WEBVTT - The Problems of Regulating Algorithms are Solvable 0:00:15.076 --> 0:00:26.276 Bushkin, This is solvable. I'm Jacob Weisberg. Typically, if a 0:00:26.356 --> 0:00:31.076 car crashes because there's say a faulty drive train, we 0:00:31.276 --> 0:00:35.876 can point to the engineering and say there's a problem 0:00:35.916 --> 0:00:40.356 with this system. With these adaptive systems, they're reacting and 0:00:40.476 --> 0:00:45.836 learning and responding to human society and human behavior, and 0:00:45.956 --> 0:00:51.156 we're still developing the scientific tools to understand what it 0:00:51.236 --> 0:00:56.396 means to have those feedback loops. Algorithms are adaptive systems. 0:00:56.876 --> 0:00:59.956 They're pieces of computer code that shape many aspects of 0:00:59.956 --> 0:01:05.036 our digital lives. They're closely guarded trade secrets and powerful tools, 0:01:05.676 --> 0:01:11.036 and they're regularly blamed for amplifying cultural and political divisions. 0:01:11.796 --> 0:01:16.836 We often hear technologists say we couldn't have known, and 0:01:17.796 --> 0:01:23.116 the idea that they haven't turned those lenses on questions 0:01:23.316 --> 0:01:27.196 impacting the common good. It's a scandal if they haven't 0:01:27.236 --> 0:01:30.276 ask the question. It's a scandal if they've asked it 0:01:30.316 --> 0:01:32.876 and they're not telling us what they found. This is 0:01:32.876 --> 0:01:36.316 a fourth chapter in our Solvable series examining solutions for 0:01:36.436 --> 0:01:41.876 America's polarization problem. Today we're talking about social media algorithms 0:01:42.196 --> 0:01:44.356 and how to deal with them. You can think of 0:01:44.396 --> 0:01:48.236 social media companies as fancy restaurants. The cooks behind the 0:01:48.236 --> 0:01:51.636 most successful one often don't want to reveal their recipes, 0:01:52.156 --> 0:01:54.636 but customers have a right to know what they're eating. 0:01:55.476 --> 0:01:59.116 It turns out we've been down this road before. The 0:01:59.236 --> 0:02:04.356 Good Housekeeping labs started just around the turn of the century. 0:02:04.876 --> 0:02:08.156 People were concerned about what was in their food what 0:02:08.276 --> 0:02:11.916 was in others. This was before the creation of the FDA, 0:02:11.956 --> 0:02:16.396 and so people subscribed to Good Housekeeping. Those labs would 0:02:16.916 --> 0:02:21.036 test common products and tell people if they were safe 0:02:21.196 --> 0:02:24.916 or not. Ultimately, the federal government stepped in to regulate 0:02:24.996 --> 0:02:31.876 food safety, including disclosure requirements around ingredients, nutrients, and calories. Similarly, 0:02:32.316 --> 0:02:37.756 establishing algorithmic safety and accountability will take a variety of players. 0:02:38.356 --> 0:02:40.876 I want to live in a world where digital power 0:02:41.076 --> 0:02:44.356 is both guided by evidence and accountable to the public. 0:02:44.996 --> 0:02:48.876 Nathan Mathias teaches at Cornell University and leads the Citizens 0:02:48.916 --> 0:02:53.796 and Technology Lab. The problems of regulating algorithms are solvable. 0:02:58.356 --> 0:03:01.716 My co host an Apple Bomb, spoke with Nathan Mathias. 0:03:01.756 --> 0:03:07.516 Here's their conversation. So Nathan, tell me what's an algorithm. 0:03:08.396 --> 0:03:12.236 Algorithms can be thought of as a recipe, a series 0:03:12.356 --> 0:03:16.716 of steps often programmed into a computer that determine how 0:03:16.756 --> 0:03:22.236 a machine behaves. But the challenge, as any cook often finds, 0:03:22.596 --> 0:03:24.476 is that when you put them out into the world, 0:03:24.636 --> 0:03:28.876 especially something of sufficient complexity, they often behave in ways 0:03:29.076 --> 0:03:32.636 that are different from what we expect. Can you just 0:03:32.636 --> 0:03:36.436 take a minute to explain how that's problematic and why 0:03:36.556 --> 0:03:40.756 why should we care that algorithms are deciding which piece 0:03:40.796 --> 0:03:44.716 of content you see on Facebook or which video you're 0:03:44.716 --> 0:03:49.396 being recommended on YouTube. Algorithms happen at all levels, from 0:03:50.196 --> 0:03:54.076 exactly how the electrons go from one point to another 0:03:54.236 --> 0:03:57.316 on the Internet to the much more high level things 0:03:57.436 --> 0:04:02.116 that we think about in our direct experience. For example, 0:04:02.356 --> 0:04:07.196 an algorithm determines what your email imbacts decides is spam, 0:04:07.596 --> 0:04:12.076 an algorithm on Twitter decides which faces to show when 0:04:12.156 --> 0:04:17.476 it's displaying a photo. And algorithms also and critically make 0:04:17.556 --> 0:04:23.356 decisions about what information to prioritize when showing us feeds 0:04:23.436 --> 0:04:28.676 on Facebook, on Twitter, when determining which adds we see 0:04:29.076 --> 0:04:33.716 which adds we don't, and those are often some of 0:04:33.756 --> 0:04:38.716 the uses of algorithms that people worry about in society 0:04:38.716 --> 0:04:43.236 and policy circles. YouTube makes a recommendation system to help 0:04:43.316 --> 0:04:46.516 us find the videos we like, and suddenly we're worrying 0:04:46.556 --> 0:04:52.676 about recommendations of extremism. Microsoft makes a fun chat assistant 0:04:53.116 --> 0:04:56.716 that will have interesting conversations with you, and now we're 0:04:56.716 --> 0:05:01.236 worrying about it learning racism and hatred. So we find 0:05:01.356 --> 0:05:05.276 that although we have the simple building blocks of an 0:05:05.276 --> 0:05:09.716 algorithm that an engineer can imagine, they often grow to 0:05:09.796 --> 0:05:14.676 be something larger than we might omit initially imagine. I've 0:05:14.676 --> 0:05:18.556 written and others have written about the problem of algorithms 0:05:18.556 --> 0:05:25.036 on Facebook favoritizing or preferring content and posts that are emotional, 0:05:25.236 --> 0:05:29.196 that are negative, that are divisive. You know, there's been 0:05:29.196 --> 0:05:32.116 an argument that that's one of the reasons why we 0:05:32.196 --> 0:05:34.956 have so much division and polarization in our societies, that 0:05:35.356 --> 0:05:39.836 we are being fed more and more excitable and angry 0:05:39.876 --> 0:05:43.436 content because the algorithm tests and guesses that that's what 0:05:43.436 --> 0:05:45.036 we're going to want to see or anyway that's what's 0:05:45.036 --> 0:05:47.516 going to keep us online or keep us using Facebook. 0:05:47.956 --> 0:05:49.956 Is it accurate? Is that how they work? We do 0:05:50.036 --> 0:05:53.436 live in a world where many of the systems that 0:05:53.556 --> 0:05:57.436 determined what we see and give our attention to our 0:05:57.556 --> 0:06:02.116 learning from our behavior, our preferences, and from the collective 0:06:02.156 --> 0:06:05.636 behavior of many others, some of who aren't paying. Some 0:06:05.676 --> 0:06:09.956 of them have motivated, coordinated campaigns to influence the algorithms, 0:06:10.476 --> 0:06:14.796 and they're adapting in real time. And so because we've 0:06:14.836 --> 0:06:19.276 never really faced a situation like this at such scale, 0:06:19.796 --> 0:06:22.876 people have a lot of concerns about how those algorithms 0:06:22.916 --> 0:06:27.596 are behaving and what they're doing to society. One of 0:06:27.636 --> 0:06:32.356 the fundamental challenges that I think scientists are still wrestling 0:06:32.396 --> 0:06:37.636 with is this challenge of influence. Typically, if a car 0:06:37.756 --> 0:06:42.916 crashes because there's say a faulty drive train, we can 0:06:42.956 --> 0:06:47.116 point to the engineering and say there's a problem with 0:06:47.196 --> 0:06:51.876 this system. With these adaptive systems, they're reacting and learning 0:06:51.916 --> 0:06:57.236 and responding to human society and human behavior. And we're 0:06:57.316 --> 0:07:02.756 still developing the scientific tools to understand what it means 0:07:02.836 --> 0:07:06.076 to have those feedback loops. And in the meantime, we 0:07:06.156 --> 0:07:08.116 have to live in a world where these things have 0:07:08.516 --> 0:07:12.676 very real power. If the Facebook algorithm is designed to 0:07:12.756 --> 0:07:15.196 keep all of us on Facebook as long as possible, 0:07:15.756 --> 0:07:18.476 who's able to watch that, who's able to control it, 0:07:18.956 --> 0:07:22.316 who's following the science? Almost no one is in a 0:07:22.436 --> 0:07:28.676 position right now to regulate and manage those algorithms. For example, 0:07:28.956 --> 0:07:34.516 in February, Facebook announced that they would be reducing the 0:07:34.676 --> 0:07:40.116 political content appearing in people's news feeds in several countries. 0:07:40.636 --> 0:07:43.516 We don't really know the details of what they're doing. 0:07:43.916 --> 0:07:46.956 We also have evidence, because they say they're doing tests, 0:07:47.276 --> 0:07:51.036 that they're not necessarily sure themselves what the impact is 0:07:51.076 --> 0:07:54.876 going to be. When you think about who currently has 0:07:55.676 --> 0:07:59.596 some power to shape what algorithms do, I think there 0:07:59.636 --> 0:08:03.516 are some people at different levels of society who have 0:08:03.596 --> 0:08:06.756 a little bit of influence. We've seen, for example, European 0:08:06.836 --> 0:08:12.836 regulators step in around antitrust around what kinds of products 0:08:13.036 --> 0:08:17.396 get promoted by search engines, for example, So governments have 0:08:17.476 --> 0:08:22.436 been doing a little bit. We definitely have companies themselves 0:08:22.436 --> 0:08:26.516 are being seen as almost government like and having policy teams, 0:08:26.516 --> 0:08:29.156 so they're trying to understand how their own systems work 0:08:29.636 --> 0:08:33.236 and manage them in some way without the rest of 0:08:33.316 --> 0:08:36.476 us having that much transparency into their values or goals 0:08:36.596 --> 0:08:40.836 or even their results. And then in some areas there 0:08:40.876 --> 0:08:44.956 are other actors who have power to manage and govern 0:08:45.116 --> 0:08:47.956 algorithms in a constraint way. So if you've ever been 0:08:47.996 --> 0:08:52.276 a Facebook group administrator, for example, or you know someone 0:08:52.316 --> 0:08:55.116 who's a Reddit moderator. They have a little bit of 0:08:55.116 --> 0:08:58.996 an ability to tweak what gets promoted or how they 0:08:59.316 --> 0:09:02.036 given algorithm works, even though they don't have a lot 0:09:02.076 --> 0:09:06.916 of visibility into the underlying code or necessarily the power 0:09:06.956 --> 0:09:09.476 to tell a big company to change what they do. 0:09:10.756 --> 0:09:13.356 A couple of weeks ago, I had reason to talk 0:09:13.396 --> 0:09:17.756 to a Facebook spokesman, and the topic was the experiments 0:09:17.796 --> 0:09:21.156 that Facebook does with its algorithms, the way in which 0:09:21.196 --> 0:09:23.876 they test different things. As you say, they try and 0:09:23.956 --> 0:09:27.196 use more or less political content. You know. Actually, after 0:09:27.236 --> 0:09:31.876 the events on January sixth at the Capitol, they came 0:09:31.956 --> 0:09:34.756 up with a way of moderating the news feed so 0:09:34.836 --> 0:09:37.036 that there wouldn't be so much disinformation in it. But 0:09:37.156 --> 0:09:40.076 of course, as you say, they don't publish the results 0:09:40.076 --> 0:09:43.916 of these experiments or of these changes. One of the 0:09:43.996 --> 0:09:46.996 solutions that I know that you have suggested is that 0:09:47.036 --> 0:09:51.396 there should be outside moderators, or there should be citizens 0:09:51.396 --> 0:09:55.676 scientists who are studying these algorithms, you know, either with 0:09:55.756 --> 0:09:58.636 the cooperation of Facebook and Google or maybe not with 0:09:58.396 --> 0:10:02.796 their cooperation. What step one The place to start is 0:10:02.836 --> 0:10:07.036 often with your own experience. I'll tell you a story 0:10:07.116 --> 0:10:10.876 just to illustrate this about six years or ago, I 0:10:10.916 --> 0:10:15.916 was approached by a group of women who faced online harassment, 0:10:16.236 --> 0:10:19.636 threats of violence, and other kinds of risks. For them. 0:10:20.076 --> 0:10:22.756 The first step was to acknowledge that it was a 0:10:22.756 --> 0:10:28.156 problem and to find other people who had the same problem. 0:10:28.196 --> 0:10:30.916 They were able to realize that they had common needs 0:10:30.996 --> 0:10:33.836 and common goals, and they actually came up with a 0:10:33.876 --> 0:10:38.516 way to record their experiences, both the kinds of harassment 0:10:38.796 --> 0:10:43.396 that they were facing, and also to record how Twitter 0:10:43.756 --> 0:10:47.996 did or often didn't handle their reports. That was the 0:10:48.036 --> 0:10:50.476 point actually that they then reached out to me and said, 0:10:51.356 --> 0:10:55.276 this is clearly a systemic problem. We've all experienced it, 0:10:55.436 --> 0:10:59.036 we want to see change. We know that better understanding 0:10:59.156 --> 0:11:03.196 data and science will help us think of better solutions 0:11:03.476 --> 0:11:07.156 and also, if necessary, to create pressure for those solutions. 0:11:07.716 --> 0:11:10.196 That was a great moment then for me and the 0:11:10.196 --> 0:11:14.076 team of researchers I led to develop a methodology and 0:11:14.156 --> 0:11:17.396 analyze the data they were collecting, and that report that 0:11:17.436 --> 0:11:21.916 we ended up creating has been influential in industry. It's 0:11:21.916 --> 0:11:25.276 helped law enforcement understand how to better support people who 0:11:25.316 --> 0:11:29.476 experience online harassment, and it's also been useful in policy 0:11:29.476 --> 0:11:33.076 debates in this country about online harassment. We're at a 0:11:33.156 --> 0:11:37.956 moment where we're still building the lines of communication and 0:11:38.116 --> 0:11:42.236 the idea of citizen science as a mode of understanding 0:11:42.236 --> 0:11:46.636 and improving our digital lives. So at this stage, I 0:11:46.676 --> 0:11:49.396 think the best first step is really to find other 0:11:49.436 --> 0:11:53.156 people who care about the thing you care about. So 0:11:53.196 --> 0:11:56.516 we need to identify the problems that have to be studied, 0:11:56.596 --> 0:11:59.236 and we need the labs where they can be studied. 0:11:59.756 --> 0:12:04.236 That's the first step. Absolutely, there's another important step at 0:12:04.316 --> 0:12:08.836 the ecosystem level. There's a funding challenge. Most of the 0:12:09.156 --> 0:12:12.356 search that goes into funding that goes into social computing 0:12:12.716 --> 0:12:16.356 comes from the tech industry, like hundreds of millions of dollars, 0:12:16.676 --> 0:12:19.716 and if you look at the money that comes into 0:12:19.876 --> 0:12:23.756 industry independent research, it's a tiny drop in the bucket. 0:12:24.196 --> 0:12:28.996 So as policymakers debate ideas like taxing tech companies, I 0:12:28.996 --> 0:12:33.276 could imagine they're being funding within that for industry independent 0:12:33.356 --> 0:12:38.716 accountability research. We're also finding ourselves having to invent new 0:12:39.396 --> 0:12:42.556 funding models for this kind of research as well. And 0:12:42.596 --> 0:12:48.716 then presumably at some point, some regulatory mechanism that makes 0:12:48.756 --> 0:12:52.796 sure that the Internet platforms will work with you and 0:12:52.836 --> 0:12:55.756 we'll listen to you exactly. So I think we're seeing 0:12:55.796 --> 0:13:00.316 more and more researchers in this space say that we're 0:13:00.356 --> 0:13:04.436 going to need some kind of regulation to provide protections 0:13:04.476 --> 0:13:08.396 and support for independent research to go on even when 0:13:08.436 --> 0:13:12.996 companies find it uncomfortable. One of the crises, you know 0:13:13.036 --> 0:13:16.116 at the moment in American life is the fact that 0:13:16.156 --> 0:13:19.676 a part of the country now lives in a completely 0:13:19.716 --> 0:13:22.236 alternative universe from the rest of the country. And we 0:13:22.276 --> 0:13:25.116 all saw on January the six that there are people 0:13:25.116 --> 0:13:27.556 who are so convinced that Donald Trump won the election 0:13:27.596 --> 0:13:31.156 that they were willing to attack the capital and even 0:13:31.356 --> 0:13:36.596 murder policemen and other in an attempt to disrupt Congress's 0:13:36.636 --> 0:13:40.516 work and prevent the naming of verifying of Joe Biden 0:13:41.036 --> 0:13:44.396 as president. How do you relate that to this problem 0:13:44.436 --> 0:13:46.796 of algorithms. I mean, if we had if we could 0:13:46.836 --> 0:13:49.916 solve the algorithm problem, if we if we were able 0:13:49.956 --> 0:13:53.716 to structure algorithms so that they favored civic discourse and 0:13:54.276 --> 0:13:58.596 civil conversation instead of promoting division and anger, could that 0:13:58.756 --> 0:14:03.876 help us heal this deep divide, this epistemological divide whereby 0:14:03.876 --> 0:14:06.916 we all live in different realities. You know, we know 0:14:06.996 --> 0:14:09.516 that when crowds of people get involved in stuff that 0:14:09.556 --> 0:14:12.236 doesn't necessarily mean that the outcome is good or better. 0:14:13.036 --> 0:14:16.596 So why should we be so sure that citizen participation 0:14:16.676 --> 0:14:19.676 in the regulation of the internet will give us good regulation. 0:14:20.196 --> 0:14:25.716 It's important to differentiate between who's making decisions and who's 0:14:25.796 --> 0:14:30.356 producing evidence. Evidence is something that you can put into 0:14:30.556 --> 0:14:34.076 the conversation about what to do, and so long as 0:14:34.116 --> 0:14:39.036 that evidence is produced in a reliable way, it has 0:14:39.156 --> 0:14:42.596 value to bring to the conversation. So your feeling is 0:14:42.636 --> 0:14:46.316 that this is a question. It's not just important for 0:14:46.796 --> 0:14:48.756 I don't know the future of social media. It's really 0:14:48.836 --> 0:14:52.436 the question it's important for democracy giving that power, giving 0:14:52.596 --> 0:14:58.316 some of that oversight ability to citizen scientists, to outside groups, 0:14:58.436 --> 0:15:03.396 maybe to some government ombudsman, maybe to some regulators, that 0:15:03.476 --> 0:15:08.116 this would democratize that power that social media companies have. Yeah, 0:15:08.796 --> 0:15:12.716 one of my personal heroes in the social sciences is 0:15:12.796 --> 0:15:17.076 Kurt Lewin, one of the founders of social psychology, who 0:15:17.196 --> 0:15:22.596 himself barely escaped Nazi Germany with his life and went 0:15:22.636 --> 0:15:27.036 on to influence so much in science and society. And 0:15:27.076 --> 0:15:30.556 he had this great quote which says, it's essential that 0:15:30.636 --> 0:15:35.236 a democratic commonwealth and its educational system apply the rational 0:15:35.316 --> 0:15:40.516 procedures of scientific investigation to its own processes of group living. 0:15:40.876 --> 0:15:44.956 And Lewin believed that that needed to be done in 0:15:44.996 --> 0:15:48.916 a democratic way if we were going to maintain the 0:15:49.076 --> 0:15:53.516 values that we have as democratic societies. That it wasn't 0:15:53.636 --> 0:15:58.876 just enough to do research that supported democracy. You needed 0:15:58.996 --> 0:16:02.836 the research itself to be democratic in nature. And I 0:16:02.876 --> 0:16:06.476 think in an era where so much of what we 0:16:06.676 --> 0:16:13.196 do is influenced by design and algorithms, that reality is 0:16:13.636 --> 0:16:18.356 clearer than it even was in Lewin's time. There's a 0:16:18.396 --> 0:16:23.996 long tradition of citizens, scientists, and outsiders working outside the 0:16:23.996 --> 0:16:27.116 government are sometimes in tandem with the government in order 0:16:27.156 --> 0:16:30.796 to push regulation or particular direction. Do you see yourself 0:16:30.836 --> 0:16:32.876 belonging to that tradition and can you describe it a 0:16:32.916 --> 0:16:35.716 little bit? You're asking me a question about something that 0:16:35.836 --> 0:16:42.156 I absolutely love and obsessed by, so question. Yeah, you know, 0:16:42.236 --> 0:16:45.076 I grew up, you know, in the United States as 0:16:45.116 --> 0:16:50.036 a Guatemalan American, with this sense that science was this 0:16:50.116 --> 0:16:54.156 tool of like powerful people in institutions that didn't always 0:16:54.636 --> 0:16:58.956 include or pay attention to the general public or the 0:16:59.036 --> 0:17:03.476 marginalized as anything other than research participants like you can 0:17:03.516 --> 0:17:06.476 be a subject in the research and we will call 0:17:06.516 --> 0:17:09.636 you a subject. But when I was a graduate at 0:17:09.636 --> 0:17:12.556 the MIT Media Lab, I started to learn about this 0:17:12.596 --> 0:17:18.436 amazing tradition of citizen science in different places and times 0:17:18.436 --> 0:17:22.676 over the last really two hundred years. In the mid 0:17:22.796 --> 0:17:26.596 nineteenth century, there was a group of people who went 0:17:26.636 --> 0:17:31.036 around London and bought bread from different shops and used 0:17:31.076 --> 0:17:36.396 this new idea of a microscope to count what was 0:17:36.476 --> 0:17:42.036 actually in the bread and found widespread food adulteration. This 0:17:42.076 --> 0:17:47.316 set of studies ended up helping launch the trajectory of 0:17:47.316 --> 0:17:50.116 what is now the Lancet, one of the premier medical 0:17:50.196 --> 0:17:54.756 journals in the world. Another example I really love is 0:17:54.796 --> 0:17:59.956 the story of the Good Housekeeping Labs, which was started 0:18:00.436 --> 0:18:03.796 just around the turn of the century. People were concerned 0:18:03.836 --> 0:18:07.236 about what was in their food, what was in other products. 0:18:07.276 --> 0:18:10.196 This was before the creation of they DA. There really 0:18:10.276 --> 0:18:14.716 wasn't that much regulation of what went into the mass 0:18:14.796 --> 0:18:20.556 production ecosystem, and so people subscribed to Good Housekeeping. Those 0:18:20.636 --> 0:18:25.596 labs would test common products and tell people if they 0:18:25.636 --> 0:18:28.636 were safe or not and use the good Housekeeping seal 0:18:28.636 --> 0:18:32.156 of approval, and often in fact in the late nineteenth 0:18:32.156 --> 0:18:37.396 early twentieth century, because there was this convergence of the 0:18:37.556 --> 0:18:42.036 rising women's movement and a passion for science, you would 0:18:42.116 --> 0:18:48.036 have women's organizations actually leading a lot of citizen science efforts. 0:18:48.076 --> 0:18:52.836 And then later on when the US established the FDA, 0:18:52.876 --> 0:18:55.876 it was actually the scientists from the Good Housekeeping Lab 0:18:56.316 --> 0:19:01.716 that built up the FDA's initial scientific capacities and leading 0:19:01.796 --> 0:19:05.436 us to where we are today, where we have more 0:19:05.676 --> 0:19:10.596 organized and supported regimes of testing and science and regulation. 0:19:11.396 --> 0:19:15.676 So when you think that's how algorithm regulation or social 0:19:15.716 --> 0:19:19.916 media regulation could evolve with teams of citizen scientists like 0:19:19.996 --> 0:19:22.476 the people at your lab, or is the idea that 0:19:22.556 --> 0:19:25.716 eventually this is something the government would do or is 0:19:25.756 --> 0:19:28.676 this something that will some other kind of civic body 0:19:28.756 --> 0:19:30.676 will do. Do you have a kind of trajectory of 0:19:30.676 --> 0:19:32.756 how this could work in the long term. In the 0:19:32.836 --> 0:19:36.996 short term, citizen science and work from the outside is 0:19:37.036 --> 0:19:41.516 a necessity. We're currently at a moment where if you 0:19:41.716 --> 0:19:45.116 want to look at what tech companies are doing from 0:19:45.156 --> 0:19:49.076 the inside, you have to sign these NDAs, you have 0:19:49.276 --> 0:19:53.636 to do work that they feel comfortable with. And like 0:19:53.796 --> 0:19:59.556 many other citizen scientists in other domains, we find ourselves 0:19:59.676 --> 0:20:03.636 inventing methodologies to answer urging questions that people need to 0:20:03.716 --> 0:20:08.236 understand now, and I think, you know, we have a 0:20:08.276 --> 0:20:11.796 small but growing number of institutions that are starting to 0:20:11.836 --> 0:20:15.796 do that work. The Barkup Consumer Reports Digital Labs has 0:20:15.836 --> 0:20:19.236 been building a team that are initiatives like Joey boil 0:20:19.276 --> 0:20:23.156 and Wine's Algorithmic Justice League that all do work of 0:20:23.196 --> 0:20:27.876 this kind. In the longer term, I would love to 0:20:27.876 --> 0:20:31.636 see a healthy ecosystem. I draw a lot of inspiration 0:20:31.876 --> 0:20:36.396 from the work of Eleanor Ostrom, the Nobel Prize winning 0:20:36.436 --> 0:20:41.276 political scientist who wrote about how you incorporate science into 0:20:41.916 --> 0:20:48.596 complex governance scenarios where you have competing interests. I think 0:20:48.636 --> 0:20:51.356 we're likely, I hope, in the long term, to get 0:20:51.396 --> 0:20:54.196 to a point where companies are going to be more transparent. 0:20:54.276 --> 0:20:57.276 They're going to actually publish their protocols and research on 0:20:57.316 --> 0:20:59.556 the issues we care about, and that's going to be 0:20:59.596 --> 0:21:05.076 an important part. I think we really desperately need more 0:21:05.676 --> 0:21:09.036 government supported efforts, and I'll leave it to the policy 0:21:09.116 --> 0:21:12.036 makers to figure out what that actually looks like. And 0:21:12.076 --> 0:21:15.876 I think will continue to see citizen scientists trying to 0:21:15.916 --> 0:21:20.196 make sense of and improve their own contexts and environment, 0:21:20.356 --> 0:21:24.636 just like we have in the arena of environmental protection, 0:21:24.756 --> 0:21:29.916 consumer protection. Those are all mature ecosystems where you have 0:21:30.036 --> 0:21:35.676 science happening from different perspectives and different points in the ecosystem. Right, 0:21:35.716 --> 0:21:38.996 So it's not just government scientists. They're also independent scientists. 0:21:39.036 --> 0:21:42.516 And there's the Sierra Club, and they're individuals and they're 0:21:42.956 --> 0:21:45.236 you know, so there there are lots of different perspectives 0:21:45.236 --> 0:21:48.396 on the same environmental problem. And you imagine that that 0:21:48.636 --> 0:21:53.196 would eventually be possible in monitoring and regulating the social 0:21:53.196 --> 0:21:56.916 media companies too, exactly, and in democracy, we hope that 0:21:56.956 --> 0:22:01.036 having multiple perspectives helps us get to a better solution. 0:22:01.516 --> 0:22:04.196 At least that's that's the vision of democracy I want. 0:22:04.676 --> 0:22:08.956 I want to cling to in how I imagine the work. 0:22:09.156 --> 0:22:13.756 And so I think we need that for governing social media, 0:22:13.916 --> 0:22:17.676 for governing the role of digital technologies in our lives, 0:22:17.956 --> 0:22:19.956 and we have a lot of work to build up 0:22:19.956 --> 0:22:24.796 the industry independent part of that ecosystem. Nathan, I know 0:22:24.916 --> 0:22:28.836 that you started your education in the humanities and you 0:22:29.356 --> 0:22:32.596 moved later on to technology. Can you tell me a 0:22:32.636 --> 0:22:35.716 little bit about how that happened? How does an English 0:22:35.756 --> 0:22:39.156 major become part of this other world? When I was 0:22:39.196 --> 0:22:45.156 a teenager, I had this amazing opportunity to meet with 0:22:45.236 --> 0:22:48.876 and talk to a local computer science professor. I was 0:22:49.076 --> 0:22:52.636 really passionate about the arts. I was really passionate about computing, 0:22:53.356 --> 0:22:56.596 and he said, computing as a lens on the world. 0:22:57.076 --> 0:23:02.156 If you really care about understanding technology, you need to 0:23:02.316 --> 0:23:08.076 understand society. You need to pay close attention to the 0:23:08.116 --> 0:23:13.596 world around you, because computing without that has no heart, 0:23:13.676 --> 0:23:17.836 it has no moral compass. With his encouragement, I felt 0:23:17.996 --> 0:23:24.276 empowered and prompted to spend my undergraduate time reading literature, 0:23:24.396 --> 0:23:29.596 studying the humanities, asking myself the big questions. Was really 0:23:29.916 --> 0:23:34.196 during my second undergraduate degree, when I was a student 0:23:34.316 --> 0:23:38.356 at Cambridge University that I started to ask questions about 0:23:38.996 --> 0:23:43.156 literature and what we read, and its impact on democracy, 0:23:43.276 --> 0:23:49.076 its impact and connections to psychology. I realized that not 0:23:49.276 --> 0:23:54.636 only were we collecting massive amounts of data about human 0:23:54.676 --> 0:23:58.156 experience and behavior that could help us answer some of 0:23:58.156 --> 0:24:03.156 those questions. I also realized that those enduring questions about 0:24:03.196 --> 0:24:06.876 what it means to live well together in society that 0:24:06.996 --> 0:24:11.636 we've been asking as long as we've had written records 0:24:11.796 --> 0:24:16.196 are incredibly important to the present time. And that's what 0:24:16.396 --> 0:24:18.876 led me to actually go back to grad school and 0:24:19.716 --> 0:24:23.876 study those questions further. And those aren't questions that are 0:24:23.916 --> 0:24:27.596 normally asked in Silicon Valley, presumably. And I don't know 0:24:27.676 --> 0:24:31.516 if I can speak for all of Silicon Valley, but 0:24:31.596 --> 0:24:37.116 I do think that I think we often hear technologists 0:24:37.196 --> 0:24:42.636 say we couldn't have known, and I can't really tell 0:24:42.796 --> 0:24:49.636 whether that's true or whether it's a rhetorical line to take, 0:24:50.516 --> 0:24:56.356 because the reality is that companies have built some of 0:24:56.396 --> 0:25:02.516 the world's most sophisticated social scientific research endeavors in the 0:25:02.596 --> 0:25:07.316 history of humanity, and the idea that they haven't turned 0:25:07.396 --> 0:25:12.996 those lenses on questions impacting the common good is just 0:25:13.556 --> 0:25:18.156 unimaginably astonishing. That it's a scandal if they haven't ask 0:25:18.276 --> 0:25:21.356 the question. It's a scandal if they've asked it and 0:25:21.436 --> 0:25:24.196 they're not telling us what they found. I want to 0:25:24.196 --> 0:25:27.436 live in a world where digital power is both guided 0:25:27.476 --> 0:25:31.196 by evidence and accountable to the public, and so I'm 0:25:31.356 --> 0:25:35.636 very dissatisfied when people tell me they haven't asked the 0:25:35.716 --> 0:25:39.916 question before. Nathan, what are a few things that you 0:25:39.956 --> 0:25:44.876 could ask our podcast listeners to do to help solve 0:25:44.956 --> 0:25:47.876 this problem themselves? So are there books you think they 0:25:47.916 --> 0:25:50.916 should read? Are there, you know things they should watch 0:25:50.956 --> 0:25:54.316 to get a better understanding these ideas or their organizations. 0:25:54.516 --> 0:25:58.996 You can suggest they be involved with things they can do. Yeah. First, 0:25:59.196 --> 0:26:02.516 there are some organizations that are building up this kind 0:26:02.516 --> 0:26:07.236 of work. You can join, subscribe, or give to organizations 0:26:07.316 --> 0:26:11.796 like the Markup, like Consumer Reports, the Algorithmic Justice League, 0:26:12.276 --> 0:26:15.916 or the Citizens and Technology Lab, which I lead. In 0:26:15.956 --> 0:26:20.836 addition to that look out for opportunities to participate in research, 0:26:21.316 --> 0:26:24.476 kat Lab will be announcing some new studies later this year. 0:26:25.116 --> 0:26:29.236 Many other researchers, some of whom I've mentioned, will announce 0:26:29.276 --> 0:26:34.156 public calls asking people sign up and help us measure 0:26:34.276 --> 0:26:38.316 or test a new idea. For example, the Mozilla Foundation, 0:26:38.316 --> 0:26:41.916 who run the Firefox browser, have a volunteer program for 0:26:41.956 --> 0:26:45.756 people to sign up and collectively monitor what kinds of 0:26:45.756 --> 0:26:50.316 recommendations YouTube is making about the role of that algorithm 0:26:50.356 --> 0:26:57.196 in our lives. That was Nathan Matthias, who leads Cornell 0:26:57.356 --> 0:27:01.796 University Citizens and Technology Lab. Will include links to his 0:27:01.876 --> 0:27:05.196 suggestions for ways that you can get involved with evaluating 0:27:05.196 --> 0:27:09.876 algorithms and improving the social media ecosystem. This is the 0:27:09.956 --> 0:27:12.436 last episode of our mini series about dealing with the 0:27:12.476 --> 0:27:15.636 problem of political polarization. I'd urge you to go back 0:27:15.636 --> 0:27:19.476 and listen to previous episodes if Eli pariser with former 0:27:19.516 --> 0:27:23.196 President Juan Manuel Santos of Columbia and of course my 0:27:23.276 --> 0:27:26.556 co host Anna Applebaum, who you've just been hearing from. 0:27:26.676 --> 0:27:28.636 When you listen to them, I think you'll come away 0:27:28.676 --> 0:27:33.076 with an understanding that polarization doesn't have to keep getting worse. 0:27:33.436 --> 0:27:36.156 It's not a one way street, and there are societies 0:27:36.156 --> 0:27:39.076 we can point to where it has gotten better. But 0:27:39.196 --> 0:27:43.036 to diminish polarization, we need to address factors propelling it 0:27:43.076 --> 0:27:48.196 in technology, media and politics. Next week I'm Solvable. We'll 0:27:48.236 --> 0:27:51.756 talk with Catherine Coleman Flowers. She's the founder and director 0:27:51.796 --> 0:27:55.556 of the Center for Rural Enterprise and Environmental Justice. We'll 0:27:55.596 --> 0:27:59.196 discuss how poor sanitation in America is solvable. Yes, it's 0:27:59.236 --> 0:28:02.236 still a problem here in the United States. I hope 0:28:02.276 --> 0:28:06.996 you'll join us. Solvable Senior producer is Jocelyn Frank. Research 0:28:07.036 --> 0:28:11.236 and booking by Lisa Dunn. Managing producer is Katherine Girardou. 0:28:11.596 --> 0:28:15.996 Mia Lobell is the executive producer of Pushkin Podcast. Solvable 0:28:16.076 --> 0:28:18.956 is a production of Pushkin Industries. If you like the show, 0:28:18.996 --> 0:28:22.076