You open TikTok for a few minutes. Then Instagram. Maybe YouTube. Before you know it, half an hour—or even an hour—has disappeared.
Sound familiar?
What feels like a simple break from work, boredom, or everyday life is often being shaped by something working quietly behind the screen: the social media algorithm.
Algorithms decide what appears in our feeds, what videos are recommended next, which posts are placed in front of us, and sometimes even what we discover without actively searching for it. They are designed to understand our interests and predict what will keep us watching, clicking, commenting, sharing, or coming back.
Now, Australia is considering new rules that could give people more control over these personalised feeds while placing greater responsibility on technology companies to protect young users from harmful online content.
But this raises a bigger question for everyone who spends time online:
Are social media algorithms simply making our feeds more interesting—or are they making it harder for us to stop scrolling?
What exactly is a social media algorithm?
Whenever you open platforms such as TikTok, Instagram, Facebook, YouTube, or other social networks, you are presented with a carefully selected stream of content.
You might see a video about something you searched for yesterday. A post related to a hobby you frequently engage with. A creator whose content you watched several times. Or even something completely unexpected that the platform believes you are likely to find interesting.
Behind those recommendations are algorithms.
In simple terms, algorithms are sets of computer-powered rules that analyse information and make predictions about what you are likely to engage with.
They can consider things such as:
- What posts you like, comment on, save, or share
- Videos you watch and how long you watch them
- Accounts and topics you follow
- Searches you make
- Content you post or interact with
- Your approximate age or location
- Other information provided through your online activity
The goal is relatively simple: show people content they are likely to respond to.
The complicated part is understanding what “respond to” actually means.
Sometimes it means something you genuinely enjoy.
Sometimes it means something surprising.
And sometimes it means something that makes you angry, worried, shocked, or curious enough to keep watching.
The technology companies behind these platforms generally keep the details of their recommendation systems private. The algorithms are considered a crucial part of their businesses because keeping users engaged can translate into more time spent on the platform and, ultimately, more opportunities to generate revenue.
Why is scrolling so difficult to stop?
Most people have experienced it.
You pick up your phone intending to check one notification.
One video appears.
Then another.
Then another.
You tell yourself, “Just one more.”
Suddenly, 45 minutes have passed.
Researchers and digital wellbeing advocates argue that this behaviour isn’t simply about people lacking self-control. Social platforms have developed features specifically designed to make continued engagement easy.
Endless scrolling removes a natural stopping point. Notifications pull users back into the platform. Recommendations constantly provide something new to watch. And because you don’t always know what will appear next, there is an element of anticipation.
That unpredictability can encourage people to keep checking.
Mark Williams, honorary professor of cognitive neuroscience at Macquarie University, has argued that platforms have developed highly engaging features that can encourage both children and adults to remain online longer than they originally intended.
Other researchers have pointed to the role of the brain’s reward system. When something unexpected or particularly interesting appears, users may experience a sense of anticipation that encourages them to continue looking for the next rewarding piece of content.
The result can become a familiar cycle:
Scroll. Discover. React. Scroll again.
And because there is no obvious end to the feed, there may never be a natural moment that tells you it is time to stop.
When engagement becomes the priority
The biggest concern isn’t necessarily that algorithms recommend content.
Personalised recommendations can be genuinely useful. They can help users discover new creators, learn about topics they care about, find communities, and come across content they would otherwise never have encountered.
The concern arises when keeping someone’s attention becomes more important than the quality or consequences of what they are being shown.
Algorithms learn from engagement.
If people consistently react strongly to controversial, emotional, frightening, or provocative content, that information can influence future recommendations.
This can create what many describe as a “rabbit hole.”
You watch one video.
The platform recommends something slightly more extreme.
You engage with it.
The next recommendation moves further in that direction.
Eventually, someone may find themselves looking at content that is dramatically different from what they originally intended to watch.
Steven Roberts of Monash University has argued that recommendation systems can favour provocative and extreme material because emotional reactions often generate strong engagement.
And engagement is extremely valuable to social media platforms.
The problem with extreme content
One of the more worrying areas is the potential impact of recommendation systems on teenagers and young adults.
Researchers and advocacy organisations have raised concerns about young people being exposed to harmful communities and ideas through algorithm-driven recommendations.
One example is the so-called “manosphere”—a broad collection of online communities and influencers associated with misogyny, hostility towards women, and, in some cases, extreme or violent views.
Research highlighted by Melbourne-based men’s health organisation Movember has suggested that recommendation systems can lead young men from relatively ordinary interests, such as fitness or sport, towards increasingly extreme content.
The concern is not simply that harmful content exists online.
It is that users may be guided towards it without deliberately searching for it.
That distinction matters.
Someone may never type a search for extremist or misogynistic content. They may simply interact with a few videos that appear harmless. The recommendation system then uses those interactions to predict what else might hold their attention.
Over time, the content can become more extreme.
Roberts has also pointed to concerns about the influence of online manosphere personalities in Australian schools. A national survey involving more than 2,600 teachers found that 70% reported frequently or occasionally seeing the influence of manosphere social media personalities on student attitudes and behaviour.
For parents, teachers, and policymakers, this raises an uncomfortable question:
If an algorithm can influence what a young person sees every day, who should be responsible when that influence becomes harmful?
Australia’s proposed response
Australia is moving towards giving users greater control over personalised social media recommendations.
Under the proposed approach, social media companies would be required to provide users with an option to opt out of algorithmically personalised feeds.
The idea has been described as a “my feed, my way” approach.
Instead of constantly receiving content selected according to a user’s predicted interests, people could potentially choose a simpler feed focused primarily on updates from friends, family, and accounts they have deliberately chosen to follow.
The proposed rules also place stronger responsibilities on platforms to protect users under 18 from harmful material, including pornography, online bullying, content that glorifies criminal behaviour, and material that encourages eating disorders.
It represents a broader shift in how governments are thinking about social media.
For years, the conversation largely focused on what individuals should do to protect themselves online.
Now, more attention is being placed on the technology itself—and on whether platforms should be responsible for the systems they build.
But will people actually turn the algorithm off?
This may be the most interesting part of the debate.
Giving people a choice is one thing.
Getting people to use it is another.
A 2025 survey involving more than 6,000 people across six countries found that only around one in five respondents said they would choose to switch off personalised recommendations if given the option.
Why?
Because algorithms are good at what they were designed to do.
They make feeds feel busy, relevant, entertaining, surprising, and highly personalised.
A feed made up primarily of posts from friends, family, and selected creators might be healthier for some people—but it could also feel noticeably quieter.
And that quietness may be exactly what makes it difficult for people to choose.
After years of personalised recommendations, users have become accustomed to opening an app and immediately finding something designed to capture their attention.
The algorithm knows what keeps us watching.
The question is whether we know when to stop.
Are algorithms the problem—or how we use them?
It would be too simple to say that algorithms are entirely good or entirely bad.
They have transformed how people discover information, entertainment, education, businesses, communities, and new ideas.
A well-designed recommendation system can introduce someone to a useful expert, an inspiring creator, an educational video, or a community that genuinely improves their life.
But the same technology can also amplify misinformation, outrage, harmful behaviour, unrealistic expectations, and extreme content.
The difference often comes down to what the system is optimised to achieve.
If the primary objective is to maximise engagement, the most valuable content isn’t necessarily the most accurate, balanced, healthy, or constructive.
It may simply be the content that keeps someone watching.
And that’s where the debate becomes much bigger than social media.
The future may be about having a choice
Perhaps the most important part of Australia’s proposal isn’t whether everyone eventually turns personalised recommendations off.
It is whether people should have the right to choose.
Imagine opening an app and being able to decide:
Do I want a feed designed to predict what will keep me engaged?
Or do I want a feed showing me what I deliberately chose to follow?
For some users, personalised recommendations may remain the preferred option.
For others, switching them off could provide a healthier digital environment with fewer distractions and less exposure to content they never intentionally asked to see.
The important change is giving users greater control over the experience.
Because social media is no longer simply a place where we choose what to look at.
Increasingly, it is a place where technology chooses what to put in front of us.
And perhaps the biggest question isn’t whether algorithms are amusing or addictive.
It’s whether we are still the ones deciding what deserves our attention.
The human side of the story
Behind every scroll is a real person, with limited time, attention, emotions, relationships, responsibilities, and a life happening away from the screen.
Technology can make our lives easier and more connected, but our attention is still ours.
Maybe the healthiest relationship with social media isn’t about deleting every app or rejecting technology altogether.
Maybe it is simply about having enough awareness to pause and ask:
“Did I choose to watch this, or did the feed choose it for me?”
That small question could be more powerful than we realise.
