My auto insurer, Progressive, knows when I (rarely) hit the brakes hard. But why just use that information to determine my premium? Why not amalgamate data on hard braking and provide it to the people who design and maintain roads?
If I'm the only one hitting the brakes hard at a specific spot, that's a me problem, but if loads of people are doing the same at the same spot, that's a systemic problem that some government agency can and should fix, heading off accidents.
A recent report makes the case persuasively and, I hope, will lead to the amalgamation and sharing of near-crash information by insurers. Doing so could save a lot of lives and avoid a lot injuries and property damage.
Two researchers at Google looked at 10 years of public crash data, compared it against aggregated data on hard braking and found an extremely high correlation. They then used the hard braking data, separate from the crash reports, to predict trouble spots and again had very strong results. For instance, the intersection of Highway 101 in California with Interstate 880 in San Jose was in the top 1% of all road segments for hard braking — and police reports show a crash every six weeks, on average, for decades.
Google is making its data available to transportation agencies — and they should use it. While the U.S. has traditionally viewed auto safety as the responsibility of the individual driving the vehicle, European countries have shown the importance of system design.
Using features such as roundabouts, protected bike lanes, lower speed limits and narrower lanes (which prompt drivers to go more slowly), European countries have far fewer traffic deaths per capita than the U.S. does. For instance, the U.K. reports 2.6 traffic deaths per 100,000 people per year; France, 4.9; Germany, 3.3; Spain, 3.7; and Italy, 5.3; while the U.S. reports 14.2.
The U.S. has such a car culture, including a love for pickups and massive SUVs, so I'm not sure U.S. roads will ever be as safe as those in Europe. But using hard braking, rather than police reports, provides information rapidly and overcomes the inconsistencies that arise because different police agencies handle traffic reports differently. The telematics also can extend the use of data to roads that, unlike the 101/880 interchange, aren't so heavily traveled and aren't such obvious outliers — simply because of randomness, an accident may not happen for a long time in a less-traveled spot, but hard braking can still alert authorities that a big problem exists. In addition, the telematics data can be more precise — you don't just notice that accidents happen in a certain spot but can see that hard braking picks up at a certain time of day, in certain weather or at a certain time of year.
In general, I wish the insurance industry had been faster to use the full capabilities of telematics. For years, they were just used to tell people after the fact that they had been recorded doing something dangerous. The incentive to do better was there but remote, because the incidents would only affect premiums somewhere down the line. It's only in recent years that telematics devices are being used to coach drivers in real time about being drowsy, following too closely, etc., and even now the focus is mostly on fleets of drivers, not individuals.
I understand technology adoption curves, so I know the industry couldn't just wave a magic wand. I also realize that part of the issue is critical mass — you can't do something like aggregate data on hard braking if you don't have enough cars on the road using telematics that can report instances of the behavior.
But I think back to how magical it seemed 25 years ago when I wrote something about how it was going to be possible to learn about traffic jams in real time, because authorities were going to track mobile phones in cars. If phones in an area were stopped, you had a problem. If they were all going 75mph, things were all clear.
And I think we're at this sort of place now with sensors in cars. Hard braking is actually just one example. Sensors will be able to report on potholes or other problems with roads. Dashcams can monitor for other safety-related issues, including crazy driving. (Yes, privacy will be a thorny problem.)
But, for now, I'll be happy if we can just get information on potential accidents into the hands of the right people so they can do what they can to head off fatalities, injuries and property damage. Traffic deaths in the U.S. have been falling in recent years, but more than 40,000 people lost their lives on U.S. roads in 2024, and that's far too many, even for a country that loves its cars.
Cheers,
Paul