UTS student completing a Masters in Advanced Journalism. Areas of interest include style, subculture, education and community action.
There is much conjecture about the 'sinister' algorithms social media giant Facebook uses to moderate its Newsfeed, so Central News went to the source to find the truth.
Vice President of Product for Facebook Newsfeed, Adam Mosseri, said the algorithm simply exists to give Facebook users a more “delightful” experience.
Mr Mosseri was speaking to journalism students, scholars, and professionals at the UTS Centre for Media Transition on August 7 about Facebook’s role in the modern news media landscape.
Facebook Newsfeed does not reveal all the items friends follow, share and post. Rather, it provides a select portion of posts based on a complicated algorithm.
“Newsfeed is, in a lot of ways, about helping people find things that they’re going to be delighted in, that they’re going to find meaningful,” Mr Mosseri said.
“We’re looking at finding new and interesting ways to connect people with stories that they may not know exist yet but might find meaningful.”
The algorithm at its simplest level, he says, is a “formula or set of steps, designed to solve a particular problem".
The problem in Newsfeed's case is how to bring users the posts that they’d be most interested in so they don’t have to scroll through so many posts.
Firstly, the algorithm looks at “what’s on the menu”.
The algorithm then looks for a series of “signals” such as who posted it, what time it was posted, the type of media you interact with and even your current internet browsing speed.
Filter bubbles are human nature they exist on and off Facebook
“There’s actually hundreds of thousands of different signals that we use which we then compile into a set of predictions,” Mr Mosseri said.
These predictions include simple things such as how likely you are to comment on, click, watch or share a story. Facebook also makes more complicated predictions such as “how likely would you be to say this was informative if we asked you?”.
All of this data is then compiled into a “relevancy score” which shows how interested Facebook’s algorithm thinks you will be in a story. This is done for every story on the news feed and they are ordered accordingly, with the “most interesting” stories appearing at the top.
These scores are personalised, which means that different users who are friends with the exact same group of people and follow the exact same pages may have a very different looking news feed.
Facebook and its algorithm been heavily criticised in the media for creating a “filter bubble”, where people aren't exposed to things that they don't agree with. This, critics claim, leads to polarisation and has been blamed for influencing the result of the US election and other major elections around the world.
“Filter bubbles are human nature they exist on and off Facebook,” Mr Mosseri said.
He said that amongst the hundreds of friends people have on Facebook there are at least some opposing viewpoints.
Mr Mosseri will also be speaking to Australian politicians in Canberra this week regarding the senate inquiry into the future of journalism. Tech giants such as Facebook and Google have been blamed for loss of business in many news companies.
Facebook is currently working with news publishers with ways to monetise the content they produce on Facebook, with one possibility being a subscription model.
But, questions remain over whether Facebook users, who are used to getting content for free, will be willing to start paying subscription fees for content from specific publishers.
“People are always going to gravitate towards what’s free… but it’s also really important to meet people where they are,” Mosseri says.
“Even if it’s difficult, it’s still a value add for publishers who might have difficulty getting a large enough audience to install their own app or go to their website directly.”
“It’s definitely a worthy challenge to pursue.”