Mark Zuckerberg Says Facebook Will Compete Directly Against Google As A Search Engine
- Facebook has a bigger "index" of data on hand than any other search engine.
- Facebook's artificial intelligence unit is working to make that data available through Graph Search, Facebook's search engine.
- The final product will be rolled out on mobile devices so that people can ask Facebook for advice and solve their problems using voice recognition on their phones.
The caveat here is that this effort will take 10 years, he said.
This vision emerged when Zuckerberg was asked about Graph Search, which is somewhat clumsy to use and has received lukewarm reviews. The different between Graph Search on Facebook and Google search is that Google will return concrete results for discreet keywords, whereas Graph Search is intended to help you find answers to questions where the answer isn't fixed or clear, such as, "What are my friends' favorite restaurants?"Here's what Zuckerberg actually said:
So the first release [of Graph Search] indexed more than a trillion connections between all the people and interests and events and groups and things that everyone was connected to. The second release that we did recently was around all the updates. So there are more than a trillion status updates and unstructured text posts and photos and pieces of content that people have shared over the past 10 years, and indexing that was a really big deal, because as the number of people on the team who have worked on web search engines in the past have told me, a trillion pieces of content is more than the index in any web search engine.
That last phrase - "more than the index in any web search engine" seemed deliberate. It obviously implies that Zuckerberg thinks he has a database on hand that is bigger than Google's. And he didn't say "search engine," he said "web search engine." A cynic might interpret that as meaning he believes Google search is somewhat confined to internet browsing whereas Facebook's in-app experience might provide something more personal that is separate to, or in challenge with, the overall web.
(There's a debate to be had about how Zuckerberg is defining "index," and Google - which indexes every page available on the entire web - probably believes its index is larger.)
But that is the scale of Facebook's ambition, to be "more" than "any web search engine." It's not clear that people get just how important this is to Facebook. (Business Insider first wrote about the plan in October last year, but few paid attention.)
It's a huge vision. But given that Zuckerberg has gathered 1.2 billion users onto a single platform and is single-handedly funding the development of the internet itself in the developing world, it fits within the scale that Zuckerberg thinks at.Doubtless Google has its own A.I units and its own grand plans for search. But until recently, Microsoft's Bing and Yahoo's search engine had taken only small slices of Google's business. What Zuckerberg seems to be saying is that in 10 years time, Facebook will be capable of providing a research and question-and-answer service that will be better than "any" other search engine.
Here's some more detail on what Zuckerberg said:
The way that we're thinking about this, there's just so much content that people have shared on Facebook that simply building the infrastructure to index all of it and start ranking it is a multiyear effort, which we're making our way through.
... And I can speak to some of the deep learning work. In the last quarter, Yann Lecun, one of the really earliest folks, and one of the founders of deep learning, and a professor at NYU, joined us to lead our AI group. And this is a long term research group that we have. It's going to fit into our strategy over a longer, maybe five or ten year, period.
And their goal is really just to try to understand how everything on Facebook is connected, by understanding what the posts that people write mean, and the content that is in the photos and videos that people are sharing, help people with tasks like if you're sharing a voice clip in messenger, being able to transcribe that for people, so that they can receive it more easily.
So these are some pretty big tasks in AI that are things that we have teams that are working on that will need to be researched over time, and will have obvious implications for the products that we do, but over time the real value will be if we can understand the meaning of all the content that people are sharing, that we can just provide much more relevant experiences for people across everything that we do.
Internally, we talk about our strategy, and there's a three-year strategy, a five-year strategy, and a 10-year strategy. And the three-year plan is really all about building new kinds of experiences for sharing, like so many of the questions on this call have been about. The five-year approach is really mostly about helping people use their network to answer interesting questions or solve problems that they have.
And that's where all the Graph Search work and the open graph work and some of the early parts of the AI work that we're doing you're going to start to see over that period of time.