Unsafe Safety

No one has ever stood up in a public meeting and argued for danger. That is precisely what makes "safety" the most useful word in politics — not because it describes something everyone wants, though it does, but because invoking it is one of the only moves available in public life that ends an argument instead of starting one. Object to a policy sold as safety and you are no longer arguing about the policy. You are arguing, as far as the room is concerned, against safety itself, which is not a position anyone can hold in public and be taken seriously. The word doesn't win the debate. It cancels it.

That cancellation is the actual subject of this essay, not safety as such. Nobody sensible is against seatbelts, food inspection, or bridges built to code — those are safety measures that survive ordinary scrutiny because they can point to a mechanism, a body count avoided, a cost anyone can weigh against the restriction it buys. What this essay is after is a narrower and more consequential thing: the move where "safety" stops being a claim that can be checked and starts being a trump card that ends the hand, deployed specifically at the moment when checking the claim would be inconvenient.

Securitization, or how an issue leaves politics

International-relations scholars gave this move a name before anyone else did. The Copenhagen School's securitization theory, developed mainly by Ole Wæver and Barry Buzan in the 1990s, describes exactly the mechanism at work: an issue gets treated as politics as usual — contestable, debated, subject to normal procedure, reversible by the next election — right up until someone successfully frames it as a security matter. Once that framing sticks, the issue exits ordinary politics and enters what Wæver called an existential, emergency register, where extraordinary measures that would never survive normal deliberation become not just permissible but obviously necessary. The securitizing move doesn't have to be true to work. It has to be accepted — by an audience willing to grant that this particular threat justifies suspending the process that would otherwise scrutinize the response to it.

The theory was built to explain state behavior — border closures, surveillance expansions, wars justified by threats that later turned out to be overstated or invented outright. But the mechanism it names isn't specific to states, and it isn't specific to the word "security." "Safety" does the identical job at a smaller, more domestic scale, and arguably does it more effectively, because "security" at least sounds like it belongs to generals and diplomats — something a citizen might reasonably question. "Safety" sounds like it belongs to everyone, which means questioning it reads as questioning your own children, your own neighbors, your own body. The securitizing move dressed as safety doesn't need an existential threat to a nation. It only needs a plausible harm to a person, and a person is easier to conjure than a nation-state's survival.

Safetyism as a culture, not just a tactic

Jonathan Haidt and Greg Lukianoff gave the domestic, cultural half of this phenomenon its own name — safetyism — in their account of how American institutions, particularly universities, came to treat physical and emotional safety as a sacred value: not one good among several to be balanced against others, but a trump that ends discussion the instant it's invoked. Their diagnosis wasn't only political. It was developmental — a culture that overprotects, in their account, produces exactly the fragility it claims to be preventing, the way a child never exposed to germs develops a weaker immune system, or an ecosystem where every threat is a violation of a hedge fund's risk tolerance instead of a decision made under uncertainty. Nassim Taleb's antifragility argument sits directly behind this: systems that are shielded from every stressor don't become safer, they become brittle, primed to fail catastrophically the first time a stressor they were never allowed to encounter finally arrives.

What Haidt and Lukianoff were describing on campuses generalizes past campuses. Once a culture accepts that safety is a value that isn't supposed to be weighed against other values — only asserted and complied with — the word stops functioning as a description of a real, measurable risk and starts functioning as a permission slip. And permission slips, unlike risk assessments, don't need evidence. They need only be signed by someone the audience already trusts to be acting in good faith on their behalf, which is exactly the posture "safety" claims by default, before a single fact has been offered.

Where the trump card actually gets played

The pattern shows up wherever an interest that would lose an open argument finds it more convenient to win a safety argument instead.

"If it saves just one life." This phrase is securitization in a single sentence — a standard so absolute that no cost, however large, however certain, can outweigh a benefit that is merely possible. Taken literally it would justify banning cars, since traffic fatalities are a certainty at scale; taken as an actual argument it is never applied that consistently, which is the tell that it isn't really a standard at all. It's a stopping point, deployed exactly when its proponent doesn't want the actual trade-off — cost against probability, restriction against benefit — examined in the room.

"Trust and safety." Content-moderation regimes at major platforms are named, deliberately, to borrow the same rhetorical shield securitization theory describes: a decision to remove speech, deplatform an account, or suppress a topic is framed as a safety function rather than an editorial judgment, which moves it out of the register where editorial judgments get argued about — bias, viewpoint, who decided and by what standard — and into the register where objecting reads as objecting to safety. The team doing the moderating is rarely called an editorial board, even when editorial judgment is exactly what it's exercising, because an editorial board can be second-guessed and a safety team, rhetorically, cannot.

Zoning and building codes. Fire codes, structural codes, and setback requirements began as genuine, checkable safety measures — a fire escape's dimensions can be tested against how fast people actually move down one. But the same regulatory apparatus, once established as a safety authority, becomes a durable veto point that outlives the specific hazard it was built to address, repurposed by incumbent homeowners to block new housing under the same "safety" language originally used to prevent tenement fires. The category never has to be re-justified once it's installed; it only has to be invoked.

National security and secrecy. Classification regimes, surveillance authorizations, and encryption-backdoor proposals are argued in almost every case as a safety measure against an enumerated harm — terrorism, child exploitation, foreign interference — precisely because that framing forecloses the question a plain civil-liberties argument would have to answer: what is this power actually being used for, by whom, and who checks that. "Think of the children" and "national security" are the same move at different registers, both engineered to make the asker of that question look like the one who doesn't care about the harm.

Public-health emergency powers. Emergency public-health authority is a real and sometimes necessary tool — the securitization framing isn't wrong that some situations genuinely are emergencies. What the framing obscures is the difference between an emergency, which is by definition temporary and subject to sunset, and a permanent expansion of executive discretion justified by an emergency that quietly never gets formally closed, because closing it would require reopening a debate the emergency framing was used to avoid in the first place.

The actuary behind the curtain

There is a second hidden ruler in all of this, and it rarely gets named because it doesn't even need the securitizing move — it operates a level below rhetoric entirely, in the terms of a policy nobody in the room has read. Insurance underwriting and liability law have become, in practice, one of the most powerful unelected safety regulators in modern life, setting de facto rules that no legislature voted on and no public ever debated, enforced not by police but by the much quieter mechanism of a premium going up or a policy simply refusing to renew.

Playgrounds are the clearest case. The disappearance of merry-go-rounds, tall slides, seesaws, and monkey bars from American playgrounds over the last four decades tracks the liability-insurance market far more precisely than it tracks injury data — municipal insurers and their loss-prevention consultants set equipment standards well past what the actual, fairly low base rate of serious playground injury would justify, because an insurer's exposure isn't the injury itself, it's the lawsuit the injury might trigger, and the cheapest way to manage lawsuit exposure is to remove the equipment rather than defend it. The word attached to the removal is always "safety." The actual client being served is the insurer's loss ratio.

Medicine runs the same circuit under the name defensive medicine: a physician orders an extra scan, a hospital adopts a blanket policy, a specialist declines to treat a borderline case, not because the medical evidence calls for it but because malpractice carriers and hospital risk-management offices have made the conservative choice the only one that doesn't jeopardize coverage. Workplaces run it through workers'-compensation premiums, which quietly dictate procedures no OSHA regulation requires, because the insurer's actuarial table, not the regulator's rulebook, is what the employer is actually optimizing against. Contractors and builders run it through liability riders that exceed the building code itself — the code is public and can be challenged in a hearing; the rider is a private contract term, and there is no hearing for those at all.

This is securitization's quieter cousin. Where the state's version needs an audience to accept a threat framing before the extraordinary measure is justified, the insurance version doesn't need public acceptance of anything — it only needs a signature on a policy nobody expected to become a rulebook. The premium schedule and the claims-history model are the actual rulemaking body; "safety" is simply the word painted on the door of an office that is, underneath, run by actuaries pricing risk for a company's own balance sheet, not by anyone accountable to the people living under the rule they've set. It is, in this essay's terms, an especially effective hidden ruler precisely because it never has to make a securitizing move in public at all. There is no debate to shut down, because the public was never told a decision had been made.

The fear driving those premiums upward — that any given incident is one lawsuit away from a ruinous verdict — deserves the same scrutiny this essay has applied to everything else called "safety," and honesty requires admitting the evidence here doesn't point one direction cleanly. Studies and judicial surveys have repeatedly found that purely groundless litigation is viewed as a limited problem by many of the judges who actually see the caseload firsthand, not the epidemic the premium-setting logic seems to assume. Where real abuse concentrates, it concentrates unevenly — in specific high-stakes categories like mass torts and particular personal-injury niches, and in specific jurisdictions with reputations as plaintiff-friendly venues, rather than uniformly across the civil justice system an insurer's actuarial model treats as a single undifferentiated risk pool. Which reforms are warranted in response — damage caps, loser-pays rules, tighter pleading standards, litigation-funding disclosure — remains genuinely, seriously contested along ideological and interest-group lines, and this essay isn't the place to settle it. What can be said without settling it is narrower and still does the needed work: an underwriter pricing playground equipment or a hospital's malpractice exposure is pricing against worst-case tail risk and reputational exposure, not against the actual, contested, unevenly distributed empirical picture of where litigation abuse really lives — which means the "safety" standard that results is calibrated to a fear of litigation, not to litigation as it actually occurs, one more step removed from anything a citizen, a patient, or a parent on a playground ever got to weigh in on.

Genuine safety and safetyism, told apart

None of this is an argument against safety, any more than the previous essay's argument was against every inherited category. It's an argument for the same discipline this site keeps returning to: a claim earns the deference "safety" currently gets for free only if it can survive being checked — a stated mechanism, a measurable harm, a cost that's been weighed rather than declared unweighable, and a sunset or a review process that puts the claim back on the table once the circumstances that justified it have passed. A load-bearing safety measure can produce all four on request. A securitizing move producing only the fourth thing — the demand for immediate, unquestioning compliance — has already told you which kind it is.

The tell is identical to the one this site's last essay described for reified categories generally: a genuine safety claim gets defended with a mechanism when asked. A safetyist one gets defended with a restatement of the harm, louder, and a faint suggestion that asking the question was itself the suspicious act. That shift — from here is the mechanism to how could you even ask — is the exact moment safety stops describing a risk and starts functioning as a hidden ruler, unelected, unaccountable, and, for as long as the word keeps working, completely unquestioned.

The circumvention, named plainly

Put plainly, this is what "safety" accomplishes when it's used this way: it lets a decision that would lose a public argument skip the public argument. Public opinion is not overridden by force in these cases — it's overridden by a word that makes forming an opposing opinion feel, to the person forming it, like a small act of self-endangerment. That is a more efficient circumvention than censorship, because it doesn't need to silence the objection. It only needs to make the objector silence themselves, and hand the decision to whoever was persuasive enough to make the word stick — which, as the securitization theorists observed of states and Haidt and Lukianoff observed of institutions, is rarely the person with the best evidence and reliably the person with the most to gain from the argument never happening at all.