My Feed, My Way: who designs the default

The phone buzzes and the feed picks tonight’s argument for you. Not your mates. Not the news editor. The ranking engine. That is the street signal.

On 8 September 2026 the Albanese Government put Digital Duty of Care exposure drafts out for consultation, sold hard under the banner My Feed, My Way. Platforms would have to tell users they have a choice, then make that choice real: keep the personalised recommender, or run a feed built from accounts you actually follow. Breach the duty and the draft points toward penalties in the ballpark of $109 million, with eSafety as the enforcement shop. The same package reaches into games, messaging, and AI chatbots aimed at under-18s. For under-16s it also tells platforms to kill personalised ranking and endless-scroll design features.

Hold the under-16 social media minimum-age law in the other hand. Do not zip the two into one statute. One regime targets the account. This draft targets design duties and recommender choice. Same country. Same summer of “online safety.” Different machines. Keep them apart.

Five keys, as the system actually presents them.

One. Follow-only or chronological choice for users sixteen and over. Opt out of “recommended for you” and you see what you chose. That is not a skin. It hits the business model that sells prediction.

Two. Duty of Care as risk-test-before-harm, not only complaint-after-harm. Document foreseeable risks. Maintain mitigations. Nine-figure civil exposure if you treat the paperwork like cosplay.

Three. The under-16 feature stack inside this package: personalised feeds and addictive design off for kids. Paper strength is high. Street strength still depends on age assurance and whether platforms get paid to obey.

Four. Parliamentary gearing. Coalition voices in the run-up, including Angus Taylor and Sarah Henderson, heard censorship risk in ministerial power to widen what counts as harm. The Greens, via Sarah Hanson-Young and Fix Our Feeds, want genuine opt-in: algorithm off unless you actively switch it on. Labor sells choice. One ear hears a speech leash. The other hears a dark-pattern loophole.

Five. UX theatre. Europe already forced similar toggles. The known failure mode is burial, degraded opt-out product, and a slow train back to default. A pop-up is not sovereignty. A sticky, visible, costly-to-reverse default is power.

Mechanism.

This is credential-class governance of attention. Ministers, eSafety, advocates, and platform counsel negotiate what a nation is allowed to notice at scale. Parents and kids supply the moral warrant. Adults get a toggle. The durable power sits in three places: who defines harm, who audits the risk register, and who can lengthen the list without recalling parliament every time a new model ships. Anika Wells has said the draft is about systems and features that cause harm, not a government misinformation police. Take her at her word on day one. Still watch the discretionary list on day one hundred. That is where a safety duty becomes a permanent ministry of mood.

Australian Interest test is blunt. Does this make Australia more able to govern its information environment, or more dependent on offshore platforms to perform safety while the ad engine stays intact? A follow-only default would be the harder sovereign move. An opt-out notice is the softer one. Soft can still matter if eSafety enforces and the choice is not a buried ritual. Soft fails if Australians get the language of control while the feed still owns the evening.

Landing.

Australia is not choosing free speech versus safety as abstractions. It is choosing who designs the default. Defaults are power. Everything after that is the press conference.

Rico Holt  ·  ricoholt.com