Posted on

How Misinformation ‘Superspreaders’ Seed False Election Theories

On the morning of Nov. 5, Eric Trump, one of the president’s sons, asked his Facebook followers to report cases of voter fraud with the hashtag, Stop the Steal. His post was shared over 5,000 times.

Image

By late afternoon, the conservative media personalities Diamond and Silk had shared the hashtag along with a video claiming voter fraud in Pennsylvania. Their post was shared over 3,800 times.

Image

That night, the conservative activist Brandon Straka asked people to protest in Michigan under the banner #StoptheSteal. His post was shared more than 3,700 times.

Image

Over the next week, the phrase “Stop the Steal” was used to promote dozens of rallies that spread false voter fraud claims about the U.S. presidential elections.

New research from Avaaz, a global human rights group, the Elections Integrity Partnership and The New York Times shows how a small group of people — mostly right-wing personalities with outsized influence on social media — helped spread the false voter-fraud narrative that led to those rallies.

That group, like the guests of a large wedding held during the pandemic, were “superspreaders” of misinformation around voter fraud, seeding falsehoods that include the claims that dead people voted, voting machines had technical glitches, and mail-in ballots were not correctly counted.

“Because of how Facebook’s algorithm functions, these superspreaders are capable of priming a discourse,” said Fadi Quran, a director at Avaaz. “There is often this assumption that misinformation or rumors just catch on. These superspreaders show that there is an intentional effort to redefine the public narrative.”

Across Facebook, there were roughly 3.5 million interactions — including likes, comments and shares — on public posts referencing “Stop the Steal” during the week of Nov. 3, according to the research. Of those, the profiles of Eric Trump, Diamond and Silk and Mr. Straka accounted for a disproportionate share — roughly 6 percent, or 200,000, of those interactions.

While the group’s impact was notable, it did not come close to the spread of misinformation promoted by President Trump since then. Of the 20 most-engaged Facebook posts over the last week containing the word “election,” all were from Mr. Trump, according to Crowdtangle, a Facebook-owned analytics tool. All of those claims were found to be false or misleading by independent fact checkers.

The baseless election fraud claims have been used by the president and his supporters to challenge the vote in a number of states. Reports that malfunctioning voting machines, intentionally miscounted mail-in votes and other irregularities affecting the vote were investigated by election officials and journalists who found no evidence of widespread voter fraud.

The voter fraud claims have continued to gather steam in recent weeks, thanks in large part to prominent accounts. A look at a four-week period starting in mid-October shows that President Trump and the top 25 superspreaders of voter fraud misinformation accounted for 28.6 percent of the interactions people had with that content, according to an analysis by Avaaz.

“What we see these people doing is kind of like setting a fire down with fuel, it is designed to quickly create a blaze,” Mr. Quran said. “These actors have built enough power they ensure this misinformation reaches millions of Americans.”

In order to find the superspreaders, Avaaz compiled a list of 95,546 Facebook posts that included narratives about voter fraud. Those posts were liked, shared or commented on nearly 60 million times by people on Facebook.

Avaaz found that just 33 of the 95,546 posts were responsible for over 13 million of those interactions. Those 33 posts had created a narrative that would go on to shape what millions of people thought about the legitimacy of the U.S. elections.

A spokesman for Facebook said the company had added labels to posts that misrepresented the election process and was directing people to a voting information center.

“We’re taking every opportunity to connect people to reliable information about the election and how votes are being counted,” said Kevin McAlister, a Facebook spokesman. The company has not commented on why accounts that repeatedly share misinformation, such as Mr. Straka’s and Diamond and Silk’s, have not been penalized. Facebook has previously said that President Trump, along with other elected officials, is granted a special status and is not fact-checked.

Many of the superspreader accounts had millions of interactions on their Facebook posts over the last month, and have enjoyed continued growth. The accounts were active on Twitter as well as Facebook, and increasingly spread the same misinformation on new social media sites like Parler, MeWe and Gab.

Dan Bongino, a right-wing commentator with a following of nearly four million people on Facebook, had over 7.7 million interactions on Facebook the week of Nov. 3. Mark Levin, a right-wing radio host, had nearly four million interactions, and Diamond and Silk had 2.5 million. A review of their pages by The Times shows that a majority of their posts have focused on the recent elections, and voter fraud narratives around them.

None of the superspreaders identified in this article responded to requests for comment.

One of the most prominent false claims promoted by the superspreaders was that Dominion voting software deleted votes for Mr. Trump, or somehow changed vote tallies in several swing states. Election officials have found no evidence that the machines malfunctioned, but posts about the machines have been widely shared by Mr. Trump and his supporters.

Over the last week, just seven posts from the top 25 superspreaders of the Dominion voter fraud claim accounted for 13 percent of the total interactions on Facebook about the claim.

Many of those same accounts were also top superspreaders of the Dominion claim, and other voter fraud theories, on Twitter. The accounts of President Trump, his son Eric, Mr. Straka and Mr. Levin were all among the top 20 accounts that spread misinformation about voter fraud on Twitter, according to Ian Kennedy, a researcher at the University of Washington who works with the Elections Integrity Partnership.

Mr. Trump had by far the largest influence on Twitter. A single tweet by the president accusing Dominion voting systems of deleting 2.7 million votes in his favor was shared over 185,000 times, and liked over 600,000 times.

Like the other false claims about voter fraud, Mr. Trump’s tweet included a label by Twitter that he was sharing information that was not accurate.

Twitter, like Facebook, has said that those labels help prevent false claims from being shared and direct people toward more authoritative sources of information.

Earlier this week, BuzzFeed News reported that Facebook employees questioned whether the labels were effective. Within the company, employees have sought out their own data on how well national newspapers performed during the elections, according to one Facebook employee.

On the #StoptheSteal hashtag, they found that both The New York Times and The Washington Post were among the top 25 pages with interactions on that hashtag — mainly from readers sharing articles and using the hashtag in those posts.

Combined, the two publications had approximately 44,000 interactions on Facebook under that hashtag. By comparison, Mr. Straka, the conservative activist who shared the call to action on voter fraud, got three times that number of interactions sharing material under the same hashtag on his own Facebook account.

Jacob Silver contributed reporting.

Posted on

The Only Safe Election Is a Low-Tech Election

After Monday’s Iowa caucus debacle, I’ve decided that Americans should vote by etching our preferred candidate’s name into a stone tablet with a hammer and chisel.

Or maybe by dropping pebbles into a series of urns, as the ancient Greeks did.

Or possibly just by voting the way we voted for much of the 20th century, on analog punch-card machines that spit out paper ballots to be hand-counted by election workers, with zero iPhones in sight.

Basically, we should be begging for the most analog election technology possible. Because what happened on Monday night — a long and confusing delay in vote counting, due in part to a mobile app that was hastily designed and inadequately tested before being deployed in one of America’s most important elections — was an inexcusable failure. It caused distress and confusion, set off innumerable conspiracy theories, and started the 2020 election season by undermining trust in the democratic process.

And all because Iowa Democrats wanted a new app.

The app, whose name was kept secret by Democratic officials, was compared to a “fancy calculator” that was supposed to help Iowa caucus chairs send their results to the state Democratic Party. But as my colleagues reported, it posed problems for caucus precinct chairs all day.

Some chairs weren’t able to use it at all. Others had connectivity issues, or simply didn’t know how to work the app, and were forced to endure long hold times on a phone hotline instead.

There is no indication that any of these technical issues changed the results of the caucuses, or that any systems were hacked or compromised. And there were nontechnical issues that may have added to the chaos, such as new rules and work sheets that were designed to simplify the caucus process but seemed mostly to have sown confusion.

Regardless, the damage was done. The hours spent waiting for overdue results created an information vacuum, which was quickly filled by conspiracy theorists. By nightfall, liberals and conservatives alike were tossing around allegations of vote tampering and election rigging, and casting doubt over the legitimacy of the caucuses.

Democrats quickly began blaming Shadow, the tech start-up that built the app, and Acronym, a Democratic digital strategy operation that invested in Shadow. These firms do deserve scrutiny, not least because it appears that they neglected to quickly respond when reports of user issues began surfacing on Monday.

(Late Monday, Acronym put out a statement, saying that it did not provide technology to the Iowa Democratic Party and that it was merely an investor in Shadow.)

But Democrats should also blame their party’s leadership for entrusting such an important process to new technology in the first place — not just in Iowa but in places like Nevada, where Democrats are reportedly planning to use a similar mobile app to tally votes in that state’s caucuses this month.

It’s enough to make you wonder: Have these party officials ever been to a polling site or a caucus venue? They are not pristine WeWorks with blazing fast internet connections and an army of Geek Squad workers on call. They are mostly high school gyms, nursing homes and church basements with cinder-block walls and horrible cellphone service. The people who work at them are volunteers, and many are — how can I put this delicately? — members of the generation that still refers to the TV remote as “the clicker.”

Using a proprietary app to report vote totals is the kind of thing that sounds simple on a start-up’s whiteboard but utterly falls apart in a chaotic real-world environment, where connections drop, phones malfunction and poorly tested apps strain under a surge of traffic. Add an army of frenzied poll workers, impatient voters and twitchy news media, and you might as well have asked the caucus workers to whip up their own JavaScript.

I’m not opposed to technology in political campaigning. Want to use Facebook ads to drum up donors? Go for it. Want to put your voter database on the blockchain? Be my guest.

But when it comes to the actual business of registering and counting people’s votes, many of the smartest tech experts I know fiercely oppose high-tech solutions, like “paperless” digital voting machines and mobile voting apps. After all, every piece of technology involved in the voting process is a possible point of failure. And the larger and more interconnected the technical system, the more vulnerable it is to an attack.

“Many of the leading opponents of paperless voting machines were, and still are, computer scientists, because we understand the vulnerability of voting equipment in a way most election officials don’t,” said Barbara Simons, a computer scientist and board chair of Verified Voting, an election security nonprofit, in an interview with The Atlantic in 2017.

In Iowa, there is a silver lining: The caucus system doesn’t use voting machines at all, and its public, open-air nature means that it is less susceptible to tampering than a secret ballot. In addition, state officials this year required caucusgoers to fill out paper “presidential preference cards,” which could be used in case of a recount. So despite the delays, there is at least some certainty that the results will be accurate when they are finally announced.

And it may be too close to November’s elections to overhaul state election technology, or to institute the kind of much-needed election security reforms that have been repeatedly proposed by Democrats (and blocked by Republicans) since 2016.

So unless something changes, get used to the era of the voting day software glitch. Because we’ve invited tech into our elections, whether or not it’s up for the task.