One of the earliest studies Facebook conducted to measure its influence centered on American electoral politics. On the day of the 2010 midterm election, the social media site encouraged users to vote. It was a controlled experiment, and the results were convincing. About 340,000 people voted in the election who otherwise might not have after seeing a message indicating that their friends had. Researchers have known for years that social pressure can increase turnout, as demonstrated in offline experiments. But here was evidence that a site used by millions of Americans might be able to get a large percentage of them to vote.
2021 Election: Complete coverage and analysis ArrowRight
In the abstract, it’s an interesting finding and one that seems like a boon in a country where people often skip elections when the presidency isn’t on the ballot. But in practice, Facebook turned out less likely voters from its user pool, meaning, it’s safe to assume, a disproportionate number of Democratic-leaning voters. Facebook knows better than anyone how its users view politics, given the information that’s shared. A backlash emerged: Could Facebook turn its dials to pick electoral winners?
As it turns out, this was a minor dust-up in the fight over Facebook’s role in politics. A few years later, it removed human moderation of its news feed because of complaints that right-wing publications were being excluded. The moderators argued that they were simply removing false information and, sure enough, when the process was automated, fake news quickly propagated. All of this preceded the 2016 election, in which foreign intervention and abuse on the platform quickly attracted massive scrutiny. Neither was why Donald Trump was elected — that credit is probably more because of his campaign’s huge investment in Facebook’s targeting tools — but each meant new scrutiny on the company.
Advertisement
Story continues below advertisement
During Trump’s time in office and in part because of his victory, Facebook and Twitter faced new scrutiny for user behavior, including abuse and misinformation. Each, in turn, increased content moderation, leading to sanctions against various people. For Trump and his allies, this was useful. They elevated examples of content removal as evidence that the Powerful Silicon Valley Liberals were trying to silence conservative voices.
This was simply an online version of a long-standing practice on the right, trying to police media gatekeepers by implying bias or exaggerating negative consequences. It took decades before mainstream media outlets learned to treat such assertions with skepticism; even now, bad-faith attacks are often not treated as such. But for Facebook, the pressure seems to have had an effect.
Documents leaked by an internal company whistleblower show how effective it was. The Financial Times reported Monday that a memo written by a Facebook employee in December criticized the way in which repeat offenders on misinformation were nonetheless allowed to continue to use the platform.
Advertisement
Story continues below advertisement
“The employee wrote that decisions were made to exempt ‘publishers on the grounds that they were ‘sensitive’ or ‘likely to retaliate,’ ” the paper’s Hannah Murphy and Dave Lee write. “The note added: ‘In the US it appears that interventions have been almost exclusively on behalf of conservative publishers,’ citing Breitbart, Diamond and Silk, Charlie Kirk and PragerU in particular as receiving special treatment.”
At the same time, other documents obtained by the Wall Street Journal last year showed how the company dealt with anticipated complaints from right-leaning outlets about efforts to show fewer political posts: Facebook weighted the algorithm against left-leaning ones.
What brings this entire conversation full circle, though, is another detail from the new set of documents, as reported by The Post’s Elizabeth Dwoskin, Tory Newmyer and Shibani Mahtani.
Advertisement
Story continues below advertisement
“Ahead of the 2020 U.S. election, Facebook built a ‘voting information center’ that promoted factual information about how to register to vote or sign up to be a poll worker. Teams at [Facebook-owned] WhatsApp wanted to create a version of it in Spanish, pushing the information proactively through a chat bot or embedded link to millions of marginalized voters who communicate regularly through WhatsApp. But [CEO Mark] Zuckerberg raised objections to the idea, saying it was not ‘politically neutral,’ or could make the company appear partisan, according to a person familiar with the project who spoke on the condition of anonymity to discuss internal matters, as well as documents reviewed by The Post.”
Zuckerberg — often cited as the final stop on decisions linked to politics — objected to a tool that would offer Spanish-language information on registering to vote because it might not seem sufficiently “neutral.” This is not a difficult thread to unwind: Hispanic voters vote more heavily for Democrats, so encouraging them to vote may be a civic good, but it also runs the risk of further riling up the right.
Again, a central reason for the complaints about bias in the first place was to get the company to act to the right’s benefit in precisely this way. Facebook will be no more able to shut down complaints about bias than The Washington Post because they, like us, are a useful foil to the right and because the criticism itself is seen as a way to neuter unhelpful things. Facebook wanted to get more people to vote, and it worried that it would be seen as partisan in doing so. The (right’s) system works.
Meanwhile, a private philanthropic organization funded by Zuckerberg and his wife contributed money to fund county-level election systems across the country. The result? Constant attacks by Trump targeting Zuckerberg and Facebook.
Advertisement
Story continues below advertisement
“Mark Zuckerberg alone spent $400 million on election meddling,” he said during a speech in Texas earlier this year. This is not true, of course. “In virtually all of the key swing states, he funded unmanned and unprotected drop boxes that were deployed in Democrat-run cities and heavily Democrat precincts to scoop up ballots,” Trump said.
“Zuckerberg broke the law spending millions of dollars,” he said at a dinner in North Carolina. This is not true, either. “Don’t you think he broke the law? Millions of dollars to get out the vote efforts in highly Democrat areas,” Trump said.
After the Journal’s recent report about how Facebook’s content moderation mechanisms had given Trump and his son a pass on its rules, Trump attacked Facebook, which he conflated with the election-funding effort.
Story continues below advertisement
Part of this is the fact that Trump probably doesn’t understand and certainly doesn’t care about the distinction between these different entities and what they’re trying to do. Part of it, too, is that he and his allies have learned that attacking Facebook as biased will have the intended effect of getting the company to act to its benefit even when they don’t say a word.
Facebook could turn out more voters, as its own research has shown. But in 2020, it decided against directly supporting the effort to increase turnout because it didn’t want to be attacked for helping Democrats. So the company did nothing — and got attacked for helping Democrats anyway.
This is how it works.