What is algorithmic gatekeeping?
Algorithmic gatekeeping is the automated form of the oldest power in the news: deciding what reaches the audience and what does not. Where an editor once chose the front page, a recommendation system now ranks each person's feed. On October 5, 2021, Frances Haugen, a former Facebook product manager who had worked on exactly these systems, told a U.S. Senate subcommittee that the company had become a $1 trillion business by resolving the conflict between profit and safety in its own favor. She described her own job first:
Primary source 01 / 07
My job has largely focused on algorithmic products like Google+ Search and recommendation systems like the one that powers the Facebook News Feed.
The gatekeeper, she testified, is now the ranking system, and it reaches almost everyone:
Primary source 02 / 07
Right now, Facebook chooses what information billions of people see, shaping their perception of reality.
That reach would matter less if the system ranked for accuracy. Haugen's account, drawn from internal research she copied before leaving, was that it ranked for engagement, and that engagement rewarded conflict:
Primary source 03 / 07
The result has been a system that amplifies division, extremism, and polarization
Profit over safety
Haugen's central charge was not that Facebook's employees were careless. It was that a business paid by engagement faces a standing conflict of interest, and that the company kept resolving it the same way:
Primary source 04 / 07
I saw that Facebook repeatedly encountered conflicts between its own profits and our safety. Facebook consistently resolved those conflicts in favor of its own profits.
The scale of what was being traded, she argued, was enormous:
Primary source 05 / 07
Facebook became a $1 trillion company by paying for its profits with our safety, including the safety of our children.
She reduced the finding to a line she wanted the senators to remember:
Primary source 06 / 07
Facebook chooses profit over safety every day
A gate no one can inspect
What makes algorithmic gatekeeping different from the newsroom kind is that the rules are private. An editor's judgment could be argued with in public. A ranking model is invisible, even to regulators, and Haugen argued that only the company can see how it works:
Primary source 07 / 07
Only Facebook knows how it personalizes your feed for you.
This is the gatekeeper the earlier chapters were tracking. The filter bubble named the personalized feed, the attention economy explained its motive, and algorithmic gatekeeping is the power it now holds over the record. The last chapter turns to the moment the machine stops sorting the record and starts producing it. Facebook publicly disputed Haugen's account.