Australia wants to give you back control of your internet

Proposed laws would allow Australians to turn off algorithmically selected social-media feeds—but the attempt to restrain Big Tech also raises an important question about government power

Australians could soon be given something that social-media companies have been reluctant to provide: a genuine choice about who decides what appears on their screens.

Under the Federal Government’s proposed “My Feed, My Way” reforms, digital platforms would be required to let users aged over 16 choose between an algorithmically personalised feed and one based on the friends, organisations and content creators they have deliberately chosen to follow.

Platforms that fail to comply could face penalties exceeding $100 million.

Prime Minister Anthony Albanese says the proposal is intended to transfer power from global technology companies to ordinary Australians.

That description captures the significance of the idea—but not all its complications.

An algorithm does more than recommend the next entertaining video. It observes behaviour, identifies patterns and predicts what will keep a person watching, clicking, reacting and returning.

It does not know the user in any human sense. It characterises the user through data and then constructs a stream of content around that characterisation.

Gradually, the internet can begin to narrow itself around the individual.

The user may believe he or she is exploring the world. Increasingly, however, the world being presented has already been selected.

What would change?

The proposal would require social-media platforms to give Australians an enduring choice over their principal content feeds.

A user could continue receiving recommendations selected by a platform’s algorithm or choose a feed centred upon accounts the user has intentionally followed.

That distinction matters.

Presently, following a person, publisher or organisation does not necessarily mean its material will be prominently displayed. The platform determines what receives priority, what disappears and what unrelated content should be inserted between chosen sources.

The commercial purpose is straightforward. Platforms benefit when users remain engaged for longer periods. Content that generates anger, fear, fascination or argument can therefore be particularly valuable, irrespective of whether it leaves the audience better informed.

Under the proposed system, Australians would theoretically be able to step away from that continuous experiment in behavioural prediction.

The reforms are part of wider proposed digital-duty-of-care legislation intended to make online services responsible for identifying and reducing foreseeable harms.

The measures would extend beyond conventional social media to parts of the digital economy including applications, online games and artificial-intelligence services.

Protections proposed for children include action against pornography, bullying, material encouraging eating disorders, content glorifying crime or dangerous behaviour and material capable of causing serious psychological harm.

The reforms would also address design features intended to prolong engagement, including endless scrolling and some forms of personalised content delivery.

The legislation is expected to be introduced to Parliament later this year and remains a proposal rather than settled law. Its eventual form could change substantially during parliamentary consideration. Reuters, Associated Press

The algorithm is not neutral

Social-media algorithms are sometimes described as though they were merely efficient filing systems.

They are not.

They are commercial decision-making systems operating at enormous scale. They determine which voices are amplified, which subjects become unavoidable and which emotions are most likely to be repeatedly stimulated.

Two Australians can enter the same social-media platform and encounter almost entirely different versions of their country.

One may see economic opportunity, community activity and practical information. Another may be presented with an unending procession of political outrage, social breakdown and personal grievance.

Neither feed necessarily represents society as a whole.

Each represents the platform’s calculation of what is most likely to retain that particular person’s attention.

That creates a circular process. A person pauses to examine a subject, the system interprets the pause as interest and then supplies more of the same. Repetition reinforces attention, while attention produces further repetition.

Curiosity can be mistaken for conviction.

Before long, a passing interest can become the organising theme of an individual’s online world.

The great power of the algorithm is not that it commands people to believe something. It makes particular ideas appear prevalent, urgent and inescapable.

A matter of personal agency

The strongest argument for the Government’s proposal is therefore not that algorithms are inherently evil. Recommendation systems can be helpful, entertaining and remarkably effective at finding material a user would otherwise never discover.

The problem is the absence of informed and continuing consent.

Most users do not know why a particular post has appeared, which observations contributed to its selection or what alternative material has been excluded.

They are nominally operating the application, but the application is also operating upon them.

Providing a meaningful choice would recognise that Australians should be able to decide whether they want a platform to predict their interests or simply show them the sources they selected.

It would not abolish algorithmic feeds. People who enjoy them could continue using them.

The essential change would be that personalisation becomes a visible choice rather than an invisible condition of participation.

Protecting users or controlling speech?

There is nevertheless a second issue that should not be dismissed.

The proposed legislation combines individual control over algorithms with wider powers relating to harmful content.

Those are connected matters, but they are not identical.

Giving a person the ability to switch off recommendations expands individual control. Giving government authorities greater power to define or remove material can transfer control in a different direction.

The Government says the proposal is not intended to regulate political opinion or ordinary public debate. Its stated targets include illegal material, sexual exploitation, violence, terrorism, serious harassment and other readily identifiable harms.

Few Australians would object to effective measures against such content.

The difficulty emerges at the boundary.

Who defines harm when the material is legal but confronting? How much discretion should rest with a minister or regulator? What parliamentary safeguards will apply when categories are expanded? Will platforms remove legitimate material excessively because deletion is safer than risking a substantial penalty?

These are not arguments for abandoning the legislation. They are reasons for Parliament to examine its language and accountability mechanisms carefully.

Australia should be capable of protecting children and confronting demonstrably harmful conduct without creating a loosely defined system through which lawful disagreement can be suppressed.

The final legislation must preserve that distinction.

Can the platforms work around it?

Implementation will be another test.

Several major platforms already provide some version of a “following” or chronological feed, but it may be placed behind secondary menus while the algorithmically curated feed remains dominant.

A nominal choice is not necessarily a meaningful choice.

Platforms could comply technically while making the alternative difficult to find, less functional or repeatedly subjecting users to prompts encouraging them to return to personalisation.

The legislation will therefore need to address interface design as well as the mere existence of an off switch.

A person’s selection should remain in place until that person changes it. Declining algorithmic recommendations should not disable unrelated functions or result in a deliberately inferior service.

If Australia is serious about returning control to users, compliance must be measured by practical experience rather than the presence of an obscure setting.

Australia is confronting a global system

Australia has increasingly been willing to challenge the operating assumptions of global digital companies.

Its restrictions on social-media accounts for children under 16 attracted international attention. The proposed algorithm reforms would extend that intervention from access to the architecture of the platforms themselves.

The consequences could reach beyond Australia.

Digital companies prefer uniform global systems. If a major market requires an effective non-personalised option, the necessary technology may eventually become available elsewhere.

Australia would not be abolishing the algorithm. It would be establishing the principle that the algorithm serves the user—and that the user does not exist merely to serve the algorithm.

The Times View

For much of the internet’s history, the argument has been framed as a contest between freedom and censorship.

The rise of algorithmic feeds reveals another form of power.

Freedom of expression means little if an unseen commercial system determines which expression will actually be encountered.

Australians should be permitted to choose an algorithmic feed if they find it useful. They should also be able to reject it without abandoning the modern public square.

The proposed laws deserve support for attempting to restore that choice. Their content-control provisions, however, must be precisely written, transparently administered and subject to proper parliamentary oversight.

The question is not whether somebody will influence what Australians see online. That already happens every minute of every day.

The question is whether that power belongs to the technology company, the government—or the individual.