Facebook Explains Process for Trending Topics Feature Under Fire From Congress for Bias

As Facebook faces congressional inquiries into allegations that one of its website features is biased against conservative content, Vice President-Global Operations Justin Osofsky in a Thursday blog post sought to shed more light on how the Trending Topics module works (see 1605100032 and 1605110048). He said an algorithm, not people, identifies popular topics, "ones that have a high volume of mentions and a sharp increase in mentions over a short period of time." The algorithm also uses an external RSS website crawler to spot breaking news, but it doesn't "consider perspective or politics" and has checks and balances regardless of a story's ideological slant, he said. Then there is a team of Facebook employees who review the algorithm's results, he said. The team confirms the topic is tied to a current real-world news event, writes a topic description, applies a category label such as sports or science, and checks whether it's a national or international breaking news story covered by most or all of 10 major media outlets, Osofsky said. The list of Trending Topics is then personalized for each user through an algorithm based on several criteria. "Not everyone sees the same topics at the same time," he wrote. The social network's guidelines don't "permit the suppression of political perspectives" nor prioritize one viewpoint over another, and the company hasn't directed any reviewers to suppress conservative news, he said, directly contradicting allegations in a recent Gizmodo story from ex-contractual employees. Osofsky said Facebook is taking the reports seriously and "found no evidence to date that Trending Topics was successfully manipulated, but will continue the review of all our practices."