An Ugly Truth: Inside Facebook's Battle for Domination [calibre 4.99.5]

Over the course of six weeks, the security team constructed a map of the Internet Research Agency’s activity on the platform. The Russians had bought more than 3,300 advertisements, which had cost the IRA roughly $100,000 to promote on Facebook; some of those advertisements did little other than promote the 120 Facebook pages the IRA was running. They also produced more than 80,000 individual pieces of organic content. In total, an astonishing 126 million Americans had been reached.

Wherever there was a “seam line issue,” a divisive position that could turn Americans against one another, there was the IRA. Gun control, immigration, feminism, race—the IRA ran accounts that took extreme stances on each of these issues. Many of their pages supported the Trump campaign and conservative groups across the United States, but they also ran pages in support of Bernie Sanders. “It was a slow, slow process, but what we found that summer, the summer of 2017, it just blew us away,” recalled one member of Facebook’s security team. “We expected we’d find something, but we had no idea it was so big.”

While over the years some Facebook employees had speculated that the IRA was focused on spreading disinformation in the United States, no one had thought to go looking for a professional disinformation campaign run by the organization. The security team had not believed the IRA audacious or powerful enough to target the United States. Facebook’s assurances to Senator Warner and other lawmakers that they had fully uncovered Russia’s efforts to influence the 2016 elections were based on the assumption that they were looking for Russian military intelligence activity, but they had missed the forest for the trees. While they were chasing state-backed hackers spreading emails from the Clinton campaign, they had overlooked the IRA’s highly effective troll army.

The discovery of the IRA ads prompted Stamos’s team to double down. In Menlo Park, they were already isolated in their building on the outskirts of the Facebook campus; Moran and a small number of other members of the threat intelligence team were working in a dedicated room in Facebook’s DC offices. They were all instructed not to discuss their findings with any of their colleagues. They were told that executives wanted to produce a comprehensive report on the IRA activity before they went public with their findings.

So, the team was forced to watch in silence as Facebook’s own spokespeople unknowingly misled the press. On July 20, by which time some clusters of Russian ads had already been found but had not been linked directly to the IRA, a DC-based Facebook spokesperson informed a CNN anchor that the company had “seen no evidence that Russian actors bought ads on Facebook in connection with the election.” The spokesperson worked less than one hundred feet from where the threat intelligence team sat, but he had not been briefed on its findings.

On August 15, “Project Campaign,” a dedicated team that had been formed following the discovery of the ads, published an internal report that highlighted hundreds of Facebook and Instagram accounts that had been linked to the IRA. The report revealed that the team had also found, but had not analyzed, real-world Facebook events created by the IRA, such as protests and other gatherings, as well as a number of IRA-run pages that were co-run by Americans. The IRA had even sent emails and messages to Facebook employees, asking to learn more about the company’s policies and have their pages and accounts reenabled when they ran into technical problems.

Project Campaign’s report recommended that, going forward, Facebook immediately change its policies to allow the security team to remove any Facebook account found participating in “coordinated inauthentic behavior”—defined as any accounts that “mislead users about the origin of content on Facebook,” “mislead users about the destination of links offsite,” and “mislead users in an attempt to encourage shares, likes, or clicks.”