Traffic deaths in the U.S. are increasing at a rate that population growth cannot account for alone. The spike in traffic deaths has forced the to consider big policy changes that sound empty but may actually help, according to a new book by . Even if goals like “” seem naive, Singer says these policies can make a difference.

Singer spoke to about her book, There Are No Accidents, which argues that in order to address the spike in U.S. traffic deaths, we have to start looking for causes beyond human error and look at the conditions that compound risk. Those conditions, it turns out, are under our control.

The is worth a read, but two points stick out. Namely, that the word “accident” implies inevitability (which is misleading), and that car crashes happen inside the context of layers of risk that stack:

We die by accident because of risk exposure. The layers of risk compound, increasing the likelihood that mistakes aren’t survivable. As our regulatory systems have declined since the Reagan administration, we’ve lost a lot of ability to police how corporations expose us to unsafe conditions. At the same time, economic inequality drives accidental death in a variety of ways.

A good way to think of it is that you’re more at risk to get killed in a car crash if you drive an older car. You’re also more at risk if you live in a place where everyone drives older cars. You’re also more at risk if the roads are in worse repair. Those are three layers of risk, and if you’re only exposed to one of those layers, you’re less likely to die in an accidental death than someone who is exposed to all three. And that’s part of the reason we see such strong correlations between accidental deaths and the economies of local and state populations.

The problem, Singer , also traces to how we think of of crashes as “accidents.” Singer says “accident” has a hint of passivity. Like an accident is something that happened and no one is to blame. The majority of the time someone did mess up, and the risk of that mistake killing someone relates to the conditions of the crash. These are well within the control of policy and to some degree, carmakers, too. Singer on:

“Accidents happen when errors occur under dangerous conditions, but you can create conditions that anticipate errors and make those mistakes less of a life-or-death equation,” she writes. “Or you can focus all your energy on errors, and let the same accidents happen again and again.”

A lot of the time, companies aren’t open to regulation or oversight, and that attitude is what we should push back against, Singer argues.

Imagine if said that developing their automated driving systems and making them safer would cost too much, or end up bankrupting the company. Well then tough shit. If a feature isn’t safe, it’s not a feature; it’s a bug. Singer says tried pulling that move, claiming would shut the company down, and yet Ford is still around today.

Carmakers may try to skirt responsibility (well, except for ) but it’s up to policy changes enacted by people to not let the car companies get away with it.

I can get onboard with Singer’s argument because the idea of risk being layered sounds more complete than that of population growth. If this alone explained the rise in traffic deaths, then there should be a linear progression to that increase. Bloomberg traffic deaths have grown at three times the rate of U.S. population growth. There’s more at work here.