What You Need to Know About Facebook’s Latest Content Transparency Reports
Abby Long is the Senior Managing Editor at PMG.
A few days ago, Facebook released a new report that outlines the most popular content viewed on the Facebook platform throughout the last quarter. Titled the Widely Viewed Content Report, the data analysis was presented as a step in the right direction in upholding Facebook’s commitment to transparency, but was soon met with criticism from analysts and industry watchdogs, as the data presented in the report prompted more questions than answers about what information Facebook users are engaging with most on the platform.
The vast majority of the most popular content outlined in the widely viewed content report was innocuous, relating to recipes, memes, and images of cute animals, supporting Facebook’s argument that the platform has gone to great lengths to reduce the spread of misinformation on the platform. For years, Facebook has been criticized for being “deliberately opaque” about how content is shared on the platform, and this report was likely aimed to address those concerns. The widely viewed content report was also meant to “offer context” that contrasts with the popular Facebook’s Top 10 Twitter feed that’s run by Kevin Roose, a New York Times journalist. The Twitter feed sources information from CrowdTangle, a Facebook-owned analytics platform, and is often used by policy experts to argue that political bias is ever-present on Facebook as the majority of popular content is linked to Facebook Pages managed by political news outlets and pundits.
Screengrab of information presented in Facebook’s Q1 widely viewed content report
As reported by The Verge, “Facebook says that the report shows what people actually see on the platform, as opposed to what content on Facebook gets the most interaction.” In other words, Facebook chose to highlight its “most viewed” content over “most engaged” content in this report.
Eighty-seven percent of posts that people viewed in the U.S. during Q2 2021 did not include an outside link.
Fifty-seven percent of posts in news feeds come from the friends users have chosen to follow, while 19.3 percent comes from Facebook Groups and 14.3 percent originates from Facebook Pages followed.
The top 20 domains included in the Facebook data represent only 1.9 percent of the stories in Facebook newsfeeds, with those associated with traditional news sites representing only 0.3 percent.
It’s no surprise to learn that YouTube, Amazon, Gofundme, and TikTok are among the top domains viewed on the platform.
One surprising discovery in the data set was playeralumniresources.com, which was the ninth most popular domain in the Facebook news feed, with its homepage serving as the most common URL shared on Facebook in Q2 of this year, garnering 87.2 million views last quarter. This information, along with other signals, led many to believe the data was incorrect.
Soon after Facebook’s widely viewed content report was released, The New York Times reported that Facebook had shelved a previous version of the report fearing public outcry, as the report contained evidence that misinformation about COVID-19 was among the most viewed articles on the platform. After the New York Times article was published, Facebook communications executive Andy Stone tweeted a link to the previously unreleased Q1 content transparency report (PDF) and cited the complexities of categorizing misinformation as the reason for shelving the report.
Discrepancies between the two reports include skewed analyses, varying data sets, and more, with the situation continuing to spark growing criticism against Facebook’s data collection and transparency practices. For now, Facebook plans to release a content transparency report on a quarterly basis, citing plans to be more transparent moving forward.
“We’re guilty of cleaning up our house a bit before we invited company. We’ve been criticized for that; and again, that’s not unfair.”
Stay in touch
Subscribe to our newsletter
— Andy Stone, Policy Communications Director at Facebook