Dr. Safiya Umoja Noble is an Associate Professor at the University of California, Los Angeles (UCLA) in the Department of Information Studies where she serves as the Co-Founder and Co-Director of the UCLA Center for Critical Internet Inquiry (C2i2). She also holds appointments in African American Studies and Gender Studies. She is a Research Associate at the Oxford Internet Institute at the University of Oxford and has been appointed as a Commissioner on the Oxford Commission on AI & Good Governance (OxCAIGG). She is a board member of the Cyber Civil Rights Initiative, serving those vulnerable to online harassment.

Previously, she was a visiting faculty member to the USC Annenberg School for Communication and Journalism, and began her academic career as an Assistant Professor in the College of Media and the Institute of Communications Research at the University of Illinois at Urbana-Champaign.

In this week’s episode, we talk about her work, her book, and the intersection of race and technology. We also talk about the concept of “technological redlining” and what this kind of discrimination means for people of color today and in the future.

Episode Notes

Safiya’s book, Algorithms of Oppression: How Search Engines Reinforce Racism

Safiya’s website

Safiya on Twitter

UCLA Center for Critical Internet Inquiry

UCLA Center for Critical Internet Inquiry on Twitter

Hypervisible Exchanges blog from Chris Gilliard

Other Books

Stamped from the Beginning, Ibram Kendi

Color of Law, Richard Rothstein

Lower Ed: The Troubling Rise of For-Profit Colleges in the New Economy, Tressie McMillan Cottom

Related Episodes

Episode #181 with Virginia Eubanks

Episode #179 with Kandrea Wade

Support the Show

This show is completely listener-supported. There are no ads on the show notes page or in the audio. If you would like to financially support the show, please check out my Patreon page, where just for a few bucks a month, you can get a sneak peek at guests, grab stickers, or even a podcast mug. Your support helps me cover audio editing services, transcription services, and more. You can also support the show by sharing it with others and reviewing it on iTunes or your favorite podcast provider.

Transcript

Welcome back to the PolicyViz podcast. I’m your host, Jon Schwabish. I hope you and your families are all well and healthy in the strange times. And obviously today here in the United States, it’s election day. So it’s a pretty big day. I remember just a few years ago when people would argue that there wasn’t a big difference between either the major presidential candidates or the political parties. That has certainly changed over the last several years; the differences between the candidates couldn’t be larger, the differences between the parties couldn’t be larger, both in terms of policy and politics in the approach to governing.

My hope today is that everyone who wants to vote can vote and does vote, voter suppression strikes at the very core of any democratic country and for any party, any person, any organization to threaten voters is, to me, quite upsetting and count outrageous. So I hope today we have a smooth day in our elections. But okay, enough about politics. And before we get to the show a few things about what’s going on at PolicyViz.

First, I’ve restarted my newsletter. Up until this fall, I sent maybe one or two newsletters out each year. It’s a lot of work to do. I didn’t think folks really wanted to hear from me even more than they already do on the blog and the podcast, a plus the way I was writing them, they just got kind of like big advertisements. But I have finally found I think what is a good balance. I have the newsletter going out now every other week. It comes out the day before the podcast. And it talks a little bit about the podcast that people see the next day. So it’s a little bit of a sneak preview. I also write behind the scenes post. It’s kind of a short blog post. I’ve only done two or three already. But it’s pretty much a short blog post. I’ve already written about the last phases of my book writing process for the book that comes out in January. And I’ve written about some database hacking I’ve been doing in Excel. So if you’d like to sign up for the newsletter, the new newsletter, do head over to PolicyViz and check it out.

I’ve also added a few more things to the PolicyViz shop; a couple of more shirts. I’ve updated the match at data visualization game, which is now in stock. And if you’re into the R programming language, you’ll find a couple of shirts that I think will strike your fancy just to say. Okay, enough of all that and on to the show.

Okay, let me just say, I love all my guests. My guests are great. They do great work. They take time out of their schedules to come sit and chat with me. And I love all of them. I will say, however, that this week’s guest is special. It was a great conversation, super fascinating and I was really excited to have this guest on the show with me. So this week Safiya Noble joins me on the show. Safiya is an associate professor at UCLA, where she serves as the co- founder and the co-director of the Center for Critical Internet Inquiry. She is also the author of the best selling book Algorithms of Oppression. I picked it up over the summer, and I could not put it down and I’m just so happy that she took time out of her schedule to come chat with me on the show.

Her book covers racist and sexist algorithmic bias and commercial search engines. It’s a problem that has not yet gone away, as you’ll hear about in the interview. So it was just a pleasure to chat with Safiya Noble. I had a really great time during this interview, as you will certainly hear in the audio. Her work is so interesting. Her perspectives on the intersection of race and technology are so interesting. They’re so important, and they’re invaluable for anyone who works and communicates data. So I hope you will enjoy this week’s episode as much as I did with the interview and putting it together. So here’s my conversation with Safiya Noble.

Jon Schwabish: Hi, Safiya! Thanks for coming on the show. It’s great to chat with you.

Safiya Noble: It’s great to be here. Thank you.

JS: I’m really excited to talk about your book that links are on the show notes page, Algorithms of Oppression: How Search Engines Reinforce Racism. And I want to talk both if we can about the book, and also your current work and what you’re looking at now and the activities and things that you’re looking at, especially in the current era in which we’re in, the moment that we’re in. So, but maybe we can start by having you talked a little bit about yourself and your background, and then you can start talking about the book and where it came from and how you went through the process of writing it and what you’re trying to accomplish with it.

SN: Sure. So, , this book was really the outgrowth of my dissertation when I went to grad school and I always like to tell people that I spent my first career in advertising and marketing for about 15 years, and everything that probably isn’t good for us like [Inaudible] [00:05:01] and booze. I was a part of selling all those things. And no, I think of going back to grad school, it’s kind of like, for me. Yeah. When I went to grad school at the University of Illinois at Urbana Champaign, I was thinking about advertising and marketing. That’s the thing that I really knew super well inside and out. I’ve been an expert in multicultural and urban marketing and public relations. And when I went back to grad school, it was kind of a time when, of course, Google was on the rise, Facebook was on the rise. These platforms were becoming platforms before our very eyes. And people were kind of enamored in a way that I thought was kind of curious in academia, because when I was in industry, I was really clear that platforms like Google and Yahoo, in particular, were advertising platforms. I mean, they were places where we were doing pretty significant media buying. We were definitely working with public relations teams to create what we call like advertorial right to make it to the front page of search and to make our clients products and services seem like they were credibly third party verified so to speak. And yet, in academia, people were starting to talk about Google, like it was the new public library. And this disconnect, really was fascinating to me. And that was the impetus for starting to think about well, wait a second, let’s double click a little bit here on Google, in particular, really, as a case, I mean, it could have been Microsoft or Yahoo, but no one really uses them the way they use [Inaudible] [00:06:50].

JS: Right.

SN: Fortunately, Google got the attention because they were the monopoly leader. But that’s really how I kind of got to it. And I was because I’d worked in the multicultural and urban marketing space for so long, I was attuned to thinking about how vulnerable communities, marginalized communities, under represented communities, and even trend setting and trend leading communities which I think are of African Americans, for example, my community as being both often politically, socially and economically marginalized, and at the same time being the trendsetters and the trend leaders in popular culture, for example. So it’s like this, interesting duality plurality. That’s where I kind of started looking at how communities were represented in search engines and that really opened up I think, the Pandora’s box that became Algorithms of Oppression.

JS: Right. Do you find it odd even today how many people seem to think that these services that we get for free are just free, and there is no other component? I love that word advertorial.

SN: Yeah. I mean, it’s still surprises me. On one hand, it surprises me that people who I think should know don’t know. And that kind of that sliver in particular. So I remember recently giving before we were all locked down, I was giving a talk at a big conference and a public librarian approached me. And she said, , I always thought Google was a nonprofit. And I was really [Inaudible] [00:08:30]. And I thought, Oh, my gosh, it was so important to learn that from her, and it made me realize that even the people that we think of as being highly trained and astute, they are also just as susceptible to the kind of marketing discourses that come out of Silicon Valley, as anybody else right. So Silicon Valley has really, for 30 years, tried to convince us that their products are just tools, and that anything that happens in them, is the fault of the public, not the fault of the faulty tool. So we have a lot of work to do for sure, with all kinds of audiences, but I like to focus on data scientists and computer scientists and librarians and teachers and professors and people who I think have kind of like an exponential type of impact on helping us understand what we’re really dealing with.

JS: Right. Do you want to talk about an example or two from the book on how the algorithms particularly in the search engines in Google have or still do I guess, misrepresent people of color, underrepresented groups? Is it just to give people if they haven’t registered a flavor of what? And I’ll give people a warning like it’s the examples you show are pretty shocking of how these algorithms work and misrepresent different groups.

SN: Yeah, this is probably the point where we should warn everybody that if you are listening to this podcast in the car with the kids,

JS: Yeah, this is where you want to turn the volume down,.

SN: Pause it and then pick it up with a car [Inaudible] [00:10:12]

JS: Yeah.

SN: Okay. So what got me going in this study, like the first real study was looking at a variety of keywords that represent women and girls and I, and not just women and girls. I mean, I looked at a lot of different keyword combinations, because the first, let’s say, well crafted study that I did was I took all of the identities represented in the U.S. Census. So I took the gender categories and the racial and ethnicity categories, and I paired them and combo them in 80 different ways. And what was shocking to me and by shocking, I mean, not really shocking, but still shocking, was to seeing how when you paired any kind of ethnic marker, except for white, with the word girls, so African-American, black-black, in particular, and I thought black was important because most , black girls identify as black girls, right? I mean, African-American is kind of a way we might characterize ourselves in more formal settings, but in our families and with our friends many of us just identify as black as kind of a cultural and political identity. And so it was shocking to see that for black girls, Latina girls, Asian girls, the whole first page of search was almost exclusively pornography or hyper sexualized content. And the first of these being I think in 2009, when I was looking at this, the first hit for black girls was hotblackpussy.com and then by 2011, it was sugaryblackpussy.com and you ask yourself, oh, my God, how can this be when you don’t have to add the word sex, or porn, but black have become synonymous, Asian girls, Latina girls just become synonymous without adding any sexualized keywords. And this really is what opened up what became a much longer inquiry into all of the ways in which these systems really fail vulnerable people and that we have a lot of mythology around why this happens.

 So for years, I would talk about my work, and people would say, well, that’s just because that’s what’s most popular. And I would say, well, how can we say that this is what’s most popular without adding these kind of hyper sexualized keywords and how fair is that? I mean, this is tyranny of the majority that has girls of color in the United States will never be in the majority in a way to affect the way they are misrepresented in search engines. But more importantly they will also in the near term future, as children never have the capital to be able to use AdWords and keyword planning tools and all of these, the back end, that helps optimization happen. And of course, the porn industry is masterful at many-many technologies that we have are really due to the research and investment that they’ve made. But also, is it fair? And is it moral? And is it ethical? And that’s really what is the kind of opening salvo to the book, Algorithms of Oppression, which follows with many-many more kind of gruesome tales of misrepresentation arm.

JS: You know, it’s really interesting, I read Ibram Kendi’s book Stamped From The Beginning after reading your book, and the entire I mean, at least the first half of the book, he talks about how the way that white people would describe the black community and people in bondage was hyper sexualizing, both men and women. And so for someone to say that what happens in the algorithms or what happens in their Google search, is because that’s what’s popular, is sort of ignoring a long history of the way the white power structure has viewed, and I guess pushed down the black community in this country.

SN: It’s really true. I mean, I try to give a whole historical and kind of sociological context like Ibram’s book does. I mean, his book came out around the same time. I wasn’t able to read it before. And while I was writing my book, I wish I could have I mean, it’s, but these histories many black feminists have written for decades about the gruesome, hyper sexualization of African peoples and of course, one of the reasons for that is because it helps justify the enslavement of people, right? The degradation is part of the dehumanization of people. And when you dehumanize people, you are much less attuned to their loss of rights, the ways in which they are discriminated against, the way in which they are oppressed. You lose empathy and the ability to care. And so talking about how to humanization happens in the digital space is very important, because it is also a many of the ways that we’re engaging online are dehumanizing. And they’re racialized and gendered in that dehumanization. And if we think there isn’t a relationship between people coming across disinformation, propaganda, racist ideologies, that seems so subtle, right? I mean, I can’t tell you how many thousands of people have told me over the last decade, that what shows up on the first page of search is just what it is. And there is no politics, right? There is no power. there is nothing nefarious happening. It’s just what it is. I actually had a professor once say to me, when I was a grad student in presenting my work, he said, maybe black girls just do more porn. I mean, the rarity, right of the way in which people would justify these kinds of problems, is really why I mean, I try to expose people in this book to how to think about the way in which we’ve all been socialized in racist systems that then make it very difficult when we are designers and makers of technology also to even be willing to look at these problems.

JS: Right. I want to pivot a little bit and sort of look, because we’ve looked back a little bit, I want to look forward. So in this book, and also elsewhere, and some of your writings, you’ve sort of coined this term, technological redlining, which I think is really a fascinating concept. And I’ll just for a brief moment, talk about what redlining is, for listeners who don’t know. So in the early part of the 20th century, there was a federal government agency whose job it was to assess housing values in major metropolitan areas around the country. They created these maps where that assessed risk and the red areas of those maps are the highest or the highest risk areas, but those red areas could have just one black household living in that area. And those will be assessed the highest risk and those the impact of those maps has persisted for generations to impact the accumulation of wealth, accumulation of income and neighborhood segregation in the U.S. And so, Safiya I’m interested, from your perspective, when you think about technological redlining how does that affect groups now? And then also in the future does it have the same intergenerational effect going forward?

SN: It’s such a great question. You know, as I was writing the book, I didn’t know about the genius of my colleague, Chris Gilliard [ph], who also writes about digital redlining. And I think of our concepts of kind of technological or digital redlining as being quite similar in that. What we’re arguing is that these systems, all kinds of digital engagements, not just search. I mean, every type of digital engagement is really tied to tracking, surveilling, categorizing, and predicting us into certain types of futures. And those futures often and the way in which we’re classified digitally, is usually something we have no ability to affect for the most part, but we really don’t know what our digital profiles are. We don’t know, just like, African-Americans in the past, did not know that they lived in zip codes that were being redlined and keeping them out of mortgages, right, and financial services. In many ways, those same processes are happening now. Our digital traces, information gathered about us is also used when we are doing things like looking for insurance quotes online or looking for different kinds of banking and educational products. There is a great book called Lower Ed, by Tressie McMillan Cottom, where she talks about how over and over again poor women and especially poor women of color, black women, whose digital traces are embedded in these systems that they’re using all the time. When they go to look for things like higher education, they are targeted with predatory ads, that push like the Trump universities that the like for profit predatory kinds of educational, I use acquitting fingers right; scams, quite frankly, and are targeted for all kinds of predatory products.

So these are the kinds of ways that we’re talking about digital or technological redlining. The thing that I worry about the most that I’m studying are really are the predictive analytics that underlies so many of the systems we’re engaging with because I don’t think that it’s far fetched to see how systems of social credit, which really originate in the United States, with our own kind of credit scoring systems, certainly are in play in the UK. We put a lot of focus on China. But we want to remember, a lot of these models started, before China picked them up. So these social credit, these predicting models about who has access to all kinds of opportunities, goods and services in the future, really are the fundamental logics of so many products now. And I think these are things that are hidden from view. People don’t understand them. When they hit the news, like algorithms predicting who would get to go to college or not in the UK, which hit about a month ago, then people kind of realize, like, whoa, whoa, wait, what, but until people start to understand that many of the future opportunities for our kids and our next generations are going to be over determined by algorithms and AI then we haven’t done enough work yet. And I think we need regulation, we need policy and we need a huge amount of social awareness about what these technologies, I think, are going to do to our quality of life and the way we live.

JS: Yeah. So on that note, so what can people do to combat the algorithmic racism, and especially for people who might be listening to podcasts who are sort of immersed in data, immersed in technology, what can we do if anything to address some of these inequities?

SN: It’s really a good question. I mean, I think there is intervention on a couple of levels. Well, first, let me say, for data scientists and people who work with data, one of the things I always try to teach my students at UCLA is that data, like race and gender are social constructs. Data come from somewhere, and they are incredibly subjective. People think that data is somehow neutral, that it just is what it is, that it has no politics, but see, those of us who make data, people who are researchers, we know, for example, the sociologist news that where they kind of mark the beginning of one category and the end of the next category it’s sometimes arbitrary, it’s sometimes a guesstimate, it’s sometimes kind of like the best you can do. It’s sometimes just a number. It’s just 25% and that’s what it is. And, and that kind of data making is also often especially we’re talking about big data sets, it’s representative data. It isn’t a close reading of people representative data is, it’s somebody’s grandma, it’s it’s simply his family member. It’s people who are vulnerable, who can’t speak for themselves who can’t clarify and nuance the way in which they’re represented. So this kind of representative data gets used and taken up like a truth teller. And that is very dangerous.

So I think we got to pull back and think about what we’re doing when we’re doing things with data. There are people who obviously, there are so many tech workers who are also intervening at the level of just saying, I’m not going to work on certain products, because I think these are unethical products, or these are going to change the future in a way that we don’t want. We only wish for example, you know, I look at people like climate scientists now, who lament the fact that all they did was kind of put the information out and say, hey, the way human beings are living in an organized society is really going to kill the planet. Here is the research, but they didn’t take an activist stance about that work like there.

JS: Yeah.

SN: I think we have to those of us who work closest to these models, we have to take a more activist stance about the harm and say, wait a minute, we don’t want to do this, this is dangerous, and not lay back, like the climate scientists until it’s too late. And then I think for everyday people, obviously, there is so many ways that we should be thinking about protecting our data, but these things in our privacy and like how we engage with the internet, but I think we need to really exercise our power by getting candidates educated. We need to vote on policymakers that are astute about these issues. Many of the interventions can happen at local and state levels. We certainly see like ban on facial recognition, greater privacy laws in California where I live; those things don’t have to all just go to the federal government. And I just want to remind people that we still have a lot of power to connect and organize locally where we live.

JS: Yeah. I wanted to ask, and you may not know the answer to this, but are there states or countries or are there jurisdictions that are or even individual policymakers that are leading in these efforts to address these issues?

SN: Well, I think Senator Warner’s office, the Senator from Virginia has been way out front on things like antitrust and harm. That’s coming from big tech, let’s say broadly as a category. And I think we want to support and watch what’s happening in his office. I think that on the Federal Trade Commission, we have Commissioner Rohit Chopra, who is one of the best and smartest thinkers about consumer harm. So he, for example, is taking on the cases where automotive dealerships, for example, not only tell their sales people that when you see black and Latino customers approach to mark the price of the car up no matter what, but also, that are using predatory software that gives higher interest rates to African-American and Latino customers. He has taken them on, and took on a case out of New York. And so I think that there are individual policymakers who are definitely trying to create a critical mass on the federal level. I think California has been a place where we can test a lot more thinking and it’s a place where we need to. I mean we are, here we are the heart of Silicon Valley is right in Silicon beach, right down the street from where I live. And here you have, like the largest industry in the world, doesn’t pay taxes, doesn’t pay back into the system, takes the cream of the crop, students into their employee ranks, and then really gives back so little to society. So I think there are people, certainly those of us in the UC system at UCLA, our Center for Critical Internet Inquiry, we’re really trying to track these things and your subscribers can certainly follow us. And we can try to keep pushing out resources. I think in Europe, the EU has been the most probably aggressive; Germany and France, around thinking about the harms of data. But it’s very complicated, because the scale of these systems is so intense that when you talk about taking on these systems, you’re talking about taking on the entire financial services sector, which is all digital now. You’re talking about taking on the markets, the global financial markets. You’re talking about taking on state governments that are now so deeply intertwined into digital systems that it’s quite difficult to see where big tech companies begin and end and states begin and end. And then of course, you have companies themselves, like Facebook, who are operating at the level of nation state unto themselves trying to make their own currency, trying to make their own laws to govern themselves. So there are many points of pressure that I think we need to be paying attention to.

JS: Right. I can’t decide which final question I want to ask so I am going to ask both of them. This is so fascinating. Let me ask this one first. So do you think now we’re in October of 2020. So it’s a it’s an odd time for many reasons. But I want to ask whether you think things are changing for the better or changing for the worse? And I can’t really tell based on what you said so far, whether you think we’re going uphill or downhill on this?

SN: Well, I think that we haven’t bottomed out yet. And so that makes me nervous because they think one of the things that I say in the book is that we have more data and technology than ever.

JS: Yeah.

SN: We’ll have more global social kind of political and economic inequality to go with it. So one of the things we know, for example, is that the promises of the internet and the promises of digital technologies was that they were going to even out the world, right, make things more equitable, make democracy more plausible, accessible. What we’ve seen, though, is that many of these technologies are being used in service of the kind of rise of authoritarian regimes, including in the United States, I’m sorry to say. We’re seeing again, a level of control and lack of transparency by so many of these actors between the tech sector and governments that I’m not actually feeling hopeful yet like that everything will turn on the next election. What I will say, though, is that we are in the era of kind of the tech lash and that people are starting to think more critically. More films and television shows, and scholars and thinkers and podcasters and public intellectuals, like yourself, are taking up these issues in a way that they really did not 10 years ago. So in that way I’m super energized, that we’re building a critical mass of people. And you know, I like to think of it this way that right now I’m writing about the relationship between three different eras; the era of big cotton, the era of big tobacco and the era of big tech. And one of the things that they all have in common is that it really was a small group of people who were kind of abolitionists, who took on huge industries that people thought could never be taken on. And they shifted our, the paradigm of how we think about the enslavement of African peoples. They shifted the paradigm of how we think about the public health crisis associated with big tobacco is just what it is, and there is nothing we can do about it. And I think we’re in a moment where some of us are thinking about what technologies should be abolished? What should be made illegal? What is too harmful? What are the secondary, tertiary effects of some of these technologies? Even if you aren’t on Facebook, how you’re affected by those kinds of platforms? So I think that in that way, if we pull back and take a long view, instead of a short view, I feel hopeful that there will be enough of us who will talk about the ethical and moral feelings of the institution of big tech, and shift is towards something better. And I believe we’ve already done that. Human beings have already done that in different moments. I think we’ll do it again around this moment.

JS: Well, like the way you shifted from the pessimistic outlook to the optimistic outlook. So that’s..

SN: That’s the best [Inaudible] [00:32:26]

JS: Yeah. I wanted to ask us one last question on on what you’re working on now and what you and your colleagues are looking at for future work.

SN: Okay. So we have, at UCLA, we have this new center, it’s the UCLA Center For Critical Internet Inquiry. We’re part of a network of centers that have been funded by the Minderoo Foundation. And they include kind of critical technology centers at NYU, New York University, at Cambridge, at the University of Western Australia, us at UCLA, and some friends at Oxford, at the Oxford Internet Institute and one of the things that, I guess there is kind of two things we’re all thinking about, which is how to strengthen the research so that people can access it and pick it up and touch it and do things with it. That’s so important. We’re thinking about public policy, what are the ways that we can inform policymakers with the best evidence based research because we think that’s what should inform policy, not just raw power. And we’re thinking about how to shift culture. So my colleague, Sarah Roberts at UCLA, she works, she’s kind of one of the world’s authorities on commercial content moderation. So we’re obviously thinking about policies that affect technology workers and the traumas that they experience in moderating content. That’ll always be core. And I’ll always be kind of working on algorithmic discrimination and oppression. But we’re taking up and trying to link arms, I think with other researchers around the world who also care about these issues. Most of the places that have studied the Internet and society have really been advocates for the tech sector, and have taken a lot of funds and resources from the tech sectors. So we’re kind of like, we’re on a shoestring compared to those places. But we don’t get big tech money into the center, because we don’t want it to influence the research we do. And we think that gives us a space for people who are interested again, in studying the most vulnerable and those who are most harmed and we need help and support. So people who are interested in that kind of work, again, should link up with any of these places where this work is being done.

JS: Yes, absolutely. And I’ll put links to the center to your work so that people can check that out. Safiya, thanks so much. This has been really interesting. The book is great. Your work is great, and I really appreciate you taking the time to come chat with me.

SN: Thanks so much, Jon. It’s really my honor and my pleasure, appreciate you.

JS: Thanks to everyone for tuning in to this week’s episode of the podcast. I hope you will check out Safiya’s book and her work at UCLA. I have linked to all the things that we talked about in the episode in the show notes. If you’d like to support the podcast, please tell your friends and colleagues about it, write a review on iTunes or head over to my Patreon page. So okay. Anyway, until next time, this has been the PolicyViz podcast. Thanks so much for listening.

A number of people help bring you the policy of this podcast. Music is provided by the NRIS. Audio editing is provided by Ken Skaggs and each episode is transcribed by Jenni Transcription Services. The PolicyViz website is hosted by WP Engine and is published under WordPress. If you would like to help support the podcast, please visit our Patreon page.