Algorithm opt-in is really a platform opt-out
Australia's Greens want to make social media algorithms opt-in. But what happens when the algorithm is the whole point of the app?
Algorithm opt-in is really a platform opt-out
Social media platforms have spent the past decade perfecting a machine that knows what will keep you looking at the next piece of content, then the next, then the one after that. The Australian Greens now propose to make that machine optional. Their Fix Our Feeds bill would flip the default so that users must actively choose algorithmic curation rather than having it thrust upon them. On paper this sounds like handing control back to ordinary people. In practice it asks them to opt out of the very product they keep coming back for.
The bill diagnoses a real harm but misidentifies its source
The bill arrives wrapped in urgent language about misogynistic content, men's violence against women and girls, and the real-world harm that flows from endless exposure to rage and toxicity. Senator Hanson-Young is right that these problems exist and that parents, teachers and young people have described them plainly. The platforms have built billion-dollar businesses on engagement, and engagement rewards outrage. Every extra minute a user spends scrolling delivers another impression to an advertiser. The system is working exactly as engineered.
Yet the engineering runs deeper than any single piece of content. The algorithm does not merely select videos or posts. It constructs the entire environment in which those posts appear, learning from thousands of micro-behaviours, what makes a particular user pause, share, comment or keep scrolling at 1am. Remove that loop and you do not get a gentler version of the same app. You get something closer to a list of newspaper headlines delivered in the order they were published, stripped of the variable rewards that make the experience compulsive. Users do not open TikTok to read the equivalent of the front page in chronological order. They open it to be surprised, seduced, amused or productively distracted from whatever else is happening in their lives.
Opt-in does not neutralise demand — it just forces users to declare it
This is why the opt-in framing misdiagnoses the product. People are not reluctantly submitting to algorithmic feeds. They are choosing them, often knowing the cost. The hand that reaches for the phone in the dark is not confused about what it will find. It is seeking the hit. That desire for escape, for the next dopamine micro-dose, is the demand the platforms have scaled to meet. Telling the platforms to hide the engine behind an active consent screen does not eliminate the demand. It merely forces the user to declare it.
The hand that reaches for the phone in the dark is not confused about what it will find. It is seeking the hit.
The behavioural response is predictable. Some fraction of users, perhaps those most concerned about time wasted or content encountered, will stay with the chronological default and find the experience flatter. Discovery becomes harder. The sense of an infinite, tailored world disappears. Many others will opt in quickly, especially younger users for whom the non-algorithmic version feels like a half-built app. The Greens' emphasis on avoiding dark patterns is understandable, yet platforms have every commercial reason to make the opt-in path obvious and the default path dull. Their revenue depends on it.
Disrupting the calibration may just move the attention elsewhere
This is not a new tension. Similar logic has appeared in other regulated markets where the government tries to alter a core feature while expecting the rest of the service to remain unchanged. When supply is shaped by a particular incentive structure, blunt changes to that structure produce workarounds or reduced participation. Here the supply of user attention is calibrated to the algorithm. Disrupt that calibration and the attention may simply migrate elsewhere, to less regulated corners of the internet or to new apps that figure out how to deliver the same loop without falling foul of the letter of the law.
The current defaults exploit the gap between what users say they want (less time online, less toxicity) and what they actually do at 11pm on a weekday. A genuine digital duty of care would need to grapple with age verification, enforcement against illegal content, and the commercial incentives that reward scale above all else. Those are harder problems. They require governments to accept trade-offs rather than pretend that flipping a default restores agency without changing the product.
The Fix Our Feeds bill reflects a common instinct: if the machine is harmful, make the machine optional. But the machine is not bolted onto the platform. It is the platform. Australians who use these services for connection, humour, news or simple mental escape have built habits around its current form. For many, the chronological alternative will feel like choosing to drive with the handbrake on. They will either accept the brake or find another car.
In the end the reform risks delivering a Pyrrhic victory. Those most determined to escape the scroll may benefit. The platforms will adapt, users who crave the old experience will opt in, and the underlying market for attention, outrage and addiction will persist in slightly altered form. The algorithm will still be there for anyone who wants it, because that is what the platform was built to be. Changing the opt-in box does not rewrite the code that made the box necessary in the first place.
Sources
Productivity Commission — Inquiry into Social Media and Online Safety
Frequently Asked Questions
What would the Fix Our Feeds bill actually change about how social media works?
The bill would require platforms like Meta, TikTok and Google to serve users a chronological, non-personalised feed by default. Users who want the algorithm-driven experience they currently get would have to actively opt in. The change targets the recommendation engine, not the content itself.
Why won't making algorithms opt-in fix the problem of harmful content?
Because the algorithm is not an optional add-on — it is the core product users return to. Most users, particularly younger ones, are likely to opt in quickly, and platforms have strong commercial incentives to make the opt-in path easy and the default experience deliberately flat. The underlying market for attention and engagement remains unchanged for everyone who opts in.
Could users just move to other platforms to get around this rule?
Yes, and that is one of the central risks. If the regulated platforms degrade the default experience, attention may migrate to less regulated apps or new platforms that replicate the same engagement loops without falling under the bill's scope. Regulatory arbitrage is a known failure mode when rules target a feature rather than the commercial incentive behind it.
What would a more effective alternative to opt-in regulation look like?
More structurally targeted options include mandatory algorithmic audits, enforceable harm thresholds on how content is amplified, and reforms to the commercial incentives that reward engagement above all else. The EU's Digital Services Act requires large platforms to conduct risk assessments on their recommendation systems — a model that addresses the incentive, not just the interface.
Does this bill do anything useful at all?
It may benefit a small subset of users — those genuinely motivated to reduce their screen time or limit exposure to toxic content — who would stick with the chronological default. The concern is that this group is likely small, the platforms will adapt, and the reform will be claimed as a win without materially changing the experience for the majority of Australians.