Kevin Drum is asking an interesting question and coming to a conclusion that I think is completely wrong:
But here’s a more interesting question: after driverless cars become widely available, how long will it be until human-driven cars are made illegal? I say ten years. It will vary state to state, of course, and there will likely be exceptions of various kinds (specific types of commercial vehicles, ATVs meant for fun, etc.). Still, without a special license they’ll become broadly illegal on streets in fairly short order. The proximate cause will be a chart something like the one on the right.
I think this is an interesting question, but when I went to visit my in-laws last month, there were still horse and buggies on the road. And those have been technologically obsolete for a century now.
There are a few things that I think Kevin is getting wrong. First, there is a massive distributional issue. Driverless cars will by definition be new cars. The first wave of driverless cars won’t be 100% adapted. Some people will be technophobic, others will like driving sticks, others will be reluctant to put their life into the hands of a piece of software even if that software is statistically a much better driver than the average human (as we are all above average drivers in our own internal estimation it’s just those assholes who are honking at me that can’t drive). And others will decide that they don’t want to spend the money.
Even assuming that there is a fairly rapid shift in the share of proportion of driver controlled and driverless vehicles sold over a couple of years so that in five or ten years from the first good autopilot to 90% of new cars being sold are driverless or driver minimized vehicles, there will be millions of new vehicles that require drivers on the road.
No state government is going to tell tens of thousands of middle class or better voters that they need to junk their $20,000, $30,000 or $40,000 capital investments for safety reasons.
Furthermore, the used car market lags the new car market. My primary used car in high school was made several months before I could walk. Factory fresh, it had sixty six horsepower and by the time I bought it for $50 it could just hit 65 MPH going down hill with a good tail wind. It did not help my social life in high school but retrospectively that car kept me out of a lot of bad decisions simply because the car simply would not allow me to show off and be stupid.
The typical American car has at least a fifteen year lifepan.
There is no way any state government is going to successfully tell most of its working class voters that they need to scrap a $5,000 to $20,000 capital investment for a marginal safety improvement.
What is far more likely once there is good data on operational usage of driverless cars is that they will be treated like anti-lock brakes and skid-control features by the insurance companies. Driverless cars will receive a massive insurance discount because they’ll be far less risky as they remove the most common source of error (human error) from the equation. But driven cars will still be available and still be insurable but at a higher rate.