So that's what he was building in there.
Yesterday came the much-hyped Facebook announcement: a new live search engine within the site: 'Facebook graph search'. The skinny: A search bar will allow users to find what their friends have liked or recommended, as well as photos of friends. The search functions in three main areas: Photos, People (via their interests and profile information) and Places (i.e. venues,activities locations). Facebook users can type a query into the search box, such as "Friends who live in Norwich and like Skiiing," or "Restaurants my friends like," and get instant results. Facebook will rank the answers depending on the number of "likes" or the level of your interaction with their chosen person, place or photograph.
The idea is to get recommendations from your friends and friends of friends, based on their privacy settings for content they have posted, and to find people within Facebook who share your interests. You could say it's DatingDirect meets TripAdvisor meets Yelp meets your Facebook newsfeed ..?
See Facebook's presentation of it here. and AllFacebook's good post here for more detail. It's in Beta (desktop only) with a limited number of US accounts at the moment. You can sign up to be notified when it's live for you here.
Interestingly, graph search won't be able to search for posts and status updates that match a particular topic. (Take note, community managers) Apparently that's on the roadmap. But Facebook says the loss shouldn't be noticed given that this type of search is rarely done, in part because the current search feature for it isn't very good, and this the first step towards improvement.
The Guardian have a look at industry opinion of the development and its commercial implications here.
How does a graph search differ from a web search?
The search is entirely personal and customisable. Mark Zuckerberg explained the differences between a web search and the graph search as follows:
This is infrastructure we've been building up for years, and it's something that nobody else has ... In general, Web search is designed to take any open-ended query and return to you links that may have answers to the questions you might have. Graph search is very different. Graph search is designed to take a precise query and return to you the answer, not links to other places where you might get it from.
For example, a Google search can show you 'cinemas in Twickenham', bringing up a list of them. (Until recently, you would also have been able to see which of your Google+ contacts had +1′d any of those cinemas, which echoes this new Facebook search, but alas, no longer). However, a Facebook graph search can show you cinemas in Twickenham liked by your friends. Or you can drill right down: nurseries liked by your friends who actually live in Twickenham, or those liked by your friends who work for a certain company, or your married friends, or ... well, you get the idea.
Facebook have partnered with Bing to give users the option to bring up results beyond Facebook too: according to Bing, open graph is a potentially powerful combination of social and web search:
When you do a web search on Facebook, the new search results page features a two-column layout with Bing-powered web results appearing on the left-hand side, overlaid with social information from Facebook, including how many people like a given result. On the right hand side, you will see content from Facebook Pages and apps that are related to your search. We think this is a powerful combination.
How do those three areas of Facebook graph search work?
PHOTOS
Users can search for photos of their friends using search terms, for example:
- Photos my friends took in New York City
- Photos I like
- Photo before 1990
Here's my search for 'Photos of me'
PEOPLE
Users can also search for friends and friends of friends based on the info they have put in their profiles: location, religion, relationship status, education, likes etc. Want to find a concert partner or start a foodie group? Connect with friends who like the same activities-and meet new people (friends of friends) too. For example, you could search for:
- People who like cycling and live near Seattle
- Music my friends like
Here's my search for 'People who live in my city':
PLACES (and things)
Discover things like restaurants and films through your friends and connections.
Explore new places to eat and new films to see -all through people you know. You could search for:
- Restaurants in London my friends have been to
- Cities my family visited
People can also discover music, films, TV shows, and other interests based on what their friends like. If you type in something generic like, "films my friends like," graph search will return not just links to the films' Facebook pages, but clips, too. There are several filters that users can take advantage of, such as friends in California who went to U.C. Berkeley and like sushi. Graph search has also integrated Bing technology, so users can also use graph search as a general search engine.
Here's my search for 'restaurants my friends have liked':
Graph Search and Privacy
Facebook have explained it like this: Your privacy choices determine what's searchable. Here are Facebook's tips to control what information about you shows up in a graph search:
People: searches may be location based. To control who can see your current city, edit this setting in your About tab on your Timeline, or watch this video to learn how to edit this setting. I SERIOUSLY RECOMMEND that you watch the video: it's very helpful.
Photos: You can review photos you've shared or have been tagged in. Watch this video to learn how to report or remove photo tags of you. (This video goes into more detail than the one above on the photo privacy options). Photos that are untagged but not deleted can still be discovered by others on Facebook.
Places: To review or remove location tags of you at any time, go to the Posts You're Tagged In section of your Activity Log.
Apparently, when graph search goes live for everyone, there will be another prompt to review your privacy settings. But why not just do it right now?