The Australian government is considering forcing social media platforms to enable an algorithm opt-out, in order to address polarization fueled by content matching in social apps. The move comes after the country spent months working on its teenage social media ban, which has thus far had little impact on the negative behaviors connected with social media use.
As reported by The Guardian, Australian social services minister Tanya Plibersek said that the government is exploring the potential algorithm opt-out following a recent teenage strangling death and a rise in reports of sexual violence among teenagers.
As per Plibersek, in The Guardian: “If you set up a social media account as a young boy in Australia today, it takes an average of 23 minutes for you to be fed misogynistic, violent misogynistic content online. That is a deliberate decision of the social media companies that is wrecking our young boys, giving them all of the wrong ideas about what healthy relationships look like.”
Plibersek said on the ABC’s Insiders program that the government is exploring various options to address concerns with algorithmic amplification, including both opt-in and opt-out approaches that would enable more control over what people are shown in the app.
Addressing algorithmic amplification is an important consideration, and based on a range studies, it could have a much bigger impact than banning teen users from social apps.
The negative impacts of algorithms has been covered extensively in academic research, including:
- In 2016, a study published in Harvard Business Review found that “high arousal emotions,” such as anger and fear, generally drove the biggest social media response, particularly in terms of viral sharing.
- A study published by the National Library of Medicine in 2023 found that the presence of negative words in news headlines increased consumption rates among social media users.
- Another report, which studied Instagram engagement, and was published by Science Direct in 2021, found that posts that elicited negative emotions received up to 70% more likes than positive emotions, and accounts that drove more engagement had “a significantly greater number of likes and video views.”
- An analysis conducted by the Korean Advanced Institute of Science and Technology, which was published by Taylor and Francis Online in 2023, found that posts expressing anger reached more people than those expressing anxiety, while social media users who were angry indicated a higher intention to share emotional posts.
- Back in 2012, a study published by Wharton Business School found that content that evokes anger is likely to be shared more, with the amount of anger inspired by a post proportionately driving the virality of that comment.
Angst and fear drive algorithmic amplification because these systems are based on binary calculations of post engagement. More likes and comments inform algorithmic decisions, regardless of the actual content itself, which then leads to broader exposure of these posts.
Given the evidence, it’s clear that algorithms play a major role in driving emotional triggers, which, in turn, can lead to more division and angst. And given that the majority of people also rely on social media for news and information, according to Pew Research, that creates a dangerous mix.
The challenge, then, would be in enacting restrictions, and forcing platforms to enable algorithm options.
Meta, for example, has repeatedly touted the value of algorithmic matching when it comes to driving more user engagement, which directly ties back to its business model of showing people ads.
As such, any restriction on algorithms would impact its business process, which could lead to bigger clashes around the restriction of trade.
But given that Australia is a relatively small market, that could make it an ideal testing ground for algorithm opt-outs, in which Meta and other regulators could measure the impacts of switching people to a purely chronological feed of posts from pages that people have chosen to follow.
Meta has already experimented with this in the EU, where users are now able to pay a monthly fee to opt out of algorithmic matching.
Though, of course, few people have opted to pay, making this a minor test case. The real test would be if the Australian government can make algorithmic amplification opt-in, ensuring that all users are cut off from the never-ending feed of recommended posts.
A full-scale test of that approach would showcase its true potential, and it’ll be interesting to see if Australia’s regulators look to take that path.
