Earlier, we wrote about the first fatal crash of a partly-self-driving car. A Tesla, operating on autopilot mode, failed to detect a semi-trailer crossing in its path, and the resulting collision killed its human driver.

The crash has provoked a great deal of discussion in the media about safety data, the potential for future technology, and the problems of human-interaction with partially automated systems. A key challenge is that current (and at least foreseeable) implementations of self-driving technology are likely to require human supervision and intervention, and the man-machine combination likely poses its own risks. The seeming inerrancy of the machine systems in route circumstances may lull humans into a sense of complacency or inattention, with disastrous consequences when the machine fails. (Indeed, according to at least one account the victim of the Florida crash was watching a Harry Potter video at the time of the crash).

A parked Tesla. Credit: AJ Batac, Flickr
Tesla (price $101,500) Parking Space (Free!) . Credit: AJ Batac, Flickr


While we may instinctively regard the safety issues surrounding autonomous vehicles as being primarily technological in nature, they also depend critically on institutional arrangements we establish and the policy choices we make about transportation and public space.  Safety will be determined as much by rules of the road as by any safety device.

To many achieving safety will come in the form of fully automated vehicles that would eliminate any human role in the driving process. Not only are such vehicles still years—if not decades—away, but then there’s the challenge navigating a transitional period in which self-driving vehicles share the road populated mostly with human-piloted vehicles.

To avoid these problems altogether, Jerry Kaplan writing in the Wall Street Journal, has suggested that self-driving vehicles be given their own infrastructure. Lanes on highways would be reserved for self-driving vehicles.

Taking the initiative in this way would better foster innovation and let the free market work its magic. What might such a plan look like? Perhaps we could start by reserving high-occupancy-vehicle lanes or certain roads at specific times for automated vehicles.

It’s breathtaking that neither the Kaplan—nor apparently his editors at the Wall Street Journal—grasped the glaring contradiction between “free market” and “government prohibiting non-automated vehicles from using roads.”

There’s no such thing as a free market for transportation: transportation hinges directly on public policy, particularly spending on roads and the rules that govern their use. A technologically advanced car is essentially useless without a network of public roads on which it can operate. The (largely) private market for vehicles depends directly on a public policy of building roads and regulating them in ways favorable to vehicle travel.

How we get from today’s technology to tomorrow’s, whatever form it takes, will be very much about public policy choices. That’s the way its always been.  It’s worth spending a moment thinking back about this process. In the early days of the automobile, the rules of the road were quite different. There were no traffic signals, and wagons, streetcars, and pedestrians freely mixed on city streets. In the earliest days, some cities required cars be preceded by a flag-bearer to warn other travelers of its approach.

A key victory of what Vancouver’s Gordon Price calls “motordom” came in literally re-writing the rules of the road. By law and custom, other users were marginalized or completely excluded from the roadway. Pedestrians not crossing at marketed crosswalks were branded “jay-walkers”—a derisive and entirely manufactured term, designed to shame what had long been common behavior on city streets.

There’s no such thing as a free market for roads

While it’s cloaked in the rhetoric of markets, Kaplan’s call to dedicate a portion of the public right of way exclusively to autonomous vehicles is really the latest incarnation of Asphalt Socialism: We ought to give massive public subsidies to private vehicle movement, privilege these cars over other forms of transportation, and generally subordinate the quality of place to the movement of vehicles.

It may make sense for fully automated vehicles to have their own right of way: but if they do, they ought to pay for the privilege. The root of many of our transportation and urban problems is the consistent under-pricing, and consequent massive subsidies to private vehicle transportation. Our decision to subsidize freeway construction in cities—ostensibly to speed travel—has chiefly led to much more sprawling development patterns. Its an open question whether fleets of cheap, comfortable authonomous vehicles would lead people to choose to commute longer distances from even more far-flung locations.  Subsidized infrastructure for automated vehicles would recapitulate this mistakes of the past with a whole new set of technologies.

Likewise, the advent of driverless vehicles raises important issues about the legal liability for damages when car crashes occur.  Will driverless vehicle manufacturers or software providers or map makers be subject to legal recourse under product liability laws?  At least one study has suggested a fairly sweeping legal framework that would shift liability to vehicle owners.  How the legal architecture of liability is re-written will likely have a profound impact on the deployment of this technology.

The advent of highly instrumented vehicles should instead be treated as an opportunity to revisit archaic and failed choices about how we regulate, price and pay for roads. If vehicles had to pay for the cost of the public roads on which they travel (and the public street space that is routinely used for free car storage), drivers would make very different decisions about when, where and how much to travel.  Realizing the safety and other benefits that can potentially come from self-driving vehicles will be just as much a matter of working out the right public policies as it is tackling the technological challenges.