Self-Driving Cars and the Virtues of Federalism
The new Department of Transportation guidelines for self-driving cars have arrived after a buildup of complaints about how autonomous vehicles are somehow being held back by a confusing "patchwork" of government regulations on self-driving cars.
I don't buy that this is a problem, by the way, because by far the biggest roadblocks for self-driving cars are the extremely complex technological problems they still have to solve. As for the patchwork of regulations, I would argue that it isn't a bug, it's a feature. Variation in government responses to new technology highlights the virtues of federalism and makes us less likely to go down a regulatory blind alley.
State and local government reactions to self-driving cars have varied from enthusiastic to hostile but have mostly been enthusiastic. Pittsburgh beat everyone to be the first to get self-driving cars on the road, so now Boston is saying that they need some, too. It has become hip for forward-thinking cities that want to be futuristic to float plans that involve the use of self-driving cars to transform their downtowns.
In Michigan, a law is making its way through the state legislature that would open the roads to self-driving cars. The state is being so accommodating to automakers that they actually stopped to amend the law at the request of Google. It's almost as if the state feels it has some kind of special interest in making sure it stays at the forefront of automotive innovation. Meanwhile, California's existing law is pretty restrictive, so they're looking to make exceptions to accommodate Detroit's rivals and/or partners in Silicon Valley.
In Chicago, by contrast, two aldermen have proposed an ordinance to ban self-driving cars. Naturally, it is a total coincidence that the guys proposing this ban get a lot of money from the taxi industry. (This is one of the few cities where it's against the building code to install PVC plumbing in most structures, because PVC is simple and cheap and anyone can install it—which would deprive the plumbers' unions of work. In Chicago, what Da Unions want, Da Unions get.) Similarly, in New York City, local laws require that a driver keep at least one hand on the wheel at all times, which negates the whole point of self-driving cars. And drivers' groups—including a group representing Uber's own drivers—have pledged to raise a firestorm of opposition if anyone tries to change the law. Uber drivers worked hard to steal their jobs away from the taxi drivers, so they're not about to let a bunch of robot cars do the same to them.
So yes, this is quite a patchwork, and that's actually how the system is supposed to work. It's called "federalism," and it's the idea that the national-level government in Washington, DC, does not have to impose uniform regulations on the entire United States from the top down. The big virtue of this federal system is that is allows state and local governments to craft their own legal response to any new practice, problem, or invention, so we can see how those responses work out and ideally adopt the ones with the best results. At the very least, it lets us vote with our feet (or with our self-driving wheels) for the jurisdictions whose approach we prefer.
The best case for federal involvement is the fact that self-driving cars will be crossing state lines and using the interstate highways. Yet this is not likely to be the first place where a detailed legal response to self-driving cars is necessary. Driving on interstates is the simplest problem for self-driving cars to handle. Many automobiles already on the road are equipped with features like adaptive cruise control and systems to keep you in your lane. It's the off-the-highway driving that poses the big questions when it comes to safety: what happens when you have to negotiate narrow, winding side streets with lots of pedestrians and cyclists. Which is precisely what you encounter in places like Pittsburgh, Boston, New York, and Chicago.
At the same time, the downtown districts of big cities are where the technological problems of this kind of self-driving are likely to be solved first. Getting a self-driving car to work properly requires detailed maps that allow the car to recognize exactly where it is on the road and what it should be looking for. Densely populated downtown districts are where it makes the most economic sense to compile that kind of detailed driving data. So self-driving cars are likely to be primarily a local phenomenon at first, dealt with by city governments.
At this early stage, we're going to see a need for customized local responses. So states like Michigan and California are going to have a special interest in moving forward to allow larger zones for early testing, because they are home to the companies that are in the forefront of the new technology. (The same goes for Pittsburgh, which was chosen for the Uber test because that's where Uber poached its whole self-driving unit from the robotics faculty at Carnegie Mellon University.) Some Western states with lots of long, straight highways might end up being more interested in enabling the development of self-driving trucks first, as Nevada has been doing.
In an article making the case for federal regulations, Adrienne LaFrance starts by pretty much making the case against them.
Here’s how it usually happens: Technology comes first, and the law follows.... It’s difficult, bordering on impossible, to predict all the ways in which transformative new machines will change the world.
That strikes me as the overriding consideration. Why should we be deciding on universal, standardized federal regulations when we still don't quite know what self-driving technology is going to look like?
Moreover, I don't think we can assume that we will need any particular new regulations to govern self-driving cars. When it comes to concern for the safety of the public, I suspect the vigilance of what LaFrance dubs a "federal DMV for self-driving cars"—not the most encouraging way of describing it—is going to be way less important than the immediately self-interested vigilance of the people who are actually trusting their lives to those cars. Or the vested interest of the companies who are developing the cars—Tesla, Ford, Uber, Lyft—who don't want to be sued into oblivion by the victims of preventable self-driving accidents.
In fact, so far the main regulatory role in self-driving cars has been to offer exemptions from the myriad existing regulations that govern automobiles, such as that one about keeping one hand on the wheel. Even the new federal guidelines talk about the need to allow autonomous vehicles to violate the rules of the road in situations where it is required and normally accepted, such as crossing over lane divisions to get around a broken-down car that is blocking the road. We'll know driverless cars have truly arrived when states loosen laws against operating a vehicle while intoxicated, because the occupant of the car will no longer really be operating it beyond telling it to drive him home. (Next up: self-driving cars that use machine learning to decipher slurred speech.)
It is best for the federal government to tread lightly in this unknown and still developing field. If it needs to say anything at all, let it stop with the kind of general guidelines it has offered so far and leave more specific legal responses to the state and local level, allowing the flexibility and experimentation made possible by a system of federalism.