On: Australia is eyeing a world-first ‘fair and reasonable’ test for data...

1 September 2026

Australia proposes to ask, before collecting a person’s information, whether the collection is fair and reasonable. The phrase sounds modest. It is not. It smuggles in an entire moral theory, and the drafters appear not to have noticed which one.

“Fair and reasonable” by whose standard? Two senses hide in the phrase. Reasonable to the company, whose interest in the data is genuine and measurable; or reasonable to the person, who would not sell that information at any price he understood. The test is being written as if these coincided. They coincide only when the person is fully informed and genuinely free to refuse - conditions that the consent forms, forty pages of dense prose, were invented precisely to defeat.

The right to be forgotten raises a harder question, and I confess I do not know the answer. A debt is forgotten when paid; a fact about me is not a debt. If I behaved badly at forty, the record at fifty is true, and truth has claims. Yet the claim of truth is not unlimited - no one is obliged to publish every true thing. The question is where the burden lies, and a statute that does not say who must prove what will be settled by whoever can afford the lawyers.

Still, I find myself more hopeful than usual about legislation, which is a rare condition for me and should be treated with suspicion. The alternative regime - consent by checkbox, enforced by regulators who never read the checkbox - has been tested for two decades and has failed its own null hypothesis. Any new test, even a vague one, outperforms a fiction.

My reservation is proportional, I hope: a vague standard invites arbitrary enforcement. But vagueness corrected by adversarial courts is preferable to precision in the service of extraction. I would rather have an unclear rule honestly applied than a clear one applied to nobody’s benefit but the collectors'.