What is Actually Going On
I wrote a whole book—Confessions of a Recovering Engineer: Transportation for a Strong Town—to explain why our transportation system continues to kill astounding numbers of people and how addressing it will also fix the finances of our cities, along with making many of our other broader social struggles a little easier to manage.
To summarize: We have overbuilt, overengineered, and over-designed our roads and streets based on the belief that having greater buffers, more maneuvering room, and increased recovery area is how we improve safety. Except in very specific situations, such as interstates outside of urban areas, this belief is wrong.
Don’t feel bad if this is not intuitive to you; most transportation professionals struggle with the insight, as well. It feels like having more buffer room provides more margin for error, but the reality is that more margin for error induces higher speeds. It signals to drivers that there is a greater margin for error (in a sense, there is) and so they feel comfortable speeding up. This is very human behavior that is far from anything that should be considered deviant.
High speeds are not dangerous, per se. The real danger comes when high speeds are combined with randomness. When lethal speeds occur in an environment where drivers enter and exit the traffic stream, where there are intersections, drivers pulling in and out of driveways and parking stalls—all of these things create random and unexpected interactions, and the higher speeds reduce the margin of error, with tragic consequences.
You can make such environments safe in one of three ways. You can eliminate driver error, which for obvious and self-serving reasons is the favored approach of the experts in the field. Fixing humans and their embedded flaws is a longstanding hubris of a certain mindset of people that have never read, and been astonished by, the work of Daniel Khaneman and Amos Tversky. As any biologist will tell you, humans are slightly evolved chimpanzees. There is a lot of beauty in that insight, but it only makes us human, not perfectible potential-gods.
More realistically, the second thing you can do is to remove randomness (close intersections, eliminate parking, etc.) and thus eliminate the statistical likelihood of a tragedy occurring. The third and final thing that can be done is to slow speeds, thus increasing the margin for error and lowering the stakes of any collision. (I’ll acknowledge a fourth option—the long-promised revolution in automated vehicles—which is almost as much a fantasy as it is a distraction.)
Here’s my theory of why we have seen a spike in traffic deaths that is not abating, despite traffic levels returning to pre-pandemic levels.
Prior to the pandemic, overwhelming levels of traffic congestion artificially reduced speeds for much of the day and, in doing so, artificially reduced the number of traffic deaths compared to what would be experienced in free-flow conditions. The more congestion and the longer it lasts, the more fatality rates go down.
When the pandemic began, that congestion went away, allowing the drivers that remained to exploit the full operating capacity of the roadway that had been overengineered for them. Speeds went up, along with the number of random interactions, which is the fatal combination.
While traffic has returned to near-normal volume overall, what has not returned to normal is the duration of intense congestion. With work-from-home being normalized, more drivers have the flexibility to choose when they want to take an auto trip. They are choosing to drive during non-peak times, which has the effect of spreading out the overall traffic volume, while reducing the length of those periods of stifling congestion. The result is that a greater volume of traffic is now being exposed to the most dangerous conditions: high speeds and randomness.
And that is how we get more deaths with less traffic, no assumption of individual derangement or selfishness required.
It is also how we get weird sampling data showing more people in crashes showing up impaired, distracted, or without seatbelts. They are the ones more likely to be in a crash in these dangerous conditions, but they were always there. There aren’t meaningfully more of them, there are just statistically more opportunities now for their trip to wreak havoc.
Hypotheses to Test My Theory
Every good theory generates a testable hypothesis. I suggested earlier that we could test the “selfishness” theory, as defined by the experts, by looking at crash data correlated with vaccination rates. It’s not being tested because it’s a silly theory.
The theory that a meaningful portion of our population has become reckless, rebellious, and fatalistic and that mental shift has induced them to speed, forgo a seatbelt, and drive impaired, is more of an assertion of beliefs than something data can confirm. I’d be interested in how cognitive psychologists might test this hypothesis, but I suspect they might find it easier to test the defensive attribution error of the experts.
My theory can be tested through a hypothesis, however. If I am correct, we should be able to evaluate crash data and plot the time of day that fatalities occurred. Before the pandemic, those crashes should cluster outside times of peak congestion. For the last year, the times of fatal crashes should have become more diffuse throughout the day.