach is better -- and, should Facebook's take on search prove popular with users, what Google should do about it.
Even though they built it, Rasmussen and Stocky say they can only guess at what social search will mean for the masses.
"This all remains to be seen," said Rasmussen, who previously started the company that became Google Maps and later created the more divisive Google Wave. "We're also curious about this question. Obviously we think there is something here -- otherwise, we wouldn't have spent so much time on it."
The power of social
One day Rasmussen opened Facebook to see that his friend Zuckerberg was listening to a song, which appeared in his news feed. Rasmussen, curious, clicked the link so he could listen along in real time. That prompted an embarrassed message from Zuckerberg: This song is terrible, Zuckerbeg said. Sorry about that!
"That interaction was much more valuable than just listening to the song," Rasmussen said. "Watching terrible movies with your best friends can be better than watching an awesome movie alone. Watching awesome movies with your best friends is where we're trying to get to."
In other words, social search promises to bring us more than mere answers -- it can also bring us into conversation.
"There's a little bit more to it than just finding the best place to have a meal tonight," Rasmussen said. "Knowing which of your friends recommended it, or which of your friends liked it, will help you have a more social experience."
Google has also introduced ways of making recommendations more social. Apps you purchase in the Google Play store will be displayed, along with your Google+ profile, when a friend searches for them. The same goes for reviews of places on Google Local. And if your Google+ friends +1 Web links, those links will rise higher in search results for you. For most people, though, Facebook better reflects a person's actual friends and family members than Google's social products do. Even if a healthy chunk of your friends and family members are on Google+, they might be lumped in with brands, publications, celebrities, and other people the user follows but doesn't have a personal connection with.
In short, it's easy for an average user to grok why they might search Facebook for "plumbers my friends like." Thanks to the fact that Graph Search is embedded at the top of Facebook, and can interpret natural language queries, it's easy to use, too. Google's social recommendations, scattered across various products, look weak by comparison.
Hey, 1 billion, meet 30 trillion
But what if the value of social search is overstated? Google sees your friends as important to answering search queries -- that was the point of Search Plus Your World, introduced last year -- but they're only one signal.
What are the other signals? Well, for starters, there are the 30 trillion Web sites Google has indexed, across 230 million domains. There's the Knowledge Graph, its database of 570 million people, places, and things, which now has mapped more than 18 billion connections. Google is betting that for most questions, that enormous database will provide better answers than the random sample offered by the average Facebook user's 150 friends.
Google declined to comment for this story. But its philosophy on search is readily apparent. Google wants to answer your question no matter who you are -- and, unlike Facebook, no matter who you know. The fact that you would have to rely on your friends having visited a good Chinese restaurant to get a decent recommendation is a huge bug, in Google's mind. Sure, it will show you suggestions from friends if it has some to share. But Google wants to answer your question well even if it doesn't.
At its announcement, Facebook made much of the difference between Graph Search and "Web search," which returns not answers but links to other Web pages that hopefully do. The thing is, Google has been trying to move beyond "10 blue links" for years now -- the Knowledge Graph is only the latest effort to deliver answers on the search results page itself. Even Facebook isn't above the blue links -- thanks to its , you can perform Web searches inside the social network.
The battle to come
It's important to note that several major categories of Google search are as yet unaffected by Facebook's entry into the space: video search, product search, flight search, and maps.
But Graph Search is still in beta. Rasmussen and Stocky said they both have "years" of work ahead of them. So far Graph Search is available only in English, to a select few beta testers. It can't search status updates or notes. The recommendation engine for things like plumbers appears to rely heavily on "likes," and how many people have ever liked their plumber on Facebook?
Still, it's easy to imagine. And even in beta, there's one search Facebook does better than anyone else: . Facebook is the world's largest storehouse of pictures, and Graph Search makes them searchable in a way that is not only functional but fun. It handles vanity searches ("photos of me"), creeper searches ("photos of friends of my friends who are single"), and searches designed for pure exploration: photos of Paris, photos of puppies, photos from 1980. It's a rabbit hole every bit as fun to fall down as Wikipedia -- or, more to the point, Google Images.
If Facebook builds the rest of Graph Search as well as it did the photo components, Google may actually have something to worry about. In the meantime Google will continue building its version of the "Star Trek" computer, one that answers questions perfectly no matter your social connections. And while it does, a team led by its former employees will be busy upending our expectations for search.