Cleaning up social media won’t be easy, particularly since banning or significantly throttling more prominent accounts even after repeated violations of policy or common decency would be bad for business. Top accounts appear to be treated more leniently than the general public, forcing Facebook, in one recent episode, to explain why it wasn’t giving Steve Bannon the boot after he suggested that Dr. Anthony Fauci and Christopher Wray, the director of the F.B.I., should be beheaded. Facebook said Mr. Bannon hadn’t committed enough violations.
It’s really about money. Divisiveness brings more engagement, which brings in more advertising revenue.
Users should worry that Facebook and Twitter won’t maintain the same level of vigilance now that the election has passed. (Facebook’s chief executive, Mark Zuckerberg, said as much, according to BuzzFeed.) And the incentives for posting misleading content didn’t disappear after Nov. 3.
If the companies truly care about the integrity of their platforms, they’ll form teams of people to monitor the accounts of users with the most followers, retweets and engagement. That includes those of Mr. Trump, both today and later as a private citizen, but also of President-elect Joe Biden and President Jair Bolsonaro of Brazil, and other influential accounts, like those of Elon Musk, Bill Gates and Taylor Swift. Facebook says it has software tools to identify when high-reach accounts may violate rules, but they clearly are not catching enough quickly enough.
Think of these frontline moderators as hall monitors whose job is to ensure that students have a pass, but not necessarily to issue penalties if they don’t. The monotony of refreshing Justin Trudeau’s social media feed is worth it for the preservation of democracy and promotion of basic facts.
“For the platforms to treat all the bad info as having the same weight is disingenuous,” said Sarah Roberts, an information studies professor at the University of California, Los Angeles. “The more prominent the profile, the higher the accountability should be.”
With such a system, the companies could ensure the swiftest possible response so that posts are vetted by actual people, including outside fact checkers, familiar with company policy, nuance and local customs. When they rely too much on software to decide what to examine, it can happen slowly or not at all. Particularly in the heat of an election, minutes count and dangerously false information can be seen by millions immediately. If enough people believe an unfettered lie, it gains legitimacy, particularly if our leaders and cultural icons are the ones endorsing it.
Posts, tweets and screenshots that lack a warning label are more likely to be believed because users assume they have passed Facebook’s and Twitter’s smell tests, said Sinan Aral, a Massachusetts Institute of Technology professor who studies social media.