Facebook Used to Be Fun—Graph Search Makes It Useful

1/18/13Follow @wroush

I spend a lot of time on Facebook. Probably more than the average user, who spends around 400 minutes on the site per month, or about 13 minutes per day, according to data from comScore. It isn’t exactly wasted time—I’m usually gathering or spreading news and keeping up with my friends by browsing their news feeds, profiles, and photos—but it’s not quite what I’d call productive. It doesn’t help me plan my time or get stuff done in the real world. So I usually feel a little guilty about it. That’s something I’d never say about other online tools like Google or Gmail or Twitter.

But I think all of this is about to change. For better or worse, Facebook isn’t just for socializing anymore. With the addition of Graph Search, the new search utility that the company began rolling out to its English-speaking users this week, it’s about to become a whole lot more useful.

Graph Search, in a nutshell, turns the signature blue bar at the top of every Facebook page into a big search box. Using natural language, I can type queries relating to the stuff Facebook knows about—people, places, photos, and entities with a Facebook page, to start—and get results custom-filtered for me, based on the connections I’ve built and the preferences I (and my friends) have expressed inside Facebook.

Say I’m having trouble remembering which of my high-school classmates went to Michigan State, the university closest to my home town. I can just type, “My friends from high school who went to Michigan State University” and see the whole list. (It turns out there were 10 of them.)

Facebook co-founder and CEO Mark Zuckerberg at this week's Graph Search announcement. Lead project engineers Tom Stocky (left) and Lars Rasmussen (middle), both ex-Googlers, are in the background.

Facebook co-founder and CEO Mark Zuckerberg at this week's Graph Search announcement. Lead project engineers Tom Stocky (left) and Lars Rasmussen (middle), both ex-Googlers, are in the background.

And that’s just a simple example. By stringing together clauses, I can get results of amazing specificity. For instance: “Movies liked by people who like movies I like” or “Friends of my friends who are Web designers and live in San Francisco.” Facebook has spent years accumulating the data needed to answer such questions. But before Graph Search, only programmers inside the company could ask them.

Do these changes mean Facebook will also become less fun? Quite possibly. It depends on exactly how people end up using the feature, and how Facebook goes about monetizing it. I’ve been playing with Graph Search since Tuesday, and I don’t think it’s ridiculous to predict that it will be game-changing. Web entrepreneurs have been talking for years about using social data as the foundation for a new type of online search. Facebook is the first to pull it off.

Here’s the really big picture: From now on, if you’re trying to decide what movies to watch, what albums to buy, what books to read, what restaurants to eat at, or where to go on vacation, Facebook—not Google, not Yelp, not Amazon—could be your first stop, and possibly your only stop. That’s about as big a shift as we ever see in the Internet business, and there is no doubt that it’s going to upset the established order.

I’m not saying that Facebook is about to displace Google. But, by creating a new way to look for stuff, Graph Search also creates a new way to be found, which will change the way businesses think about search engine optimization and search engine marketing. Up to now, SEO and SEM have been all about Google, and how to get the search giant to rank your links higher on Google search result pages. But for the first time since the late 1990s, there’s another serious player in search. (You might count Microsoft’s Bing as a serious player, of course, but Microsoft has allied with Facebook on Graph Search.)

Nobody knows yet what the new tools for optimizing one’s exposure within Graph Search will be. Getting lots of “likes” will certainly become more important than ever, and it’s also possible (John Battelle thinks it’s probable) that Facebook will introduce some way for companies to buy paid Graph Search listings. For the moment, Facebook is staying mum, and says it will focus first on expanding Graph Search so that it includes more types of data, is accessible in other languages, and works on mobile devices (right now it’s Web-only).

But I’m getting ahead of myself. All I really want to do with this week’s column is convince you that Graph Search, a project led by ex-Google engineers Lars Rasmussen and Tom Stocky, is a big deal. Facebook says it’s limiting the feature to “a very small number of people” for now, to give it time to test and optimize the feature, so it may be a while before you get to try it out yourself. (You can sign up to join the waiting list for Graph Search here.) Meanwhile, I’ll share some of the queries I’ve tested and give you a taste of the results that came back, along with some commentary.

For context: I have 475 friends on Facebook, which is about twice the average number. About 70 of those friends are people I know from high school, 34 are people I know from college, and most of the rest are colleagues, former colleagues, or people I’ve met in the course of my journalistic work.

My friends who are musicians—1 result. This was odd, because I have way more musician friends than that. I think the problem is that most of them don’t explicitly describe themselves as “musicians” in their profiles.

Hotels in Las Vegas visited by my friends—24 results. Including Excalibur. I’m going to unfriend the guy who stayed there.

Photos taken at hotels in Las Vegas uploaded by my friends—more than 100 results. Apparently what happens in Vegas doesn’t stay in Vegas after all.

Movies liked by my friends who liked Midnight in Paris—Fewer than 100 results, including some unexpected ones: John Carter, The Big Lebowski, and Run Lola Run.

Restaurants in San Francisco liked by people who like Cha-Ya Vegetarian Japanese Restaurant—More than 100 results, including Herbivore, Millennium, and Gracias Madre. This was a truly useful query, since I hadn’t heard of any of those places.

Photos taken at vegetarian & vegan restaurants in San Francisco liked by my friends—nearly 100 photos, believe it or not. That’s why I love it here.

Religious views of my friends—more than 100 results, including one Jedi and one Pastafarian.

My friends who work at Google—6 results. I actually know a lot more people than that at Google, but maybe they hang out on Google+ rather than Facebook.

My friends who work at Google and went to Stanford University—2 results. You can see how recruiters will love Graph Search.

Photos taken in Paris, France, uploaded by my friends—about 94 results. Half of them just show the Eiffel Tower, but some of them are really good.

Games liked by my friends who like Angry Birds—more than 100 apps listed. But they’re all Facebook apps, and I’m not a big Facebook gamer. This feature would be more useful if it listed iOS apps. Which it never, never will.

My photos liked by my friends—More than 100 results. Facebook sends a notice to your News Feed every time a friend likes one of your photos, but it’s easy to lose track of this type of information. Graph Search helps you bring it all back together.

Bookstores visited by my friends—15 results, topped by Kepler’s Books in Menlo Park, CA. I guess I’ll have to check it out.

Movies liked by people who are journalists—more than 1,000 results, topped by The Godfather. Go figure. All the President’s Men or Citizen Kane would have been my predictions.

Music liked by people who like Imogen Heap—more than 1,000 bands listed, with Radiohead, Björk, Depeche Mode, and Coldplay at the top of the list.

As you can see, Graph Search understands many different types of queries, and it’s pretty good at grammar too. You can construct searches relating to about 20 types of pages (athletes, books, cities, colleges, countries, employers, games, grad schools, high schools, interests, languages, movies, musicians, political views, regions, schools, sports, sports teams, and TV shows) and 16 types of public places (bars, bookstores, brunch places, cafés, gyms, hair stylists, hotels, landmarks, movie theaters, museums, parks, restaurants, schools, shopping malls, supermarkets, and theaters).

You can refine your search to results “liked” by you, your friends, your family members, or specific friends; to specific cities; or to places you, your friends, or your family members have visited. (Facebook’s information about your whereabouts comes from check-ins and the geotags on your photos.)

Graph Search has a few shortcomings—which is to be expected, considering that it’s an early beta product. The biggest problem is probably this: the results are only as good as the information users have given to Facebook. If you’re a musician but you haven’t made that clear in your profile, you’re not going to show up in searches for musicians. If you were ever snookered into “liking” something that you don’t actually like (which happens all the time, according to tech blogger Steve Cheney), it will skew the results of your friends’ searches.

Other problems also worry me. It’s not clear how results are ranked within Graph Search result pages, and how much I should trust them. Why was Kepler’s at the top of my bookstore search? Because it was visited by more of my friends than any other bookstore on the list, or because they visited it more often? Or because Kepler’s paid Facebook for the placement? That last one isn’t a possibility right now, but it might be in the future.

On top of all that, there are a lot of queries that yield only one or two results or none at all (e.g. vegetarian restaurants in San Diego liked by my friends), presumably because my network is too small. Yet I have more friends than most Facebook users. So I’m wondering how useful Graph Search will be to people with the average number of friends (around 230) or fewer. It seems clear that in a Graph Search world, it’s advantageous to have more Facebook friends, the better to draw on their likes and experiences. But by definition, the more friends you have, the shallower each connection will be. I’m not sure that’s the direction anyone wants to go in their social networking activity.

According to biographer George Dyson, the pioneering computer scientist John von Neumann refused to predict how the electronic computer he was building at Princeton in the 1940s would be put to work. The machine was “so radically new that many of its uses will become clear only after it has been put into operation,” von Neumann said.

He was right about that—and I think the same thing is true of Graph Search. It will take a while for Facebook users, and Facebook itself, to figure out the software’s most interesting applications, but they will be many and profitable. Facebook just evolved from a toy into a tool.

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

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

  • Kelly

    Facebook is a complete joke! how people get wrapped up in that garbage I’ll never know. The things people will share with complete strangers never ceases to amaze me. http://www.ficksitall.blogspot.com

  • TDT

    Better unfriend that person who liked Coldplay, too . . .