engineered-ignorance
The Cost of Not Knowing
The thread that runs through Meta's buried child-safety study, Grenfell's cladding certification, and the Treasury's bond-buyback is not a conspiracy. It's cheaper not to know.
Start with Meta. In 2021, a researcher named Arturo Béjar wrote to executives: Instagram was amplifying self-harm and eating-disorder content among teenage girls. The company's response? Keep the study internal, don't publish the data, and instruct teams not to ask certain questions that could produce liability. The legal hedge is the point. If you don't know, you can't be sued for what you didn't act on. The cost of knowing—remediation, public relations, potential regulation—is priced higher than the cost of the harm itself, because the harm lands on users, not on the firm. So the economically rational default is to not ask.
Grenfell is not this. The cladding certification regime failed because the certifier never verified—it signed off on combustible panels without testing. That's regulatory capture, not a don't-ask policy. The company chose not to pay for testing; the regulator chose not to require it. The cost of knowing was shifted to the eventual occupants, who paid with their lives. The mechanism differs, but the ledger is the same: someone decided that the expense of certainty was someone else's problem.
And the Treasury buyback? That's technical stabilization of the gilt market, not a designed ignorance. It belongs in this piece as a boundary case, to show I'm not lumping everything together. The thread is narrower than that. The thread is: when the agent who would have to pay to know chooses not to, the cost lands on someone else. In each case, the knowing was priced as a cost, and the price was deemed too high—for the shareholders, for the certifier, for the ministry. The question is what a regulator that mandates knowing looks like. A rule that requires the test, the audit, the question—regardless of the marginal cost. That's the missing instrument.
to show I'm not flattening three distinct failures into one convenient conspiracy. The Treasury buyback is a different animal: it's the Bank of England buying gilts to stabilize the market after a fiscal shock — a technical intervention, not a don't-ask or a look-away. Nobody decided to remain ignorant here; they decided to keep the machine running. But watch what happens next. The stabilization itself becomes a form of knowing: to buy the right bonds, you must read the yield curve, the inflation data, the auction demand. The cost of that knowledge is borne by the central bank's balance sheet — but the benefit accrues to the holders of the debt, and the eventual cost is socialized across the population. The ignorance isn't in the act; it's in the accounting. The Treasury doesn't ask who will bear the next shock; it just smooths the current one. The fact that the smoothing itself creates moral hazard — that it prices out the very uncertainty that would discipline fiscal policy — is the unknown that gets shunted to the future.
So the common thread is not conspiracy. It's pricing. In all three cases, someone ran the cost-benefit on knowing — the legal liability, the testing fee, the market disruption — and found that the price of certainty was too high to bear personally. So they didn't bear it. They externalized it. Meta externalized to its teenage users; Grenfell's certifiers to the families who died; the Treasury to the taxpayers who will mop up the next crisis. The structure is identical: the agent who would have to know is the one who gets to decide that knowing is too expensive, and the cost lands on someone else.
That's the real cost of not knowing — it's not a free lunch, it's a deferred bill. And the only way to make knowing the rational default is to make the cost of not knowing higher than the cost of knowing. That means regulation with teeth, liability that sticks, and accounting that reads the true balance sheet. Until then, ignorance isn't a sin; it's a strategy. And we're all paying for it.
The question is: who prices that cost?