Accidents Are Intentional

Our roads are deadly because we refuse to remove the most dangerous drivers

In America, the biggest killers that aren’t a disease or drug are either a gun or a car.

I’ve been thinking a lot about an excellent series in CalMatters called “License to Kill.” It’s well worth your time to read, but TL;DR California is unusually lax in removing dangerous drivers from our roadways. Drivers with multiple DUIs were allowed to keep their licenses until the predictable thing happened and they killed somebody.

Preventing crime depends on how you conceptualize criminality. Is crime something we are all capable of or is criminality relegated to a small subset of the population? The former implies that crime reduction should come from broad-based policies; the latter that society should focus on the incorrigible element.

Put even more abstractly, is criminality distributed according to a normal distribution or a power law distribution? If it’s the former, then we should focus on suppressing the average level of criminality across society. But if it’s the latter, then focusing on the small subset of problem actors would fix the vast majority of a specific crime problem.

A few weeks ago, I saw a presentation by University of Chicago criminologist Jens Ludwig on homicide in that city. In trying to prevent murders, he argues, we miscategorize the decision to kill as a cost-benefit analysis. By imposing draconian penalties—treating minors as adults, mandatory minimum sentencing and the like—we hope to stop future homicides by both deterring potential murderers or removing them from society.

But murder, according to Ludwig, flows from a more primitive form of thinking. The decision to kill is more reflex than conscious decision and is made by people with still undeveloped brains, easy access to guns and a lifetime of training that aggression must be responded to with equal aggression as there is no larger authority coming to your rescue. He told a story about a Cook County jailer supervising juvenile inmates. About 20 percent of these kids are psychopaths and if we let them out, they’d kill again. But 80 percent would go on to live normal lives if we could just give them a do-over on the five minutes of bad decisions that put them in his custody.

Homicide, or more specifically the type of homicide that plagues cities like Chicago, is a normal distribution problem. By fixing young males’ “System I” thinking through violence intervention programs, Ludwig claims we can reduce the baseline impulse to kill by rewiring the reflex that responds to a perceived threat with overwhelming violence. And based on the evidence he describes in his book on the topic, it seems like he has a point.

The interventions Ludwig describes in his book are important because America treats homicide as if it’s power law distributed. That is to say, the median number of homicides an American commits is practically zero and the vast majority are committed by a small subsection of the population. Therefore, the strategy has focused on either deterring would-be murderers by promising to lock them up for decades—even life—or by simply doing so after they kill so that they cannot kill again.

But according to Ludwig’s assessment, we’re going about this all wrong. While society does indeed contain psychopaths who need to be separated from the rest of us, murder is something that all of us are capable of given the right societal conditions and easy access to firearms. After all, the decision to kill is very rarely preceded by a lengthy cost-benefit analysis.

Aside from guns, cars make America a uniquely dangerous nation compared to others of similar levels of wealth. About 44,000 Americans die from gun fire each year, about 35 percent of those deaths being homicides (though easy access to firearms often makes an impulsive decision to end one’s life tragically permanent.) An equal number of Americans die annually because of cars—either because they were struck by one or because they were within one that collided with another. About 14.2 Americans per 100,000 will die because of a car, a rate comparable to that of Bosnia and Herzegovina or Armenia and nearly three times that of Canada, Australia or France. What’s more, many other advanced nations have seen their automobile fatality rate plummet over the last 30 years. In the US, it’s actually increased since 2010 and has now surpassed that of Russia.

Are car crashes normally distributed or power law distributed? That is to say, is there some number of crashes that all of us will instigate at sometime in our lives or is a small minority of drivers responsible for the overwhelming majority of crashes? The former requires a broad set of policies that target everyone who gets behind the wheel, the latter requires identifying and targeting that small subset so that they can be removed from the roadways.

The answer to this question is surprisingly hard to find. Search for it online and you’ll find an insurance industry statistic that the average driver files a collision claim once every 17.9 years. Thus, they extrapolate, the typical American driver will be involved in between three and four crashes in the 60-something years they drive in the course of a lifetime.

However, this might be a situation where Bill Gates walks into a bar and turns everyone into a billionaire on average. For instance, a 2015 study from Louisiana found that over an eight year period, drivers with two or more crashes accounted for 34 percent of all collisions in the state even though they represented just five percent of Louisiana drivers. Another study using Kentucky data from 1995 to 2002 found that only about five percent of licensed drivers in the state were involved in any crash during that period. Of those who were, 56 percent had exactly one crash and an additional 28 percent had two over the eight years studied. The remaining drivers—those with three or more crashes—represented less than 1 percent of all licensed Kentucky drivers.

Studies also show that most drivers involved in a crash have a history of poor driving prior to the incident. A meta analysis of 99 road safety studies found that prior tickets were an important predictor of future crashes, but that they only explained a small portion of the total variance. A study of 140,000 driver records in California found that prior collisions and traffic citations are an important predictor of future crashes.

Driving is an exercise in rule following. Cars do not collide because we set rules on which cars drive in which direction, at what time and at what speed. We’ve been conditioned to refer to crashes as “accidents,” implying that they are unavoidable calamities, rather than the result of deliberate human action. The grievous injury or death that often results from vehicle collisions becomes a vagary of fate rather than someone else’s choice to harm you. It makes the act of driving feel safer than it would to realize fully that your body and your life are at the mercy of the person with the worst self-control on the roadway with you at a given time.

Driving also takes place in legally privileged spaces. To exist on a roadway requires you to be in a vehicle and to operate that vehicle requires you to pass a certification process, register your vehicle with the government and purchase insurance from a private carrier. Anyone is allowed on a bus, anyone can walk along a sidewalk, but only a subset of society can use a roadway for its intended purpose even though roads and other car infrastructure comprise a sizable chunk of American urban space.

In seeking to prevent deaths and injuries resulting from vehicle collisions, the government can exercise large amounts of discretion in who it allows on its roadways by policing who receives and who can keep their driver’s license. New drivers are tested on their ability to operate a car safely and drivers can lose their licenses temporarily or permanently after committing some infraction. Insurance companies also price the risk of a given driver—someone with many infractions may eventually have to pay a premium that they can no longer afford—which serves as a de-facto license suspension.

The biggest flaw in this system is that it relies on faulty information. Traffic enforcement relies on police officers to witness an infraction and issue a citation. However, there are far fewer police officers than motorists which makes the probability of being issued a citation for breaking the law infinitesimally small. Further, California law enforcement agencies are stopping far fewer drivers than they were prior to COVID-19. The drop has been precipitous in some cities: San Francisco police issued 96% fewer citations in 2023 compared to 2014. On the other hand, Black motorists are far more likely to be stopped relative to their proportion of the population.

Automated enforcement is the solution to this problem. A traffic camera does not discriminate against motorists of a certain race and does not go on a silent strike because it’s displeased with the city government. Later this year, six California cities will unveil automated speed monitoring within city limits as part of a five year pilot study. In addition, speed cameras will also be deployed at freeway work zones to ticket those who speed near construction workers.

California’s freeways are an excellent venue to test automated enforcement. They are drab and featureless spaces where driving is mostly uniform. There is no vegetation to obscure a camera nor is there any reason to deviate from mostly keeping to the posted limit within a single lane of travel. Speeding is an obvious violation that can be monitored, but so too can dangerous merging and passing. Because the space is so homogenous, erratic behavior stands out that much more.

These cameras are a good first step, but they’re based on deterring speeders in areas with a history of fatal collisions. These areas are better served with traffic-calming infrastructure that limits the ability to drive recklessly. Narrow roadways, bulb-outs and other physical barriers will do more to save lives and prevent injuries—European nations such as the Netherlands have already proved how effective this approach is.

The true value of a traffic camera is that it can be everywhere, all the time. In theory, we could cover our city streets and freeways in traffic cameras to develop nearly complete data about the driving habits of every vehicle traveling upon them. In theory, we could increase the probability to nearly 100% that traffic infractions like speeding would be met with a punishment.

This would obviously serve a deterrent function, but the true purpose of these cameras isn’t just to bring down average speeds. Rather, these cameras could identify problem drivers and suspend their licenses before they have the chance to kill or injure somebody. A habitual speeder, for instance, has a far higher likelihood of causing a collision than one who rarely exceeds the speed limit. An incorrigible speeder—one who continues to speed despite multiple violations—is a prime candidate to have their license revoked. As it stands now, the only way to identify a problem driver is after they cause a collision. Ubiquitous cameras mean we no longer need to wait until there’s a mangled body to identify who should no longer be on the road.

Automated traffic enforcement is a rare instance when a paltry investment in already cheap technology will pay dividends in thousands of lives. The civil liberties case against automated enforcement is likewise paltry: the abuses of companies such as Flock can be obviated by keeping the data with a statewide law enforcement agency like the California Highway Patrol and requiring judicial warrants to access vehicle location data beyond moving violation enforcement. Moreover, the alternative to automated enforcement is not no enforcement, but biased enforcement done by human beings—often with deadly consequences.

Homicides persist because we treat the act like it’s something only a small number of monsters are capable of committing. Recognizing that “normal people” in a fit of rage and with access to a pistol are capable of killing helped save lives. Likewise, we should shift our understanding to acknowledge that a person’s lack of impulse control and inability to follow directions makes them orders of magnitude more deadly on the roadways. We must recognize that road safety is not just a matter of the built environment but of being more selective about who should retain the privilege of driving.

We call it an “accident” to obscure the fact that a series of intentional decisions led to a terrible outcome. A collision is not an accident and we need to stop treating this human carnage as if it’s somehow inevitable.