Australia Mandates Social Media User Opt-Outs

There is a fence across the feed. The modern man says, “I see no reason for this algorithm; let us remove it and give people back their own eyes.” The wiser man says, “Before you remove it, tell me why it was put there. If you cannot tell me, I am not sure you have understood what you are removing.” The Australian government proposes to let users opt out of algorithm-driven content feeds, and every decent instinct in me wants to cheer, and every instinct trained by thirty years of watching reformers wants to ask what the algorithm was actually doing before we knock it down.

The obvious answer is that the algorithm was doing something sinister: capturing attention, radicalising the lonely, feeding children images no child should see. This is true, and it is the reason given in the stakes of this very proposal - online safety, protection of the young, reduction of exposure to dangerous and divisive material. I do not dispute a word of it. But notice the peculiar shape of the argument. The algorithm is blamed for showing people what they did not choose. The remedy is to let people choose. And the assumption buried in the remedy is that what people would choose, left to themselves, is safer than what the algorithm chose for them. This is precisely the point on which I would like the reformer to pause, because it is not obviously true, and the whole history of human vice suggests it is often false. Men did not need an algorithm to invent the gladiatorial arena. The feed did not invent the mob; it only found the mob’s address more efficiently than the town crier ever managed.

Here is the fence properly examined. The algorithm was built, whatever its sins, to do one thing that a purely chronological or user-chosen feed does not do: it stands between the user and the worst of himself. A man who follows nothing but his own untutored appetite on the internet does not drift toward moderation; he drifts, with the same gravity that pulls water downhill, toward whatever is most inflaming, most confirming, most comfortable in its rage. The algorithm, for all its manipulations, was at least in principle answerable to something - advertisers who dislike controversy near their products, regulators who dislike lawsuits, engineers who dislike being hauled before parliament. Remove it, hand the wheel to the raw appetite of the user, and you have not removed the danger. You have merely removed the one part of the system that had any obligation to notice it.

The Australian government, to its credit, seems to sense this, which is why the legislation is aimed at platforms and not at algorithms as such - it does not abolish the feed, it merely insists the citizen may step outside it if he wishes. That is a wiser fence than the one I feared they were building. It does not say “tear down the gate”; it says “leave a gate in the fence, so that whoever wishes to walk around may do so without being told the fence itself does not exist.” This is the distributist in me speaking rather than the reactionary: the danger was never the algorithm, it was the absence of a choice, and a law that restores the choice without pretending the underlying appetites have changed is a law that has, for once, understood the ground it stands on.

But watch what happens the day the opt-out is exercised by everyone the safety advocates most wanted to protect. The teenager who is being radicalised by an algorithm will, given the option, generally choose more of exactly what radicalised him, because the algorithm was not lying to him about his own appetite - it was flattering it. The fence being removed protected him from himself as much as from the platform. When is not stated in the article; I do not know if this law arrives next year or the year after. But I know that the day it arrives, some quiet official in Canberra will discover that the hardest part of giving people freedom is that they use it.