A year and a half ago, as the world erupted in mass protests in response to the murder of George Floyd and racist police violence more broadly, workplaces and institutions across the United States faced a reckoning over the country’s history of systemic racism, and their roles in it. One of the first companies rocked by this movement was Facebook.
On May 29, 2020, just a few days after Floyd’s death on May 25, and in response to reports of occasional fires and building break-ins during the initial protests in Minneapolis, U.S. President Donald Trump posted a contentious message on both Facebook and Twitter that read in part, “when the looting starts, the shooting starts.” Outraged observers immediately noted the racist history of the phrase. Twitter soon hid the tweeted version of this statement with a warning message, noting that Trump’s comment “violated the Twitter Rules about glorifying violence.” Nevertheless, it elected to leave the post up (albeit behind a scrim) in favor of the public interest.
Facebook took a different tack, leaving it up without any warnings. An initial review found the post to be “non-violating” of Facebook standards, and CEO Mark Zuckerberg defended the decision to leave the post up, even as he claimed to have “a visceral negative reaction to this kind of divisive and inflammatory rhetoric.” This led to much internal dissent, with senior-level Facebookers criticizing Zuckerberg; other employees staged a “virtual walkout” in protest and demanded answers. In June of last year, racial justice advocate Brandi Collins-Dexter directly cited Zuckerberg’s actions as a key example of why Black activists tend to distrust Facebook.
Now, new disclosures made to the Securities and Exchange Commission and provided to Congress in redacted form by whistleblower Frances Haugen’s legal counsel show that this intra-Facebook conflict rocked the company far more than the world realized. The redacted versions received by Congress were reviewed by a consortium of news organizations, including Slate.
Among the documents is a series of charts showing the results of two years’ worth of internal surveys probing employees’ perceptions of Facebook. The surveys were taken twice a month from January 2019 through December 2020, and they measured four factors: how optimistic employees felt regarding the company, how confident they were in the executive leadership, how proud they were to work at Facebook, and how much of a social good they felt the platform offered.
Near the beginning of March 2020, overall employee optimism in the company stood at about 58 percent—but near the end of the month, it shot up to 75 percent. Pride in the company undergoes a similar trend, going from 60 percent at the beginning of March to about 80 percent by the end of it. Confidence in Facebook leadership and perception of the network as a social good floated at similarly high levels during this time. One can reasonably assume these good feelings about Facebook were fueled by the sudden burst of traffic—a 50 percent surge over the course of just one March week, according to the New York Times—that hit the site as the pandemic began to spread within the U.S. That spike could be attributed to any number of factors, including coronavirus news on users’ feeds and the use of groups by health care workers to share stories and advice about treating COVID. There was also added confidence in Facebook’s subsidiary products, including the utility of Messenger and WhatsApp for audio and video calls between distant loved ones. Likely, Facebook’s workers held the sentiment expressed in a representative Times report from March 23, 2020: that “big tech could emerge from [the] coronavirus crisis stronger than ever.”
But this honeymoon didn’t last long. After optimism in the company peaked at 82 percent in mid-May 2020, it plummeted in early June to about 60 percent and further sank throughout July. It lifted again in August, sank throughout September, rose again slightly in late October, and then dropped sharply after November, hitting about 50 percent in December—just before the Jan. 6 Capitol insurrection, which, as the Washington Post reported as part of its own examination of the Facebook Papers, the platform played no small part in fomenting.
The other measures of internal company approval also changed throughout this time. By May 25, confidence in leadership was measured at 75 percent—before dropping to about 47 percent just a couple weeks later. For the rest of the year, faith in leadership followed similar trends to the measures of optimism (lifting slightly in August and October, dropping in September and early December), never again even reaching close to that May peak. Pride in the company rose from 60 percent in early March 2020 to about 80 percent throughout the rest of the month—only to collapse to about 54 percent after May. It followed arcs similar to those of optimism and faith in leadership throughout the rest of the year. Belief in Facebook as a social good, measured at 80 percent near the end of March, slightly declined through the spring, and dove from 73 percent at the end of May to about 48 percent by early June.
Though it is difficult to declare causation, the timing across all four charts clearly suggests one turning point: Facebook executives’ inaction on Trump’s “looting/shooting” post after May 29, 2020. These internal surveys demonstrate that Facebook quickly squandered the goodwill it had rebuilt for itself in the early months of the pandemic, very likely due to its ambivalence regarding Trump’s Facebook-supported and -spread rhetoric—especially in comparison with Twitter’s actions for the same post.
Testimony from current and former employees and other documents from the Facebook Papers show concrete examples of the impact of the “looting/shooting” controversy. In September 2020, the first results of a survey of Facebook users that “measures [the] perceived legitimacy” of the platform’s harm-reduction efforts, was published internally to FB’s Integrity team, which Frances Haugen was a member of at the time. The survey results displayed three “beliefs” that led to users undermining the legitimacy of Integrity’s work, with one stating that “FB is complicit in perpetuating societal harm, including ruining Democracy + offline violence.” One of the examples cited as a likely contributor to this belief was the “Trump ‘looting + shooting’ post.” Interestingly, the survey cites other examples of how Facebook’s poor handling of racial issues may have negatively affected perceptions of its harm-reduction work, including a July 2020 Recode piece on civil rights leaders’ anger at the network over hate speech.
I also found another screenshot in which a data scientist (name redacted) who worked in monitoring and preventing “Violence and Incitement” on Facebook created a much-viewed, undated Google Doc about why they were leaving the company. While they said they were not departing because of any anger at Facebook or the individuals working at the company, they did write:
Working on Violence and Incitement throughout the US election year was taxing. We made some decisions during this time, particularly in the months leading up to the election starting in around May, that I vehemently disagree with, and that did make work during this time even more challenging. …
Trump’s “Looting and Shooting” post was viewed orders of magnitude times more than the total number of views that we prevent in a day, and it was incitement of violence in the clearest sense of the term. It’s not hard to draw a straight line from that post to actual shootings that took place at protests. …
One other has made me uncomfortable for the last few months, which again is not the chief reason for my departure but I’d be lying if I said it didn’t contribute to my starting to look around. I wrote a note in June called We’re Not Allies Until We Do Something. … Despite a lot of lip service being paid to the concept and label of allyship, as a company we had done very little to back it up. This was at the same time that the “Looting/Shooting” post went up, and I wrote, “Personally, I have a very hard time looking my Black colleagues in the eye and asserting my commitment to allyship knowing that the system that I contribute to building is used to incite violence against them.” … Since then, I really don’t think much has changed for the better.
In another example, my colleague Aaron Mak and I recently interviewed former Facebook engineer Behdad Esfahbod about how Facebook responded when he was detained in Iran for seven days. He told us that after months of frustration with Facebook, he finally resigned after HR confronted him about his tweets calling out bad actors in the tech industry—even as the platform had allowed Trump’s “looting/shooting” post to stay up.
What the Facebook Papers show is that Donald Trump’s incitement to violence against anti-racist protesters may have been excused by Facebook—but not by many of its employees.