Hacker News Reader

Comment by kator | Hacker News Reader

This seems interesting but they recorder is missing a lot of context. If you drive Los Angeles freeways you'll be hitting the breaks a lot for random people who cut you off in bumper-to-bumper traffic, but the data will only show you stomping on the breaks all the time and the insurance company might conclude you're a bad driver because you're not paying attention.

Also how does it distinguish between drivers? My kids often "borrow" the car and I'd like to be able to "advise" them when they're causing our insurance to go up by driving poorly.. :-)

View on HN

> If you drive Los Angeles freeways you'll be hitting the breaks a lot for random people who cut you off in bumper-to-bumper traffic, but the data will only show you stomping on the breaks all the time and the insurance company might conclude you're a bad driver because you're not paying attention.

That is the correct conclusion though. The insurance company could care less whether you're a bad driver—what they really care is how much of an accident risk are you.

Even a good driver frequently caught in LA traffic is a higher risk than a good driver elsewhere. Hence the device is reporting exactly what the company needs.

I've used one of these devices in my car for the last couple of years and have enjoyed really cheap car insurance as a result (variable based on my driving but usually $30-$35 CAD/month).

It is very true that if you are driving in rush hour traffic you have no choice but to appear to be a bad driver. The irony of this is that the accident happening in rush hour is probably minor, yet when I'm barrelling down the highway at a steady pace I appear to be a great driver whereas an accident would be very costly (fatal perhaps).

Also they normalize your infractions (sharp breaking, sharp acceleration, speeding, etc) by distance driven so long highway driving is even more beneficial (few infractions, long distances) than traffic jams (many infractions, short distances).

Insurance companies have a strong financial incentive to have their models be well calibrated with reality, so you should expect it more and more with time that when an insurance company thinks you're a bad driver then it really means that your driving sucks.

True, but they'll happily take 10 horrendous accidents over 1100 minor bumps if the bumps cost them a thousand each and the bad accidents cost them a hundred thousand each.

Our incentives are different. We would rather have many tiny incidents (up to a limit) than a few life threatening incidents.

Therefore, while an insurance company would happily put into place a process that turns a small number of drivers into maniacs and everyone else into placid sheep they'd take it.

Not that they are, and it sounds unlikely that such a process can even exist, but it's important to know that our incentives aren't aligned.

Insurance companies usually want to have accurate models and accurately predict the future. Hence, I presume they'll figure out the "LA freeway braking" issue.

I don't think it needs to distinguish between drivers. Your kids probably should raise your rates if they are driving poorly.

Kids on your policy pretty much raise your rates, period.

And the insurance companies have been doing this for a long, long time. Many moons ago (pre-internet) I worked summers doing telemarketing for a Very Large insurance company. When people signed up for auto policies the agents would casually ask the ages of the children during conversation. That information was stored away for years, sometimes a decade, until the kid turned 16.

My job was to call and "casually" ask questions like "so, which of your cars is Johnny driving when he gets the keys?". You can guess the rest.

Reply on news.ycombinator.com