An employee writes a note on the message board at the new headquarters of Facebook in Menlo Park, California January 11, 2012
An employee writes a note on the message board at the new headquarters of Facebook in Menlo Park, California January 11, 2012

Facebook ‘Trending Topics’ controversy leads site to publish details of how site chooses what people want to read

The company claims that humans are a part of the process – but mostly to filter out hoax and other stories and decide how important they are

Andrew Griffin
Friday 13 May 2016 10:51
comments

Facebook has responded to mounting controversy about its Trending Topics feature by publishing the full details of how every story gets chosen.

The company has come under fire over the last week for allegedly encouraging the humans involved in the trending module to avoid stories from conservative outlets and promote progressive ones like Black Lives Matter. It has repeatedly denied that it has manipulated that algorithm and has now laid out precisely how it works, in an attempt to stop further outrage.

It has also said that it has protections in place to ensure that it shows off “the most important popular stories, regardless of where they fall on the ideological spectrum”. It said that its team aren’t allowed to suppress particular political perspectives or to prioritise one ideological viewpoint.

In order to try and shut down the criticism, Facebook laid out the steps that a story goes through when it is on its way to the Trending Topics bar.

The company starts by using a special algorithm that identifies topics that have spiked in popularity on Facebook, it said. That looks out for things that people are talking about a lot, and a lot more, and then shows them to its team.

The same algorithm also scans a huge list of websites for stories that are breaking, so that the site can watch out for news. Doing so presumably means that it doesn’t have to wait until people start talking about a topic, but can identify it before they do.

Things that the algorithm picks out are shown to members of Facebook’s “Trending team”, it said.

They are required to perform a series of checks, which includes ensuring that the topic is a news event, rather than just a lot of people saying they are about to eat lunch, and that it has been corroborated by reporting from a list of news outlets. They also categorise news to say what sort of topic it is about and how important it is.

That is then fed back into another algorithm that personalises news for each person. That algorithm pulls in information including how important news is, where they live, and whether they have clicked on similar topics in the past.

Join our new commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies

View comments