by Steven Rosenfeld February 15, 2021
How Online Propaganda Grew More Insidious as 2020 Election Day Approached
(Image: Nased.org/Stanford Internet Observatory/Matt Masterson)
Partisan disinformation to undermine 2020’s presidential election shadowed every step of the voting process last year but took an unprecedented turn when the earliest false claims morphed into intricate conspiracies as Election Day passed and President Trump worked to subvert the results, according to two of the nation’s top experts tracking the election propaganda.
At the general election’s outset, as states wrapped up their primaries and urged voters to use mailed-out ballots in response to the pandemic, false claims began surfacing online—in tweets, social media posts, text messages, reports on websites, videos and memes—targeting the stage in the electoral process that was before voters. These attacks on the nuts and bolts of voting, from registration to the steps to obtain and cast a ballot, began as “claims of hacking and voter fraud… [that] honed [in] on specific events,” said Matt Masterson, who helped lead the Department of Homeland Security’s election security team.
“This is a lot of what we talked about with you at CISA [the U.S. Cybersecurity and Infrastructure Security Agency] in the lead-up [to Election Day], anticipating that were there were problems experienced, and then in the contested elections, those would be used to blow out of proportion or lie about what was actually taking place,” Masterson said, speaking to the nation’s state election directors in early February at a winter 2021 conference.
But as November 3’s Election Day approached and the vote counting continued afterward in presidential battleground states, Masterson and a handful of teams working inside and outside of government to trace and track disinformation, and to urge online platforms and sources to curb their false content, saw an unexpected development. The narrowly focused threads that attacked earlier steps in the process of running elections swapped out purported villains and protagonists and became a full-blown conspiratorial tapestry attacking the results.
“They all got combined into one big narrative… one large lie to try to undermine confidence in the election,” said Masterson, whose presentation at the National Association of State Election Directors’ (NASED) meeting traced this evolution.
“Misinformation is the frontier in election security and election integrity,” said Aaron Wilson, senior director for election security at the Center for Internet Security, which tracked 209 cases of misleading or deliberately false attacks on voting, at the same NASED forum.
Early Predictable Attacks
Masterson’s and Wilson’s presentations were some of the most detailed analyses yet tracing the evolution of propagandistic attacks on 2020’s voting process and election administration. The Stanford Internet Observatory, where Masterson is a fellow, will release a full report—including naming the biggest purveyors of 2020 election disinformation, both the platforms and their highest-volume users—later this winter.
Election officials knew they would be targets for partisan misinformation (mistaken claims) and disinformation (intentionally distorted claims) in 2020. Their first lines of defense, following the cyber intrusions by Russia in 2016, were hardening their infrastructure—the computers that run elections—and creating clearinghouses to rapidly communicate about threats and responses among the nation’s more than 8,000 election jurisdictions. Masterson and Wilson led efforts within this sphere, where, by all accounts, 2020 saw no major cybersecurity breaches.
But while election officials were pleased with the steps they were taking inside their state and local offices, the outside attacks on the nuts and bolts of voting kept building during the general election. The early attacks were narrowly focused but shrewd, noted Wilson, who said election disinformation’s purveyors exploited the public’s lack of knowledge of how elections are run. As Masterson noted, the initial wave cited procedural steps to claim voter fraud and hacking.
For example, as voting rights groups sent absentee ballot applications to voters in swing states, posts appeared on social media falsely stating that voters—and dead people—were receiving multiple ballots, Wilson said. As states put up online voter registration portals to assist voters during the pandemic, online posts falsely asserted that voter information could be sabotaged—altered by political opponents to block would-be voters. When early in-person voting began, false claims erupted about when and where to vote, using ballot drop boxes, the results (before there were any), and votes being thrown out.
“A theme that permeated the misinformation that was reported to us was that it really resulted from or took advantage of people’s lack of knowledge of how elections are run,” Wilson said.
In the week before Election Day and in its immediate aftermath, the volume of misinformation and disinformation increased. Half of CIS’s cases emerged in this period. As the process shifted to casting ballots and counting votes, more conspiratorial narratives emerged where the vote-counting process became the target—and the villains were swapped to fit the new storyline.
Masterson offered two examples of this transition. The first showed how ex-President Trump’s fervent supporters rapidly embraced the false claim that Trump votes were being disqualified because of the pens they used to mark their ballots. The second example showed how older false claims were updated and expanded in ways not seen in prior election propaganda.
More Than Echo Chambers
The first example was “Sharpiegate,” which emerged on November 4, a day after Election Day.
“Sharpiegate was, of course, the claim that using Sharpies [pens] on ballots either invalidated the ballots or the votes weren’t counted,” said Masterson. “It originated in Arizona. And what you saw was these messages really begin to take off. Of the 100 messages that were shared [on Twitter], they got 200,000 or more retweets, or likes, or furtherances.”
Within hours, fact-check organizations like PolitiFact posted responses on Twitter, he said. But those posts barely drew more than 10,000 viewers on November 4 and the next day, whereas the “#Sharpiegate” retweets escalated to 20 times that volume. Masterson said Sharpiegate showed “how quickly a narrative can take off, and despite really good efforts to push back, how fast people will latch onto a false or misleading narrative.”
The episode didn’t stop there—and showed how the architecture of online communications amplified a patently false claim to an audience primed to receive it.
“It started in Arizona, but it didn’t take much time to then have those claims alleged in other states, other jurisdictions, Michigan specifically, even if the same [voting] systems, the same pens, Sharpies, weren’t used at all,” Masterson said.
Election officials did not sit idly by. The secretaries of state in Arizona and Michigan, and county election departments in those states, all responded with their own tweets on November 4, he said. Their rebuttals took the best form at dispelling misinformation: first stating facts, then addressing the disinformation’s claims, and then laying out other information, he said.
Notably, two Arizona NBC TV affiliates reported on the fabricated controversy and posted on its Twitter page, “Sharpies do not invalidate ballots,” Masterson recounted, showing slides of the posts. “Those are two local television stations pushing back, offering facts, [yet] along the side there, in the comments, people are basically saying, ‘You’re lying,’ ‘You’re incorrect,’ ‘You don’t know what we are talking about…’ ‘You’re covering it up.’”
Masterson then turned to his second case study, which showed “the building of the conspiracy or narrative around a fraudulent election.”
Hammer and Scorecard Becomes Dominion
Masterson started with a November 2018 post on a social media page for QAnon, which is an increasingly popular and layered far-right conspiracy that, among other things, baselessly accuses leading Democrats of operating pedophile rings and drinking the blood of children. The 2018 post includes “claims or theories—false, incorrect—that DHS [the Department of Homeland Security—where Masterson was a senior adviser for election security] was putting watermarks and isotopes on ballots in order to track those ballots to track voter fraud.”
“It was not new to 2020. But it built. It grew. The narrative got more and more complex as it went on,” he said. “And ‘Ballotgate’ became a phrase that started with that tighter conversation around how DHS was going to track voter fraud and crack down on it, and began to be used to describe any claims of any manipulation of any ballots… which then grew into ‘Hammer and Scorecard.”
“For those of you not familiar, Hammer and Scorecard was a claim that there were two pieces of software that U.S. intelligence had developed to use internationally to rig [voting] systems, and the two pieces of software allowed for the manipulation of the systems in it,” he continued. There were several versions of this claim. One said that foreign governments were using the software against Trump. Another claimed that federal officials were using this software against Trump. Another blamed unspecified domestic “bad actors.”
The “obviously, demonstrably false” Hammer and Scorecard story drew hundreds of thousands of retweets and shares in the week after Election Day, Masterson noted. By the next weekend, the related traffic on Parler—an unregulated platform favored by Trump supporters until it was taken offline in January—escalated into hundreds of thousands of messages. Those conversations then began to blend with claims that Dominion Voting Systems, whose balloting and counting machines were used in a few swing states, were secretly stealing Trump votes.
“Hammer and Scorecard morphs into, instead of just this CIA-, intelligence community-focused theory, to then begin to talk about Dominion and the various conspiracies about Dominion,” he said. Theories about unadvertised counting features on voting systems being used by insiders to steal votes have circulated among the political left for two decades, Masterson noted.
“But now they got all combined into one big narrative that used Hammer and Scorecard, and Dominion, and other systems into one large lie to try to undermine confidence in the election,” he said, “to the point… [where] there were conversations [by Republican legislators in Arizona] about the seizure of voting systems. That should make any state election official’s skin crawl and shudder. Because none of it is true, and yet there’s this push to use the lie to undermine the results completely.”
More Transparency, More Propaganda
Masterson and Wilson discussed other 2020 trends that had unexpected consequences that fed the stolen election narrative. By many measures—live video streams, public and press viewing areas, partisan election observers—the 2020 presidential election was the most transparent ever. But some of those public images were put forth as false evidence of election theft, they said. Images of ballot drop boxes and storage bins were top examples, where pro-Trump pundits and bloggers claimed that the images showed vote theft in progress.
“The same goes for data,” Masterson said. “We have more data available around elections than we’ve ever had before. I think that’s only going to increase. That is, again, a positive, a good thing. But we saw over and over again the misapplication of election data, whether it was election night reporting… [or] vote totals—you know, the claims that there were dumps of votes, even though election officials had messages over and over and over again [about] how the vote count was going to proceed.”
More insidious were statistical reports that purported to show vote count irregularities from academics and others who had little experience running elections or that made big assumptions—such as that registered Republicans would only vote for Trump. “I know [MIT election scholar] Charles Stewart and the folks at Stanford [Internet Observatory and its partners] just picked apart all these statistical inaccuracies and claims. But [those making the claims were] using the data to create really good-looking but completely misleading and incorrect charts… to undermine confidence.”
Masterson said that foreign adversaries, such as Russia and Iran, both overtly and covertly drew on the domestic disinformation campaigns to fan an already chaotic election—for example, there were reports of Iranian intelligence officials posing as the far-right Proud Boys and sending threatening emails to Democrats.
He praised some statewide election officials, such as Wisconsin’s Meagan Wolfe and Georgia’s Gabriel Sterling, for being constant presences that debunked disinformation. He said that a constant media presence was needed in 2020 and would be needed in future elections.
“The more avenues that it’s coming at people, the more likely they are to both see it and digest it, because they are seeing it [disinformation] from multiple sources,” he said. “The response to these claims needs to be a continued broad push of transparency and facts, not recoiling and saying, ‘It doesn’t matter. It’s already too late…’
The Platforms Feint Response
Masterson and Wilson ended their NASED presentations on upbeat notes. But their analyses underscored that online disinformation attacking the voting process was often more effective in its ability to propel cynical partisan beliefs than factual rebuttals.
Immediately after their talk, another little-known aspect to combatting 2020’s misinformation emerged. The next panel featured representatives from Twitter, Facebook and YouTube. They defended how their platforms dealt with the falsehoods on their perches, such as posting labels on posts that were disputed or false, posting voter alerts, and occasionally taking down posts.
When the time for questions came, the very officials who avoided cybersecurity breaches in 2020, who pivoted to voting by mail and early voting in the pandemic, and who presided over America’s highest-turnout presidential election sat in grim silence.
“I’ll jump in,” said Judd Choate, Colorado’s elections director. “There was a real concentrated effort to dismiss or undercut all the basic tenets of the way we operate elections.”
“We had Sharpiegate. We had attacks on our voting systems and on our election policies, claims of fraudulent votes, dead voters, and so forth, all of which we have the facts. We have the ability and wherewithal to… attack on each one of those claims,” he said.
But after Election Day, when counting votes was under attack and new narratives emerged that attacked the accuracy of the results and election’s legitimacy, Choate said that the platforms’ ban on political ads prevented officials from responding to falsities filling their platforms.
“Colorado, in particular, made several attempts to purchase time on Google, and [we] were rebuffed every time. We were categorized as a political ad. We’re clearly not a political ad. We were the facts. We were the trusted voice,” he said, adding that Colorado met similar obstacles at Facebook.
“Going forward… we need the ability to be proactive,” he said. “And we really didn’t have it in this post-election environment.”