David Kirkpatrick (left) and Cory Ondrejka (all photos by Asa Mathat)
Vice President, Engineering, Facebook
Founder and Editor-in-Chief, Techonomy
It’s called the Internet of Everything rather than of “things,” because it is increasingly apparent that we are part of an interlaced reality. Whether it’s a Fitbit, video memory augmentation, apps that connect our friends to our whereabouts in our car, or a panoply of other kinds of intersections, Facebook is central. It is the main connection mechanism for people. In this discussion, Facebook’s head of mobile, Cory Ondrejka, talks with Techonomy’s David Kirkpatrick about how the company sees its role. Read the full transcript below.
Kirkpatrick: I was backstage, earlier, talking to Cory Ondrejka, who is head of mobile for Facebook and has been very busy with Facebook Home and a lot of other things lately, and who I happened to know from his days when he was at Linden Lab Second Life, where he was head of technology and one of the real progenitors of that. So he has some interesting background in the virtual world. But he made the point, you know, we call it “Man, Machines and the Network,” but we haven’t talked about man very much yet, or people, which is actually probably what we should have had there. We weren’t trying to be sexist. But Facebook, which is a company that I wrote a book about, and, you know, passionately a believer in the importance of, and so are 1.1 billion people, so it’s not a minor service, is clearly important because it has given people more connectivity to other people and really more personal empowerment than any other technology aside from the mobile phone, I would argue. So from a software point of view, it’s a singularly significant people tool. So if we were talking more about the first word in our subtitle, what should we be thinking about, Cory?
Ondrejka: Well we’ve spent the last couple hours talking a lot about technology. It’s been fantastic to hear about the different ways we can approach sort of bolting intelligence onto things in a lot of different ways. We’ve been talking about networking, we’ve been talking about approaches that are sort of like backpacks, like we saw at the opening, where you can take an object that is built to be inanimate and stick technology on it. Certainly discussions about first cubic centimeter-sized technology and then cubic millimeter-sized technology, and I think that’s all incredibly empowering, because that enables us to go out and allow—you know, Marc Andreessen likes to talk about software eating the world, and this is definitely the next piece of it, because that technology carries software all over the place, to places software never to got to go before.
And so you have that first layer that’s basically the sensors and some compute. And you think about the next layer, depending on whether you have smarts there or you have smarts in the cloud, you either have the apps there and then the network or you have the network and then the apps or some combination thereof. But ultimately most of these are going to be about generating information for people. Because that’s what tends to make it interesting, that I know my door is open, or that maybe you know my door is open. And one thing we haven’t necessarily talked about a ton today is how do you think about where that information goes and who you want to share it with when? I think the Nest is actually a really good example of this. So if my parents had a Nest, I’d love to be able to monitor the temperature in their home. It would be really useful for me to know that, you know, probably that they’re still around, and it’s changing and adapting to them moving room to room; that’s a good sign that they’re still ambulatory and healthy and that’s great information for me. On the other hand, if the Nest was detecting temperature that indicated their home was on fire, I’d probably like their neighbors to know, the police to know, the fire department to know, and really anybody nearby to know. And those kinds of contextual sharing about who gets the information when is something that can be relatively difficult for computers to do. These tend to be relatively complex algorithms or algorithms that are easy to break, but they’re really easy things for us to make decisions about.
And so getting people back in the loop is something that is going to happen, and it’s a place where we actually have a lot of experience. So we run platforms both for Web apps and for mobile apps, and both of them will apply to the Internet of Things. Because if you have smarts down on your sensor, and we heard a lot about different networking technologies and different approaches, or are you networking locally or are you going out to the Web first, these are all problems we’ve solved a lot over the last forty years. So I hope we’re not inventing more things that we don’t need to invent, because routing packets around is something we actually understand pretty well right now. And so if you are routing up to the Web, well then you just look like a Web app, and so whether you’re a Web app running on Amazon web services, or you’re a little sensor on my doorbell, you’re still talking, speaking Web. Which means if you want to use Facebook as part of that, that’s already done, that platform already exists for you.
On mobile we look at it a little differently, where we think about: Can we help with identity? What services can we help with? Can we move information between multiple devices? Can we help synch information for you? So if you’ve got intelligence down on the device, again, that looks a lot like a phone, conceptually, where you have a decent amount of computing sitting somewhere. Again, we already have a platform that works for that, so I expect that people are probably already using us for these applications and I just don’t know about it yet.
Kirkpatrick: So in other words, going to the fire versus movement example you gave for your parents, you envision that there would be a version of a Nest app that would interact with Facebook where you could just check off boxes or something and instruct it, as a person with specific interests, that this information can be shared with others than me. I mean, Facebook has those kinds of controls already—
Ondrejka: Yeah, that’s already how we think about it. Apps that run on our platform, you already make decisions about can this application, for example, post on your behalf? Can it just post to you? Can it post to the world? Can it post to your friends? And so we understand privacy within our friends, ourselves, groups, fairly well, and so it makes a lot of sense to route information from the Internet of things in a similar way.
Kirkpatrick: Rob, I think in the first session, from Qualcomm was talking about something similar, the idea that with 2,000 sensor devices in your home, you might want to keep that pretty close to the vest, but some things you want to send up to the cloud and share. It’s the same set of ideas, right? Basically?
Ondrejka: Yeah, I mean, this is, again, very similar to how we route information today. And so you have on the one hand a sort of practical networking question about what is your connection up to the server infrastructure. And can you afford or do you want to be lighting up that network with the 2,000 things all around your house? There’s a separate question about whether for privacy reasons you want that data leaving your house, although, frankly, with strong encryption I think that’s actually a little bit of a red herring, because there are lots of ways to move that data around that if you trust it to purchase something on Amazon, you can probably trust it for your front doorbell. So there are ways to move data around and ways that are quite trustworthy. But for practical reasons you might have enough intelligence in your house that what you want to set up is some synthetic signal that’s more of an aggregation of what’s going on.
Kirkpatrick: Yeah, I mean I think it is something that I’m often in the position of explaining to people, as something of a Facebook explainer. So often people have come up to me and started fulminating about Facebook as a privacy invader, mostly on the basis of advertising concerns and other kinds of things. But what I often explain to them, even though there may be some legitimacy to some of their point, Facebook was the first system that ever had offered any privacy on the Internet, in the sense of control of who would see your information. And it still quite singular in the architecture and infrastructure of the Internet, that it’s a system for controlling who sees what. I’m just sort of amplifying what you already said; I’m not sure there’s a question in here, except you’re welcome to explain what I’m trying to say better. But it does make sense, what you’re saying to me, that Facebook could have a unique role, as we have more and more intelligent devices to control. Because there will be an extremely significant question of who gets what information, and Facebook is uniquely structured to provide that control.
Ondrejka: Well, I think, obviously privacy is incredibly important to us, incredibly important to the community of over a billion users who use Facebook. And so we do spend a lot of time thinking about how to make concepts of privacy easy to use for people. We were talking today about transaction costs. One of the challenges with giving controls over privacy is what is the cost of changing those controls? What is the cost of making a decision about privacy? And one of the stats that we actually watch pretty closely is how often people post with different privacy settings. So on Facebook you can post a status update about what you’re doing or where you are, and you can actually control the privacy setting for that post. And a reasonably large percentage of people will use multiple privacy settings during the course of a week. And that’s one of the metrics that we track to make sure that we are keeping those controls easy to use and straightforward and comprehensible. And so the natural extension of that is as you start having access to integrated objects that are around you, that know about you, that know about what you’re doing, where you’ve been, where you’re going, potentially how healthy you are, who you see, you’re probably going to want to make pretty affirmative decisions about the privacy of that data. And building tools to allow people to do that is frankly non-trivial from a user interface standpoint. I agree, I think we’ve been doing this for a long time, we’re continuing to push on it to make sure that we’re doing the best possible job that we can, and it definitely is not the norm for systems on the Internet that allow you to share. It’s far more common to have a “world or me” kind of privacy model—
Kirkpatrick: Yeah, it’s all or nothing.
Ondrejka: Or maybe even just “world,” right?
Kirkpatrick: Right. That’s very common. Are app developers coming to Facebook already, or developing apps that provide some version of an Internet of Everything control system, using Facebook? Or is that still to come?
Ondrejka: I think that’s still nascent. I am 100% convinced, based on the number of developers we have on our various platforms, that there’s somebody out there doing it, because it’s so straightforward to do that. I don’t think we’ve seen anybody achieve a level of sort of prominence with it yet, but we are still reasonably early on this. And I think we’re seeing a lot of competing approaches. The last panel, I think, did a really good job of splitting out the closed but functioning version, so the vertically integrated solution, versus the “well let’s start with a platform and let thousands of tinkerers go out there and do amazing things.” And the reality is we need both of those models to be out there experimenting, and as we start seeing them achieve more success, I think that’s when we’re going to start seeing more integration because it will make more sense. Because you’re now, if you’re tinkering and it’s just you and a few of your friends, you may be really comfortable just sharing it, because the same group of you are all playing with it together. It’s different once this is a core part of your daily life.
Kirkpatrick: Is there any foreseeable future scenario when we have all of these quote-unquote intelligent devices like door locks or, you know, cars or air conditioners, etcetera, that your door lock would have a Facebook page, or a profile, even more dramatically?
Ondrejka: Well it definitely wouldn’t have a profile since profiles are people. But I think that having a page comes down to what is the purpose of the object. I think that it’s certainly conceivable that if you are, say, in an event location and you use the Internet of Things to put up really good photos of the event, and you’re doing a lot of automation around that, and the participants in the event understand that part of being at the event is, “Hey, you’re photo’s going to get taken and posted to the Net,” you know, a Facebook page might be a good place to aggregate some of that data. And so I think that’s definitely possible. I think the front doorbell may not be a use case that needs that.
Kirkpatrick: But the Fitbit is already doing it, actually, now that I think of it.
Ondrejka: That’s exactly right.
Kirkpatrick: That’s an answer to the question I asked before. It actually does have—it’s a Facebook app, it allows you to share with some people or not others, and I have it in my pocket. It is a little bit of intelligence that monitors my sleep and my steps and my caloric exertion, etcetera.
So we’ve got a few minutes, can we get the house lights and anybody who has thoughts or questions for Cory, love to hear them. I certainly could throw more questions at him. Anybody have anything they’d like to ask or comment?
Mark Plakias: Yeah, Mark Plakias from Orange. Just wanted to pick up on this great premise and the idea that Facebook has invested more in sharing controls than other folks. So you’re obviously a great person to ask about something that would seem very common in this space of IoT but nobody’s talked about today, and that’s ephemerality. So I’m just wondering what your thoughts are on ephemerality.
Ondrejka: Well I think you’re going to end up with ephemerality in—wow, I got that out the first time, I’m surprised at that—
Kirkpatrick: Just define what you think he means and then we confirm with him that that’s what he means.
Ondrejka: So I assume what you mean is basically time-gated data.
Ondrejka: I think that there’s ephemerality at a few places along the tool chains that are going to emerge. One is you have 200 sensors in your home and you have some more centralized hub in your home that’s aggregating. It’s very likely that you’re going to be throwing away data there and sharing average data, aggregated data. It just makes a lot of sense to be doing that. You may not want all the individual data points. You then have that data stream going up. Maybe the service you’re using has data retention policies, and so those are going to apply to the service you use and the data you send to them. Beyond that, if you share it, now you get into the very complex question of, if you have shared data, are you sharing data in a way that you have some ability to hopefully indicate that you don’t want it to be around anymore? But of course, once you’ve shared the data that becomes a pretty complicated question.
So I think that we’re going to see, in a very similar way, frankly, to how people put most things up on the Net, is you do spend time thinking about what you’re going to put there, you think about who the audience is, and I think there’s some audiences where the infinite record of every time my front door’s opened, that’s fine. I don’t mind being able to see that data forever. My family, it’s no problem. Will I share that in a non-aggregated way outside there? Maybe not. Because I don’t want somebody to be able to study the patterns of when I’m home and when I’m not home. So I think that, as with any other data that you put on line or put out into the world, you certainly want to be thinking about what the policies are that are going to be attached to the data. And I think that goes back to: Are we talking about vertically integrated solutions? Are we talking about sort of platform solutions? Are we talking about further sharing that data through Facebook or other services? And sort of at each step along that tool chain, there are going to be both technological and practical reasons to toss data away, and there’s also going to be policy and retention rules. You know, is your service, say, in the E.U., in which case you’re going to tumble into a very different set of data retention laws than if it’s a U.S.-hosted service? All of that’s going to apply as soon as you start moving this data around. And I think that’s important to keep in mind.
Kirkpatrick: Facebook’s already been sort of a proof point of some of the risks of the measurement of certain things about your home. Specifically, it’s not autonomic, and it has nothing to do with sensing, but Facebook, especially early on, before people had any real clue about how to control their privacy settings, there were a number of incidents where people were robbed because they had shared they were going on vacation on Facebook. And so if that was being done automatically for you, you had better be in pretty good control, as you pointed out, of who sees that information. Anyway, anybody else? Peter.
Peter Platzer (NanoSatisfi): Yeah, just a follow-up on that. On Wall Street for example, company’s IT departments force deletion of records, because if they are available, then they have to furnish them, even if they aren’t by law required to have them. Can you see a future where Facebook has like a feature where I can say, “this piece of information I want you to forcefully destroy after x”?
Ondrejka: We generally don’t comment on anything in the future, feature-wise, so. And we do have data retention policies that are public for how we handle data.
Kirkpatrick: It is really hard on Facebook to make anything go away if it’s been shared at all.
Platzer: Let me re-phrase. I wasn’t actually talking about Facbeook, I think—
Kirkpatrick: In fact, it’s impossible so far, there’s no architectural control on Facebook today that allows you to delete—I mean, it’s shared, it’s public, it’s on somebody else’s—been visible to them, you can’t make it invisible to them.
Ondrejka: In general there are a great of controls, if we want to go into detail here in terms of your ability to update the privacy of things you do and how you share. That’s actually baked in, and that is live on the data going forward. But that’s a core piece of how Facebook is—
Kirkpatrick: But literally, once something has been shared, you can’t be certain that someone hasn’t copied it, even if it’s just like Snapchat with these things that automatically re-photograph and that kind of thing.
Ondrejka: No, if somebody pulls text into a text file, you can do a lot of things with text. But a core part of Facebook’s architecture is to have strict privacy controls throughout our system. And that’s actually baked all the way down into the—
Kirkpatrick: I mean, I don’t want to get into a deep debate, but some of the European regulators have tried to argue for the right to be forgotten, this pretty idiotic concept in my opinion. And Facebook argued in court, as I recall, or in some legal forum that it would be almost impossible to make someone forgotten if a lot of their information had been shared across large numbers of friends and followers on Facebook. Because, you know, you don’t want to get in the habit of removing something that someone might have commented on. I mean there’s a lot of intersections and layering of interactions that happen that are very hard to undo, is all I’m saying. And maybe it’s not an Internet of Everything point. Anybody else have something? Oh, in the back.
Bradshaw: Hi, Tim Bradshaw from The Financial Times. I was just curious, the fact that you’re here and talking about this. You’re sort of talking about it in a way that is sort of, “Oh, you know, these are some things that people might do on our platform or might already be doing on our platform.” I’m not sure to what extent you’re kind of signaling that this is something that Facebook is focusing on or something that you’re already enabling with the various kind of verbs that you allow people to do, like the Fitbit example that we gave. Is this something that you see as a big kind of opportunity for Facebook, or is this just a kind of another neat idea that developers can do and have their own fun?
Ondrejka: Well, I think that part of why all of you are here is that this is such an exciting idea. The thing is, when you start thinking about bolting intelligence onto stuff—now, I can’t speak for any of you, but I find that whole concept just fundamentally exciting. I spent a lot of time building virtual worlds where it’s actually easy, because you already have code running to simulate everything, and if you want to make something a little bit smarter, that’s already baked into the system. And so the idea that we can do that in the real world, especially as a software engineer, is fundamentally very exciting to me personally.
Kirkpatrick: You’re taking Second Life and making it back to real life or vice versa or something?
Ondrejka: Well, it’s doing everything that was easy to do in video games and virtual worlds, and being able to do them in the real world is a pretty profound idea. We’ve been talking about it for the last couple hours. In terms of Facebook, I think the realization, and the Fitbit’s a great example, that between Open Graph and platform we already enable a lot of the things that we’re sort of speculating about is something that, again, I find pretty interesting. What I’m looking forward to is watching how people use that and what they do with it. Is this a major area of focus for us? Well again, we don’t comment on unannounced stuff. What I focus on is phones and mobile devices; that’s what I spend approximately 23 hours a day thinking about. And in a lot of ways, the billion plus people carrying phones around were already on the cusp of this. Because we’ve already put an always on, always connected computer, even attached it to a thing, attached it to something far more interesting than stuff—a person.
Kirkpatrick: Which is what Gordon said, also, right? The same point.
Ondrejka: Exactly. And so, in some ways I think we already are in this space, based on being a platform for mobile applications.
Kirkpatrick: Unfortunately, we’ve got to wrap. And I would be able to talk to you at great length, and luckily we’ll have you again, I hope, at another event.
Ondrejka: I look forward to it.