The whole "women drive less" angle is bizarre; most insurers already consider gender along with miles-to-work, use of the car (pleasure, farm, commute), age, and marital status in developing rates.
Using miles driven per year does make sense, and it can take into account the suspicions some on this thread have voiced that infrequent drivers aren't as skilled, in the rare instance that those drivers are unsafe but have had no accidents (which would already increase their rates). Still, even given that theory, the additional driving skill from 40k miles a year vs. 5k miles per year is highly unlikely to compensate for 8x the exposure to collisions.
As for the cost of capturing the mileage data, the insured would simply give that information. Of course, they could lie, but if they lied too egregiously that could nullify the policy and their insurer could deny any claims.
In any case, it is unlikely that such a rating factor would be very large, given the application of other criteria like # of accidents, sex, class factor, etc.
It is curious to me how some folks who are otherwise free market supporters turn on insurance companies, and somehow think it's all price-fixing and whatnot. Let me tell ya, folks, it ain't, though sometimes (okay, a lot) state regulators stick their fat noses into things and screw the consumer by refusing to let the market set the price for insurance.
You are aware, of course, of the anti-trust exemption for the insurance industry via the McCarran-Ferguson Act?
Trust me, it is price fixing. The proper solution is to get rid of the govt mandates, and allow the free market and civil liability statutes to control the price of insurance. And get rid of their anti-trust exemption.