On the surface it seems like a good idea (maybe): usage-based or mileage-based insurance where you pay less if you drive fewer miles, don't drive late at night, don't drive into bad areas, and don't make sudden starts and stops. Insurers in England and the U.S. have launched pilot programs. I hope they succeed. I wonder if they will. They're up against a Nobel-prize-winning theory called adverse selection that also explains why used cars aren't such a good deal. (If a used car is any good, you keep it or sell it to a relative; if it's a clunker, you're more likely to want to sell it.)

Adverse (or negative) selection as applied to car insurance might be explained this way: Worse drivers might not opt for pay-as-you-drive insurance because their insurance would probably go up. Some of us are dyslexic when it comes to speed signs: We see "65" and parse it as "75." Or we just love driving exuberantly. We'd likely pay more under plans that track driving habits. So why sign up to pay more? And if bad drivers don't sign on, then who underwrites the discounts for the really good drivers? The really good drivers now get discounts of, say, 10% for no accidents, yet by changing their habits only a little, under pay-as-you-drive, they might qualify for discounts of as much as 60% with a trial plan under way at Progressive Insurance.

Problems with pay-as-you-go?
Plus there are other issues:

-- Pay-as-you-drive insurance might have to pass muster with state insurance commissioners.

-- The black boxes needed to track drivers cost several hundred dollars. Usually the insurance company has to pick up the cost; sometimes drivers pay the installation fee, which cuts into first-year savings.

-- The whole family might have to sign on to the plan for a car: the teen driver just starting out as well as the parents with three decades of mostly safe driving. You don't want Junior driving the car after midnight, but would rates go up if mom or dad were out past the stroke of midnight? (If it's a car-by-car program, a multi-car family might opt to reserve one car for late-night motoring.)

-- You'd pay more if you drove in dangerous neighborhoods. What if you lived in those neighborhoods? Is that a double whammy and does it discriminate against minorities, some might wonder.

-- Some people fear this is the first step toward big brother car insurance where insurance based on your personal driving habits wouldn't be an option but the norm.

-- Some insurance companies fear pay-as-you-drive might lower insurance premiums overall, perhaps by exposing the vast number of good drivers who aren't getting discounts. That's the theory and I'm reporting it; I'm not sure I've seen those drivers where I live.

Why it might work
Thomas Hallauer and Sharon Gill of Telematics Update note one British insurer, Norwich Union, abandoned a pilot project after two years, apparently when it attracted just 10% of its goal of 100,000 users. That said, Hallauer and Gill believe there are plenty of reasons why there's still life in pay-as-you-drive insurance:

-- The necessary GPS locater gives you the basis for a low-cost navigation system, which NU offered, and an OnStar-like crash notification service, which it didn't. As a bundle of services, it may be more palatable.

-- Pay-as-you-drive works for commercial car and truck fleets and is popular, at least with the fleet owners and managers.

What about drunk drivers?
One noted source (well, Wikipedia), aruges that as we automate driver behavior through this, photo radar, and red light cameras, there may be less police presence on the highways looking out for drunk drivers. But that argument works both ways: When a cop is sitting behind a billboard with a radar gun, he or she isn't out patrolling the roads.

About that Nobel Prize
About the Nobel Prize: George Akerlof shared the 2001 Nobel Prize in economics based on his 1970 essay, "The Market for Lemons," about why it's hard to find a good used car, especially buying from an individual seller. He pointed out, with statistical clarity what's obvious in hindsight: If it's a good car, you hang on to it; if it's a bad car, you dump it. Ergo, a disproportionate number of used cars for sale are lemons.