1 00:00:15,076 --> 00:00:26,276 Speaker 1: Bushkin, This is solvable. I'm Jacob Weisberg. Typically, if a 2 00:00:26,356 --> 00:00:31,076 Speaker 1: car crashes because there's say a faulty drive train, we 3 00:00:31,276 --> 00:00:35,876 Speaker 1: can point to the engineering and say there's a problem 4 00:00:35,916 --> 00:00:40,356 Speaker 1: with this system. With these adaptive systems, they're reacting and 5 00:00:40,476 --> 00:00:45,836 Speaker 1: learning and responding to human society and human behavior, and 6 00:00:45,956 --> 00:00:51,156 Speaker 1: we're still developing the scientific tools to understand what it 7 00:00:51,236 --> 00:00:56,396 Speaker 1: means to have those feedback loops. Algorithms are adaptive systems. 8 00:00:56,876 --> 00:00:59,956 Speaker 1: They're pieces of computer code that shape many aspects of 9 00:00:59,956 --> 00:01:05,036 Speaker 1: our digital lives. They're closely guarded trade secrets and powerful tools, 10 00:01:05,676 --> 00:01:11,036 Speaker 1: and they're regularly blamed for amplifying cultural and political divisions. 11 00:01:11,796 --> 00:01:16,836 Speaker 1: We often hear technologists say we couldn't have known, and 12 00:01:17,796 --> 00:01:23,116 Speaker 1: the idea that they haven't turned those lenses on questions 13 00:01:23,316 --> 00:01:27,196 Speaker 1: impacting the common good. It's a scandal if they haven't 14 00:01:27,236 --> 00:01:30,276 Speaker 1: ask the question. It's a scandal if they've asked it 15 00:01:30,316 --> 00:01:32,876 Speaker 1: and they're not telling us what they found. This is 16 00:01:32,876 --> 00:01:36,316 Speaker 1: a fourth chapter in our Solvable series examining solutions for 17 00:01:36,436 --> 00:01:41,876 Speaker 1: America's polarization problem. Today we're talking about social media algorithms 18 00:01:42,196 --> 00:01:44,356 Speaker 1: and how to deal with them. You can think of 19 00:01:44,396 --> 00:01:48,236 Speaker 1: social media companies as fancy restaurants. The cooks behind the 20 00:01:48,236 --> 00:01:51,636 Speaker 1: most successful one often don't want to reveal their recipes, 21 00:01:52,156 --> 00:01:54,636 Speaker 1: but customers have a right to know what they're eating. 22 00:01:55,476 --> 00:01:59,116 Speaker 1: It turns out we've been down this road before. The 23 00:01:59,236 --> 00:02:04,356 Speaker 1: Good Housekeeping labs started just around the turn of the century. 24 00:02:04,876 --> 00:02:08,156 Speaker 1: People were concerned about what was in their food what 25 00:02:08,276 --> 00:02:11,916 Speaker 1: was in others. This was before the creation of the FDA, 26 00:02:11,956 --> 00:02:16,396 Speaker 1: and so people subscribed to Good Housekeeping. Those labs would 27 00:02:16,916 --> 00:02:21,036 Speaker 1: test common products and tell people if they were safe 28 00:02:21,196 --> 00:02:24,916 Speaker 1: or not. Ultimately, the federal government stepped in to regulate 29 00:02:24,996 --> 00:02:31,876 Speaker 1: food safety, including disclosure requirements around ingredients, nutrients, and calories. Similarly, 30 00:02:32,316 --> 00:02:37,756 Speaker 1: establishing algorithmic safety and accountability will take a variety of players. 31 00:02:38,356 --> 00:02:40,876 Speaker 1: I want to live in a world where digital power 32 00:02:41,076 --> 00:02:44,356 Speaker 1: is both guided by evidence and accountable to the public. 33 00:02:44,996 --> 00:02:48,876 Speaker 1: Nathan Mathias teaches at Cornell University and leads the Citizens 34 00:02:48,916 --> 00:02:53,796 Speaker 1: and Technology Lab. The problems of regulating algorithms are solvable. 35 00:02:58,356 --> 00:03:01,716 Speaker 1: My co host an Apple Bomb, spoke with Nathan Mathias. 36 00:03:01,756 --> 00:03:07,516 Speaker 1: Here's their conversation. So Nathan, tell me what's an algorithm. 37 00:03:08,396 --> 00:03:12,236 Speaker 1: Algorithms can be thought of as a recipe, a series 38 00:03:12,356 --> 00:03:16,716 Speaker 1: of steps often programmed into a computer that determine how 39 00:03:16,756 --> 00:03:22,236 Speaker 1: a machine behaves. But the challenge, as any cook often finds, 40 00:03:22,596 --> 00:03:24,476 Speaker 1: is that when you put them out into the world, 41 00:03:24,636 --> 00:03:28,876 Speaker 1: especially something of sufficient complexity, they often behave in ways 42 00:03:29,076 --> 00:03:32,636 Speaker 1: that are different from what we expect. Can you just 43 00:03:32,636 --> 00:03:36,436 Speaker 1: take a minute to explain how that's problematic and why 44 00:03:36,556 --> 00:03:40,756 Speaker 1: why should we care that algorithms are deciding which piece 45 00:03:40,796 --> 00:03:44,716 Speaker 1: of content you see on Facebook or which video you're 46 00:03:44,716 --> 00:03:49,396 Speaker 1: being recommended on YouTube. Algorithms happen at all levels, from 47 00:03:50,196 --> 00:03:54,076 Speaker 1: exactly how the electrons go from one point to another 48 00:03:54,236 --> 00:03:57,316 Speaker 1: on the Internet to the much more high level things 49 00:03:57,436 --> 00:04:02,116 Speaker 1: that we think about in our direct experience. For example, 50 00:04:02,356 --> 00:04:07,196 Speaker 1: an algorithm determines what your email imbacts decides is spam, 51 00:04:07,596 --> 00:04:12,076 Speaker 1: an algorithm on Twitter decides which faces to show when 52 00:04:12,156 --> 00:04:17,476 Speaker 1: it's displaying a photo. And algorithms also and critically make 53 00:04:17,556 --> 00:04:23,356 Speaker 1: decisions about what information to prioritize when showing us feeds 54 00:04:23,436 --> 00:04:28,676 Speaker 1: on Facebook, on Twitter, when determining which adds we see 55 00:04:29,076 --> 00:04:33,716 Speaker 1: which adds we don't, and those are often some of 56 00:04:33,756 --> 00:04:38,716 Speaker 1: the uses of algorithms that people worry about in society 57 00:04:38,716 --> 00:04:43,236 Speaker 1: and policy circles. YouTube makes a recommendation system to help 58 00:04:43,316 --> 00:04:46,516 Speaker 1: us find the videos we like, and suddenly we're worrying 59 00:04:46,556 --> 00:04:52,676 Speaker 1: about recommendations of extremism. Microsoft makes a fun chat assistant 60 00:04:53,116 --> 00:04:56,716 Speaker 1: that will have interesting conversations with you, and now we're 61 00:04:56,716 --> 00:05:01,236 Speaker 1: worrying about it learning racism and hatred. So we find 62 00:05:01,356 --> 00:05:05,276 Speaker 1: that although we have the simple building blocks of an 63 00:05:05,276 --> 00:05:09,716 Speaker 1: algorithm that an engineer can imagine, they often grow to 64 00:05:09,796 --> 00:05:14,676 Speaker 1: be something larger than we might omit initially imagine. I've 65 00:05:14,676 --> 00:05:18,556 Speaker 1: written and others have written about the problem of algorithms 66 00:05:18,556 --> 00:05:25,036 Speaker 1: on Facebook favoritizing or preferring content and posts that are emotional, 67 00:05:25,236 --> 00:05:29,196 Speaker 1: that are negative, that are divisive. You know, there's been 68 00:05:29,196 --> 00:05:32,116 Speaker 1: an argument that that's one of the reasons why we 69 00:05:32,196 --> 00:05:34,956 Speaker 1: have so much division and polarization in our societies, that 70 00:05:35,356 --> 00:05:39,836 Speaker 1: we are being fed more and more excitable and angry 71 00:05:39,876 --> 00:05:43,436 Speaker 1: content because the algorithm tests and guesses that that's what 72 00:05:43,436 --> 00:05:45,036 Speaker 1: we're going to want to see or anyway that's what's 73 00:05:45,036 --> 00:05:47,516 Speaker 1: going to keep us online or keep us using Facebook. 74 00:05:47,956 --> 00:05:49,956 Speaker 1: Is it accurate? Is that how they work? We do 75 00:05:50,036 --> 00:05:53,436 Speaker 1: live in a world where many of the systems that 76 00:05:53,556 --> 00:05:57,436 Speaker 1: determined what we see and give our attention to our 77 00:05:57,556 --> 00:06:02,116 Speaker 1: learning from our behavior, our preferences, and from the collective 78 00:06:02,156 --> 00:06:05,636 Speaker 1: behavior of many others, some of who aren't paying. Some 79 00:06:05,676 --> 00:06:09,956 Speaker 1: of them have motivated, coordinated campaigns to influence the algorithms, 80 00:06:10,476 --> 00:06:14,796 Speaker 1: and they're adapting in real time. And so because we've 81 00:06:14,836 --> 00:06:19,276 Speaker 1: never really faced a situation like this at such scale, 82 00:06:19,796 --> 00:06:22,876 Speaker 1: people have a lot of concerns about how those algorithms 83 00:06:22,916 --> 00:06:27,596 Speaker 1: are behaving and what they're doing to society. One of 84 00:06:27,636 --> 00:06:32,356 Speaker 1: the fundamental challenges that I think scientists are still wrestling 85 00:06:32,396 --> 00:06:37,636 Speaker 1: with is this challenge of influence. Typically, if a car 86 00:06:37,756 --> 00:06:42,916 Speaker 1: crashes because there's say a faulty drive train, we can 87 00:06:42,956 --> 00:06:47,116 Speaker 1: point to the engineering and say there's a problem with 88 00:06:47,196 --> 00:06:51,876 Speaker 1: this system. With these adaptive systems, they're reacting and learning 89 00:06:51,916 --> 00:06:57,236 Speaker 1: and responding to human society and human behavior. And we're 90 00:06:57,316 --> 00:07:02,756 Speaker 1: still developing the scientific tools to understand what it means 91 00:07:02,836 --> 00:07:06,076 Speaker 1: to have those feedback loops. And in the meantime, we 92 00:07:06,156 --> 00:07:08,116 Speaker 1: have to live in a world where these things have 93 00:07:08,516 --> 00:07:12,676 Speaker 1: very real power. If the Facebook algorithm is designed to 94 00:07:12,756 --> 00:07:15,196 Speaker 1: keep all of us on Facebook as long as possible, 95 00:07:15,756 --> 00:07:18,476 Speaker 1: who's able to watch that, who's able to control it, 96 00:07:18,956 --> 00:07:22,316 Speaker 1: who's following the science? Almost no one is in a 97 00:07:22,436 --> 00:07:28,676 Speaker 1: position right now to regulate and manage those algorithms. For example, 98 00:07:28,956 --> 00:07:34,516 Speaker 1: in February, Facebook announced that they would be reducing the 99 00:07:34,676 --> 00:07:40,116 Speaker 1: political content appearing in people's news feeds in several countries. 100 00:07:40,636 --> 00:07:43,516 Speaker 1: We don't really know the details of what they're doing. 101 00:07:43,916 --> 00:07:46,956 Speaker 1: We also have evidence, because they say they're doing tests, 102 00:07:47,276 --> 00:07:51,036 Speaker 1: that they're not necessarily sure themselves what the impact is 103 00:07:51,076 --> 00:07:54,876 Speaker 1: going to be. When you think about who currently has 104 00:07:55,676 --> 00:07:59,596 Speaker 1: some power to shape what algorithms do, I think there 105 00:07:59,636 --> 00:08:03,516 Speaker 1: are some people at different levels of society who have 106 00:08:03,596 --> 00:08:06,756 Speaker 1: a little bit of influence. We've seen, for example, European 107 00:08:06,836 --> 00:08:12,836 Speaker 1: regulators step in around antitrust around what kinds of products 108 00:08:13,036 --> 00:08:17,396 Speaker 1: get promoted by search engines, for example, So governments have 109 00:08:17,476 --> 00:08:22,436 Speaker 1: been doing a little bit. We definitely have companies themselves 110 00:08:22,436 --> 00:08:26,516 Speaker 1: are being seen as almost government like and having policy teams, 111 00:08:26,516 --> 00:08:29,156 Speaker 1: so they're trying to understand how their own systems work 112 00:08:29,636 --> 00:08:33,236 Speaker 1: and manage them in some way without the rest of 113 00:08:33,316 --> 00:08:36,476 Speaker 1: us having that much transparency into their values or goals 114 00:08:36,596 --> 00:08:40,836 Speaker 1: or even their results. And then in some areas there 115 00:08:40,876 --> 00:08:44,956 Speaker 1: are other actors who have power to manage and govern 116 00:08:45,116 --> 00:08:47,956 Speaker 1: algorithms in a constraint way. So if you've ever been 117 00:08:47,996 --> 00:08:52,276 Speaker 1: a Facebook group administrator, for example, or you know someone 118 00:08:52,316 --> 00:08:55,116 Speaker 1: who's a Reddit moderator. They have a little bit of 119 00:08:55,116 --> 00:08:58,996 Speaker 1: an ability to tweak what gets promoted or how they 120 00:08:59,316 --> 00:09:02,036 Speaker 1: given algorithm works, even though they don't have a lot 121 00:09:02,076 --> 00:09:06,916 Speaker 1: of visibility into the underlying code or necessarily the power 122 00:09:06,956 --> 00:09:09,476 Speaker 1: to tell a big company to change what they do. 123 00:09:10,756 --> 00:09:13,356 Speaker 1: A couple of weeks ago, I had reason to talk 124 00:09:13,396 --> 00:09:17,756 Speaker 1: to a Facebook spokesman, and the topic was the experiments 125 00:09:17,796 --> 00:09:21,156 Speaker 1: that Facebook does with its algorithms, the way in which 126 00:09:21,196 --> 00:09:23,876 Speaker 1: they test different things. As you say, they try and 127 00:09:23,956 --> 00:09:27,196 Speaker 1: use more or less political content. You know. Actually, after 128 00:09:27,236 --> 00:09:31,876 Speaker 1: the events on January sixth at the Capitol, they came 129 00:09:31,956 --> 00:09:34,756 Speaker 1: up with a way of moderating the news feed so 130 00:09:34,836 --> 00:09:37,036 Speaker 1: that there wouldn't be so much disinformation in it. But 131 00:09:37,156 --> 00:09:40,076 Speaker 1: of course, as you say, they don't publish the results 132 00:09:40,076 --> 00:09:43,916 Speaker 1: of these experiments or of these changes. One of the 133 00:09:43,996 --> 00:09:46,996 Speaker 1: solutions that I know that you have suggested is that 134 00:09:47,036 --> 00:09:51,396 Speaker 1: there should be outside moderators, or there should be citizens 135 00:09:51,396 --> 00:09:55,676 Speaker 1: scientists who are studying these algorithms, you know, either with 136 00:09:55,756 --> 00:09:58,636 Speaker 1: the cooperation of Facebook and Google or maybe not with 137 00:09:58,396 --> 00:10:02,796 Speaker 1: their cooperation. What step one The place to start is 138 00:10:02,836 --> 00:10:07,036 Speaker 1: often with your own experience. I'll tell you a story 139 00:10:07,116 --> 00:10:10,876 Speaker 1: just to illustrate this about six years or ago, I 140 00:10:10,916 --> 00:10:15,916 Speaker 1: was approached by a group of women who faced online harassment, 141 00:10:16,236 --> 00:10:19,636 Speaker 1: threats of violence, and other kinds of risks. For them. 142 00:10:20,076 --> 00:10:22,756 Speaker 1: The first step was to acknowledge that it was a 143 00:10:22,756 --> 00:10:28,156 Speaker 1: problem and to find other people who had the same problem. 144 00:10:28,196 --> 00:10:30,916 Speaker 1: They were able to realize that they had common needs 145 00:10:30,996 --> 00:10:33,836 Speaker 1: and common goals, and they actually came up with a 146 00:10:33,876 --> 00:10:38,516 Speaker 1: way to record their experiences, both the kinds of harassment 147 00:10:38,796 --> 00:10:43,396 Speaker 1: that they were facing, and also to record how Twitter 148 00:10:43,756 --> 00:10:47,996 Speaker 1: did or often didn't handle their reports. That was the 149 00:10:48,036 --> 00:10:50,476 Speaker 1: point actually that they then reached out to me and said, 150 00:10:51,356 --> 00:10:55,276 Speaker 1: this is clearly a systemic problem. We've all experienced it, 151 00:10:55,436 --> 00:10:59,036 Speaker 1: we want to see change. We know that better understanding 152 00:10:59,156 --> 00:11:03,196 Speaker 1: data and science will help us think of better solutions 153 00:11:03,476 --> 00:11:07,156 Speaker 1: and also, if necessary, to create pressure for those solutions. 154 00:11:07,716 --> 00:11:10,196 Speaker 1: That was a great moment then for me and the 155 00:11:10,196 --> 00:11:14,076 Speaker 1: team of researchers I led to develop a methodology and 156 00:11:14,156 --> 00:11:17,396 Speaker 1: analyze the data they were collecting, and that report that 157 00:11:17,436 --> 00:11:21,916 Speaker 1: we ended up creating has been influential in industry. It's 158 00:11:21,916 --> 00:11:25,276 Speaker 1: helped law enforcement understand how to better support people who 159 00:11:25,316 --> 00:11:29,476 Speaker 1: experience online harassment, and it's also been useful in policy 160 00:11:29,476 --> 00:11:33,076 Speaker 1: debates in this country about online harassment. We're at a 161 00:11:33,156 --> 00:11:37,956 Speaker 1: moment where we're still building the lines of communication and 162 00:11:38,116 --> 00:11:42,236 Speaker 1: the idea of citizen science as a mode of understanding 163 00:11:42,236 --> 00:11:46,636 Speaker 1: and improving our digital lives. So at this stage, I 164 00:11:46,676 --> 00:11:49,396 Speaker 1: think the best first step is really to find other 165 00:11:49,436 --> 00:11:53,156 Speaker 1: people who care about the thing you care about. So 166 00:11:53,196 --> 00:11:56,516 Speaker 1: we need to identify the problems that have to be studied, 167 00:11:56,596 --> 00:11:59,236 Speaker 1: and we need the labs where they can be studied. 168 00:11:59,756 --> 00:12:04,236 Speaker 1: That's the first step. Absolutely, there's another important step at 169 00:12:04,316 --> 00:12:08,836 Speaker 1: the ecosystem level. There's a funding challenge. Most of the 170 00:12:09,156 --> 00:12:12,356 Speaker 1: search that goes into funding that goes into social computing 171 00:12:12,716 --> 00:12:16,356 Speaker 1: comes from the tech industry, like hundreds of millions of dollars, 172 00:12:16,676 --> 00:12:19,716 Speaker 1: and if you look at the money that comes into 173 00:12:19,876 --> 00:12:23,756 Speaker 1: industry independent research, it's a tiny drop in the bucket. 174 00:12:24,196 --> 00:12:28,996 Speaker 1: So as policymakers debate ideas like taxing tech companies, I 175 00:12:28,996 --> 00:12:33,276 Speaker 1: could imagine they're being funding within that for industry independent 176 00:12:33,356 --> 00:12:38,716 Speaker 1: accountability research. We're also finding ourselves having to invent new 177 00:12:39,396 --> 00:12:42,556 Speaker 1: funding models for this kind of research as well. And 178 00:12:42,596 --> 00:12:48,716 Speaker 1: then presumably at some point, some regulatory mechanism that makes 179 00:12:48,756 --> 00:12:52,796 Speaker 1: sure that the Internet platforms will work with you and 180 00:12:52,836 --> 00:12:55,756 Speaker 1: we'll listen to you exactly. So I think we're seeing 181 00:12:55,796 --> 00:13:00,316 Speaker 1: more and more researchers in this space say that we're 182 00:13:00,356 --> 00:13:04,436 Speaker 1: going to need some kind of regulation to provide protections 183 00:13:04,476 --> 00:13:08,396 Speaker 1: and support for independent research to go on even when 184 00:13:08,436 --> 00:13:12,996 Speaker 1: companies find it uncomfortable. One of the crises, you know 185 00:13:13,036 --> 00:13:16,116 Speaker 1: at the moment in American life is the fact that 186 00:13:16,156 --> 00:13:19,676 Speaker 1: a part of the country now lives in a completely 187 00:13:19,716 --> 00:13:22,236 Speaker 1: alternative universe from the rest of the country. And we 188 00:13:22,276 --> 00:13:25,116 Speaker 1: all saw on January the six that there are people 189 00:13:25,116 --> 00:13:27,556 Speaker 1: who are so convinced that Donald Trump won the election 190 00:13:27,596 --> 00:13:31,156 Speaker 1: that they were willing to attack the capital and even 191 00:13:31,356 --> 00:13:36,596 Speaker 1: murder policemen and other in an attempt to disrupt Congress's 192 00:13:36,636 --> 00:13:40,516 Speaker 1: work and prevent the naming of verifying of Joe Biden 193 00:13:41,036 --> 00:13:44,396 Speaker 1: as president. How do you relate that to this problem 194 00:13:44,436 --> 00:13:46,796 Speaker 1: of algorithms. I mean, if we had if we could 195 00:13:46,836 --> 00:13:49,916 Speaker 1: solve the algorithm problem, if we if we were able 196 00:13:49,956 --> 00:13:53,716 Speaker 1: to structure algorithms so that they favored civic discourse and 197 00:13:54,276 --> 00:13:58,596 Speaker 1: civil conversation instead of promoting division and anger, could that 198 00:13:58,756 --> 00:14:03,876 Speaker 1: help us heal this deep divide, this epistemological divide whereby 199 00:14:03,876 --> 00:14:06,916 Speaker 1: we all live in different realities. You know, we know 200 00:14:06,996 --> 00:14:09,516 Speaker 1: that when crowds of people get involved in stuff that 201 00:14:09,556 --> 00:14:12,236 Speaker 1: doesn't necessarily mean that the outcome is good or better. 202 00:14:13,036 --> 00:14:16,596 Speaker 1: So why should we be so sure that citizen participation 203 00:14:16,676 --> 00:14:19,676 Speaker 1: in the regulation of the internet will give us good regulation. 204 00:14:20,196 --> 00:14:25,716 Speaker 1: It's important to differentiate between who's making decisions and who's 205 00:14:25,796 --> 00:14:30,356 Speaker 1: producing evidence. Evidence is something that you can put into 206 00:14:30,556 --> 00:14:34,076 Speaker 1: the conversation about what to do, and so long as 207 00:14:34,116 --> 00:14:39,036 Speaker 1: that evidence is produced in a reliable way, it has 208 00:14:39,156 --> 00:14:42,596 Speaker 1: value to bring to the conversation. So your feeling is 209 00:14:42,636 --> 00:14:46,316 Speaker 1: that this is a question. It's not just important for 210 00:14:46,796 --> 00:14:48,756 Speaker 1: I don't know the future of social media. It's really 211 00:14:48,836 --> 00:14:52,436 Speaker 1: the question it's important for democracy giving that power, giving 212 00:14:52,596 --> 00:14:58,316 Speaker 1: some of that oversight ability to citizen scientists, to outside groups, 213 00:14:58,436 --> 00:15:03,396 Speaker 1: maybe to some government ombudsman, maybe to some regulators, that 214 00:15:03,476 --> 00:15:08,116 Speaker 1: this would democratize that power that social media companies have. Yeah, 215 00:15:08,796 --> 00:15:12,716 Speaker 1: one of my personal heroes in the social sciences is 216 00:15:12,796 --> 00:15:17,076 Speaker 1: Kurt Lewin, one of the founders of social psychology, who 217 00:15:17,196 --> 00:15:22,596 Speaker 1: himself barely escaped Nazi Germany with his life and went 218 00:15:22,636 --> 00:15:27,036 Speaker 1: on to influence so much in science and society. And 219 00:15:27,076 --> 00:15:30,556 Speaker 1: he had this great quote which says, it's essential that 220 00:15:30,636 --> 00:15:35,236 Speaker 1: a democratic commonwealth and its educational system apply the rational 221 00:15:35,316 --> 00:15:40,516 Speaker 1: procedures of scientific investigation to its own processes of group living. 222 00:15:40,876 --> 00:15:44,956 Speaker 1: And Lewin believed that that needed to be done in 223 00:15:44,996 --> 00:15:48,916 Speaker 1: a democratic way if we were going to maintain the 224 00:15:49,076 --> 00:15:53,516 Speaker 1: values that we have as democratic societies. That it wasn't 225 00:15:53,636 --> 00:15:58,876 Speaker 1: just enough to do research that supported democracy. You needed 226 00:15:58,996 --> 00:16:02,836 Speaker 1: the research itself to be democratic in nature. And I 227 00:16:02,876 --> 00:16:06,476 Speaker 1: think in an era where so much of what we 228 00:16:06,676 --> 00:16:13,196 Speaker 1: do is influenced by design and algorithms, that reality is 229 00:16:13,636 --> 00:16:18,356 Speaker 1: clearer than it even was in Lewin's time. There's a 230 00:16:18,396 --> 00:16:23,996 Speaker 1: long tradition of citizens, scientists, and outsiders working outside the 231 00:16:23,996 --> 00:16:27,116 Speaker 1: government are sometimes in tandem with the government in order 232 00:16:27,156 --> 00:16:30,796 Speaker 1: to push regulation or particular direction. Do you see yourself 233 00:16:30,836 --> 00:16:32,876 Speaker 1: belonging to that tradition and can you describe it a 234 00:16:32,916 --> 00:16:35,716 Speaker 1: little bit? You're asking me a question about something that 235 00:16:35,836 --> 00:16:42,156 Speaker 1: I absolutely love and obsessed by, so question. Yeah, you know, 236 00:16:42,236 --> 00:16:45,076 Speaker 1: I grew up, you know, in the United States as 237 00:16:45,116 --> 00:16:50,036 Speaker 1: a Guatemalan American, with this sense that science was this 238 00:16:50,116 --> 00:16:54,156 Speaker 1: tool of like powerful people in institutions that didn't always 239 00:16:54,636 --> 00:16:58,956 Speaker 1: include or pay attention to the general public or the 240 00:16:59,036 --> 00:17:03,476 Speaker 1: marginalized as anything other than research participants like you can 241 00:17:03,516 --> 00:17:06,476 Speaker 1: be a subject in the research and we will call 242 00:17:06,516 --> 00:17:09,636 Speaker 1: you a subject. But when I was a graduate at 243 00:17:09,636 --> 00:17:12,556 Speaker 1: the MIT Media Lab, I started to learn about this 244 00:17:12,596 --> 00:17:18,436 Speaker 1: amazing tradition of citizen science in different places and times 245 00:17:18,436 --> 00:17:22,676 Speaker 1: over the last really two hundred years. In the mid 246 00:17:22,796 --> 00:17:26,596 Speaker 1: nineteenth century, there was a group of people who went 247 00:17:26,636 --> 00:17:31,036 Speaker 1: around London and bought bread from different shops and used 248 00:17:31,076 --> 00:17:36,396 Speaker 1: this new idea of a microscope to count what was 249 00:17:36,476 --> 00:17:42,036 Speaker 1: actually in the bread and found widespread food adulteration. This 250 00:17:42,076 --> 00:17:47,316 Speaker 1: set of studies ended up helping launch the trajectory of 251 00:17:47,316 --> 00:17:50,116 Speaker 1: what is now the Lancet, one of the premier medical 252 00:17:50,196 --> 00:17:54,756 Speaker 1: journals in the world. Another example I really love is 253 00:17:54,796 --> 00:17:59,956 Speaker 1: the story of the Good Housekeeping Labs, which was started 254 00:18:00,436 --> 00:18:03,796 Speaker 1: just around the turn of the century. People were concerned 255 00:18:03,836 --> 00:18:07,236 Speaker 1: about what was in their food, what was in other products. 256 00:18:07,276 --> 00:18:10,196 Speaker 1: This was before the creation of they DA. There really 257 00:18:10,276 --> 00:18:14,716 Speaker 1: wasn't that much regulation of what went into the mass 258 00:18:14,796 --> 00:18:20,556 Speaker 1: production ecosystem, and so people subscribed to Good Housekeeping. Those 259 00:18:20,636 --> 00:18:25,596 Speaker 1: labs would test common products and tell people if they 260 00:18:25,636 --> 00:18:28,636 Speaker 1: were safe or not and use the good Housekeeping seal 261 00:18:28,636 --> 00:18:32,156 Speaker 1: of approval, and often in fact in the late nineteenth 262 00:18:32,156 --> 00:18:37,396 Speaker 1: early twentieth century, because there was this convergence of the 263 00:18:37,556 --> 00:18:42,036 Speaker 1: rising women's movement and a passion for science, you would 264 00:18:42,116 --> 00:18:48,036 Speaker 1: have women's organizations actually leading a lot of citizen science efforts. 265 00:18:48,076 --> 00:18:52,836 Speaker 1: And then later on when the US established the FDA, 266 00:18:52,876 --> 00:18:55,876 Speaker 1: it was actually the scientists from the Good Housekeeping Lab 267 00:18:56,316 --> 00:19:01,716 Speaker 1: that built up the FDA's initial scientific capacities and leading 268 00:19:01,796 --> 00:19:05,436 Speaker 1: us to where we are today, where we have more 269 00:19:05,676 --> 00:19:10,596 Speaker 1: organized and supported regimes of testing and science and regulation. 270 00:19:11,396 --> 00:19:15,676 Speaker 1: So when you think that's how algorithm regulation or social 271 00:19:15,716 --> 00:19:19,916 Speaker 1: media regulation could evolve with teams of citizen scientists like 272 00:19:19,996 --> 00:19:22,476 Speaker 1: the people at your lab, or is the idea that 273 00:19:22,556 --> 00:19:25,716 Speaker 1: eventually this is something the government would do or is 274 00:19:25,756 --> 00:19:28,676 Speaker 1: this something that will some other kind of civic body 275 00:19:28,756 --> 00:19:30,676 Speaker 1: will do. Do you have a kind of trajectory of 276 00:19:30,676 --> 00:19:32,756 Speaker 1: how this could work in the long term. In the 277 00:19:32,836 --> 00:19:36,996 Speaker 1: short term, citizen science and work from the outside is 278 00:19:37,036 --> 00:19:41,516 Speaker 1: a necessity. We're currently at a moment where if you 279 00:19:41,716 --> 00:19:45,116 Speaker 1: want to look at what tech companies are doing from 280 00:19:45,156 --> 00:19:49,076 Speaker 1: the inside, you have to sign these NDAs, you have 281 00:19:49,276 --> 00:19:53,636 Speaker 1: to do work that they feel comfortable with. And like 282 00:19:53,796 --> 00:19:59,556 Speaker 1: many other citizen scientists in other domains, we find ourselves 283 00:19:59,676 --> 00:20:03,636 Speaker 1: inventing methodologies to answer urging questions that people need to 284 00:20:03,716 --> 00:20:08,236 Speaker 1: understand now, and I think, you know, we have a 285 00:20:08,276 --> 00:20:11,796 Speaker 1: small but growing number of institutions that are starting to 286 00:20:11,836 --> 00:20:15,796 Speaker 1: do that work. The Barkup Consumer Reports Digital Labs has 287 00:20:15,836 --> 00:20:19,236 Speaker 1: been building a team that are initiatives like Joey boil 288 00:20:19,276 --> 00:20:23,156 Speaker 1: and Wine's Algorithmic Justice League that all do work of 289 00:20:23,196 --> 00:20:27,876 Speaker 1: this kind. In the longer term, I would love to 290 00:20:27,876 --> 00:20:31,636 Speaker 1: see a healthy ecosystem. I draw a lot of inspiration 291 00:20:31,876 --> 00:20:36,396 Speaker 1: from the work of Eleanor Ostrom, the Nobel Prize winning 292 00:20:36,436 --> 00:20:41,276 Speaker 1: political scientist who wrote about how you incorporate science into 293 00:20:41,916 --> 00:20:48,596 Speaker 1: complex governance scenarios where you have competing interests. I think 294 00:20:48,636 --> 00:20:51,356 Speaker 1: we're likely, I hope, in the long term, to get 295 00:20:51,396 --> 00:20:54,196 Speaker 1: to a point where companies are going to be more transparent. 296 00:20:54,276 --> 00:20:57,276 Speaker 1: They're going to actually publish their protocols and research on 297 00:20:57,316 --> 00:20:59,556 Speaker 1: the issues we care about, and that's going to be 298 00:20:59,596 --> 00:21:05,076 Speaker 1: an important part. I think we really desperately need more 299 00:21:05,676 --> 00:21:09,036 Speaker 1: government supported efforts, and I'll leave it to the policy 300 00:21:09,116 --> 00:21:12,036 Speaker 1: makers to figure out what that actually looks like. And 301 00:21:12,076 --> 00:21:15,876 Speaker 1: I think will continue to see citizen scientists trying to 302 00:21:15,916 --> 00:21:20,196 Speaker 1: make sense of and improve their own contexts and environment, 303 00:21:20,356 --> 00:21:24,636 Speaker 1: just like we have in the arena of environmental protection, 304 00:21:24,756 --> 00:21:29,916 Speaker 1: consumer protection. Those are all mature ecosystems where you have 305 00:21:30,036 --> 00:21:35,676 Speaker 1: science happening from different perspectives and different points in the ecosystem. Right, 306 00:21:35,716 --> 00:21:38,996 Speaker 1: So it's not just government scientists. They're also independent scientists. 307 00:21:39,036 --> 00:21:42,516 Speaker 1: And there's the Sierra Club, and they're individuals and they're 308 00:21:42,956 --> 00:21:45,236 Speaker 1: you know, so there there are lots of different perspectives 309 00:21:45,236 --> 00:21:48,396 Speaker 1: on the same environmental problem. And you imagine that that 310 00:21:48,636 --> 00:21:53,196 Speaker 1: would eventually be possible in monitoring and regulating the social 311 00:21:53,196 --> 00:21:56,916 Speaker 1: media companies too, exactly, and in democracy, we hope that 312 00:21:56,956 --> 00:22:01,036 Speaker 1: having multiple perspectives helps us get to a better solution. 313 00:22:01,516 --> 00:22:04,196 Speaker 1: At least that's that's the vision of democracy I want. 314 00:22:04,676 --> 00:22:08,956 Speaker 1: I want to cling to in how I imagine the work. 315 00:22:09,156 --> 00:22:13,756 Speaker 1: And so I think we need that for governing social media, 316 00:22:13,916 --> 00:22:17,676 Speaker 1: for governing the role of digital technologies in our lives, 317 00:22:17,956 --> 00:22:19,956 Speaker 1: and we have a lot of work to build up 318 00:22:19,956 --> 00:22:24,796 Speaker 1: the industry independent part of that ecosystem. Nathan, I know 319 00:22:24,916 --> 00:22:28,836 Speaker 1: that you started your education in the humanities and you 320 00:22:29,356 --> 00:22:32,596 Speaker 1: moved later on to technology. Can you tell me a 321 00:22:32,636 --> 00:22:35,716 Speaker 1: little bit about how that happened? How does an English 322 00:22:35,756 --> 00:22:39,156 Speaker 1: major become part of this other world? When I was 323 00:22:39,196 --> 00:22:45,156 Speaker 1: a teenager, I had this amazing opportunity to meet with 324 00:22:45,236 --> 00:22:48,876 Speaker 1: and talk to a local computer science professor. I was 325 00:22:49,076 --> 00:22:52,636 Speaker 1: really passionate about the arts. I was really passionate about computing, 326 00:22:53,356 --> 00:22:56,596 Speaker 1: and he said, computing as a lens on the world. 327 00:22:57,076 --> 00:23:02,156 Speaker 1: If you really care about understanding technology, you need to 328 00:23:02,316 --> 00:23:08,076 Speaker 1: understand society. You need to pay close attention to the 329 00:23:08,116 --> 00:23:13,596 Speaker 1: world around you, because computing without that has no heart, 330 00:23:13,676 --> 00:23:17,836 Speaker 1: it has no moral compass. With his encouragement, I felt 331 00:23:17,996 --> 00:23:24,276 Speaker 1: empowered and prompted to spend my undergraduate time reading literature, 332 00:23:24,396 --> 00:23:29,596 Speaker 1: studying the humanities, asking myself the big questions. Was really 333 00:23:29,916 --> 00:23:34,196 Speaker 1: during my second undergraduate degree, when I was a student 334 00:23:34,316 --> 00:23:38,356 Speaker 1: at Cambridge University that I started to ask questions about 335 00:23:38,996 --> 00:23:43,156 Speaker 1: literature and what we read, and its impact on democracy, 336 00:23:43,276 --> 00:23:49,076 Speaker 1: its impact and connections to psychology. I realized that not 337 00:23:49,276 --> 00:23:54,636 Speaker 1: only were we collecting massive amounts of data about human 338 00:23:54,676 --> 00:23:58,156 Speaker 1: experience and behavior that could help us answer some of 339 00:23:58,156 --> 00:24:03,156 Speaker 1: those questions. I also realized that those enduring questions about 340 00:24:03,196 --> 00:24:06,876 Speaker 1: what it means to live well together in society that 341 00:24:06,996 --> 00:24:11,636 Speaker 1: we've been asking as long as we've had written records 342 00:24:11,796 --> 00:24:16,196 Speaker 1: are incredibly important to the present time. And that's what 343 00:24:16,396 --> 00:24:18,876 Speaker 1: led me to actually go back to grad school and 344 00:24:19,716 --> 00:24:23,876 Speaker 1: study those questions further. And those aren't questions that are 345 00:24:23,916 --> 00:24:27,596 Speaker 1: normally asked in Silicon Valley, presumably. And I don't know 346 00:24:27,676 --> 00:24:31,516 Speaker 1: if I can speak for all of Silicon Valley, but 347 00:24:31,596 --> 00:24:37,116 Speaker 1: I do think that I think we often hear technologists 348 00:24:37,196 --> 00:24:42,636 Speaker 1: say we couldn't have known, and I can't really tell 349 00:24:42,796 --> 00:24:49,636 Speaker 1: whether that's true or whether it's a rhetorical line to take, 350 00:24:50,516 --> 00:24:56,356 Speaker 1: because the reality is that companies have built some of 351 00:24:56,396 --> 00:25:02,516 Speaker 1: the world's most sophisticated social scientific research endeavors in the 352 00:25:02,596 --> 00:25:07,316 Speaker 1: history of humanity, and the idea that they haven't turned 353 00:25:07,396 --> 00:25:12,996 Speaker 1: those lenses on questions impacting the common good is just 354 00:25:13,556 --> 00:25:18,156 Speaker 1: unimaginably astonishing. That it's a scandal if they haven't ask 355 00:25:18,276 --> 00:25:21,356 Speaker 1: the question. It's a scandal if they've asked it and 356 00:25:21,436 --> 00:25:24,196 Speaker 1: they're not telling us what they found. I want to 357 00:25:24,196 --> 00:25:27,436 Speaker 1: live in a world where digital power is both guided 358 00:25:27,476 --> 00:25:31,196 Speaker 1: by evidence and accountable to the public, and so I'm 359 00:25:31,356 --> 00:25:35,636 Speaker 1: very dissatisfied when people tell me they haven't asked the 360 00:25:35,716 --> 00:25:39,916 Speaker 1: question before. Nathan, what are a few things that you 361 00:25:39,956 --> 00:25:44,876 Speaker 1: could ask our podcast listeners to do to help solve 362 00:25:44,956 --> 00:25:47,876 Speaker 1: this problem themselves? So are there books you think they 363 00:25:47,916 --> 00:25:50,916 Speaker 1: should read? Are there, you know things they should watch 364 00:25:50,956 --> 00:25:54,316 Speaker 1: to get a better understanding these ideas or their organizations. 365 00:25:54,516 --> 00:25:58,996 Speaker 1: You can suggest they be involved with things they can do. Yeah. First, 366 00:25:59,196 --> 00:26:02,516 Speaker 1: there are some organizations that are building up this kind 367 00:26:02,516 --> 00:26:07,236 Speaker 1: of work. You can join, subscribe, or give to organizations 368 00:26:07,316 --> 00:26:11,796 Speaker 1: like the Markup, like Consumer Reports, the Algorithmic Justice League, 369 00:26:12,276 --> 00:26:15,916 Speaker 1: or the Citizens and Technology Lab, which I lead. In 370 00:26:15,956 --> 00:26:20,836 Speaker 1: addition to that look out for opportunities to participate in research, 371 00:26:21,316 --> 00:26:24,476 Speaker 1: kat Lab will be announcing some new studies later this year. 372 00:26:25,116 --> 00:26:29,236 Speaker 1: Many other researchers, some of whom I've mentioned, will announce 373 00:26:29,276 --> 00:26:34,156 Speaker 1: public calls asking people sign up and help us measure 374 00:26:34,276 --> 00:26:38,316 Speaker 1: or test a new idea. For example, the Mozilla Foundation, 375 00:26:38,316 --> 00:26:41,916 Speaker 1: who run the Firefox browser, have a volunteer program for 376 00:26:41,956 --> 00:26:45,756 Speaker 1: people to sign up and collectively monitor what kinds of 377 00:26:45,756 --> 00:26:50,316 Speaker 1: recommendations YouTube is making about the role of that algorithm 378 00:26:50,356 --> 00:26:57,196 Speaker 1: in our lives. That was Nathan Matthias, who leads Cornell 379 00:26:57,356 --> 00:27:01,796 Speaker 1: University Citizens and Technology Lab. Will include links to his 380 00:27:01,876 --> 00:27:05,196 Speaker 1: suggestions for ways that you can get involved with evaluating 381 00:27:05,196 --> 00:27:09,876 Speaker 1: algorithms and improving the social media ecosystem. This is the 382 00:27:09,956 --> 00:27:12,436 Speaker 1: last episode of our mini series about dealing with the 383 00:27:12,476 --> 00:27:15,636 Speaker 1: problem of political polarization. I'd urge you to go back 384 00:27:15,636 --> 00:27:19,476 Speaker 1: and listen to previous episodes if Eli pariser with former 385 00:27:19,516 --> 00:27:23,196 Speaker 1: President Juan Manuel Santos of Columbia and of course my 386 00:27:23,276 --> 00:27:26,556 Speaker 1: co host Anna Applebaum, who you've just been hearing from. 387 00:27:26,676 --> 00:27:28,636 Speaker 1: When you listen to them, I think you'll come away 388 00:27:28,676 --> 00:27:33,076 Speaker 1: with an understanding that polarization doesn't have to keep getting worse. 389 00:27:33,436 --> 00:27:36,156 Speaker 1: It's not a one way street, and there are societies 390 00:27:36,156 --> 00:27:39,076 Speaker 1: we can point to where it has gotten better. But 391 00:27:39,196 --> 00:27:43,036 Speaker 1: to diminish polarization, we need to address factors propelling it 392 00:27:43,076 --> 00:27:48,196 Speaker 1: in technology, media and politics. Next week I'm Solvable. We'll 393 00:27:48,236 --> 00:27:51,756 Speaker 1: talk with Catherine Coleman Flowers. She's the founder and director 394 00:27:51,796 --> 00:27:55,556 Speaker 1: of the Center for Rural Enterprise and Environmental Justice. We'll 395 00:27:55,596 --> 00:27:59,196 Speaker 1: discuss how poor sanitation in America is solvable. Yes, it's 396 00:27:59,236 --> 00:28:02,236 Speaker 1: still a problem here in the United States. I hope 397 00:28:02,276 --> 00:28:06,996 Speaker 1: you'll join us. Solvable Senior producer is Jocelyn Frank. Research 398 00:28:07,036 --> 00:28:11,236 Speaker 1: and booking by Lisa Dunn. Managing producer is Katherine Girardou. 399 00:28:11,596 --> 00:28:15,996 Speaker 1: Mia Lobell is the executive producer of Pushkin Podcast. Solvable 400 00:28:16,076 --> 00:28:18,956 Speaker 1: is a production of Pushkin Industries. If you like the show, 401 00:28:18,996 --> 00:28:22,076 Speaker 1: please remember to share, rate, and review us. It really 402 00:28:22,076 --> 00:28:24,636 Speaker 1: helps to get the word out. You can find Pushkin 403 00:28:24,716 --> 00:28:28,516 Speaker 1: podcasts wherever you listen, including on the iHeartRadio app and 404 00:28:28,636 --> 00:28:31,396 Speaker 1: Apple Podcasts. I'm Jacob Weisberg.