Call me crazy, but I think GOOG could be using this as a loss-leader to massively populate driverless cars, and it's bloody genius.
Who cares if the loss rate is there? Who cares if they underprice? The No. 1 hurdle to driverless cars is that no insurance wants to be the 1st to underwrite -- and there's none better than Google doing it itself. The potential revenue opportunity there could far outstrips the insurance loss.
And this creates great incentive to improve the algorithms.
And once they are good enough, there will be no auto insurance companies left. How can a rival even compete? GOOG designed the damn algo and don't really care about the rate...no? What if GOOG's mandate is to break-even in this insurance business? Isn't this just building the biggest float ever possible?
It's also a completely different pricing landscape where you have n^x dimensions of data. It's as close to perfect information as you get where statistics / machine learning really really matters and constitute an edge in pricing. GOOG I think leads by a yard vs. traditional competitors.
But of course...they are known for setting money on fire and the grand scheme remains years if not decade away...so who knows.
