S1E74 / Techno-Racism & COVID at Home & Abroad / Mutale Nkonde, Corin Faife, Heidi Larson, Imran Ahmed
13 May 2021 · 21 min
In this episode of the Epidemic podcast, host Dr. Célène Gounder explores how disinformation networks exploit social‑media algorithms to target vulnerable populations, especially Black and other communities of color, during the COVID‑19 pandemic. The conversation weaves together historical analysis of the 2016 U.S. election—where Mueller’s report identified a coordinated Russian effort to depress Black voter turnout—with contemporary examples of COVID‑related misinformation, such as false claims about immunity, hydroxychloroquine, and anti‑vaccine propaganda. Guests Mutale Nkonde (AI for the People), Heidi Larson (Vaccine Confidence Project), Imran Ahmed (Center for Countering Digital Hate), and Corin Faife (The Markup) discuss how platform ad‑targeting, algorithmic amplification, and profit‑driven engagement models create an “information fog” that obscures truth. The episode highlights the disproportionate reach of public‑health ads to white users, the concentration of anti‑vaccine content in a handful of accounts dubbed the “Disinformation Dozen,” and the limited effectiveness of current de‑platforming efforts. Legislative context is provided through references to the 2020 House subcommittee hearing, Section 230 immunity, and recent state bills that restrict de‑platforming. The guests call for greater transparency, equitable messaging, and policy reforms to curb the export of U.S.‑originated disinformation worldwide.
Chapters
They benefit from traffic, no matter if it's there to look at good information or malignant misinformation.
Stories are really powerful means of communication, and this is where the gap between science and publics, I think, is the biggest.
You're listening to Epidemic, the podcast about the science, public health, and social impacts of the coronavirus pandemic. I'm your host, Dr. Celine Gounder. Disinformation has always been a problem since the beginnings of the internet, but it reached a fever pitch during the 2016 presidential election. Robert Mueller's investigation into election interference found some surprising results about where that disinformation was aimed.
One of the major findings that came out of the report by Special Counsel Robert Mueller into election interference where the African Americans were the most targeted by disinformation in the 2016 election.
This is Mutale Nkonde. She's the founder of AI for the People. It's a nonprofit organization whose research focuses on how technology impacts the lives of Black people, from voter turnout to vaccine confidence.
And what was happening, according to the report, was Russian agents were creating false online communities.
The goal of these disinformation campaigns in 2016 was to depress the Black vote in swing states. The campaigns would focus on an accepted truth and then springboard into something controversial.
In this case, that there is systemic racism in the United States, that it manifests itself through economics, so there is a need for reparations. And then they'll insert a massive lie. So The massive lie in this case was, unless reparations are promised day one, Black Americans shouldn't vote.
One effort asked Black people to do something called voting down ballot. This means that you either leave it blank or you write in a hashtag at the bottom of the ballot. Down ballot voting makes it difficult for election officials to determine who the voter's choice for president is, and the votes are tossed out. In 2016, this push to vote down ballot played out in the city of Detroit.
And what happened in 2016 was 70,000 people in this place didn't vote at the top of the ballot. They voted down ballot, and that swung the state from Obama in 2012 to Trump in 2016.
In 2017, many of these suspicious accounts arguing for reparations went quiet. But not for long.
All of a sudden, we see a hashtag pop up, which please forgive me if I don't name because I don't want to amplify it, but it was by a group that at that point were really interested in foreigners not being in the US.
Motala was surprised by where this hashtag started to appear.
The folks that were using this hashtag, who were Black people online, it seemed, would always really side with— it seemed to be white supremacists supremacists and conspiracy theorists, and we thought that that was super, super interesting.
Motali says this hashtag started to be associated with posts claiming Black people were immune to COVID. Once it became clear that people of color were disproportionately impacted by COVID, the hashtag started to deny the pandemic altogether. It is the perfect storm for disinformation networks to swoop in and target groups for their campaigns. Muthale describes it as an information fog.
And it just puts you into a situation where you log on to your Twitter account and you literally can't see. You don't know what's real, and that's not your fault. And we felt that the idea of waking up and walking out into a fog was really similar, where you just can't really see anything and it's not really your fault, but you somehow have to, you know, kind of make your way in the world anyway.
In this episode, we're going to hear more about how disinformation networks are gaming social media algorithms.
What they do is they find existing fissures in our society to cause a mini earthquake.
How the United States has become a hub for disinformation exported around the world.
Colleagues running the infodemic work out of WHO Africa office have said that a lot of what they're seeing has been imported from the US.
And what legislators need to do to tackle bad actors.
So it's going to take all of us— government, civil society, individuals, but most of all, the tech companies. They are addicted to the profits that come from misinformation and hate.
Today on Epidemic, how social media and disinformation algorithms target vulnerable populations. Mutale says African Americans and other people of color may be at higher risk of exposure to misinformation. One reason, she says, is historical.
One of the things that you'll find is that there is a lack of trust of the news media because of historic racial bias and wrongful kind of framing of Black people and the Black experience. And so they were creating alternative news systems online.
So for many, social media became a place where they could find news and commentary on things that spoke to their lived experience. And social media is no small front in the disinformation war.
Yeah, there's actually some really great data that came out of the Pew Research Center in 2019 that found that 55% of all American adults get their primary news from social media, which was really scary to us because social media, the news that comes to our news feeds, is curated by algorithms.
Once the algorithm learns what you engage with, it will show you more and more information similar to that.
Those algorithms are charged with identifying patterns in our engagement behaviors. So what do you like? What do you retweet? What do you comment on? And then serving you more news.
The people who are learning to game the algorithms have an agenda.
They were propagating disinformation, disinformation as a form of warfare.
And they can be very good at fooling people.
They have their own news sites that they reference stories from. So for journalists, it's pretty hard to source.
It's hard to even comprehend the sheer volume of disinformation they produce.
When we were talking about false accounts, there are two that always stand out in my mind where they apparently tweeted 147,000 times in a day.
Well, and to put that number in context, you said 140,000 tweets in a day. That would be more than a tweet per second. Yes. Obviously not a real person. Right.
Like, that's not a real person. It's not 10 people. It's not. But then there are also commercial services where we could program a system to tweet 20 times a second and it would happen.
What can make the problem even more complex is that social media ads are often targeted to certain groups. That means that bad actors can choose what demographics will see their content.
So you could just use the ad credits to say, let's spend $1,000 and show this to just a million people between the ages of 18 and 65 or something like that. The Facebook algorithm will then have to make a whole lot of choices about who exactly those people are.
This is Coryn Fife. He's a data reporter at The Markup. Earlier this year, Koren published a report that had some surprising findings about who gets access to accurate information on social media. Facebook has a documented history of allowing racist systems on their platforms. In 2016, ProPublica published a report showing that advertisers could block certain racial groups from seeing housing advertisements. The resulting fallout led to a 2019 lawsuit by the Department of Housing and Urban Development for violating the Fair Housing act.
Because of that, Facebook changed the targeting system to use something called ethnic or cultural interest.
But this change isn't that different from the previous system.
So you can't say, show this to a black person. But you can say, show this ad to someone interested in African-American culture or interested in Asian culture, or it might be a different proxy.
This means even after the lawsuit, Facebook continues to give advertisers the power to discriminate. And this divergence in who sees what ads on Facebook doesn't end with housing. Earlier this year, Koren published a report examining who saw sponsored posts from public health agencies on Facebook.
Only 3% of the Black people in our panel saw any of these sponsored posts from any kind of health agency, and that was compared to 6.6% of white panelists and 9.5% of Asian panelists, which is quite a big discrepancy.
This might sound illegal, but it isn't.
There isn't really a legal standard where we say that public health information has to reach people equally and not discriminate by racial groups or any other kind of protected classes.
Now, I can't imagine that the CDC or a local public health department would intentionally want to target, I don't know, only wealthy white people with their public health information, you know, COVID information. So how does this actually play out that this advertising is missing some of these most vulnerable communities?
We don't really know from the view that we have with the Markup Citizen Browser. We don't know that. Things that I could speculate are that perhaps the CDC is just taking out ads that aren't really targeted at all.
The findings from the report aren't definitive. It is a relatively small sample size. But Koren stresses that this is evidence enough that some changes need to be made.
At the end of the day, the point of this reporting is just to say we really need more transparency, especially if Facebook is going to play such a big job in distributing public health information.
But that alone isn't enough.
There's two big pieces to the puzzle. So on the one hand, there's putting out accurate information and making sure the right people see it. But then the other part is actually misinformation and trying to keep a lid on it.
And just like a virus, if you allow disinformation to spread anywhere, it will eventually spread everywhere. We'll hear more about that after the break. If the algorithm is indeed making many of these decisions, it could lead to inequities in the distribution of information, and that disinformation often starts in the United States.
I mean, we've seen it on our social media mapping how much the U.S. is a generator of a lot of the things we see circulating globally.
This is Heidi Larson. She's a professor of anthropology at the London School of Hygiene and Tropical Medicine and the director of the Vaccine Confidence Project. Heidi and her team spend a lot of time tracking the sources of disinformation online.
It's a bit like the weather room. We watch where things move and how fast they move. And we capture, for instance, a word or a phrase and a piece of the news or the rumor. And you can see where it goes around the world and then who picks it up and who amplifies it.
When it comes to making disinformation that sticks, you have to know your audience.
Those seeds, those beginnings of a rumor come from far away, and then it gets embellished with local culture, local politics, local anxieties, local histories.
One example of this was a rumor started by an American doctor early in the pandemic that a malaria medication was an effective treatment for COVID.
She's had a number of different rumors out there, but the other one was that hydroxychloroquine was a cure for COVID.
This drug was famously endorsed by former President Trump. Many doctors think it is extremely successful, the hydroxychloroquine. And that disinformation would spread like wildfire throughout Nigeria.
And it was the speed that it spread that was so striking to us. When you get the president of the U.S. saying, just go for this organic or chloroquine or finding these alternative cures, it just endorses at a different level with a different kind of authority.
Information voids and a lack of trust in health systems and government in both communities of color and the global South provide fertile soil for disinformation to take root.
There's not enough alternative credible information before you start getting to the potentially harmful mis- and disinformation.
A large part of the problem has been that science communicators haven't cracked the code on how to engage people with the right information. And government agencies like the CDC don't have substantial communications budgets.
So I think as a public health, as a health community, we need to do a better job of getting creative narrative narratives out there that are accurate, that are credible, that are appealing in ways that some of these alternative voices are out there.
Public health officials can only do so much when algorithms are working against them, and political leaders on both sides of the aisle are increasingly recognizing that misinformation and disinformation on social media are threats to public health. On March 25th, the U.S. House Subcommittee on Communications and Technology held a hearing on disinformation and social media. In attendance were the CEOs of Facebook, Twitter, and Google. There are two faces to each of your platforms. Facebook has family and friends neighborhood, but it is right next to the one where there is a white nationalist rally every day. YouTube is a place where people share quirky videos, but down the street, anti-vaxxers, COVID deniers, QAnon supporters, and flat earthers are sharing videos. At the end of the hearing, the CEOs made a surprising pledge.
The social media companies said that they would take action.
This is Imran Ahmed. He's the CEO of the Center for Countering Digital Hate. Imran traces how patterns of hate and misinformation spring up online and how our data is being used to feed it.
Quite often, the dynamics that allow the spread on social media, the amplification algorithmically of misinformation, has been the fact that we've engaged with bad information, whether we support it or we don't support it.
A new report released by his organization offered some insight into how our social media feeds work. The report found that more than half of the posts recommended to users in the study contained misinformation about COVID and one-fifth promoted inaccuracies about vaccines. One of the most famous propagators of that disinformation on social media is Robert F. Kennedy Jr., son of the late Robert Kennedy.
RFK Jr. has said that African American kids should not take vaccines because they have stronger reactions than white kids. He said that African blood is different. He's spoken scientific nonsense, and the end result of his misinformation, if it was believed by the African American community, would be death.
The vast majority of this disinformation comes from a surprisingly small number of accounts. Among them Robert F. Kennedy Jr. Imran's organization identified just 12 accounts that they say are responsible for more than 60% of all anti-vaccine content on social media. Imran's organization called these accounts the Disinformation Dozen in a report this spring. Since its release, 12 state attorneys general and U.S. lawmakers from both parties have asked social media platforms to take down these accounts.
However, the social media companies have once again shown their reluctance to take comprehensive action. And while there has been some deplatforming, and we're really pleased about that, nevertheless, most of those networks are still up. Most of the disinformation dozen are still on one or more of the platforms.
The solution to this seems simple. Why can't the social media companies just change their algorithms Imran says they have an incentive to keep things as they are.
You have someone who seeks to make money who writes an algorithm that says, find the most engaging content, the most chewy. I don't care if it's true or not. I don't care if it causes public harm, but I want to maximize the time spent on platform. If you can keep people scrolling for 3 more posts every day, you can show them one more ad. Let's say one ad is 5 cents. 5 cents across a billion people is actually $50 million. $50 million a day over a year, that's $20 billion.
Social media has become a crucial part of how we communicate with one another. However, the pandemic has made it ever more clear that the government needs to step in to stop the spread of disinformation online.
Typically, we regulate to manage those negative externalities, those costs that are imposed on society, to create costs back so we can find ways to either mitigate or to dissuade people from creating those harms. They have managed to finesse their way through and say, oh, can't do anything about it. It's just not good enough. No other industry is held to such low standards.
In April, the Senate held another hearing where Facebook executive Monica Bickert falsely testified that Facebook had deplatformed 10 of the Disinformation Dozen.
We have a culture of impunity now, not just for misinformation, but for the social media companies who are protected by law, by Section 230 of the Communications decency act, which says they cannot be held accountable in law, either civilly or criminally, for any content on their platforms from another party. And they have really hidden behind that. They've gamed it in an aggressive way, saying, you know what, we will obfuscate, we'll delay, we'll lie where necessary, we'll just get through the next day so we can continue to reap as many profits as possible, because they know as do we all, that the time of impunity will come to an end.
Imran says that deplatforming bad actors is one of the most effective tools we've got to fight disinformation.
We know that they fear deplatforming. We've been tracking the filings that they've been putting into courts, suing Facebook, suing YouTube for their deplatforming. And in there, they're extremely explicit. It damages their ability to fundraise, it damages their ability to recruit, and it lessens the amount of information they get about the particular likes and dislikes of their audience. So we know deplatforming is incredibly effective in disrupting the malignant activity of these actors.
But already others are taking steps to defang one of our best tools. Late last month, Florida passed a bill that would prohibit social media companies from deplatforming political candidates. The social media companies are pushing back, saying this violates their First Amendment rights as private businesses. We as voters have a crucial role to play in holding our elected representatives accountable to represent the public's best interest.
Write your senator and say, you need to be coming up with solutions on this because I will reward Anyone that can find a way to ensure that the information that is consumed by all citizens— and we are highly interdependent on the information that others consume so that we can build a coherent worldview that allows us to prevent things like violent extremism, QAnon conspiracism, anti-vaxxer disbelief, which actually causes damage to the whole of society. So write your representative and make them aware that you care about this issue.
Epidemic is brought to you by Just Human Productions. We're funded in part by listeners like you. We're powered and distributed by Simplecast. Today's episode was produced by Temitayo Fagbenle, Zach Dyer, and me. Our music is by The Blue Dot Sessions. Our interns are Annabel Chen, Brian Chen, and Sophie Barma. If you enjoy the show, please tell a friend about it today. And if you haven't already done so, leave us a review on Apple podcasts. It helps more people find out about the show. Follow Epidemic on Twitter and Just Human Productions on Instagram to learn more about the characters and big ideas you hear on the podcast. We love providing this and our other podcasts to the public for free, but producing a podcast costs money and we've got to pay our staff, so please make a donation to help us keep this going. Just Human Productions is a 501(c)(3) nonprofit organization, so your donations to to support our podcasts are tax deductible. Go to justhumanproductions.org/donate to make a donation. That's justhumanproductions.org/donate. And if you like the storytelling you hear on Epidemic, check out our sister podcast, American Diagnosis. On American Diagnosis, we cover some of the biggest public health challenges affecting the nation today. Past seasons covered topics like youth and mental health, the opioid overdose crisis, and gun violence in America. I'm Dr. Celine Gounder. Thanks for listening to Epidemic.