Australia Mandates Social Media User Opt-Outs
One notes, in the announcement that Australia will require social media platforms to permit users to opt out of algorithm-driven feeds, an absence more interesting than any presence. There is no date. There is no mechanism by which compliance will be measured. There is no named minister standing beside the sentence, taking ownership of it the way officials usually take ownership of things they intend to enforce. The item arrives the way certain lights arrive over marshland - reported, described, agreed upon by several witnesses, and then gone before anyone can determine its altitude.
I have catalogued enough regulatory announcements to notice the species this one belongs to. It is the announcement that names the actors - here, the Australian government on one side, the social media platforms on the other - with great precision, and then declines to name the calendar. When was the WHEN, the record says only that it is not stated in the article. This is not an oversight. An oversight is when a fact is missing because no one thought to ask. This is a fact missing because someone thought to ask and the answer was not yet fit for public consumption, or did not yet exist.
Consider the company this omission keeps. The stated stakes are considerable - protection of children, reduction of exposure to divisive material, a claim on the wellbeing of every social media user in the country, which is to say nearly everyone in it. One does not usually announce a policy of such stakes without a date attached, unless the announcing is doing work independent of the policy. A naturalist watching this behaviour in isolation might mistake it for legislation. A naturalist watching several dozen instances of it, across several dozen countries, over several decades, begins to suspect a different genus entirely: the announcement whose function is not to change the platforms’ behaviour on some future Tuesday, but to change the public’s mood on the Tuesday it is issued. The regulation is postdated; the reassurance is not.
The CONTESTED field admits, plainly, that the effectiveness and consequences of the legislation are not yet known. This is the most honest sentence in the file, and it is placed, notably, last - after the WHO, the WHERE, and the STAKES have already done their work on the reader. One might call this the standard sequence: name the villain, name the victim, name the harm, and only then, quietly, concede that nobody has established whether the proposed remedy does anything at all. I have seen this sequence in patent-medicine advertising, in wartime rationing bulletins, and in at least three prior decades of technology regulation. It is not evidence of bad faith. It is evidence of a recurring shape, and recurring shapes are what I collect.
There is also the matter of who benefits from an opt-out that nobody is required to exercise. A platform can comply with a law requiring it to offer a chronological feed option while quietly discovering, through the same behavioural research that built the algorithmic feed in the first place, that a comfortable majority of users never touch the setting. The government gets its headline. The platform gets its compliance certificate. The algorithm continues its work undisturbed, like a tide that has been formally notified of a new sign on the beach.
I picture a specific moment that the announcement does not contain: a user in Perth, thumbing through a settings menu three submenus deep, hunting for a toggle that the press release implied would be prominent, giving up before finding it, and returning to the feed that was, in fact, never really optional to begin with. This is the scene the WHEN and the WHO cannot supply, because it has not happened yet, and may never be recorded by anyone keeping that kind of ledger.
What the record contains, in the end, is intention without a clock. Whether that is because the clock has not been built, or because the clock was never meant to be checked, is not a question this dossier answers. It is merely a question the dossier makes unusually easy to ask.