Would Big Tech Embrace an Analytics Free System?

Australia is pushing social media platforms to offer a genuine off switch for recommendation algorithms. Users would see only posts from accounts they follow, in chronological order. The BBC reported this week that the proposal sits inside a wider Digital Duty of Care bill. Existing controls in other places have often been temporary or awkward to use.

From a pure capitalist view the question is straightforward. A company maximises shareholder value. It possibly will not freely build and promote an analytics free system if that change reduces time spent, data collected, or advertising revenue. Engagement is the core product. Algorithms that surface emotional, or novel content are designed to keep people scrolling. That scrolling generates the behavioural data needed that powers targeted ads. Removing the recommendation layer cuts both session length, and the quality of targeting data. In essence their Profit falls.

Experts on both sides of the argument recognise this incentive. But they disagree on how much it matters, how strong the evidence of harm is, and what governments should do about it.

Why Algorithms Drive Profit and Engagement

Social platforms measure success in minutes watched and actions taken. Recommendation systems increase both. They prioritise content that triggers anger, curiosity, insecurity, or desire. Those reactions keep users on the app for longer, and longer sessions produce more data points. This helps the advertisers who pay more for precise targeting based on that data.

Instagram head Adam Mosseri has been explicit about the commercial and product reality. In an Australian press briefing he said engagement can fall by as much as 50 percent when users switch to a non algorithmic feed. Satisfaction scores drop. Chronological feeds, he argued, become dominated by brands and publishers who post more frequently than ordinary friends or creators. The overall pie of attention shrinks. Creators and friends get less reach. Users become less interested and spend less time in the app.

Mosseri has also rejected the clinical addiction label. He distinguishes everyday language from recognised clinical addiction with withdrawal symptoms. He says the platform is guessing what users will value based on likes, shares, comments, watch time, and profile visits. It is an expensive educated guess, not perfect knowledge of desire. Platforms point to low voluntary take up of existing following only options as evidence that most users prefer the ranked experience.

This is capitalist logic in plain language. The system is optimised for attention and data. Anything that reliably reduces both will come as a cost. So companies may add the required toggles. however they will not design them to be more attractive than the profitable default, unless forced to do so and then monitored.

The Case Against Algorithmic Analytics

Campaigners and many researchers argue that the same engagement optimisation produces measurable harms, especially for young people. Chanel Contos, founder of Teach Us Consent, led the Fix Our Feeds campaign that helped shape Australia’s proposal. She told the BBC that platforms promote content that is outrageous, or that invokes strong emotions such as anger, fear, desire, lust, or insecurity because it keeps people scrolling. She has linked algorithmic amplification to the rapid funnelling of boys and young men into misogynistic content. Her campaign called for algorithms to be off by default, requiring affirmative consent to turn them on. More than 15,000 people signed the petition, including former prime ministers and mental health advocates.

Research supports parts of this view. A CESifo working paper by Simona Mandile used the 2016 introduction of Instagram’s algorithmic feed as a quasi experiment. Comparing Dutch users with Instagram accounts to those on other platforms, she found a negative effect on teenagers’ mental health for both males and females. Mechanisms included increased social isolation and intensified negative social comparison, particularly for females through greater need for validation and lower self worth.

Other work has linked repeated algorithmically structured exposure to body image pressure, disordered eating, and depressive content loops. Young people in qualitative studies describe tools such as marking content not interested as limited. Feeds can drift back to triggering material even after they try to steer them away. Psychologists at the University of Texas at Dallas examined how algorithm recommended content interacts with existing depressive symptoms. Participants with more depressive symptoms received more depressive recommendations. Their brain activity while viewing that content showed patterns associated with depression. Continued engagement can reinforce a negative feedback loop.

University of Bristol researchers have argued that policy has focused too much on access bans or total screen time and too little on the cumulative, repeated exposure that engagement driven systems create. Frequency, combinations of content, and emotional intensity matter more than isolated posts. Jonathan Haidt has popularised a broader version of the argument. In his book The Anxious Generation he links the post 2010 rise in adolescent anxiety and depression to the mass adoption of smartphones and algorithmically curated social media. Mechanisms include sleep disruption, social comparison, exposure to distressing content, and the displacement of in person interaction. Platforms optimised for engagement systematically surface high arousal material. Whistleblowers and internal documents from Meta have shown that the company was aware of negative effects on some teenage users while continuing to prioritise growth metrics.

Critics of the status quo also point to implementation failures. Researcher Chris Zomer has documented how platforms comply with laws such as the EU Digital Services Act in ways that preserve the default algorithmic experience. Non personalised feeds can be clunky. Limited access to messages or settings can occur. The feed can revert to the algorithmic version when the app is closed and reopened. Only after a Dutch court ruling did Meta make a more durable choice available in the Netherlands. His analysis appears in The Conversation article on the Netherlands and Meta. Australia’s draft rules try to close these gaps by requiring that following only feeds remain fully functional and that settings do not reset without an explicit user choice.

The Case for Algorithmic Analytics

Defenders of recommendation systems argue that the harms are overstated, the causal evidence is weaker than headlines suggest, and the practical alternatives create their own problems. Mosseri and other platform voices emphasise discovery and relevance. Without ranking, users miss content from friends they value and creators they chose to follow. Chronological order rewards volume of posting rather than quality or interest. Brands and news publishers can drown out personal accounts. The result, they say, is a worse product that people use less.

Some researchers urge caution about the strength of the causal claims. Reviews of the literature note that associations between social media use and mental health outcomes are often small. Direction of effect is hard to establish. Poor mental health may drive heavier use as much as the reverse. One overview concluded that evidence specifically for algorithms driving or reinforcing unhealthy dynamics is thin and supports at most a small effect on mood. Unfavourable social comparisons may matter more than the ranking system itself. Effects depend on who is using the platform and how. Not every user experiences the same pattern of exposure.

Free speech and constitutional arguments appear on this side as well. Some legal scholars contend that algorithmic ranking involves editorial judgments protected by free speech principles. Eugene Volokh has argued that treating the addictive aspects of social media differently from other engaging speech creates difficult lines. What one person experiences as compulsive scrolling another experiences as a useful, engaging product. Governments that force particular ranking defaults risk over moderation or reduced user agency, critics say. Right wing parties in Australia and voices in the Trump administration have raised censorship concerns about the broader Digital Duty of Care bill.

Platforms also point to existing tools. Users can already mute topics, mark content not interested, switch to following feeds in some apps, and adjust interest categories. Low take up of the more radical options is presented as revealed preference. Most people, when given a choice, stick with the ranked feed. Mosseri has described the minority who want pure chronological timelines as vocal but small.

Middle Grounds and Practical Limits

Not every expert falls cleanly into the view that algorithms are harmful or the view that algorithms are fine. Francis Fukuyama and collaborators have proposed middleware. This is an intermediate layer of competing curation services that sit between users and the dominant platforms. Users would choose their own ranking or filtering rules rather than accepting a single platform algorithm or a pure chronological default. The idea aims to restore user agency and reduce the power of a handful of companies without requiring governments to dictate content. It remains largely theoretical at scale because platforms control the interfaces and have limited incentive to open them.

Polling in Australia shows a generational split. Roughly two thirds of Gen Z respondents said they would keep their algorithms on, while nearly half of the broader public would turn them off. Real world experience with opt outs suggests friction and defaults dominate behaviour. People may value the idea of control more than they value the daily experience of a quieter, finite feed.

For parents the practical takeaway is narrower. Platform toggles, even when well designed, are not a complete solution. Device level controls, family agreements on screen time, and regular conversations still matter. Guidance on digital wellbeing for families and platform specific safety settings can reduce unwanted exposure even when the underlying recommendation system remains engagement optimised. An analytics free system can lower the volume of unsolicited content. It does not remove the need for parental oversight or personal habits. Parents can also review YouTube safety settings, X safety settings for parents, and Google parental controls alongside broader resources for parents.

The Capitalist Constraint Remains

The expert disagreement is real. Evidence of harm is stronger on repeated exposure to high arousal and comparison heavy content than on any single ranking choice. Evidence that pure chronological feeds improve wellbeing at population scale is thinner. Commercial incentives, however, are not in dispute. Companies that maximise attention and data will treat an analytics free system as a cost centre. They will implement what the law requires. They will protect the higher revenue path wherever the rules leave room.

Australia’s approach tries to reduce that room by demanding functional, persistent, obvious choice. Whether it succeeds depends on enforcement and on whether users actually prefer the quieter feed once the friction is removed. History suggests many will not. The capitalist calculation stays consistent. If the change costs users and profit, the company will not embrace it freely. Regulation can force the option. Design and defaults will still favour the engagement engine unless they are actively constrained.

Parents and young people therefore cannot wait for perfect corporate alignment. They can use available controls, set device limits, and treat recommendation systems as tools optimised for someone else’s bottom line. An analytics free system is one tool among others. It is not a substitute for the harder work of attention and family rules.

References

• Social media algorithms: What if you could turn them off? BBC report on Australia’s proposed algorithm switch, Contos campaign, and platform responses.
• Instagram’s head says engagement falls by half without the algorithm Mosseri’s comments on engagement, brands, and user experience.
• The Dark Side of Social Media: Recommender Algorithms and Mental Health CESifo working paper on Instagram’s algorithmic feed and teenage mental health.
• The Netherlands forced Meta to let users ditch algorithmic feeds Chris Zomer on implementation friction under the DSA.

Related Reading

• Digital Wellbeing for Families Screen limits, tech free times, and family agreements.
• YouTube Safety Settings for Parents Recommendation controls and supervised experiences.
• X Safety Settings for Parents Privacy, protected posts, and feed related options.
• Google Parental Controls Family Link and SafeSearch tools outside platform algorithms.
• Resources for Parents Full collection of platform safety guides.

About Children of the Digital Age

Children of the Digital Age provides online safety education and training for families, schools and organisations. Our work brings together practical safety guidance, cybersecurity and cyberpsychology. For education and training enquiries, Contact Us.

Send a message

Contact us to discuss education, training or practical guidance for your family, school or organisation. Use emergency and safeguarding services for urgent concerns.