Every time you get into a car, there is a non-zero chance that you might get injured, or worse, die. The probability of this happening depends largely on where you're driving and, of course, how much you drive. However, there are a few different ways to measure this statistical risk. A recent Bloomberg article by David Zipper highlights one ongoing debate.

The three most common methods are:

• Road deaths per capita

• Road deaths per registered vehicle

• Road deaths per distance traveled

In my opinion, options 1 and 3 seem the most relevant. Option 1 is useful because it measures a citizen's overall risk and allows driving risk to be easily compared to other causes of death (which tend to be measured on a per capita basis). The limitation is that it is harder to compare a country where everybody drives to a country where few people drive.

That's where option 3 comes in. In theory, it provides the best indicator of road risk by accounting for distance traveled, which is the primary argument for why it's commonly used in the US where the car is king. But it does "dilute" the fatality count the more people drive, and it hides overall car dependency. In his article, Zipper likens this approach to measuring cancer deaths per cigarette smoked.

In any event, here is how both methods appear in the International Transport Forum's 2025 Annual Road Safety Report (which is cited in the article):

On a per vehicle-kilometre basis, the data appears much more gradual. But on a per capita basis, the countries with the highest road fatalities appear much more as outliers. Here, you can more easily see that, broadly speaking, a person in Colombia is nearly ten times more likely to die in a road-related incident than a person in Norway (pretty much the gold standard when it comes to road safety).

Perhaps the answer is to just look at both figures to make sure you're not lying to yourself.

Cover photo by Tom Barrett on Unsplash

Charts from Road Safety Annual Report 2025

This is a shocking report from Smart Growth America on traffic fatalities in the US. Since 2010, the number of pedestrians struck and killed has increased by almost 75%. As of 2022, this number sat at just over 7,500 fatalities per year:

Here are also the top 20 most deadly metro areas:

Not surprisingly, these hot spots tend to be in the south, as opposed to in older northern cities. And that's because these tend to be car-oriented places. As the name of the report suggests, they are "dangerous by design." If you optimize for cars, it means you're making trade-offs in other places.

Charts: Smart Growth America

The US Department of Transportation has just finalized a new vehicle safety standard that will require all light-duty vehicles to be equipped with a more advanced form of automatic emergency braking (AEB) by 2029. (Light-duty vehicle = pretty much all passenger vehicles, including SUVs and trucks.)

Now, most light-duty vehicles on the road today already have some form of emergency braking. What's noteworthy about this ruling is that it requires a more robust version. Some might say one that works. Specifically, it will need to work at much higher speeds and at night.

Most of the AEB systems in operation today don't really work at night -- basically at all -- and many have shown to be ineffective when it comes to stopping for humans.

This new standard will require vehicles to automatically brake at up to 90 mph when a possible collision with a car is detected and up to 45 mph when a possible collision with a pedestrian is detected.

This seems like a very good thing, especially given the persistent problem we are having with cars killing too many people. But how do we do it?

From what I have read, this new standard will be pretty challenging to meet without the use of long-range LiDAR, especially since night vision is a requirement. I find this interesting because, even though autonomy is taking a lot longer to arrive than most people anticipated, there's still meaningful progress being made.

Here's to hoping it saves a lot of lives.