5 ms·

> Insurance companies may think you’re a higher-risk driver if you’re accelerating fast, braking aggressively, making sharp turns, using your phone while you dr

> Insurance companies may think you’re a higher-risk driver if you’re accelerating fast, braking aggressively, making sharp turns, using your phone while you drive, coming home at 3 a.m., taking dangerous roads or simply driving a lot. Is there any proof that any of these behaviors actually correlate with tangible risk? (I mean, I can understand that "simply driving a lot" increases risk.) Personally, I'd be more interested in cars with cameras using machine learning to build a risk profile of other cars. I'm not the only person who sees dangerous behavior all the time. It would be interesting if I could sell a stream from my cameras to a company that builds risk profiles based on observed behavior.

elif 3y ago
Anecdotally, when Tesla implemented safety score as a voluntary activity to unlock FSD, my gamer brain kicked in and couldn't not pursue a high score. The result is that I went from an extremely anxious and aggressive driver (which is 90% of american drivers) to completely ambivalent to my arrival time and obsessively approaching intersections cautiously and giving time for drivers that cut me off. I do think that a key component for success is near real-time feedback so it feels like a game and not a speeding ticket in the mail.
asoneth 3y ago
I have no idea whether the current telematics algorithms are any good at assessing risk. However, with a large enough corpus of training data including years of telematics on drivers with and without insurance claims it certainly seems like it should be possible to build models that can make much more accurate risk assessments than they can with the data they currently collect such as age, zipcode, and past accidents. Put another way, I would be surprised if there was zero correlation between driving behavior and insurance claims.
LanceH 3y ago
Casualty insurance companies just want the numbers, then to set the markup, then profit. They've filled out that elusive intermediate step. If it means finding a way to get a customer a discount then that's ok, as long as they have the customer. The biggest factor they can monitor is verifying miles driven. Miles driven, unsurprisingly, has a bigger impact than virtually anything else someone signing up to be monitored would be guilty of.
- gwbas1c 3y ago
  > The biggest factor they can monitor is verifying miles driven. Miles driven, unsurprisingly, has a bigger impact than virtually anything else someone signing up to be monitored would be guilty of. In general, no surprises. What really surprises me is that my insurance company never asks for an odometer reading. I haven't changed insurance in years. It's also hard to estimate how many miles I will drive in a year because my life situation changes every few years.
  - LanceH 3y ago
    They may have it from safety inspections or oil changes if you have those done at a mechanic.

elif 3y ago

Anecdotally, when Tesla implemented safety score as a voluntary activity to unlock FSD, my gamer brain kicked in and couldn't not pursue a high score. The result is that I went from an extremely anxious and aggressive driver (which is 90% of american drivers) to completely ambivalent to my arrival time and obsessively approaching intersections cautiously and giving time for drivers that cut me off. I do think that a key component for success is near real-time feedback so it feels like a game and not a speeding ticket in the mail.

asoneth 3y ago

I have no idea whether the current telematics algorithms are any good at assessing risk. However, with a large enough corpus of training data including years of telematics on drivers with and without insurance claims it certainly seems like it should be possible to build models that can make much more accurate risk assessments than they can with the data they currently collect such as age, zipcode, and past accidents. Put another way, I would be surprised if there was zero correlation between driving behavior and insurance claims.

LanceH 3y ago

Casualty insurance companies just want the numbers, then to set the markup, then profit. They've filled out that elusive intermediate step. If it means finding a way to get a customer a discount then that's ok, as long as they have the customer. The biggest factor they can monitor is verifying miles driven. Miles driven, unsurprisingly, has a bigger impact than virtually anything else someone signing up to be monitored would be guilty of.

gwbas1c 3y ago
> The biggest factor they can monitor is verifying miles driven. Miles driven, unsurprisingly, has a bigger impact than virtually anything else someone signing up to be monitored would be guilty of. In general, no surprises. What really surprises me is that my insurance company never asks for an odometer reading. I haven't changed insurance in years. It's also hard to estimate how many miles I will drive in a year because my life situation changes every few years.
- LanceH 3y ago
  They may have it from safety inspections or oil changes if you have those done at a mechanic.