Kolina Koltai, Andrew Beers and Rachel Moran have spent the past few months immersed in some of the most disturbing and important conversations playing out online.
The University of Washington researchers spend hours per week analyzing viral election-related disinformation and misinformation spreading in real time, using sophisticated automated tools as well as old-fashioned common sense to figure out where false narratives are coming from, how they spread and what can be done to stop them.
They're part of the Election Integrity Partnership, a group of over 100 researchers across multiple institutions working day and night to combat the deluge of online misinformation and disinformation surrounding the 2020 U.S. presidential election. They report concerning content directly to Facebook, Twitter, Google, Reddit, TikTok, Discord, Nextdoor and Pinterest through a shared software system.
And they're very stressed about Election Day. Last week, they received care packages of tea, chocolate, stickers and a nice note, a final-hour gesture of encouragement from their research institution, the University of Washington's Center for an Informed Public, as they're nearing the day they've been bracing for.
"You can only read so much content before it starts to eat at you," Koltai said.
Meanwhile, in a completely different part of the anti-misinformation ecosystem, Meghna Mahadevan has spent the past several months working to combat the aggressive and escalating spades of misinformation targeting the Latinx community. Mahadevan is a disinformation defense strategist with United We Dream, the country's largest immigrant-led youth network, and she's a part of a broad network of civil rights and justice organizations known as the Disinfo Defense League.
Mahadevan, who previously worked at Facebook, Google and Snap, says she believes the platforms are ill-prepared to protect people, particularly Black and Latinx voters, from false content that could dissuade them from going to the polls. So she and other civil rights activists have taken it upon themselves to educate and inoculate communities of color against the windfall of conspiracy theories, falsehoods and misleading content flooding their social media feeds. She's helped create PSA-style videos about the fact that election results likely won't come in Nov. 3 and the importance of talking to your family about disinformation. She helps run a WhatsApp group devoted to amplifying accurate information about the election. "We're really recruiting people into our WhatsApp group and saying if you have family members and friends spreading misinformation, we're going to give you content and resources to counter that," Mahadevan said.
The work of the Election Integrity Partnership and Disinfo Defense League are worlds apart. The organizations, which both launched in June, aren't affiliated and don't work together. They're only two initiatives among hundreds aimed at addressing domestic misinformation around the U.S. election.
But they reflect two mutually reinforcing sides of the same ecosystem: the growing collection of people devoting their livelihoods to combatting online misinformation and disinformation, which will face a moment of reckoning on Nov. 3 and the days afterward. These are their stories.
Inside the Election Integrity Partnership
Facebook's former chief security officer, Alex Stamos, thought the Election Integrity Partnership sounded like a rerun.
The idea for the partnership came from Isabella Garcia-Camargo, a Stanford researcher and rowing champion who had just graduated from an undergraduate program in computer science. She approached Stamos, who now runs the Stanford Internet Observatory, with an idea this past summer: What if they created a third-party organization to help respond to misinformation that election officials were seeing? She envisioned a real-time monitoring group that could partner directly with the social media companies.
Stamos said he was sure that a similar effort already existed.
But when he discovered it didn't, they got to work figuring out the best research institutions to partner with to get the project off the ground. They selected three institutions with trusted contacts within the social media companies and credibility in the misinformation space: University of Washington's Center for an Informed Public, the social media mapping firm Graphika and the Atlantic Council's Digital Forensic Research Lab.
"The intention of us combining our efforts was that so we could more readily communicate with election security officials in the U.S., and we could also speak with a unified voice when we were flagging concerning content and making recommendations to the social media platforms," said Emerson Brooking, a resident fellow at the DFRLab who has been involved from the start.
The leaders of the project include Stamos, Garcia-Camargo and Kate Starbird, a widely respected computer scientist focused on misinformation at the University of Washington. And their efforts are bolstered by dozens of researchers, many of them still in school.
Since the summer, for at least 12 hours every day, the group of more than 100 researchers across each of the institutions has come together to analyze and debunk hundreds of misinformation tips flooding in from local election officials, civil society groups like the AARP, social media analytics tools like CrowdTangle and their own automated systems. The researchers, who communicate primarily over Zoom and Slack, work in four-hour shifts; the number of people online per shift varies day-to-day. And they hail from a variety of backgrounds, with roots in media studies, computer science, anti-vaccine misinformation research, and more.
"This is not a five-person project," Koltai said. "We have a huge group of undergraduates, masters, Ph.D. students, post-docs, as well as people from the private sector. It has been a huge collaborative effort."
They pore over Reddit forums, Facebook groups, prominent Twitter accounts, TikTok videos and more, trying to keep track of the most prominent narratives surrounding voting and the results of the election. "We're trying to create the most comprehensive database of cross-platform mis- and disinformation incidents," Garcia-Camargo said. So far, they've compiled over 1,500 pieces of potential mis- and disinformation.
And when they identify concerning false narratives, like claims that Democrats are seeking to steal the election or that electronic voting machines are compromised, they weigh their options. Sometimes, the researchers put out a Twitter advisory or write up a rapid-response blog post. Others, they alert the Center for Internet Security, a cyber defense nonprofit. And often, when the content violates the social media platforms' policies (which the EIP tracks very closely), they elevate tips to the companies directly.
EIP stands out from other misinformation research efforts because of its uniquely close, mutual relationship with the companies. It uses Jira to send tips directly to the election integrity teams at Facebook, Twitter, Google, Reddit, TikTok, Discord and Pinterest, EIP researchers said, and communicates with other platforms directly.
"Disinformation is a challenge we face as society as a whole and as an industry specifically, and such challenges are best solved through collaboration," said Tara Wadhwa, safety policy manager at TikTok. "The Stanford Election Integrity Partnership provides actionable, timely insights about new kinds of disinformation spreading across the internet and has helped our work to maintain the integrity of TikTok."
"We're collaborating with the Election Integrity Partnership and have been since the launch of this project on Sept. 7," a Twitter spokesperson told Protocol. "In line with the Twitter Rules, our Civic Integrity Policy and our recently announced product and policy changes to slow the spread of misinformation, we'll take strong action on any attempts to undermine the integrity of the online, public conversation happening around the election."
Garcia-Camargo says that EIP has achieved remarkable closeness with the platforms not only thanks to Stamos, who is well-known and connected in the tech industry, but also because of its measured, targeted approach to reporting misinformation. Rather than antagonizing the companies and demanding they take down certain kinds of content, EIP makes sure to report mainly content that it knows is in violation of the platforms' specific policies. In other words: It does its best to speak the same language as the tech industry.
And when the EIP identified gray areas or loopholes, it wrote measured critiques and made suggestions about how the social media firms could tweak their approach.
Election Integrity Partnership researchers (clockwise from top left) Elena Cryst, Isabella García-Camargo, Emerson Brooking, Camille François and Kate Starbird — and a moderator, unaffiliated with the group — discuss EIP's work during a webinar in August.Image: Election Integrity Partnership
All of the researchers involved have their own personal opinions about how well-prepared the platforms are for this moment, of course. Moran, one of the EIP researchers who focuses on right-wing media, put it bluntly: "I'm not sure if the platforms could ever be ready. They're not designed that way. They're designed to make money, and some of that is not going to be compatible with what we need for a democratic conversation." While misinformation emanates from both political parties, researchers have found that it's not symmetrical: The right has a more robust ecosystem and structure around spreading falsehoods. And the companies have been hesitant to take aggressive action against the loudest right-wing figures, including Trump, over fears that they'll be accused of censoring the GOP.
But at the very least, all of the researchers who spoke to Protocol described the major platforms as "responsive" to their tips.
"We're increasing our cadence right now," said Camille François, Graphika's chief innovation officer and a former researcher with Google's Jigsaw, last week. "We're going to move to essentially a 24/7 tempo this week."
In order to make a difference, the coalition decided early on that it would remain laser-focused on a narrow mission: researching and analyzing election-related disinformation, and communicating that information to the platforms and the election official community. It did extensive work to educate the public on what they were finding but it would have been "intractable" to aim to inoculate the U.S. population against misinformation, Garcia-Camargo said.
Ultimately, EIP is planning to put together a 100-page report looking back at what role misinformation and disinformation played in the 2020 election and how effectively the EIP helped to mitigate that spread. Brooking, for his part, said he believes EIP's model could be a template for future elections and massive political events.
"I think for future elections and future discreet political events, this is the kind of model we'll need: something where you can have civil society groups that stand between government and social media platforms, ensuring there's speedy action whenever this sort of content is encountered," Brooking said.
"Our efforts are complementary to those of the folks on the frontlines," he added.
Inside the Disinfo Defense League
If anyone is on the frontlines of disinformation, it's the people involved with the Disinfo Defense League.
An expansive network of civil rights and media literacy groups, it's been working for months to innovate new ways to protect communities of color as they're targeted by racialized and opaque misinformation campaigns. It's created specialized memes, filmed videos, held webinars, distributed multilingual informational materials including a "Disinfo Defense Toolkit" and more, aiming to give their communities the best information possible during a confusing and frightening time.
"We saw a particular need for a project that was race-forward, that recognized and centered the fact that Black and Latinx communities were disproportionately targeted by disinformation in the previous election and would be the 2020 election," said amalia deloney, co-executive director of the Media Democracy Fund, the grant-making nonprofit that put together the effort, which she compared to DC Comics' Justice League. It launched officially in June with an on centering the work of people of color – specifically Black, Afro-Latinx and Latinx communities who are often disproportionately targeted by misinformation.
More than anything, the Disinfo Defense League — known as the DDL — has offered a vast array of advocacy groups the opportunity to communicate and organize behind the scenes against misinformation, trading tips and planning events aimed at disputing narratives they were hearing from people on the ground and experts.
"I don't think it's possible or a realistic goal to say 'we're going to get rid of disinformation entirely,' but what we can do is try to limit the impacts disinformation has in our communities," said Erin Shields, a national field organizer with MediaJustice.
This past week, the DDL held its most public event yet: a Disinformation Week of Action featuring events like "Canaries in the Coal Mine: Racialized Disinformation in the Digital Age," a webinar with leading experts in the field, and a Reddit AMA about racialized disinformation. The events and training sessions, which were paired with a nonstop social media campaign to debunk particular myths and share accurate information about the upcoming election, ultimately reached thousands of people, according to the organizers.
The Disinfo Defense League has regularly held events as part of its fight against racialized misinformation campaigns.Image: Disinfo Defense League
"It's a six-to-one ratio of false information vs. factual information," said Shireen Mitchell, the co-founder of the Stop Online Violence Against Women and one of the first researchers to raise alarms around bots and trolls posing as Black people online. "We want to make sure we're getting out accurate voting information as often as possible, as frequently as possible."
The DDL's leadership council includes representatives from a range of civil rights organizations including Latinx advocacy groups Mijente and United We Dream and gender justice group UltraViolet, which are each engaged in their own projects to fight disinformation within the communities they serve. For some of those advocacy groups, misinformation is a completely new topic.
"We've had this explosion of justice organizations and civil rights organizations coming together to combat this," said Bridget Todd, communications director at UltraViolet. "We could no longer ignore this growing, massive threat to our communities, especially if that threat was largely not examined."
The organizations are able to easily tap into their vast networks to track the inception and spread of misleading narratives in a way that researchers often aren't able to, with insight into private messaging spaces like WhatsApp and WeChat. And the key to the group's success, members said, is the ongoing partnership with misinformation experts like Joan Donovan, the research director of Harvard's Shorenstein Center on Media, Politics and Public Policy, and First Draft, a misinformation project founded in 2015 by the Google News Lab, which offered them advice and guidance around how to best approach the issue. For instance, they warned activists to ignore the urge to share misinformation even to debunk it, because it could wind up giving that narrative more credence.
Every morning, First Draft blasts out emails to the DDL with a list of the top misinformation narratives circulating within communities of color.
"One of the most popular Spanish language figures on Facebook, Evangelista Gary Lee, claims that confrontations between a Jews for Trump caravan and other New Yorkers is a sign of the end times," read one bullet in a recent email shared with Protocol. "His video has been viewed more than 89,000 times."
First Draft then holds 30-minute briefings with the groups to go over any questions or strategize about how to best handle those misleading narratives. "The league offers the chance to talk to researchers and academics, people who are studying disinformation and misinformation and the way it moves and mutates," Todd said. "That information has been invaluable."
Some of the organizations involved in the DDL have relationships with Facebook, Twitter, Google and the other social media companies. But for the most part, members said they are not speaking directly to the platforms; instead, they're focused on community education and playing defense.
"For communities of color, our greatest concern at this exact moment and leading up to the election is that we are able to vote in a free and fair election," said deloney with the Media Democracy Fund. "It doesn't mean we don't care about the platforms, we absolutely do. But we'll get to the platforms after we secure our vote."
This week is set to be somewhat of a Super Bowl for misinformation, as analysts predict the results of the election won't come in for days after Tuesday and protesters are set to take to the streets across the U.S. no matter the outcome. Trump himself is expected to prematurely claim victory on election night before all the ballots have been cast, a false claim with potentially enormous ramifications. And the situation will be ripe for bad actors and well-meaning social media users to spread dark, distorted falsehoods.
Misinformation researchers and activists who spoke to Protocol were nearly united in their message for Election Day: Keep calm. Think before you post. And take everything you see — everything — with a grain of salt.
"You might see things that aren't true, that are inflammatory, that upset you during this time," Todd said. "But remember as they are counting the vote, it might take time — and that's OK."
