Australia to Run a National Experiment on Social Media Algorithms

The Australian government is moving ahead with plans to make algorithmic social media feeds an opt-in option rather than the default experience, which would switch most users to chronological feeds unless they actively choose otherwise. The move could prove to be a trial balloon for regulators in other markets watching to see whether the change produces measurable improvements in user wellbeing or simply reduces time on platform.

The case for algorithmic sorting has always rested on engagement data. Facebook made the original argument in 2014, when users had an average of 1,500 potential stories per feed visit and read only 57% of them without algorithmic help. With algorithmic sorting, that figure rose to 70%. Meta's repeated tests since then have consistently shown the same result: when apps revert to chronological order, the number of stories people read and the interactions they generate both decline.

The question the Australian experiment may finally answer is whether declining engagement is a problem or a feature. Algorithms drive higher engagement in part by surfacing content that provokes strong reactions, which an expansive body of academic research links to more divisive and negative online behaviour. If chronological feeds reduce time on platform while also reducing exposure to rage-inducing content, the trade-off may look different from a social good perspective than it does from a platform revenue perspective.

The complication is that user behaviour has changed significantly since 2014. People are more active in curating and unfollowing than they once were, and the platforms themselves have moved toward interest-based algorithmic recommendations rather than purely network-based sorting. Removing algorithmic sorting would affect not just the content mix but the entire discovery mechanism that now drives much of what people see from outside their immediate network.

JOIN OUR AI SEARCH SERIES