[rconti]

How do you know that more aggressive driving causes more collisions?

How do you know which accelerometer-measured[1] parameters correlate with aggressive driving? High speeds? Abrupt acceleration? Hard acceleration? Abrupt braking? Hard braking? Late braking (how do you know the difference?) High lateral cornering loads? How are these differentiated from abrupt lane changes? Does the dongle know if you used your signal or not?

Studies show that driving late at night is far riskier than driving during daytime; due to visibility, alcohol, etc. Lots of late night trips? Time to punish that graveyard shift ER doctor for driving her car at such risky times!

1. Presumably most of these OBD-II dongles really only use accelerometer data to report back, even though they have access to more data if they want it, and if they're programmed to understand every model of vehicle

[lotsofpulp]

> How do you know that more aggressive driving causes more collisions?

I don’t, but insurance companies employ some pretty smart people who can figure that out. If they offer discounts for certain driving characteristics, I’m willing to bet that they reduce the insurer’s expenses. Otherwise they won’t be in business for long.

[rconti]

Most insurers that offer a tracking-based discount also offer policies without. It seems to me, all that matters is, on average, they don't offer a discount to tracker users that is larger than those users, as a pool, save them in claims.

The only incentive to get it "right" is that, if they do a really poor job at this, they will lose business to insurance companies that do a better job at it -- again, to the extent that the differences between insurers are significant, that customers are price sensitive, aware enough to shop around and compare the actual rates they pay, etc.

We already know that your non-tracker-based insurance premiums may vary wildly from one insurer to another. One may penalize you heavily for being 18. One may offer a larger discount to those who complete driver's ed courses, or own a car with airbags.

Maybe you'll be penalized for that first accident or ticket, or maybe your insurance company will give you some leeway (and, presumably, double down on the surcharge after your second accident).

I don't see how it's inherently obvious that an insurer that gets their tracker-based metrics 'wrong' will go out of business in short order. Again, all they have to do is charge more than they pay out in claims and overhead.