Skip to Content

How Facebook Could Give Its New Search the Edge

The company’s search engine will become formidable when it includes the text of comments and the vast store of Open Graph data about things outside Facebook.
January 25, 2013

Facebook’s new Graph Search—a feature that lets you search through the data shared by your friends—clearly needs some work (see “Facebook’s New Graph Search: Not Very Good”). It relies solely on the “likes,” check-ins, and profile data provided by your friends – signals that may be incomplete or unreliable unless you have friends all over the world who always faithfully check in wherever they go, and like products and brands honestly, not ironically. What’s more, queries must be structured in a way that often makes you feel like you’re talking to a database.

Fortunately for Facebook, help is on its way. The social networking giant has two valuable data stores that it has yet to connect up to its prominent new search box (so large that it displaces the company’s logo). Accessing these data stores could change how people find information online—not to mention the fortunes of arch rival Google.

First, Facebook could analyze the substance of user comments and other text people have added to the site, such as photo captions. That could help Facebook find many more recommendations not logged through check-ins or “like” button clicks, and to sift out misleading likes.

Second, Facebook’s search tool has yet to move outside the social network’s own borders (and into Google’s territory) by tapping into the Open Graph, a system developed by Facebook in 2010 as a way for it to understand the products, services, or other entities described on Web pages and in mobile apps. Open Graph allows Facebook to understand, for example, the artist, song, and album you are interacting with each time you click a song in a music service such as Spotify. That kind of knowledge could help Facebook search apply to more than just content from your friends.

Both of these improvements were mentioned in a brief note at the end of last week’s launch, but the company has not said how soon they might arrive. But it is possible to predict what they might allow. A mostly overlooked upgrade made by Facebook’s partner Microsoft to its Bing search engine last week, which lets you search the text of friends’ Facebook postings, provides a preview of how the first change might make Graph Search more useful. Enter “Friends who like tequila” into Graph Search today and you’ll find only people who have joined groups dedicated to the drink. Connect your Facebook account to Bing and type “tequila” into its new “Friends’ photos” search and you’ll find every photo that mentions tequila in a caption or a comment—so you’ll probably see a lot more images of people enjoying the beverage.

Bing’s new ability to access the text of your friends’ posts to Facebook is also on display in a sidebar on its main search page. When I searched for “Virgin Atlantic,” it showed me comments mentioning that airline even when the airline wasn’t tagged explicitly. I could see that search—or one for any brand or product—being much more informative about what friends think or have experienced than anything Graph Search could show me.

A small startup called Trove provides further evidence of the value of searching the full text of Facebook content (see “A Better Way to Mine Your Facebook Past”). The company originally created a product similar to Graph Search, says cofounder and CEO Seth Blank, but after a poor response from trial users, it refocused on using clues found in comments and other text to answer more free-form queries.

Trove has developed several tricks that underscore how much more useful inputted text can be than just like button clicks. For example, a search for “puppy photos” might reveal an unlabeled Instagram photo of a dog because the photo elicited a tweet using the word “dog” and a Facebook comment saying “adorable!”

More subtle signals like those might provide a better measure of which restaurants, products, and businesses friends really like—and why; far better than just “likes,” says Blank.

Trove is currently experimenting with a browser plug-in that displays its results when a person searches on shopping sites Amazon and Newegg. “People buy stuff based on what they find friends like using Trove,” says Blank. Facebook is likely exploring how to encourage similar buying behavior around its own search, with a view to charging for ads that appear alongside search results.

Facebook’s other untapped resource, the Open Graph, could also improve its power to recommend products—and hence offer new opportunities to position ads next to search results, as Google does. Folding it into Graph Search would also be a significant shift for Facebook, and it could change the way we use the Web.

Open Graph is powerful because it’s as much a creation of other websites and apps as it is of Facebook itself. Companies contribute Open Graph metadata to describe the things—products, brands, movies, people, and more—on their pages or in their apps because they want attention from Facebook’s billion users. Those efforts enable Facebook to understand what exactly it is people are interacting with, or are interested in, providing a kind of index of the wider Web and world.

The Open Graph is already big. Last year, entrepreneur Matthew Berk used five billion Web pages provided by the nonprofit Common Crawl (see “A Free Database of the Entire Web”) to show that the Open Graph describes some 400 million objects (he also found that 22 percent of the Web’s pages link to Facebook in some way). Publishers would likely redouble their efforts to embrace the Open Graph if it were included in Facebook’s search, Berk predicts, making it an even more powerful index of what people care about online. “People used to go out of their way to make it so that Google would find their content; now they look to Facebook,” he says. “It’s like a redistribution of wealth.” Berk’s research into the Open Graph led him to found a startup called Lucky Oyster, a service due to launch this year where people can explicitly share and recommend products and content to Facebook friends.

However, Berk predicts that Facebook will face challenges taming “noise” that pollutes the Open Graph, because contributors don’t describe their content in a consistent way. It’s a problem that also troubles Graph Search, which has made it abundantly clear that on Facebook, “like” doesn’t always mean you really like something.

All the same, Facebook’s approach to organizing online information and making it searchable seems to make sense—as evidenced by Google’s enthusiastic efforts to build a social network (see “The Man Building Google’s Social Network”) and a system that understands concepts mentioned on Web pages (see “Google’s New Brain”).

“The big shift is to information being about things in the world and people, not just Web pages,” says Berk. “At Google, you can see that social has become number one across the board.”

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.