[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.74,0:00:02.74,Default,,0000,0000,0000,,If you remember that first decade of the web,
Dialogue: 0,0:00:02.74,0:00:04.99,Default,,0000,0000,0000,,it was really a static place.
Dialogue: 0,0:00:04.99,0:00:07.24,Default,,0000,0000,0000,,You could go online, you could look at pages,
Dialogue: 0,0:00:07.24,0:00:09.75,Default,,0000,0000,0000,,and they were put up either by organizations
Dialogue: 0,0:00:09.75,0:00:11.27,Default,,0000,0000,0000,,who had teams to do it
Dialogue: 0,0:00:11.27,0:00:13.50,Default,,0000,0000,0000,,or by individuals who were really tech-savvy
Dialogue: 0,0:00:13.50,0:00:15.24,Default,,0000,0000,0000,,for the time.
Dialogue: 0,0:00:15.24,0:00:16.81,Default,,0000,0000,0000,,And with the rise of social media
Dialogue: 0,0:00:16.81,0:00:19.21,Default,,0000,0000,0000,,and social networks in the early 2000s,
Dialogue: 0,0:00:19.21,0:00:21.36,Default,,0000,0000,0000,,the web was completely changed
Dialogue: 0,0:00:21.36,0:00:24.97,Default,,0000,0000,0000,,to a place where now the vast majority of content
Dialogue: 0,0:00:24.97,0:00:28.28,Default,,0000,0000,0000,,we interact with is put up by average users,
Dialogue: 0,0:00:28.28,0:00:30.98,Default,,0000,0000,0000,,either in YouTube videos or blog posts
Dialogue: 0,0:00:30.98,0:00:34.29,Default,,0000,0000,0000,,or product reviews or social media postings.
Dialogue: 0,0:00:34.29,0:00:36.64,Default,,0000,0000,0000,,And it's also become a much more interactive place,
Dialogue: 0,0:00:36.64,0:00:39.27,Default,,0000,0000,0000,,where people are interacting with others,
Dialogue: 0,0:00:39.27,0:00:40.97,Default,,0000,0000,0000,,they're commenting, they're sharing,
Dialogue: 0,0:00:40.97,0:00:42.58,Default,,0000,0000,0000,,they're not just reading.
Dialogue: 0,0:00:42.58,0:00:44.45,Default,,0000,0000,0000,,So Facebook is not the only place you can do this,
Dialogue: 0,0:00:44.45,0:00:45.55,Default,,0000,0000,0000,,but it's the biggest,
Dialogue: 0,0:00:45.55,0:00:47.33,Default,,0000,0000,0000,,and it serves to illustrate the numbers.
Dialogue: 0,0:00:47.33,0:00:50.81,Default,,0000,0000,0000,,Facebook has 1.2 billion users per month.
Dialogue: 0,0:00:50.81,0:00:52.74,Default,,0000,0000,0000,,So half the Earth's Internet population
Dialogue: 0,0:00:52.74,0:00:54.39,Default,,0000,0000,0000,,is using Facebook.
Dialogue: 0,0:00:54.39,0:00:56.32,Default,,0000,0000,0000,,They are a site, along with others,
Dialogue: 0,0:00:56.32,0:00:59.54,Default,,0000,0000,0000,,that has allowed people to create an online persona
Dialogue: 0,0:00:59.54,0:01:01.32,Default,,0000,0000,0000,,with very little technical skill,
Dialogue: 0,0:01:01.32,0:01:03.80,Default,,0000,0000,0000,,and people responded by putting huge amounts
Dialogue: 0,0:01:03.80,0:01:05.78,Default,,0000,0000,0000,,of personal data online.
Dialogue: 0,0:01:05.78,0:01:08.33,Default,,0000,0000,0000,,So the result is that we have behavioral,
Dialogue: 0,0:01:08.33,0:01:10.31,Default,,0000,0000,0000,,preference, demographic data
Dialogue: 0,0:01:10.31,0:01:12.41,Default,,0000,0000,0000,,for hundreds of millions of people,
Dialogue: 0,0:01:12.41,0:01:14.44,Default,,0000,0000,0000,,which is unprecedented in history.
Dialogue: 0,0:01:14.44,0:01:17.00,Default,,0000,0000,0000,,And as a computer scientist, \Nwhat this means is that
Dialogue: 0,0:01:17.00,0:01:18.66,Default,,0000,0000,0000,,I've been able to build models
Dialogue: 0,0:01:18.66,0:01:20.99,Default,,0000,0000,0000,,that can predict all sorts of hidden attributes
Dialogue: 0,0:01:20.99,0:01:23.27,Default,,0000,0000,0000,,for all of you that you don't even know
Dialogue: 0,0:01:23.27,0:01:25.47,Default,,0000,0000,0000,,you're sharing information about.
Dialogue: 0,0:01:25.47,0:01:27.85,Default,,0000,0000,0000,,As scientists, we use that to help
Dialogue: 0,0:01:27.85,0:01:29.97,Default,,0000,0000,0000,,the way people interact online,
Dialogue: 0,0:01:29.97,0:01:32.47,Default,,0000,0000,0000,,but there's less altruistic applications,
Dialogue: 0,0:01:32.47,0:01:34.85,Default,,0000,0000,0000,,and there's a problem in that users don't really
Dialogue: 0,0:01:34.85,0:01:37.32,Default,,0000,0000,0000,,understand these techniques and how they work,
Dialogue: 0,0:01:37.32,0:01:40.45,Default,,0000,0000,0000,,and even if they did, they don't\Nhave a lot of control over it.
Dialogue: 0,0:01:40.45,0:01:41.94,Default,,0000,0000,0000,,So what I want to talk to you about today
Dialogue: 0,0:01:41.94,0:01:44.64,Default,,0000,0000,0000,,is some of these things that we're able to do,
Dialogue: 0,0:01:44.64,0:01:47.40,Default,,0000,0000,0000,,and then give us some ideas\Nof how we might go forward
Dialogue: 0,0:01:47.40,0:01:50.17,Default,,0000,0000,0000,,to move some control back into the hands of users.
Dialogue: 0,0:01:50.17,0:01:51.76,Default,,0000,0000,0000,,So this is Target, the company.
Dialogue: 0,0:01:51.76,0:01:53.08,Default,,0000,0000,0000,,I didn't just put that logo
Dialogue: 0,0:01:53.08,0:01:55.25,Default,,0000,0000,0000,,on this poor, pregnant woman's belly.
Dialogue: 0,0:01:55.25,0:01:57.09,Default,,0000,0000,0000,,You may have seen this anecdote that was printed
Dialogue: 0,0:01:57.09,0:01:59.15,Default,,0000,0000,0000,,in Forbes magazine where Target
Dialogue: 0,0:01:59.15,0:02:01.51,Default,,0000,0000,0000,,sent a flyer to this 15-year-old girl
Dialogue: 0,0:02:01.51,0:02:03.22,Default,,0000,0000,0000,,with advertisements and coupons
Dialogue: 0,0:02:03.22,0:02:05.78,Default,,0000,0000,0000,,for baby bottles and diapers and cribs
Dialogue: 0,0:02:05.78,0:02:07.46,Default,,0000,0000,0000,,two weeks before she told her parents
Dialogue: 0,0:02:07.46,0:02:09.32,Default,,0000,0000,0000,,that she was pregnant.
Dialogue: 0,0:02:09.32,0:02:12.03,Default,,0000,0000,0000,,Yeah, the dad was really upset.
Dialogue: 0,0:02:12.03,0:02:13.74,Default,,0000,0000,0000,,He said, "How did Target figure out
Dialogue: 0,0:02:13.74,0:02:15.57,Default,,0000,0000,0000,,that this high school girl was pregnant
Dialogue: 0,0:02:15.57,0:02:17.53,Default,,0000,0000,0000,,before she told her parents?"
Dialogue: 0,0:02:17.53,0:02:20.15,Default,,0000,0000,0000,,It turns out that they have the purchase history
Dialogue: 0,0:02:20.15,0:02:22.45,Default,,0000,0000,0000,,for hundreds of thousands of customers
Dialogue: 0,0:02:22.45,0:02:25.18,Default,,0000,0000,0000,,and they compute what they \Ncall a pregnancy score,
Dialogue: 0,0:02:25.18,0:02:27.51,Default,,0000,0000,0000,,which is not just whether or \Nnot a woman's pregnant,
Dialogue: 0,0:02:27.51,0:02:29.24,Default,,0000,0000,0000,,but what her due date is.
Dialogue: 0,0:02:29.24,0:02:30.55,Default,,0000,0000,0000,,And they compute that
Dialogue: 0,0:02:30.55,0:02:32.31,Default,,0000,0000,0000,,not by looking at the obvious things,
Dialogue: 0,0:02:32.31,0:02:34.83,Default,,0000,0000,0000,,like, she's buying a crib or baby clothes,
Dialogue: 0,0:02:34.83,0:02:37.77,Default,,0000,0000,0000,,but things like, she bought more vitamins
Dialogue: 0,0:02:37.77,0:02:39.49,Default,,0000,0000,0000,,than she normally had,
Dialogue: 0,0:02:39.49,0:02:40.95,Default,,0000,0000,0000,,or she bought a handbag
Dialogue: 0,0:02:40.95,0:02:42.66,Default,,0000,0000,0000,,that's big enough to hold diapers.
Dialogue: 0,0:02:42.66,0:02:44.57,Default,,0000,0000,0000,,And by themselves, those purchases don't seem
Dialogue: 0,0:02:44.57,0:02:47.04,Default,,0000,0000,0000,,like they might reveal a lot,
Dialogue: 0,0:02:47.04,0:02:49.02,Default,,0000,0000,0000,,but it's a pattern of behavior that,
Dialogue: 0,0:02:49.02,0:02:52.14,Default,,0000,0000,0000,,when you take it in the context \Nof thousands of other people,
Dialogue: 0,0:02:52.14,0:02:54.89,Default,,0000,0000,0000,,starts to actually reveal some insights.
Dialogue: 0,0:02:54.89,0:02:56.68,Default,,0000,0000,0000,,So that's the kind of thing that we do
Dialogue: 0,0:02:56.68,0:02:59.25,Default,,0000,0000,0000,,when we're predicting stuff\Nabout you on social media.
Dialogue: 0,0:02:59.25,0:03:02.05,Default,,0000,0000,0000,,We're looking for little\Npatterns of behavior that,
Dialogue: 0,0:03:02.05,0:03:04.73,Default,,0000,0000,0000,,when you detect them among millions of people,
Dialogue: 0,0:03:04.73,0:03:07.44,Default,,0000,0000,0000,,lets us find out all kinds of things.
Dialogue: 0,0:03:07.44,0:03:09.18,Default,,0000,0000,0000,,So in my lab and with colleagues,
Dialogue: 0,0:03:09.18,0:03:10.96,Default,,0000,0000,0000,,we've developed mechanisms where we can
Dialogue: 0,0:03:10.96,0:03:12.52,Default,,0000,0000,0000,,quite accurately predict things
Dialogue: 0,0:03:12.52,0:03:14.24,Default,,0000,0000,0000,,like your political preference,
Dialogue: 0,0:03:14.24,0:03:17.100,Default,,0000,0000,0000,,your personality score, gender, sexual orientation,
Dialogue: 0,0:03:17.100,0:03:20.87,Default,,0000,0000,0000,,religion, age, intelligence,
Dialogue: 0,0:03:20.87,0:03:22.26,Default,,0000,0000,0000,,along with things like
Dialogue: 0,0:03:22.26,0:03:24.20,Default,,0000,0000,0000,,how much you trust the people you know
Dialogue: 0,0:03:24.20,0:03:26.00,Default,,0000,0000,0000,,and how strong those relationships are.
Dialogue: 0,0:03:26.00,0:03:27.79,Default,,0000,0000,0000,,We can do all of this really well.
Dialogue: 0,0:03:27.79,0:03:29.99,Default,,0000,0000,0000,,And again, it doesn't come from what you might
Dialogue: 0,0:03:29.99,0:03:32.09,Default,,0000,0000,0000,,think of as obvious information.
Dialogue: 0,0:03:32.09,0:03:34.37,Default,,0000,0000,0000,,So my favorite example is from this study
Dialogue: 0,0:03:34.37,0:03:35.61,Default,,0000,0000,0000,,that was published this year
Dialogue: 0,0:03:35.61,0:03:37.40,Default,,0000,0000,0000,,in the Proceedings of the National Academies.
Dialogue: 0,0:03:37.40,0:03:38.69,Default,,0000,0000,0000,,If you Google this, you'll find it.
Dialogue: 0,0:03:38.69,0:03:40.56,Default,,0000,0000,0000,,It's four pages, easy to read.
Dialogue: 0,0:03:40.56,0:03:43.56,Default,,0000,0000,0000,,And they looked at just people's Facebook likes,
Dialogue: 0,0:03:43.56,0:03:45.48,Default,,0000,0000,0000,,so just the things you like on Facebook,
Dialogue: 0,0:03:45.48,0:03:47.62,Default,,0000,0000,0000,,and used that to predict all these attributes,
Dialogue: 0,0:03:47.62,0:03:49.27,Default,,0000,0000,0000,,along with some other ones.
Dialogue: 0,0:03:49.27,0:03:52.23,Default,,0000,0000,0000,,And in their paper they listed the five likes
Dialogue: 0,0:03:52.23,0:03:55.02,Default,,0000,0000,0000,,that were most indicative of high intelligence.
Dialogue: 0,0:03:55.02,0:03:57.34,Default,,0000,0000,0000,,And among those was liking a page
Dialogue: 0,0:03:57.34,0:03:59.24,Default,,0000,0000,0000,,for curly fries. (Laughter)
Dialogue: 0,0:03:59.24,0:04:01.34,Default,,0000,0000,0000,,Curly fries are delicious,
Dialogue: 0,0:04:01.34,0:04:03.87,Default,,0000,0000,0000,,but liking them does not necessarily mean
Dialogue: 0,0:04:03.87,0:04:05.95,Default,,0000,0000,0000,,that you're smarter than the average person.
Dialogue: 0,0:04:05.95,0:04:09.16,Default,,0000,0000,0000,,So how is it that one of the strongest indicators
Dialogue: 0,0:04:09.16,0:04:10.72,Default,,0000,0000,0000,,of your intelligence
Dialogue: 0,0:04:10.72,0:04:12.17,Default,,0000,0000,0000,,is liking this page
Dialogue: 0,0:04:12.17,0:04:14.42,Default,,0000,0000,0000,,when the content is totally irrelevant
Dialogue: 0,0:04:14.42,0:04:16.95,Default,,0000,0000,0000,,to the attribute that's being predicted?
Dialogue: 0,0:04:16.95,0:04:18.54,Default,,0000,0000,0000,,And it turns out that we have to look at
Dialogue: 0,0:04:18.54,0:04:20.15,Default,,0000,0000,0000,,a whole bunch of underlying theories
Dialogue: 0,0:04:20.15,0:04:22.72,Default,,0000,0000,0000,,to see why we're able to do this.
Dialogue: 0,0:04:22.72,0:04:25.64,Default,,0000,0000,0000,,One of them is a sociological\Ntheory called homophily,
Dialogue: 0,0:04:25.64,0:04:28.73,Default,,0000,0000,0000,,which basically says people are\Nfriends with people like them.
Dialogue: 0,0:04:28.73,0:04:30.74,Default,,0000,0000,0000,,So if you're smart, you tend to\Nbe friends with smart people,
Dialogue: 0,0:04:30.74,0:04:33.37,Default,,0000,0000,0000,,and if you're young, you tend\Nto be friends with young people,
Dialogue: 0,0:04:33.37,0:04:34.100,Default,,0000,0000,0000,,and this is well established
Dialogue: 0,0:04:34.100,0:04:36.74,Default,,0000,0000,0000,,for hundreds of years.
Dialogue: 0,0:04:36.74,0:04:37.98,Default,,0000,0000,0000,,We also know a lot
Dialogue: 0,0:04:37.98,0:04:40.52,Default,,0000,0000,0000,,about how information spreads through networks.
Dialogue: 0,0:04:40.52,0:04:42.28,Default,,0000,0000,0000,,It turns out things like viral videos
Dialogue: 0,0:04:42.28,0:04:44.68,Default,,0000,0000,0000,,or Facebook likes or other information
Dialogue: 0,0:04:44.68,0:04:46.57,Default,,0000,0000,0000,,spreads in exactly the same way
Dialogue: 0,0:04:46.57,0:04:49.03,Default,,0000,0000,0000,,that diseases spread through social networks.
Dialogue: 0,0:04:49.03,0:04:50.82,Default,,0000,0000,0000,,So this is something we've studied for a long time.
Dialogue: 0,0:04:50.82,0:04:52.39,Default,,0000,0000,0000,,We have good models of it.
Dialogue: 0,0:04:52.39,0:04:54.55,Default,,0000,0000,0000,,And so you can put those things together
Dialogue: 0,0:04:54.55,0:04:57.64,Default,,0000,0000,0000,,and start seeing why things like this happen.
Dialogue: 0,0:04:57.64,0:04:59.45,Default,,0000,0000,0000,,So if I were to give you a hypothesis,
Dialogue: 0,0:04:59.45,0:05:02.68,Default,,0000,0000,0000,,it would be that a smart guy started this page,
Dialogue: 0,0:05:02.68,0:05:04.62,Default,,0000,0000,0000,,or maybe one of the first people who liked it
Dialogue: 0,0:05:04.62,0:05:06.36,Default,,0000,0000,0000,,would have scored high on that test.
Dialogue: 0,0:05:06.36,0:05:08.64,Default,,0000,0000,0000,,And they liked it, and their friends saw it,
Dialogue: 0,0:05:08.64,0:05:11.76,Default,,0000,0000,0000,,and by homophily, we know that\Nhe probably had smart friends,
Dialogue: 0,0:05:11.76,0:05:14.82,Default,,0000,0000,0000,,and so it spread to them, \Nand some of them liked it,
Dialogue: 0,0:05:14.82,0:05:16.01,Default,,0000,0000,0000,,and they had smart friends,
Dialogue: 0,0:05:16.01,0:05:16.82,Default,,0000,0000,0000,,and so it spread to them,
Dialogue: 0,0:05:16.82,0:05:18.79,Default,,0000,0000,0000,,and so it propagated through the network
Dialogue: 0,0:05:18.79,0:05:21.36,Default,,0000,0000,0000,,to a host of smart people,
Dialogue: 0,0:05:21.36,0:05:23.42,Default,,0000,0000,0000,,so that by the end, the action
Dialogue: 0,0:05:23.42,0:05:25.96,Default,,0000,0000,0000,,of liking the curly fries page
Dialogue: 0,0:05:25.96,0:05:27.57,Default,,0000,0000,0000,,is indicative of high intelligence,
Dialogue: 0,0:05:27.57,0:05:29.38,Default,,0000,0000,0000,,not because of the content,
Dialogue: 0,0:05:29.38,0:05:31.90,Default,,0000,0000,0000,,but because the actual action of liking
Dialogue: 0,0:05:31.90,0:05:33.80,Default,,0000,0000,0000,,reflects back the common attributes
Dialogue: 0,0:05:33.80,0:05:36.27,Default,,0000,0000,0000,,of other people who have done it.
Dialogue: 0,0:05:36.27,0:05:39.16,Default,,0000,0000,0000,,So this is pretty complicated stuff, right?
Dialogue: 0,0:05:39.16,0:05:41.36,Default,,0000,0000,0000,,It's a hard thing to sit down and explain
Dialogue: 0,0:05:41.36,0:05:44.21,Default,,0000,0000,0000,,to an average user, and even if you do,
Dialogue: 0,0:05:44.21,0:05:46.40,Default,,0000,0000,0000,,what can the average user do about it?
Dialogue: 0,0:05:46.40,0:05:48.45,Default,,0000,0000,0000,,How do you know that \Nyou've liked something
Dialogue: 0,0:05:48.45,0:05:49.94,Default,,0000,0000,0000,,that indicates a trait for you
Dialogue: 0,0:05:49.94,0:05:53.48,Default,,0000,0000,0000,,that's totally irrelevant to the\Ncontent of what you've liked?
Dialogue: 0,0:05:53.48,0:05:56.03,Default,,0000,0000,0000,,There's a lot of power that users don't have
Dialogue: 0,0:05:56.03,0:05:58.26,Default,,0000,0000,0000,,to control how this data is used.
Dialogue: 0,0:05:58.26,0:06:01.37,Default,,0000,0000,0000,,And I see that as a real \Nproblem going forward.
Dialogue: 0,0:06:01.37,0:06:03.35,Default,,0000,0000,0000,,So I think there's a couple paths
Dialogue: 0,0:06:03.35,0:06:04.35,Default,,0000,0000,0000,,that we want to look at
Dialogue: 0,0:06:04.35,0:06:06.26,Default,,0000,0000,0000,,if we want to give users some control
Dialogue: 0,0:06:06.26,0:06:08.00,Default,,0000,0000,0000,,over how this data is used,
Dialogue: 0,0:06:08.00,0:06:09.94,Default,,0000,0000,0000,,because it's not always going to be used
Dialogue: 0,0:06:09.94,0:06:11.32,Default,,0000,0000,0000,,for their benefit.
Dialogue: 0,0:06:11.32,0:06:12.74,Default,,0000,0000,0000,,An example I often give is that,
Dialogue: 0,0:06:12.74,0:06:14.39,Default,,0000,0000,0000,,if I ever get bored being a professor,
Dialogue: 0,0:06:14.39,0:06:16.04,Default,,0000,0000,0000,,I'm going to go start a company
Dialogue: 0,0:06:16.04,0:06:17.50,Default,,0000,0000,0000,,that predicts all of these attributes
Dialogue: 0,0:06:17.50,0:06:19.10,Default,,0000,0000,0000,,and things like how well you work in teams
Dialogue: 0,0:06:19.10,0:06:21.77,Default,,0000,0000,0000,,and if you're a drug user, if you're an alcoholic.
Dialogue: 0,0:06:21.77,0:06:23.21,Default,,0000,0000,0000,,We know how to predict all that.
Dialogue: 0,0:06:23.21,0:06:24.97,Default,,0000,0000,0000,,And I'm going to sell reports
Dialogue: 0,0:06:24.97,0:06:27.07,Default,,0000,0000,0000,,to H.R. companies and big businesses
Dialogue: 0,0:06:27.07,0:06:29.34,Default,,0000,0000,0000,,that want to hire you.
Dialogue: 0,0:06:29.34,0:06:30.52,Default,,0000,0000,0000,,We totally can do that now.
Dialogue: 0,0:06:30.52,0:06:32.31,Default,,0000,0000,0000,,I could start that business tomorrow,
Dialogue: 0,0:06:32.31,0:06:34.36,Default,,0000,0000,0000,,and you would have\Nabsolutely no control
Dialogue: 0,0:06:34.36,0:06:36.50,Default,,0000,0000,0000,,over me using your data like that.
Dialogue: 0,0:06:36.50,0:06:38.79,Default,,0000,0000,0000,,That seems to me to be a problem.
Dialogue: 0,0:06:38.79,0:06:40.70,Default,,0000,0000,0000,,So one of the paths we can go down
Dialogue: 0,0:06:40.70,0:06:42.73,Default,,0000,0000,0000,,is the policy and law path.
Dialogue: 0,0:06:42.73,0:06:45.78,Default,,0000,0000,0000,,And in some respects, I think\Nthat that would be most effective,
Dialogue: 0,0:06:45.78,0:06:48.53,Default,,0000,0000,0000,,but the problem is we'd\Nactually have to do it.
Dialogue: 0,0:06:48.53,0:06:51.31,Default,,0000,0000,0000,,Observing our political process in action
Dialogue: 0,0:06:51.31,0:06:53.69,Default,,0000,0000,0000,,makes me think it's highly unlikely
Dialogue: 0,0:06:53.69,0:06:55.29,Default,,0000,0000,0000,,that we're going to get a bunch of representatives
Dialogue: 0,0:06:55.29,0:06:57.28,Default,,0000,0000,0000,,to sit down, learn about this,
Dialogue: 0,0:06:57.28,0:06:59.38,Default,,0000,0000,0000,,and then enact sweeping changes
Dialogue: 0,0:06:59.38,0:07:01.54,Default,,0000,0000,0000,,to intellectual property law in the U.S.
Dialogue: 0,0:07:01.54,0:07:04.00,Default,,0000,0000,0000,,so users control their data.
Dialogue: 0,0:07:04.00,0:07:05.30,Default,,0000,0000,0000,,We could go the policy route,
Dialogue: 0,0:07:05.30,0:07:06.78,Default,,0000,0000,0000,,where social media companies say,
Dialogue: 0,0:07:06.78,0:07:08.18,Default,,0000,0000,0000,,you know what? You own your data.
Dialogue: 0,0:07:08.18,0:07:10.67,Default,,0000,0000,0000,,You have total control over how it's used.
Dialogue: 0,0:07:10.67,0:07:12.52,Default,,0000,0000,0000,,The problem is that the revenue models
Dialogue: 0,0:07:12.52,0:07:14.25,Default,,0000,0000,0000,,for most social media companies
Dialogue: 0,0:07:14.25,0:07:18.28,Default,,0000,0000,0000,,rely on sharing or exploiting \Nusers' data in some way.
Dialogue: 0,0:07:18.28,0:07:20.11,Default,,0000,0000,0000,,It's sometimes said of Facebook that the users
Dialogue: 0,0:07:20.11,0:07:22.64,Default,,0000,0000,0000,,aren't the customer, they're the product.
Dialogue: 0,0:07:22.64,0:07:25.35,Default,,0000,0000,0000,,And so how do you get a company
Dialogue: 0,0:07:25.35,0:07:27.91,Default,,0000,0000,0000,,to cede control of their main asset
Dialogue: 0,0:07:27.91,0:07:29.16,Default,,0000,0000,0000,,back to the users?
Dialogue: 0,0:07:29.16,0:07:30.86,Default,,0000,0000,0000,,It's possible, but I don't think it's something
Dialogue: 0,0:07:30.86,0:07:33.18,Default,,0000,0000,0000,,that we're going to see change quickly.
Dialogue: 0,0:07:33.18,0:07:34.68,Default,,0000,0000,0000,,So I think the other path
Dialogue: 0,0:07:34.68,0:07:36.97,Default,,0000,0000,0000,,that we can go down that's\Ngoing to be more effective
Dialogue: 0,0:07:36.97,0:07:38.48,Default,,0000,0000,0000,,is one of more science.
Dialogue: 0,0:07:38.48,0:07:40.99,Default,,0000,0000,0000,,It's doing science that allowed us to develop
Dialogue: 0,0:07:40.99,0:07:42.74,Default,,0000,0000,0000,,all these mechanisms for computing
Dialogue: 0,0:07:42.74,0:07:44.79,Default,,0000,0000,0000,,this personal data in the first place.
Dialogue: 0,0:07:44.79,0:07:46.89,Default,,0000,0000,0000,,And it's actually very similar research
Dialogue: 0,0:07:46.89,0:07:48.33,Default,,0000,0000,0000,,that we'd have to do
Dialogue: 0,0:07:48.33,0:07:50.72,Default,,0000,0000,0000,,if we want to develop mechanisms
Dialogue: 0,0:07:50.72,0:07:52.14,Default,,0000,0000,0000,,that can say to a user,
Dialogue: 0,0:07:52.14,0:07:54.37,Default,,0000,0000,0000,,"Here's the risk of that action you just took."
Dialogue: 0,0:07:54.37,0:07:56.45,Default,,0000,0000,0000,,By liking that Facebook page,
Dialogue: 0,0:07:56.45,0:07:58.98,Default,,0000,0000,0000,,or by sharing this piece of personal information,
Dialogue: 0,0:07:58.98,0:08:00.48,Default,,0000,0000,0000,,you've now improved my ability
Dialogue: 0,0:08:00.48,0:08:02.57,Default,,0000,0000,0000,,to predict whether or not you're using drugs
Dialogue: 0,0:08:02.57,0:08:05.43,Default,,0000,0000,0000,,or whether or not you get\Nalong well in the workplace.
Dialogue: 0,0:08:05.43,0:08:07.28,Default,,0000,0000,0000,,And that, I think, can affect whether or not
Dialogue: 0,0:08:07.28,0:08:08.79,Default,,0000,0000,0000,,people want to share something,
Dialogue: 0,0:08:08.79,0:08:12.03,Default,,0000,0000,0000,,keep it private, or just keep it offline altogether.
Dialogue: 0,0:08:12.03,0:08:13.59,Default,,0000,0000,0000,,We can also look at things like
Dialogue: 0,0:08:13.59,0:08:16.32,Default,,0000,0000,0000,,allowing people to encrypt data that they upload,
Dialogue: 0,0:08:16.32,0:08:18.18,Default,,0000,0000,0000,,so it's kind of invisible and worthless
Dialogue: 0,0:08:18.18,0:08:19.61,Default,,0000,0000,0000,,to sites like Facebook
Dialogue: 0,0:08:19.61,0:08:22.24,Default,,0000,0000,0000,,or third party services that access it,
Dialogue: 0,0:08:22.24,0:08:25.48,Default,,0000,0000,0000,,but that select users who the person who posted it
Dialogue: 0,0:08:25.48,0:08:28.15,Default,,0000,0000,0000,,want to see it have access to see it.
Dialogue: 0,0:08:28.15,0:08:30.32,Default,,0000,0000,0000,,This is all super exciting research
Dialogue: 0,0:08:30.32,0:08:31.94,Default,,0000,0000,0000,,from an intellectual perspective,
Dialogue: 0,0:08:31.94,0:08:33.80,Default,,0000,0000,0000,,and so scientists are going to be willing to do it.
Dialogue: 0,0:08:33.80,0:08:37.41,Default,,0000,0000,0000,,So that gives us an advantage over the law side.
Dialogue: 0,0:08:37.41,0:08:39.13,Default,,0000,0000,0000,,One of the problems that people bring up
Dialogue: 0,0:08:39.13,0:08:40.73,Default,,0000,0000,0000,,when I talk about this is, they say,
Dialogue: 0,0:08:40.73,0:08:43.37,Default,,0000,0000,0000,,you know, if people start\Nkeeping all this data private,
Dialogue: 0,0:08:43.37,0:08:45.49,Default,,0000,0000,0000,,all those methods that you've been developing
Dialogue: 0,0:08:45.49,0:08:48.14,Default,,0000,0000,0000,,to predict their traits are going to fail.
Dialogue: 0,0:08:48.14,0:08:51.66,Default,,0000,0000,0000,,And I say, absolutely, and for me, that's success,
Dialogue: 0,0:08:51.66,0:08:53.45,Default,,0000,0000,0000,,because as a scientist,
Dialogue: 0,0:08:53.45,0:08:57.13,Default,,0000,0000,0000,,my goal is not to infer information about users,
Dialogue: 0,0:08:57.13,0:08:59.90,Default,,0000,0000,0000,,it's to improve the way people interact online.
Dialogue: 0,0:08:59.90,0:09:03.12,Default,,0000,0000,0000,,And sometimes that involves\Ninferring things about them,
Dialogue: 0,0:09:03.12,0:09:06.14,Default,,0000,0000,0000,,but if users don't want me to use that data,
Dialogue: 0,0:09:06.14,0:09:08.18,Default,,0000,0000,0000,,I think they should have the right to do that.
Dialogue: 0,0:09:08.18,0:09:10.83,Default,,0000,0000,0000,,I want users to be informed and consenting
Dialogue: 0,0:09:10.83,0:09:12.94,Default,,0000,0000,0000,,users of the tools that we develop.
Dialogue: 0,0:09:12.94,0:09:15.89,Default,,0000,0000,0000,,And so I think encouraging this kind of science
Dialogue: 0,0:09:15.89,0:09:17.24,Default,,0000,0000,0000,,and supporting researchers
Dialogue: 0,0:09:17.24,0:09:20.26,Default,,0000,0000,0000,,who want to cede some of that control back to users
Dialogue: 0,0:09:20.26,0:09:22.57,Default,,0000,0000,0000,,and away from the social media companies
Dialogue: 0,0:09:22.57,0:09:25.24,Default,,0000,0000,0000,,means that going forward, as these tools evolve
Dialogue: 0,0:09:25.24,0:09:26.72,Default,,0000,0000,0000,,and advance,
Dialogue: 0,0:09:26.72,0:09:28.14,Default,,0000,0000,0000,,means that we're going to have an educated
Dialogue: 0,0:09:28.14,0:09:29.83,Default,,0000,0000,0000,,and empowered user base,
Dialogue: 0,0:09:29.83,0:09:30.93,Default,,0000,0000,0000,,and I think all of us can agree
Dialogue: 0,0:09:30.93,0:09:33.49,Default,,0000,0000,0000,,that that's a pretty ideal way to go forward.
Dialogue: 0,0:09:33.49,0:09:35.68,Default,,0000,0000,0000,,Thank you.
Dialogue: 0,0:09:35.68,0:09:38.76,Default,,0000,0000,0000,,(Applause)