He says this is about making new connections as well. If you meet someone in real life, like I met someone named Chris -- I type "people named Chris who are friends of Lars and went to Stanford University."
The Graph Search lets you drill down -- you can change schools, locations, etc., until you find the person you want.
Another example: "Friends of my friends who are single males in San Francisco, Calif., and who are from India." That gives a very select group of people.
The people I"m seeing here have set all these things -- city, hometown in India, etc., -- so that it matches. I'm only able to search for what I'm already able to see on Facebook.
He says it's a powerful recruiting tool. Friends of current employees is a good place to start with recruiting. For example, "NASA Ames Research Center employees who are friends with Facebook employees."
And he gets back a list of people who fit that criteria.
Another example for recruiting: He's looking for "people who have been product managers and who have been founders."
And it gives him a list of such people.
Lars Ramussen is taking stage to show how graph search works with photos of friends.
So this is Facebook's shot across LinkedIn's bow.
He types "photos of my friends taken in Paris, France" and a slew of such photos show up.
Lars next searches for "photos of friends who have been to Yosemite National Park."
Then he searched "photos of my friends before 1990" and up come a number of photos of kids. There is one of Sheryl Sandberg from 1972. And the audience gives an "awwwww"
Another example: He searches "photos of Berlin, Germany from 1989," and it calls up more than 100 photos that are public -- not just from his friends -- that offer a look at Germany the year the Berlin Wall came down.
Lars Points out that you can only see the content you can already see on Facebook. Clearly, Facebook is mindful of a possible privacy backlash.
Tom is talking again. He's showing off "interest" search now.
I.e., "movies my friends like." Says he can now come here and see great movies through the lens of my friends. Now he's showing TV shows my friends like. And within that, you can pull up a video clip from one of those shows -- 30 Rock, Mad Men, etc.
Another example: "TV shows liked by Software Engineers." And suddenly there are a number of TV shows that software engineers like. Then does TV shows that doctors like ...results: Grey's Anatomy.
Tom uses another example: "Music liked by people who like Mitt Romney." Then he does the same for "Barack Obama." Common result: The Beatles.
Tom tries "Languages that my friends speak." Most of his friends prefer Python, but French and Mandarin are also there.
This is all a powerful way to slice and dice people and their interests on Facebook.
Lars is back on stage to show how search works for "places."
Lars says a few months ago he had a terrible toothache, so he types in "dentists liked by my friends." Second result: Mark Zuckerberg's dad. But then a long list of his friends and their dentists.
With that out of the way, he searches for "Restaurants in San Francisco, Calif., liked by my friends from India." And up comes a list.
This is starting to sound like a more customized Yelp. Very smart.
However, it doesn't have to be limited to your friends. You can search "restaurants in SF, California, liked by Culinary Institute of America graduates." This returns an array of restaurants -- but not based on your friends. Again, this is narrow search that Google and other don't do.
You can search for restaurants, bars, cities, countries, you name it.