[trendia]

I wonder if they use this data to price insurance -- they would easily know when their drivers are going over the speed limit (or, if such data is not so precise, if their average speed over 10 minutes exceeded the speed limit).

[ThrustVectoring]

More likely is approximating number of miles driven and price discriminating based off that. More miles driven = more risk of an auto accident. Basically pay-per-mile car insurance, but hidden.

[pathseeker]

How do they know you are driving? Seems too error-prone to be useful.

[ThrustVectoring]

They don't need to know you are driving to do price discrimination. They could just as well take the zip codes where you live and work and assume you're driving, and make a profit giving discounts to folks with a shorter commute regardless of whether or not they actually drive it.

[lutorm]

They might know how fast you're traveling, but they don't know who's driving.

[jstanley]

Just because it's not 100% accurate doesn't mean it's 0% accurate.

There's still value in a noisy signal.

[addflip]

That was my concern.