What Elevators Taught Us About the Need for Digital Safety Standards

The United States has some of the safest elevators in the world. Statistically, you are more likely to be struck by lightning than to die in an American elevator. That's a remarkable fact, and it didn't happen by accident. It happened because of a certification and inspection system that emerged from decades of deadly failures — and because someone decided that self-assessment by the industry was not good enough.

The history starts in the mid-1800s. Early elevators were genuinely dangerous. The mechanism that hoisted the cab was a rope, and if the rope broke, the cab fell. Freely. People died. The industry existed, products were sold, and for a long time the standard response to elevator deaths was essentially that this was an acceptable cost of doing business.

In 1852, Elisha Otis invented the safety brake — a device that would automatically lock the elevator cab in place if the cable snapped. He famously demonstrated it at the 1854 World's Fair in New York by cutting the hoist cable himself while standing in the cab. The cab dropped an inch and stopped. It was a turning point in the technology. But a better mechanism alone didn't make elevators safe. What made elevators safe was a system of standards and independent verification that took another fifty years to develop.

The American Society of Mechanical Engineers published the first edition of the A17.1 Safety Code for Elevators in 1921, after a series of high-profile accidents made the absence of a standard impossible to ignore. The code established uniform requirements for construction, installation, inspection, and testing. Crucially, it wasn't self-enforced — independent inspectors, not the manufacturers or building owners, were required to verify compliance. The system worked because the entity doing the certifying had no financial interest in the outcome.

Today, elevator inspection regimes vary by state, but the underlying principle is consistent: third-party verification, documented standards, and regular renewal. Manufacturers cannot simply declare their products safe. The certification has to be earned and maintained. The result is a fatality rate that is, by any international comparison, extraordinarily low.

Now think about the web.

Digital products today are in roughly the same position as elevators in 1880. The technology exists to do tremendous harm — financial manipulation, privacy violations, psychological exploitation — and the primary check on that harm is the word of the company doing it. Products are called "user-friendly" by the people who built them and profit from them. There is no independent standard. There is no third-party verification. There is very little consequence for most violations, and what consequences exist are inconsistent and hard to predict.

The dark patterns documented on this site — hidden costs, manufactured urgency, shame-based opt-outs, pre-checked consent boxes — are not edge cases. They are industry norms. They persist not because designers are uniquely malicious, but because the incentive structure rewards short-term conversion over long-term trust, and there is nothing in the environment that pushes back with any force.

The elevator analogy is useful precisely because it isn't abstract. We already know what the solution looks like. We've already built it in physical infrastructure, in food safety, in aviation, in pharmaceuticals. The pattern is always the same: an industry self-regulates until the harms become undeniable, a standard is codified, and independent verification becomes the norm. The products that meet the standard are certified. The ones that don't are not.

The web is overdue for that moment. The harms are already undeniable — the FTC has documented them, academic researchers have quantified them, regulators in Europe have started legislating against them. What's missing is a consistent, credible, independent standard that a product can meet and demonstrate.

That's what Light Patterns is building. Not because certification is a perfect solution, but because it's the solution that has worked everywhere else we've tried it. Elevators don't fall because we decided that independent verification was worth taking seriously. The same logic applies here.