YUSUF MEHDI: All right. Good afternoon. I'm Yusuf Mehdi. I'm with the Bing Team at Microsoft. And I would like to welcome you all here, and thank you for joining us.
Today is a kind of special day for us. It's the start of what we think is a new chapter in the future of search, and social. And simply put, I think in our minds, what you're going to begin to see later today is the beginning of how search gets better because of your friends. And that's a very profound concept, and we're going to have some of the best experts in the field come talk to you and share some thoughts on that.
But we couldn't even begin to accomplish this or even describe the opportunity without our very special and important partner, which is Facebook. And in particular I want to welcome the Facebook team, and Mark Zuckerberg, who is honored to come talk to us today, and spend some time with us, and tell some of his thoughts about social and Facebook.
So, before we jump into it, I just want to kind of give you a little bit of thoughts of how the session will go. First up will be Qi Lu, who is the leader of Bing, and the president of our Online Services Division at Microsoft, and he's going to talk to you a little bit about how we see the evolution of search, and how we are taking Bing in a new direction to help people better accomplish tasks, and make decisions.
Then Mark will talk a little bit about his thoughts on the future of social and Facebook. And then we'll come up and show you some of the fantastic collaboration between the two engineering teams, some of the features that are going to start shipping immediately after this session. So, I'll show you some of those. And then, Dan Rose, who is head of the platforms business for Facebook will come up and talk a little bit about the opportunity ahead.
We'll finish up with a Q&A, so the four of us will do a short Q&A. And I think there are then box lunches, and we'll have representatives from both Bing and Facebook here to talk. And in particular we're honored to have the engineering teams from both teams. So, Bret Taylor's team from Facebook who did a lot of the work with us is here himself, and his team. So, welcome. And then Sean Suchter's team is here with the Silicon Valley campus who did a lot of the work on the Bing side for social engineers here as well. So, we're really happy to have engineers from both sides.
So, before I bring up Qi Lu, I want to start by saying how grateful we are about the partnership with Facebook. And how important that has been for us over the last couple of years. We started our collaboration back in 2006 approximately, and I forget exactly what the numbers are, but I think Facebook had something like 50 million users. Mark would know better than I would. But even then, we knew it was a special company, and we knew that they were going to change the world in ways that we could only imagine. And over the last couple of years, we have really done a lot of work together in a number of areas, and in particular we have really admired and been impressed about how Facebook has really pressed the Web to be more open, and to be more social. And in particular the fantastic sort of benefit we have of being able to connect with our friends, and our broader social network, and create a richness of that community is an amazing thing that touches over a half a trillion people in the world today.
Which brings us a little bit to today, which is one of the great things Facebook is now doing is allowing that richness to be brought to other areas, and in particular with us with Bing and with search. And I think that's really at the heart of the partnership, and what we have tried to do today, which is we share a common vision between Bing and the Facebook team, which is it isn't just about the connections of data that help search, or frankly help our worlds today in the offline world. It is about also the connection between people. And in many cases, it's those connections that actually help improve your daily life, whether you're doing some basic activities or not.
And, let's give some common examples, just to make it real for you. So, one will be, for example, in a couple of days I'm going to go out with my wife to see a movie, and I would like to know, hey, should we go see that movie Inception? Is it a movie that I'm going to like? That might be one. Otherwise, we're going to go to a wedding in South Carolina soon, and we're going to want to take a day for vacation, what will be the tourist sites that will be interesting to us? Or even now, we're in the middle of a remodel, where is the best place to buy furniture online?
Now, the answer those questions aren't necessarily going to come just from websites, or from expert reviews. They're going to come from people in many cases, and most importantly they're going to come from people who know something about you. And that's where the real value-add will come. And so, in those scenarios, you can imagine, for example, that my friends know that my wife, maybe she's a Leo DiCaprio fan, but she's not big on sci-fi. So, they might give me a different recommendation. Or in South Carolina, they may know, hey, they like some history, but the truth is they also just like to wander around the city, and just see the sights, and the local flavor. When it comes to buying furniture online, they may say, hey, well, we've been to their home, we know what your style is, here is a place to buy furniture that is your style that we recommend, and/or it has good deals.
So, it will be more than just links and more than just information that will change how with Bing we help people accomplish tasks. As I said, in that great Beatles song, it will be a little help from your friends. So, really we're energized. We have a lot of excitement about what's possible. We're going to show you some features, but it's only the start. There's really a great canvas of what we can do.
And with that I'd like to invite up Qi Lu to share some of his vision for where we're going with search. Please welcome Qi Lu.
QI LU: Thank you. Thanks.
Thanks, Yusuf, and thank you all for joining us today.
What I will do in the next few minutes is to share some thoughts, to tell you about how we think about search and how we will take Bing forward. And the purpose is to set the contest for today's announcement. We take fundamentally a broad holistic view about search. To us search at its very core, it's about understanding user intent and to bring information and knowledge to help fulfill the intent. And the intent here is defined broadly to meet the purpose of a user.
And it was that, as the fundamental end goal and the viewpoint, if you look at all the surface area for search, today's phone is a Qwerty box, you type in keyboards, you get a list of results to pick from. It doesn't have to be that way. In the future as the Web evolves the different forms of capturing user intent, surface information, surface knowledge can all change and evolve. By and large, in today's form of a search it serves as a gateway for users to access information, to discover what's available in the Web and to interact with all the digital experience that's in the consumer Web.
For Bing our aspiration is to go substantially beyond that. Our quest is to build the capabilities and experience in Bing to fundamentally enable our users to make more informed decisions faster to better accomplish their purposes.
So, that's fundamentally how we think about the product strategies going forward for Bing. And it is granted on how we think about the Web has evolved so far and how the Web is going forward, heading towards.
So, if you take a historical look back, think about the intellectual heritage of today's consumer Web, you can say it really comes from the way Tim Berners-Lee designed HTTP and HTML, because the purpose at that time was to publish hypertext documents. And the implication is found in that the prevailing structure of today's Web is fundamentally topical in nature, because think about what's the connective tissues that hooks all the billions of pages together. It's the hyperlinks. But, what's most salient in those hyperlinks are the short text segments, it's called anchor text.
It turns out that's among all things on the Web this is the most important, signals, the competition enables us to understand the topical relevance of sites in a portion of the Web pages. And that is why for today's search engine by and large you will get good experience for navigational queries if you're looking for sites. But, if you go beyond looking for a person, looking for a product, experience really varies. And by and large it's generally not very good, because today's topical structure of the Web isn't powerful enough, does not give the search engine enough predictive powers to map user queries towards what the user purpose really is.
At the same time, what is in the Web going far beyond a Web of documents, literally into a Web of the world. In addition to all the topical information that's available on the Web now we have digital representation of people, places, things, piece-by-piece our industry is literally building a digital society online. And as a result the structure of the Web is rapidly evolving, becoming richer, and there's more salient signals that will give us the ability to discern and predict.
And among all the forces in our view driving the forward progress of the Web, the most important is Facebook, it's 500 million strong and rapidly growing user base, and the billions and billions activities of people sharing information, using the platform, using the product, because by those sharing activities new structures are being created, and the structures are being enriched at the same time.
If you visualize today's Web, and you think about there's a topical graph which is today's Web that links that hold the Web together. Now, on the other side, you have a social graph, and you have products such as light buttons that starts to overlay a person and interest, topics, events, information, and it's just the beginning, and the structure will become richer and more meaningful. There's a rich source of salient signals that the search engine can tap into, have more predictive powers, and enable our users not only to find, discover access to information, but make better decisions, complete tasks more efficiently.
And that's the central theme about today's announcement, because with a partnership of Facebook and Bing, we will be able to harness and unlock the tremendous potentials of social, the Facebook platforms, and start a journey taking today's search experience to the next level.
Yusuf later on will talk about some of the early product concepts and show some demos. But this is really the beginning, the initial steps for how people and a social relationship among people can truly become first class citizens of a search experience. In many ways, you can say this is the unfolding of a new era of better search with people. The surface area is so rich. The opportunity for innovation for unlocking consumer value is tremendous.
At the very high level, if you think about it, there are a few dimensions just even to visualize. The first dimension is, we're going to be able to create search capabilities that just weren't largely available. For example, a classic search problem a lot of people struggled with for many, many years is name queries. Today's Web search engine, if you type in a name queries, the experience generally is not very good unless you're looking for public figures. But by putting Facebook and Bing working together, that experience will be dramatically better. You can just visualize, and Yusuf will show you examples.
And those are classic search problems, challenges, finding who knows what. Again, by Facebook and Bing working together, we will be in a position to tackle that kind of problem over time, and all those examples whereby substantial consumer value and economic value can be unlocked by working together like that.
But, even more importantly is to make social people first class citizens of a search experience in their way, in a coherent way. When you go to Bing to search, you're bringing your social context, you bring the power of the Facebook platform with you. It's very simple to imagine how powerful that can be. One suggestion I have is, let's just, for each individual one of us, look at all the queries let's say in the 30 days you typed into a search engine, and after each of them ask the simple question, for this query, can the experience be much, much better if Facebook, my social content, all the people that I trust, is part of the experience? And you will find out, a substantial amount of queries, the experience will be much, much better, because we, as a human, when we try to collect the information and make decisions, typically there are several sources of information we need to assess, popular opinions, expert opinions, and trusted opinions.
Today's Web, by and large, gives you access to popular opinions, some expert opinions, but in many, many questions that we try to answer, decisions we try to make, trusted opinions, the viewpoints, the information, the knowledge for the people we know, we trust, if this information is available, the experience will be substantially better.
But all this is just the beginning. The opportunity to innovate, the opportunity to build a consumer experienced by Facebook and Bing working together, we are genuinely excited about the opportunities ahead.
I want to echo Yusuf's comment just to use this opportunity to offer my sincere thanks to Mark, the Facebook leadership team, for your vision, your leadership and support for this partnership. My heartfelt thanks to the Facebook engineering team, and certainly also Sean's team at Microsoft. There is so much work these teams have been cranking as of late. I just met with Dilshad (ph), he's also from CMU. So, it's great to see him over there. So, really, really grateful for the Facebook engineering team for all your hard work and effort that goes into this.
And with that, let me turn it over to Mark, and he will share some of his thoughts about this partnership.
MARK ZUCKERBERG: Thank you.
So, I'm only going to talk for a few minutes. I'm really honored to be here. This is, I think, one of the most exciting partnerships we've done on platform so far. So, I just want to talk for a minute about this principle that we have of designing products around people, and designing them to be social.
And then, for a second, I want to talk about why I think social search is going to be really important, and why this partnership between Facebook and Microsoft is, I think, the right one to drive that innovation forward.
So, I just want to go back a few years to when we were getting started. I started Facebook in my dorm room at college. I've studied psychology, and computer science. I was really interested when we were getting started not in just kind of helping people share any kind of information, or organizing any kind of information, but information around people.
And from studying psychology, I knew that a huge amount of people's brains is focused entirely on processing information about people. So, a lot of our visual cortex is designed to process faces, and kind of understand emotions, and expressions that people make. And we're just hardwired to make it so that information about people is by far the most interesting kind of information that we track in the world. So, that was how Facebook started off, making it so people could share a bit about themselves.
Then we got into building other applications, and early on we built a couple of things like a very early Photos app with just two or three engineers decided that they wanted to do this, and a very early Groups app. Again, a team of two or three engineers threw that together. And what a lot of these apps had in common was that there were whole companies that build photos apps, or groups apps, and with only two or three people early on, we didn't have the resources to build the most feature rich applications in either of these areas.
Over the last five years, we've filled them out, but early on these products only really had one major advantage, which was that they were extremely, deeply socially integrated. And these were the only products that were on the Web that understood the fundamental importance of designing products around people. The photos product was designed so that when you shared photos, you weren't just posting photos of anything, you were tagging photos, so that you can go to someone's profile and see all the photos of a person, which is really what you want to see a lot of the time anyway.
So, with just those social features, the Photos product now is four or five times more used than I think every other photos product on the Web. And over the last five years we've worked on kind of filling out the features on it. So now, we have high resolution photos, and all the other features that you would expect in other photos apps. And last week we announced a new version of Groups, which also in its very simple form was the most used groups application on the Web.
But we decided in 2007 that what we wanted to do was focus on building a platform. We decided we could keep on building out these small versions of these apps that were very deeply socially integrated, or we could built a platform so that anyone, all these companies that are just experts in their area, can integrate social products, and design their products around people. So that way you get both people who are experts in an area, and good social integration. And that's giving you the best of both worlds.
And so far we've just seen excellent results. Right. First we had this Canvas platform within Facebook.com, and we have more than a million developers building on that. And we launched Connect a couple of years later, and now we have more than million sites that are using Connect, and that's great.
And what we've really focused on now is working with companies that we think are going to be really scrappy, right, who are underdogs, who are incentivized to really go deep and try to do something innovative that other companies aren't doing. So, one of the first areas that we saw this in was games, right. There were these great big games companies that have existed for a while, and then are these companies that start off really small on Facebook platform, like Zynga, and Playfish, and Playdom. And using these social dynamics to design their products around people, they built some of the fastest growing companies in the technology industry today, and much faster growing than any incumbent game companies with a much smaller number of people. So, I think that's one of the most exciting things that's going on.
A lot of people look at this and they're like, OK, so Facebook, there have been social games that have been built, but what else is next? And when we look at what's going to happen over the next five years our view is that there have been many great platforms that have been built in the technology industry and games are usually an early adopter. So, kind of a leading indicator of what's going to be possible.
You saw that on iPhone. You even saw that on PCs early on. My personal theory on this is that people who build platforms are so focused on utility that they kind of overlook the opportunity to build really fun things like games. But, our view is that this is clearly not just going to be about games. There are so many applications that we've built with Facebook and that other people are doing that are cool that our view is that over the next five years we expect that almost every industry is going to be disrupted by someone building a great product that's deep in whatever area that industry is, plus is extremely socially integrated.
And one of the most important things on the Web is search, right. So, we started thinking about what would social search look like, and who would be the best partner for us to work with. And we looked around, had a lot of conversations and we have this great partnership with Microsoft dating back to 2006 when Yusuf was talking about this, we actually had 7 million users at the time. So you guys got us very early. And we're really thankful for the partnership, because it's been one of the things that's really helped us grow and it's been great over the years. And we've worked on ads together, we've worked on search, on Facebook, we've worked on maps, we're working on a bunch of other stuff that will be exciting when we can talk about that.
And the thing that makes Microsoft a great partner for us is that they really are the underdog here, right. And because of that they're in a structural position where they're incentivized to just go all out and innovate, right. And when you're an incumbent in an area, no matter how smart there's just always this tension between trying to innovate and push new things and just trying to preserve what you have. And we've never felt that with Microsoft, right, because in all the things that we've with Microsoft on they have all these smart people, but they're just trying to rapidly gain share by doing awesome stuff that no one else has talked about doing before.
So, that's actually made Microsoft a really good ally for us and a really good partner in a lot of these complex areas that we have no interest in building things out around, but are really complex like search. So, we've seen good results on ads. Search on Facebook, the map stuff is going well so far. And I don't know I just couldn't think of anyone better to be working with to build the next generation of search and I have no doubt that a great social integration in search will do for search what social integrations have done for games and photos and groups and events and all these other things before it. So, as you guys said, this is just the beginning. So, we're launching a few exciting things today, but there's going to be a lot more to come over time and it's really an honor to be working with you guys. So, thank you.
YUSUF MEHDI: OK. So, I want to show you then a couple of the things that we are shipping shortly today and then we can talk a little bit more about what's going to come. So, the first thing I want to do is I'm going to start before I show you the new things, I want to show you a little bit about what we have done with Facebook in the past as Mark referenced it, just to give you some sort of context of how we are tackling this issue of social search. So, let me start first on just the thing that we did initially with Facebook, which is just enabling great Web search within the Facebook experience.
So, for example, here I am in my account and I can come in and type something like Toyota Prius, and what I would get is I would get information about some fan pages, for example. I can see posts from my friends, or I can see now Web results right within the Facebook experience and if I click on those what I get is through our Bing iFrame implementation, just the native Web search experience right into the Facebook. This is the full fidelity Bing offering that we offer ourselves and we offer to our partner Facebook and you get some nice things. So, for example here we have this great e-card, or instant answer on the Prius, which has a bunch of nice information, photos, reviews, et cetera.
So, that was the first step we did was to really integrate and make Web search a native, seamless thing inside of Facebook. And as we've learned is that has been good for some set of users, but clearly people don't go to Facebook to do Web search, they go to Facebook to interact with their friends and their social community. So, it was a start, but there was an opportunity to do more. The second phase came about when we said, well, what's the next opportunity in social search, which is how do I find out what's going on with not only my community, but maybe what's going on with the community at large.
Today among those 500 million users it might surprise you that over 25 billion pieces of content get created or propped up everyday. And if you're interested to find out, hey, what's going on, what is the pulse there, because that content is not really available on the Web. That's all in a whole separate place. So, what we did is we said, let's combine together and make that happen.
So, we built a dedicated vertical called Bing Social. And this has, for example, the hottest topics that are trending, based on what people are talking about on Facebook and you can see here, for example, this latest episode about the miners. There's a lot of news about the miners, or there's something about The Situation, the guy from the Jersey Shore apparently got kicked off Dancing With The Stars.
So, this is the hottest thing that's going on. And if I want to come in, though, and I'll just stick with this Toyota Prius example and I say, let me go ahead and take a look at Toyota Prius, I can see the public updates, the status updates and what people are saying about Toyota Prius on Facebook, or even more powerfully I can see the shared links. These are the links that are being shared on Facebook on that topic. So, I can come in, for example and see, hey, one of the top things is about this new ad that's being put out there by Prius, or there's some stuff about the latest versions, or there's even some buzz about Google and robot cars.
So, you can go and see what's going on through the power of Facebook and the collaboration we did. So, this was sort of, if you will, the second phase of how we start to tackle search. But, then it brings us to today. And so really the opportunity we said, and I think it's more of aligned, really the powerful thing is, how do we bring the social context into the main search experience in Bing and Web search, because that's where we can also add some value. And to this point we hadn't really done that.
So, now what we do is we bring that in to help you accomplish some of those scenarios that I talked about up front. So, for example, sticking with the example I'll say, I'm looking for the Toyota Prius and when I come in and look this is what we provided before. So, this is the before, if you will in Bing. So, this is the answer we do for Prius. And we provide a bunch of information about you that helps you make decisions, because we know when people type this in they're looking to research to buy a car more often than not.
So, we'll have reviews and photos, similar models, specs, et cetera, with one click I can see news images. That's all-powerful. But, oftentimes what I really want to know is how what do my friends think about that, what reviews do they like? Do they like the car? So, here's what it will look like after. So, what we have now is we've included here a module. So, this is the module here that is the Facebook module that we've collaborated. And we show some links.
Sometimes it will be one link, sometimes it will be two links, and what we've got here is we have now the likes to articles that have been written about this subject Toyota Prius from my friends, from my social network. So, you can see for example here Paul, Adam, and Linda and a number of other people like this particular link, which is about the comprehensive coverage on the Toyota Prius.
Then there's a second story around Auto Week that a few others have liked. And so what's going on here is a couple of interesting things. And there's a lot going on behind the scenes. The first is that we have now actually brought the likes of your social group and brought extra information into the SERP that you just can't find anywhere else, that will help make this experience better.
The second thing we've done is we've actually relevancy ranked this module. So, sometimes it will show up here, sometimes bottom of the page, sometimes it won't show up at all. And so we're doing a bunch of ranking to make sure that we're bringing the right information at the right time. And we'll know that in this case if you're looking to buy a car that it might be important to have these types of likes show up in this place. So, that's the beginning of a lot of hard work.
And lastly, in case it's sort of lost, is this is particular for me, but each one of you may have a completely different search experience, which again, search has been one-size-fits-all up until this point. Now, what you're seeing is basically more personalized search results for you. So, those three things are profoundly going to change the way that we use search.
Let me show you this in a couple of examples. So, this is one example, let me give another one. Let's say I'm looking for a steakhouse in San Francisco. This is what we normally provide in Bing, which is a list of steakhouses. We rank based on reviews, based on proximity, based on a bunch of different signals. But, now by adding the module here with the likes of your friends, we can actually bring a new sort of metaphor. What you'll see here in this case, Alexander Steakhouse is one that didn't make the top 5. But, for me it's actually potentially more relevant, because my social network liked it.
So, in this case it's a review of an open table that people liked and this restaurant opened about a month ago. So, it could be that this is a newer restaurant that just opened up. People haven't had time to review it, but all I need to know is that four of my best foodie friends love it and it's on my list. OK. So, that is a way that you get relevancy.
Another example is we talked about that movie. So, there's a great one coming out, "Waiting for ‘Superman.’" And one of the things we do with Bing is we help you accomplish decisions and tasks. So, here as soon as you type that in, we know, hey, that's a movie. You probably want to go see a movie. So, we'll bring up the movie theaters, the times. With one click you can get to the map. We'll show you the reviews of that movie. That's all stuff we do today. But, now we have the Facebook module here, we're able to add like. So, in this case I can see, for example, a link to the IMDb that summarizes about the movie. And I have, again, some of my friends who have liked that. And again, here what's powerful is not only do I know the people who I trust, who I've seen, hey, this person he and I, we've probably seen 30 movies together. He or she knows if I'm going to like this movie. It doesn't matter what the experts say. And the fact that they've seen it and like something: a) it's helpful; or, b) with one click I can click and I can actually say, oh, Todd has seen that movie. Let me just call Todd up and find out, or ping him on Facebook and say, hey, what did you think of that movie? Was it really good? Should I go see it? So, another example of how we move the bar forward.
And then, a final example is, helping you cut through the clutter. So, one of the exciting things about the Web is there are a lot of videos. And we had this funny video, we did this promotion with Stephen Colbert with Bing, where he made some fun of us, but it was a great promotion. And I liked the video. Now, the thing is, to try and go find that video, a lot of times you'll type in "Stephen Colbert Bing video" and you'll get a bunch of them. But this will save us a bunch of time, because I'll have a bunch of friends who say, "this is the one I like. Yusuf, you should go check out that Colbert video."
Now, when I come to Bing, and I just type in the name of the video, if my friends have liked it, immediately I get the exact one I want on the page. I don't need to go and sort of navigate through all the different versions, and flavors that people sort of mashup on YouTube. I can get right to the one I want. So, there really are several examples of how the entire search, the core main search results page, of which we do billions every month, will change radically with the help of your friends.
OK, so that was one very big scenario. The other scenario that Qi and Mark both talked about is people search. Today, 4 percent of all the queries are people search. That's basically over a billion queries every month of people looking for other people. And the satisfaction on those is very low. Only 20 percent of the time are people satisfied for some of the reasons we've discussed.
What we've done is, we've collaborated to make a better experience. So, for example, I can come in and type in, let's say I'm looking for a colleague that I used to know from school, or from a while back, and I'm looking for him. If I type in Brian Lee, what we'll get today before the new experience is this page, which is I'll get links to some Brian Lees and some photos. How we rank this thing is based on popularity? It's really more about who is popular.
For example, here is Ottawa Senator defenseman Brian Lee, he's actually the popular one that shows up. And we have some great stats for him. And that's great if I'm looking for him. But, if I'm looking for that person, as we talked about, that's not easily done. No search engine does that today.
What we plan to do now, as of this afternoon, is start to bring in some of the Facebook profiles right into the search results. And here what's great is, we can bring in the photos. Now, how we do this is also impressive, so what we've done is, working with Facebook, we'd use a number of signals. So, we know social proximity, first of all, which is these Brian Lees are somehow related to me, either because I'm part of their network, or they're a friend of a friend. And I'm able to surface these up, and then we try and find who are the most relevant.
So, for example, in this scenario, I've got two Brian Lees, one maybe from the Microsoft network, and one maybe just from my Seattle network. And in this case, it says one friend, but this one has five. This might be the one I was looking for. Maybe I was at a party, and I met him through a bunch of friends, but I didn't remember his name. I didn't take my time to get his e-mail. I can come in and see if this is the right person by basically saying, hey, let me click on mutual friends, and see who are those mutual friends. And I can see, OK, great, Alex, Bill, John, these are the people I know about. So, that is the guy I'm looking for. And then with one click I can either just click and send a message right from this page to reach out to him and say, hey, it was great talking to you, we should get together, or I can click add as a friend, and then I'll go into the Facebook experience, and then I can effectively add him as a friend.
And so, that capability, which is for many people a very powerful and needed capability, we're now going to open up with that. And so those two classes of queries are things that we're going to start to flight starting in about probably 15 minutes, if the code is ready, and those will start to go out, and you'll start to see how those will work.
I'm going to show you two final things to wrap-up. One is, I showed you how easy all this works, but I do want to go back and talk a little bit about how we enable this in a powerful way for you to control your data, because we all know that privacy is a super-important topic, and we want to make that really good for people. So, what I want to show you is, this is the experience of what happens before any of that occurs.
So, now just kind of wind back to the beginning. As I'm coming to Bing, my very first time ever, and I have either a Facebook session that's currently live, or I have a Facebook cookie that's currently live. And what we'll do is, we pop up here in the upper left corner a popup that says, hi, Yusuf, Bing got better with your Facebook friends. And I can either choose to click to learn more, or I can think no thanks, in which case it will just turn the feature off. And, most importantly, as you can see here, we have not provided any of that capability that I showed you.
So, it isn't like you show up and that has occurred. None of that occurs, and if you don't want it to occur, you just click no thanks. So, we make that very simple. That popup will happen the next four times, and then, at any given time, through the powerful controls that Facebook has on privacy, you can always go in and delete Bing as an application if you want, so that you never get access to those settings. So, we make it very easy, and very powerful for you to control your data.
And then lastly, I want to show you two things of what's to come, because the two things I showed you we're shipping now. But, as we've talked about, there's really many places to go. One is sort of a near-term thing, which is what we'd like to do soon is be able to not only just have the Facebook module, but to be able to bringing the likes, and the faces of my friends who liked it anywhere in the links across any of the search results pages.
So, for example, here's an example of Iron Man, and I've got some on this link, and some here. So, you can see how we can just really have a different form of bringing relevancy. So, we're going to work to play and test that feature.
And then one that's further out is even how you make decisions and, as Qi talked about, how you start to know who are experts about certain topics. So, in the case where I'm looking for a restaurant, let's say a Meritage restaurant in Sonoma. One of the things we can do is, we can start to try and figure out who experts are on that topic by figuring out whether they've checked in at that location, or whether they've put up photos, and tagged things. And you can imagine we can start to build a better understanding of who is expert, and how that can help me make better decisions.
So, with that, that's just a little bit of a look of where we are, where we're going. I'm going to invite up Dan Rose to talk a little bit about the Facebook platform, and where we go from there, and then we'll take Q&A. (Applause.)
DAN ROSE: Thanks.
I wanted to share with you a little bit of the history of our relationship with Microsoft, because I think it helps in terms of context around what we're doing with them on Bing, and where we might go in the future with this partnership. In order to do this, we have to go back four years, and four years in the world of Facebook is a long time. The company is only six-and-a-half years old. So, winding the clock back four years, you get to a time in our company's history where we had 100 people. We were a small startup in downtown Palo Alto on University Avenue. As Mark said, we had seven million users. I remember that time well, because it was when I joined Facebook. And one of the first things that I was asked to do when I got there was think about what kind of partnerships we might want to enter into as a company to help us accelerate our growth, and grow our business.
And one of the areas that we originally identified as a potential partnership area that would really be good for us was advertising, a place where we had been experimenting early on, but hadn't yet felt like we reached out full potential, and thought we could potentially, with a partner, really accelerate that business. And so, in my first couple of weeks at the company, I started talking to potential partners around what we might do together in advertising, and in the very first conversation that we had with Microsoft, it became extremely clear to us that this was a company that would be a great partner for us for a couple of reasons.
No. 1, they were completely aligned with us at the highest level around what they wanted to do, and what they wanted us to do, and how we could help each other. And, No. 2, they were completely committed to working really fast to get things in place, and to help us achieve the things that we were trying to achieve.
And for a company that has 90,000 people working with a company that has 100 people, one of the important things to us was do we have a partner here that can operate at the speed that we were operating at at that time. That really came home to us when we started negotiating that contract. For the size and complexity of that first initial relationship with Microsoft, most companies would take at least a couple of months to work through all of the details in a contract. And in this case, from start to finish, the moment that we sat down to negotiate to the moment that the contract was signed, it was under a week.
And we knew in that experience that we had a partner that was going to be a great long-term partner for Facebook, because they understood the things that were important to us, and they were aligned with us on the vision that we were trying to achieve as a company, and how they could help us, and how we could help them.
Now, if you think about durable, deep, lasting relationships on the Internet, you'd be hard-pressed to find partnerships that have lasted as long as this one. Four years is an eternity on the Internet. And things change so quickly that it's almost impossible to predict exactly where the world is going to be one year out much less four years out.
And so the critical thing for a partnership like this to really work is that you have to be willing to evolve, and change, and flex, and move as things change, and flex, and move. And especially in a world where we were 100 people and seven million users on our way to a couple thousand people and 500 million users, for us to be able to work with somebody through that kind of enormous growth, we needed to have a partner that was going to be willing to kind of move with us. And that's exactly what Microsoft has done.
A year after that initial deal, we expanded the relationship, we made it global in scope, and Microsoft brought us even closer together by investing in Facebook. And shortly after that, Microsoft became one of the first big companies to really embrace the Facebook platform, and start integrating Facebook into some of their core services, like Windows Live Messenger, and Hotmail, and MSN.
In fact, in our very first FA conference, where Mark announced the Facebook platform, one of three companies that were on stage with us was Microsoft with Dan'l Lewin, who represented Microsoft's excitement to engage and participate in the Facebook platform.
Shortly after that we started doing things with Microsoft around Outlook. We've worked with them on Xbox and bringing social into the console games experience. As Mark mentioned, we've worked on Maps, and the Facebook Places experience on the Web is powered by Bing Maps.
From the very beginning four years ago one of the things that we talked about was could we some day potentially do something cool around Web search. At that time Bing didn't even exist as a brand. And we were as a company still too small to really contemplate what we might be able to do around Web search. We had that kernel of an idea and we thought there could be a world where maybe someday we could bring our assets together and really do something innovative around Web search.
And as Yusuf mentioned, a couple of years ago we started down that path when we integrated Bing results into Facebook.com and we started to show people what Web results when they were searching for non-people searches on Facebook. And over time what we realize is that ended up being one of the only pockets of the experience on Facebook where your experience wasn't social.
So, we started to talk to Bing about how can we bring more social context into that part of the Facebook experience, because everything else you do on Facebook has your friends with you, but when we show the Bing results it's the one place where you don't have your friends. And as we started to talk about and bounce ideas around how we could do that we realized that would be cool to do on Facebook.com and we should do it on Facebook.com. But, there's not that many people who search for things on Facebook.com where they're going to be relevant to them.
There's far more Facebook users who are searching for things on Bing and everyday more and more Facebook users are using Bing. So, we took that early idea and we extended it to Bing. And we said how can we do this in a way that's really seamless and frictionless, so we can touch the maximum number of users and give them the social experience with Web search.
And as Yusuf showed you we have implemented our instant personalization program here, a program that we've only done with a handful of select partners and we've implemented it with Bing to make this experience today extremely lightweight for a user and bring the context, the social context of their friends into their Web search experience.
Now, if you think back to four years ago that original partnership was just the beginning, and as you can see we've done a lot together since then. And I would say today is very much like that moment four years ago. This is just the beginning, the things that Yusuf showed you here today are just the first few things that we're able to do. In the same way that we're incredibly impressed with Microsoft's business team and being able to get that original partnership done in less than a week, we were equally impressed in this context with Microsoft's engineering team, and with Bing's team that works on the stuff, Sean Suchter and his group, and Satya's team, in terms of how quickly they were able to implement this.
We started this conversation with Bing less than two months ago, and their ability to execute an incredibly complex implementation on top of an already incredibly complex system was really impressive to us and the kind of light speed operation that we have some to appreciate about Microsoft generally, and in this case in particular the Bing team. And so, we're excited to work with them to continue to innovate and iterate, add more and more use cases, add more and more coverage, so that more and more of your experience when you're on Bing can be social. You can bring your friends with you and we'll together over the next four years get to see what we can do together next.
So, with that, I'll invite these guys back up and we'll be happy to take questions for a few minutes. (Applause.)
YUSUF MEHDI: Okay. I think Terry has some microphones if you guys if you all have any particular questions, or you can just shout them out. They have questions on their minds, I guess.
QUESTION: Harry McCracken with Technologizer. This looks really cool, but I wonder whether at some point it's a little bit less about a module and more about just sort of being everywhere as one of the factors that's always taken into account on Bing?
YUSUF MEHDI: I think the implementation we have here is a start, and I think we're going to get a lot of feedback to see how that works and as I showed you, they're one of the future things, as we're thinking about bringing the likes to all of the results that are on the page. But, over time one of the things I think you've seen with Bing is, we're really not just about the blue links any more. We have a much different visual approach. So, in the cases of visual search, or on mapping, or in a number of other areas, the whole concept of links goes away and it's a whole different visual experience. And so I think it's possible that we could do a number of different implementations.
MARK ZUCKERBERG: I mean, you showed this third one today, which is not something that's going live today, right. But, this is one of the things that we're already working on for the future. And I think one of the things that people often don't think about when they think about social context is it's not only a feature that goes into the algorithm for determining what to show, right, to decide what is going to be the best thing to show. Actually, the existence of that context in showing it influences how good the content is that you show, right.
So, it's a bit different than most other signals that you'd put into a machine learning system and I think we'll explore that, because I think that will be valuable for that, too. But, just seeing your friends' faces and their names, we're hard wired to have that influence us in a different way. So, I think one of the things that I've just been really impressed with Bing on is their willingness and ability to really innovate on interface. How much had the search interface changed in the last eight years before these guys came along. Not much, but I think a lot of the social integration that's going to happen is going to be in the interface.
QI LU: Yes, so I think that's a terrific comment from Mark and I will just add to it, if you look at Bing today the surface area is really expanding. We have this notion of instant answers, second page domain task pages, because we're focusing on making it easier for people to complete tasks, and these are the basic areas. There's more surface area that we're adding to the overall visual experience. But, even more important today is the appearance. Make those appearances coherent to enable the social awareness. There will be a lot more powers to generate engagement and the usage values and not consume experience values is really the holy grail of this partnership.
QUESTION: Jeff Fowler from the Wall Street Journal. I was wondering if you could walk us through some of the privacy implications of this. First of all, is it instant personalization, or do you actually have to go in and click allow to connect in with Facebook like you do with apps, and second of all can you tell us what exact information Bing is receiving from your Facebook profile and about your network and then what information is Facebook getting back from Bing, if any?
MARK ZUCKERBERG: Sure, so this is instant personalization. The same basic thing that we have rolled out with, I think that this is the you guys are the fifth partner, because one of our launch partners was also Microsoft. So, we have five companies that we've worked with on this two of the implementations are Microsoft. The way that instant personalization works, I mean, I think there's a lot of misconceptions around this that I think are good to clear up.
What instant personalization is is when you go to another site that's enabled, if you have instant personalization turned on then that site gets to see who you are, and gets no other information about you. Knowing who you are then allows it to query Facebook for any public information that you've shared. Right.
So, if you've decided that you want your musical interest to not be public, then the site can't get access to those. If you decide that you want your musical interests to be public then the site can access those. So, really Bing can see no other information about you that any other person out of the 500 million people who are using Facebook could see if they went to your profile, right.
So, from that perspective just because it's all public information about you, our perspective is that this is actually really good from a perspective of the type of information that you want to share and not sharing any more about that. So, I don't know if you guys want to go into more of the details around that, but this is basic instant personalization.
QI LU: Yes, precisely it's designed around the same principles as the instant personalization on the Facebook platform is being designed, and we are just one of the many partners to implement it in that spirit.
MARK ZUCKERBERG: But, I do think this is an important thing to make sure that's clear is people seem to have this notion that instant personalization is about you go to a site then all of a sudden Facebook sends all this information to that site. And that's just not true. The way it works is that you go to that site and then that site can ask for your public information.
DAN ROSE: Yes, one thing to add to that is Bing is not storing the user's specific information against that specific search. So, once you've returned the results you are not going to know that that user searched for that particular thing beyond that initial search itself.
QUESTION: Hi, Joe Menn with Financial Times. So, on the example you showed where it said, hi, Yusuf, we're going to tell you about we're going to tell you about the Facebook integration, and it said no thanks or learn more. Is it by default it's popping up, unless you hit no thanks is it first question. The second question is what percentage of Bing users, raw numbers if you'll share them, are actually signed in as who they are when they're using Bing and if it's a small number, as it is, I think, with a lot of search engines, how will your prompt them to sign in?
YUSUF MEHDI: A couple of different things, and feel free jump in, guys. The first question, Joe, is that popup shows up by default if you are logged into Facebook, or have an active Facebook cookie, the popup always shows. And, as I said before, it will show up, and none of that personalization has occurred on the first time that you come to it. So, if you click no thanks, you're taking out of it.
QUESTION: (Off mike.)
YUSUF MEHDI: If you're not logged on, then you won't even get those results. So, if you're not logged into Facebook, there's no Facebook.
QUESTION: (Off mike.)
YUSUF MEHDI: Yes, that's how it works.
MARK ZUCKERBERG: More and more sites aren't even creating their own account systems. I mean, Bing does, because Microsoft has one of the biggest in the world. But for the rest of instant personalization, a lot of the ways that a lot of sites are using this is to replace having to build their own. So, yes, fundamentally, if you're logged into Facebook, then you get a good social experience. We just view this as the future of how all these applications are going to work.
QUESTION: (Off mike.)
YUSUF MEHDI: Yes, so basically, as we said, we'll prompt you five times if you want to opt out or learn more. Then the feature is enabled, and you can turn it off at any given time by going into your settings and turning it off if you'd like.
QUESTION: Jason Kincaid, TechCrunch.
So, it seems with instant personalization that you guys have done before with Pandora, and Rotten Tomatoes, it's more entertainment, consumery stuff. With the search engine, people tend to search for things that can be more private. And I'm wondering, are you guys doing any kind of filtering as far as which queries you're going to show someone's friends. If I did a search for like a disease symptom, and I see a picture of my buddy who has recommended this story, that might kind of weird me out a little bit. Is this something you guys are thinking about?
MARK ZUCKERBERG: One thing that I think it's important to be clear on is, someone asked this question before, is Bing sending data back to Facebook. And, no. So, when you search on Bing that does not get shared with any of your friends. If you search for a disease, and then one of your friends has shared a piece of content about that disease in the past, then I don't think that will be creepy at all. They proactively chose to share that with all of their friends. So, having it come up in the context of when you're looking for it I think is exactly the optimal thing to show.
So, yes, our view is, we wanted to kind of start off, rollout the program in a very controlled way with only a handful of partners that we really trust are going to do good implementations, and we're still doing that, and we're still in that phase, but our view is that everything is going to be social eventually. One of the best ways to do that is to decrease the friction, right, so that way instead of having to go through some steps when you go to a site, everything is immediately social, again, only with information that's completely public about you anyway. So, if you choose to make any of this information not public, it wouldn't be a part of this. And I think that more and more applications are going to work this way in the future.
DAN ROSE: One way to think about this is the result that you see on Bing could look very similar to something that you might see in your news feed on Facebook, three of your friends have shared this particular link, or have liked this particular movie. That's a story that we see everyday in our news feeds on Facebook, and in this case you can get that when you're actually looking for that movie as opposed to the kind of serendipity experience that you get on Facebook because it gets pushed to you through the news feed.
QUESTION: Josh Constantine from Inside Facebook.
If I search for Brian Wilson, music, Beach Boys, and I'm clearly trying to find the musician, but I have a friend named Brian Wilson, where is it going to show the social results compared to the traditional search results?
QI LU: So, I can take that one. It depends on the specific implementation. It's an ongoing improvement process. Fundamentally, we need to understand the intent of the query. So, simply using a simple linguistic processing you can say what is the linguistic head? Are you looking for a person, looking for music? So, once you have ascertained about those, then you can say, from ranking algorithm standpoint, doesn't this make sense to bring up social context as a part of the search experience. So, it depends on how the algorithms understand the meaning of the query, or the intent of the query. That's No. 1.
The second is, the beauty of the search product is fundamentally data driven, meaning that you get feedback groups. So, we essentially have an ongoing, tuning learning process. So, as we trial and error to improve our capabilities incubating more and more social context into the Bing experience, we'll be very, very closely watching the improvement cycles to make sure that we do fundamentally improves and enhances the user experience.
QUESTION: Nick O'Neil, AllFacebook.com.
I was wondering, for all likes, so all like data is public, is that just a fact of life?
MARK ZUCKERBERG: No. You can change the privacy on that. And if you decide to make it so that it's not open to everyone, then it's not available for instant personalization. The basic way that the program works is, we have this idea which is basically 500 million people on Facebook, they can all go look you up, and see whatever is public about you in your profile. Given that, why shouldn't applications be able to do that in order to give you an awesome experience, too? So, that was kind of the birth of instant personalization.
But there are all these things that an application can ask you for if they want to access them that aren't public, but they have to ask. So, if they want the ability to access your e-mail address, certainly, which is never public by default, or any information that you've changed the privacy settings on, that ends up not being public. Or, if they want to be able to write on your wall, then they need to get specific permissions for that.
QUESTION: So, is there a way that likes for articles can actually be private then?
MARK ZUCKERBERG: Yes.
QUESTION: You can make that private? Yes.
QUESTION: Ben Parr of Mashable. Curious a little bit about if there is any talk or discussion about business models, revenue models? Was any money exchanged? Or is there any kind of financial incentive for Facebook especially in the future for this type of integration?
DAN ROSE: We're not sharing any of that information today. But, suffice it to say this is, as Mark said, an extension of a program that we've already rolled out with a handful of partners, and we'll be rolling out with more partners over time.
QUESTION: Brian Singel with Wired.com.
A quick question about how this relates to the recent change to Groups. So part of that change was to sort of actually have a lot of the conversations inside Facebook move out of purely public and to move them into groups that they were more relevant for. So, is that going to affect the kind of results you're actually going to see as more information moves from being publicly available to being inside silos inside Facebook?
And, secondly, in testing can you have a sense of what percentage of queries actually find relevant results from Facebook?
QI LU: So, I would take the second part. We have instead of testing what you can do is what we call a query server process to draw in large enough query samples just to look at, just to say for this query, if you have the type of Facebook social context, or the social context really has material for those queries. So, we have some ideas roughly how much percentage a query can be improved. I can say it's very, very substantial, and that's why I was so excited. That's No. 1.
The second is, today's query distribution is also in many ways the result of the experience. If the experience is not good, people are not going to keep trying. If we enable great experiences let's say for name queries, Yusuf mentioned it's 4 percent, and in our view is that the name queries will increase, because once you get great results, then people will search more. So, it's an ongoing evolving process.
By and large, this is just a tremendous potential ahead of us in the opportunity for innovations.
MARK ZUCKERBERG: On the Groups question, we're focused on enabling people to share with anyone who they want to share with. A bunch of people asked us last week, OK, so you're launching this group product for smaller group communication, does that mean that you're not as focused on making it so people can communicate one-on-one, or everyone? And the answer is no. We're trying to help people share with whoever they want to share with.
So, some things are just basic information that we think that people generally want to share broadly. They can change those settings if they want, but that's kind of the default mode there. Some things we think are kind of you help people share by giving them a small group. So, for example, I have this family group, and there are things that I share there with my family that I just wouldn't want to share with everyone else. So, having that group makes it so that we can share that. And then there are things that you want to share with one person at a time.
But a lot of what we focus on at Facebook is building this general social platform that enables you to share whatever you want with exactly the right people. If you don't have that, then often you just don't share it, and the information isn't used. And we're focused on building all of these different tools, and building blocks, and hopefully over time more and more of them will be used for building better social applications.
QUESTION: (Off mike.)
MARK ZUCKERBERG: There's no philosophical objection. Right now, we've made the decision for this product that we're just going to use public information. So, if they wanted to use information that wasn't public, then they would just have to ask users for permission for any of that. And when we launched Groups, we launched an API with it the same day, so that any application can get access to who is in your Groups, and do different kinds of things to show you different social context.
But search is so much about giving people a lightweight experience, and speed is so important that for at least this first version, we're just focused on not asking users for any extra information. There's a lot of information that people do want to share with everyone that we can use as context to make a large set of queries a lot more valuable, and that's what we're doing.
QUESTION: Gavin Clark from the Register.
Just a question for Microsoft, I wondered are you going to be looking at extending the People Search to other social networks like LinkedIn or Okra, and also then for Facebook, are you kind of contemplating some of these features for tying those into other search engines as well?
YUSUF MEHDI: The thing we're super-excited about is that Facebook is really the fantastic opportunity for us today, and all the work we've done here, I think, is where our time and focus is. And so that's what I would say is our focus.
And what was the second part of your question?
QUESTION: (Off mike.)
MARK ZUCKERBERG: So, I think that this question gets down to why Microsoft is such a good partner for us. We're trying to build a platform, so fundamentally this is not about working with just a single company. Over the long-term, we would love to work with everyone. But the question that I think you have to ask is, why is Microsoft the company that's being innovative and doing this before everyone else. And I think that that's fundamentally tied to the culture that they have now in their market position, and what they're trying to accomplish. And they see this as a huge opportunity. And if you look at a lot of the great companies that have done social integrations, they tend to not be the incumbents. They tend to be the disruptive force that's trying to do something different. And I think that there's just something really valuable to working with those folks, and those are the people that we seek out. And that's one of the reasons why I think that this is going to be such a good partnership for a while.
YUSUF MEHDI: So, with that, I'd like to thank you guys all.
There are box lunches, I think, outside of the room and, as I said, we'll have representatives from Bing, and the Facebook team available to answer some of your questions.
So, thank you very much.