Update: It’s all about search! See below for details. But for what it’s worth, investors are not impressed. Facebook shares are down 2% in midday trading. No doubt that’s a short-sighted view–this socially infused search service is a big deal, if only to show how Facebook will keep engaging people more and more deeply, not to mention provide a foundation for the kind of search ads that made Google king of the Web–but that’s the market for you.
There must be a couple hundred press here, plus a crowd of Facebook employees standing in the back of the room–a sure indication that Facebook considers this a big deal.
And now we are underway, with CEO Mark Zuckerberg introducing us once again to Facebook’s mission and the social graph. First he goes into Facebook’s early history, before the news feed that now dominates the service.
Now he’s introducing the third pillar of Facebook’s service beyond the news feed and Timeline. What’s most interesting is graph search.
So what is graph search? It’s not Web search. We’re not indexing the Web. We’re indexing our map of the graph. There are more than a trillion connections in this graph, with billions more added every day–likes, comments, photos, etc. Indexing this is a really hard problem and we’ve been working on it a long time.
Graph Search is privacy-aware, he says. Every piece of content has its own audience, and most is not public. You can only search for content that has been shared with you–a very tough problem to solve.
Let’s say you do a Web search for hip hop. You get links. Graph Search is different–it’s intended to provide answers: Which of my friends are in San Francisco?
One of the big design problems we had to solve was how to make this natural. We’ve come up with an interface we think works. The answer is filters–not! That’s a joke–he shows a screen full of filters that look ridiculous, of course, because they can’t scale up.
He shows a video of queries that show how they come up with answers as you type in words.
For this first beta version, he says, Facebook focused on four use cases: people, photos, interests, and places. There’s a lot more, but these are especially useful. …