You only meant to scroll through TikTok for five minutes, but before you knew it, an hour had passed. This is not simply a matter of weak self-control. The algorithm is carefully designed to keep you hooked. Why are these platforms so good at stealing our time?
This addictive design is quietly reshaping our attention, habits and even our thinking. But oddly, policymakers have put a lot of energy into something else: requiring platforms to stop users under 16 from having accounts. They seem to believe that if they just tighten the door a little, the problem will go away.
One is desperately trying to keep you inside. The other is desperately trying to keep you out. The former, an enticing system that influences hundreds of millions every day, remains almost unregulated. The latter, though controversial, has barely any real effect. In other words, we are using a microscope to inspect a flimsy door, while turning a blind eye to the machine humming away inside the room.
Maybe we have been asking the wrong question from the start. We constantly argue about how to lock the door tighter, but rarely ask: who designed that room that makes it so hard to leave?
This design is not accidental. Platforms create a continuous mechanism that guides user behavior. In other words, user choices do not happen in a neutral space. They happen inside a carefully engineered environment.
What Age Restrictions Claim and Where They Fail
Australia’s social media age restriction policy requires platforms to take reasonable measures to stop users under 16 from having accounts. The government says this is to protect minors from harm caused by design features like algorithmic recommendations and infinite scrolling. On the surface, this sounds reasonable.
But the policy’s effectiveness is far less impressive than advertised. According to the eSafety Commissioner (2026), over 4.7 million suspected underage accounts have been deleted, and about 70% of users under 16 still access social media without significant barriers. Their methods are simple:
- faking their age
- borrowing a parent’s account
- using a VPN
One 15-year-old user put it bluntly: “This ban has had almost no impact on my daily use” (Taylor, 2026).
In other words, platforms are indeed enforcing the policy: deleting accounts, showing pop-ups, and asking for age. But for those determined to stay on social media, these measures create almost no real obstacle. A system that relies on users’ honesty or strict parental supervision reveals its limits from the start. Rather than effectively restricting minors, it creates an illusion of action, a pretence that we are doing something. Policymakers can hold press conferences citing the numbers, and parents may believe the problem is solved, but in reality, children are still scrolling through the same content every day.
More importantly, this design quietly shifts responsibility. Platforms can claim they have provided safety tools. Parents can believe they have done their part. As Nicolas Suzor (2019) observes, digital platforms operate through private rules that users cannot negotiate or challenge. Once the restrictions are bypassed, the blame falls on the people least able to control themselves. It is a carefully engineered chain: when the child gets around them, it becomes a case of the child being disobedient or the parent not supervising properly. The platform’s design logic remains untouched. Under this structure, the mechanisms that truly drive users to stay, the very designs that make it hard to stop, are barely touched.
And here lies the deeper question: if bypassing these restrictions is so easy, what is the point of this door? If a door cannot even stop the people it claims to block, is it solving a problem or just producing good statistics?
The answer is not at the door. It is in the system behind it.
From Data Collection to Behaviour Shaping
The answer lies precisely within the platforms themselves.
Apps owned by Meta and Google use complex AI systems to continuously analyse vast amounts of user data. This data includes not only basic information but also records of every click, how long you linger, and even the fraction of a second you pause on an image. These seemingly tiny traces come together to form an extremely accurate digital portrait of you. They are used to build your user profile, enabling the system to constantly predict your preferences and optimise your engagement. In other words, the platform knows better than you do what you want to see next.
As algorithms keep running, the information you receive becomes more and more homogeneous, ultimately forming an information bubble. What you see is not the whole world, but a filtered, ranked and tailored version of it. You may think you are exploring freely, but in reality you are just gliding along a track laid out by the algorithm.
In other words, these systems are not only predicting your choices; they are also gradually narrowing the range of choices available to you. They compress infinite possibilities into a narrow band that best suits you, and then keep reinforcing that band.
Even on the same platform, different users can see completely different information. A 2025 study found that TikTok’s search algorithm directs users to entirely different information environments based on subtle differences in search terms (Matlach et al., 2025). This shows that algorithms are not neutral conduits. They are filters that actively shape the structure of information. As Just and Latzer (2016) argue, algorithmic selection shapes not only what users see but also how attention is distributed and habits are formed. And that stance is to keep you engaged for longer.
More importantly, this process is almost invisible to users. Most platforms do not disclose how their algorithms work, nor do they explain why certain content is prioritised. This opacity makes it hard for users to realise that their information environment is being shaped, let alone to reflect on it or resist it.
As Kate Crawford (2021) argues, AI systems are never neutral tools. They are embedded within specific economic logics, typically aimed at maximising user engagement and advertising revenue. Under such a structure, user behaviour is not only recorded but also systematically guided. You think you are swiping the screen, but in reality, the screen is swiping you.
Ironically, while we focus our regulatory efforts on the entry points of mainstream platforms like TikTok and Instagram, the systems that truly and continuously shape the attention and behaviour of hundreds of millions of people continue to operate under relatively limited constraints. We constantly try to reinforce a gate that is easily bypassed, while overlooking the engine that reshapes our perceptions every day. The problem may not be whether the gate is strong enough, but whether we have been trying to regulate the wrong target all along.
The Narrowing of Public Debate
Public debate on digital regulation tends to focus heavily on technical issues. Take age verification. Discussions usually revolve around which technology to use, how accurate it is, and whether it can truly prevent minors from accessing content.
These questions are not unimportant, but they set an overly narrow frame. When attention is locked on how to lock the door tighter, a more fundamental question gets left out. Why are platforms designed to be so addictive? Why do algorithms only care about how long you stay?
We endlessly argue about whether an age gate is strict enough yet rarely ask why the addictive machines that keep users hooked for hours are left almost unregulated.
As Terry Flew (2021) has pointed out, existing regulatory frameworks often struggle to keep pace with the scale and complexity of digital platforms. But the problem is not just about keeping up. It is about how this power imbalance shapes regulatory choices. Policymakers tend to focus on visible, measurable targets that are easier to intervene on, like age and account numbers. They sidestep more structural and difficult issues, such as how algorithms allocate attention and shape behaviour. Under this logic, what gets regulated is not power itself, only its most visible surface.
In this situation, public debate quietly narrows. People gradually come to accept that digital governance should focus on technical details: whether verification is accurate enough, whether the threshold is high enough. Meanwhile, the question of whether platforms should be allowed to design addictive machines fades into the background, becoming almost invisible.
This is not necessarily a deliberate distraction. It looks more like an institutional choice. Governance systems tend to tackle problems that can be solved. If an issue is too complex or too thorny, it gets shelved, and attention shifts to areas where results come easier. But the outcome is clear. We pour massive energy into reinforcing a door that was never secure to begin with, while never seriously asking whether the room that keeps us hooked needs to be redesigned.
The Dual Failures of Automated Governance
This misalignment in governance logic ultimately leads to a dual failure.
The first failure is targeting the wrong objective. Policies pour vast resources into fortifying a gate that can be easily circumvented, while neglecting the carefully designed system of enticements. No matter how sturdy the gate is built, as long as that enticing mechanism keeps running, people get pulled back time and again, like being caught in a whirlpool.
The second failure is a systemic tendency to avoid hard problems. Regulatory systems naturally gravitate towards issues that are easy to measure, easy to explain, and easy to use as proof that action is being taken, such as age verification and account deletion. Meanwhile, the truly thorny issues, like how algorithms allocate attention and shape behaviour, have long been excluded from the governance agenda. As Andrejevic (2019) notes, automated systems obscure responsibility while shaping user behaviour.
This is not merely a matter of technical implementation. It is a deeper bias in governance logic. We keep trying to use technical fixes to address problems that arise from the very design of the system, yet we remain unwilling to confront the more fundamental question: how platforms are designed to make it difficult for people to leave.
Looking in the Right Direction
Let’s go back to the question we started with. We have been arguing about how to lock the door tighter, yet we rarely ask whether the room itself was designed to make it impossible to leave. Age verification gets patched up. Millions of accounts get deleted. But the lure mechanisms keep running. The vortex never stops. As long as the platform’s business logic remains focused on maximising your time on screen, even the highest barriers are nothing more than a symbolic defence.
This is not simply a technical problem; it reflects a deeper misalignment in governance logic. What we need is not just more precise age verification or stricter account deletion. It is a shift in direction. From regulating the entry point to scrutinising the design logic of the system itself. In other words, stop staring at the door. Step inside and see how the system actually works. We need to start asking: should platforms be allowed to optimise their systems with the goal of keeping users hooked? When attention is treated as a resource to be allocated, who is responsible for that allocation?
At its core, this is a question of power: who controls attention, and to what end?
Pursuing these questions does not mean regulation will become easier. On the contrary, governing algorithmic design is far harder than setting up an age gate. It requires greater transparency, stricter accountability, and a rethinking of platform business models. For example, platforms could be required to disclose whether maximising dwell time is a core metric, and independent audits could assess the impact on user attention.
The next time you find yourself scrolling endlessly, pause and ask. Is this truly your choice, or the system’s default setting? And as members of the public, perhaps we should stop repeatedly checking whether the door is high enough. Instead, let’s seriously consider whether the room that keeps us hooked needs to be redesigned from the ground up.
References
Andrejevic, M. (2019). Automated Media. Routledge. https://www.taylorfrancis.com/books/9780429242595
Crawford, K. (2021). The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (1st ed.). Yale University Press. https://doi.org/10.2307/j.ctv1ghv45t
eSafety Commissioner. (2026, March 31). Social media minimum age: Compliance update. https://www.esafety.gov.au/about-us/industry-regulation/social-media-age-restrictions#compliance-update-march-2026
Flew, T. (2021). Regulating platforms. Polity Press.
Just, N., & Latzer, M. (2017). Governance by algorithms: reality construction by algorithmic selection on the Internet. Media, Culture & Society, 39(2), 238-258.
Matlach, P., Castillo, A., Drath, C., & Hevesi, E. F. (2025). Recommending hate: How TikTok‘s search engine algorithms reproduce societal bias. Institute for Strategic Dialogue. https://www.isdglobal.org/publication/recommending-hate-how-tiktoks-search-engine-algorithms-reproduce-societal-bias/
Suzor, N. P. (2019). Lawless : the secret rules that govern our digital lives. Cambridge University Press.
Taylor, J. (2026, April 11). Fifteen-year-old Noah hasn’t been kicked off any social media platforms – he’s still fighting Australia’s under-16 ban in court. The Guardian; The Guardian. https://www.theguardian.com/australia-news/2026/apr/11/australia-social-media-ban-under-16-teenager-experience
Be the first to comment