Fact-checking Trump, Twitter enters uncharted waters

Two sets of tweets from President Donald Trump show the cracks in Twitter’s evolving policy on policing false information.

For the first time ever, the social media platform added a cautionary flag to a Trump tweet about election fraud. But a baseless tweet alleging murder, Twitter left untouched.

Trump’s May 26 posts about rampant election fraud with mail-in ballots prompted Twitter to tack a link at the bottom that said, "Get the facts about mail-in ballots." That link took users to a page where Twitter wrote, "Trump makes unsubstantiated claim that mail-in ballots will lead to voter fraud." The page mentioned CNN and Washington Post reports that debunked the claim.

In contrast, Twitter let stand Trump’s tweet suggesting that MSNBC host Joe Scarborough murdered a staffer about 20 years ago. We rated that Pants on Fire, and every other news organization that looked at the assertion also found it lacked a scrap of supporting evidence.

University of Notre Dame professor Tim Weninger said Twitter’s moves don’t add up to a coherent policy.

"If the voting tweet was fact-checked by Twitter, then surely the factually false conspiracy tweet implying that Joe Scarborough was somehow involved in the death of a staff member should be fact-checked as well," Weninger said.

Twitter said there’s more to come on the policy front, saying  it hopes to have "changes in place shortly."

Meanwhile, Trump has charged, on Twitter, that the company is "stifling his free speech" (which earned another Pants on Fire rating from PolitiFact). On May 28, he signed an executive order to increase regulation of Twitter and other social media platforms.

Until now, Twitter has given political messages, and certainly any of Trump’s tweets, a wide berth. 

Its current rules aim to provide a free and safe space for public conversation, and focus on clear cut matters of protecting people from real harm. As the presidential election season began, it announced it would not accept paid political ads.

But it targeted two areas for policing — elections and COVID-19. 

It banned tweets for "the purpose of manipulating or interfering in elections." This included "posting or sharing content that may suppress voter turnout or mislead people about when, where, or how to vote."

Twitter’s press office said in an email that Trump’s mail-in ballot tweets didn’t violate its standards, but did "contain potentially misleading information about voting processes," and merited "additional context."

To provide that context, it borrowed the approach it had crafted two weeks earlier to help with COVID-19. When a mixture of people and machine power spotted a tweet that ran counter to what doctors were saying about the disease, Twitter would add the "Get the facts" link to deliver more information. That is the note that appeared on Trump’s election fraud tweets. 

The controversial claims of high-profile politicians have been the third rail of social media, and Twitter stepped on it. Facebook, which partners with PolitiFact and other fact-checkers to police false posts, refuses to downgrade statements from politicians as part of its overall third-party fact-checking program. Facebook CEO Mark Zuckerburg told Fox News that private companies "shouldn't be the arbiter of truth."

With Twitter entering uncharted waters, analysts of social media and society offered some guideposts as the platform navigates the path ahead.

The analysts we spoke to said Twitter should focus on users with the greatest reach. Someone with a large following, like @RealDonaldTrump with 80 million followers, has more potential for doing damage by spreading false information. 

"By virtue of his position, President Trump fulfills both of those requirements," said Emily Vraga, a University of Minnesota professor of  communication. "His tweets are likely to reach a lot of people and to be believed and potentially acted upon. As a result, more scrutiny, not less, should be applied to the information he — or anyone in a similar position — is sharing."

Bhaskar Chakravorti, dean of global business at Tufts University, said Twitter should be free to apply standards of accuracy.

"A private company like Twitter has no obligation to create a platform that gives license for unmoderated free speech that can cause public harm," Chakravorti said.

The two Trump tweets, one about elections and the other about Scarborough, illustrate where things get tricky.

The widower of the woman Trump implied was murdered  asked Twitter CEO Jack Dorsey, to pull down Trump’s tweets.

"I'm asking you to intervene in this instance because the President of the United States has taken something that does not belong to him — the memory of my dead wife and perverted it for perceived political gain," Timothy Klausutis wrote May 21.

But the analysts agreed that elections involve a broader social impact than the personal pain for the friends and family of the stricken Scarborough staffer.

"Misinformation that has the potential for harm to people, society and democratic institutions should be scrutinized most closely," said Chakravorti.

For Vraga, it isn’t clear that the Scarborough tweet crossed that threshold.

"Believing that a prominent public figure may have committed heinous acts does not have the same immediate potential for public harm," she said. "Claims about election fraud have a potentially huge public impact, undermining public faith in our system of democracy. The harm from this distrust is incalculable."

But for Weninger, factually, Trump’s tweets about Scarborough were more blatantly false than the questions Trump raised about mail-in voting.

"For slander and baseless murder accusations, Twitter has to try to handle all or most of those cases," Weninger said.

Our experts were in full agreement that especially with public figures like Trump, Twitter’s best approach would be to do exactly what it did — add context.

"Respecting the First Amendment and the public’s right to speak in defense of their position is a foundational principle that should be stringently protected," Vraga said.

At the same time, she added, claims can’t be divorced from the facts.

"Warning people about misinformation and giving them an easy way to access the facts can limit misinformation's effects while reducing concerns about censorship," Vraga said.

Chakravorti backed that idea, but warned that Twitter has to accept the role it has come to play.

"The social media platforms have to give up the pretense that they are a pure public square with no editorial responsibility," he said. "Twitter needs to have a fact-checking system and continue to improve it over time. Also, it needs to be transparent about the rules that determine whether a tweet is flagged."

To date, Twitter has tried to wall off the areas where it will assert its judgment. It designated elections and COVID-19 for special attention.

Its decision to flag Trump’s election tweet suggests that it might not be able to pick and choose.

"I don't see how you can ring fence some topics for moderation and flagging and not others," Chakravorti said.

Having taken the step to intervene on some issues and not others is unsustainable, he warned. 

None of the analysts we reached said Twitter had done a good job of policing falsehoods, although they saw signs it was getting better. 

In March, Twitter said it would monitor for tweets that were a public health threat. In April, a study found that it allowed twice as many false posts to remain as Facebook and YouTube.

About a month after that study, Twitter announced it would flag questionable COVID related tweets.

And two weeks after that, it applied the same method to Trump’s tweets on elections.

As the company faced blowback from Trump and his supporters, Twitter CEO Jack Dorsey struck a defiant tone.

"We’ll continue to point out incorrect or disputed information about elections globally. And we will admit to and own any mistakes we make," Dorsey tweeted.

UPDATE, 5:10 p.m.: This piece was updated after publication to note Trump's action on the executive order for social media companies.