Skip to main content

The Pointlessness of M-City

Why test cars on fake roads when they'll need to operate on real ones?
  • Author:
  • Updated:
(Illustration: Susie Cagle)

(Illustration: Susie Cagle)

By the numbers, cars are more dangerous than guns; they kill, on average, more than 90 Americans every day. Automated car technology promises to greatly reduce those terrible numbers. But there’s still a long way to go before automation is truly road-ready.

To ostensibly speed up that progress, the University of Michigan recently unveiled its new Mobility Transformation Facility automated vehicle test track—complete with robot pedestrians and cyclists. According to university researchers, “a realistic off-roadway environment is an essential step before a significant number of highly automated vehicles can be deployed safely on actual roadways.” The 32-acre M-City is a kind of grown-up Autopia, designed with a variety of traffic signals, road types, and complex obstacles, all surrounded by movable building facades.

This kind of methodical research and development seems prudent when coordinating computerized metal death traps capable of speeds up to 100mph. But it also seems to kind of miss the point of how driverless technology will actually be implemented.

The project appears to be Michigan’s attempt at reclaiming the car industry for its region—and in a decidedly different way than Silicon Valley, where Google has run self-driving cars on public roads for the last several years, racking up more than one million miles and the dents to prove it.

M-City is not for moving fast or breaking things. This kind of methodical research and development seems prudent when coordinating computerized metal death traps capable of speeds up to 100mph. It’s kind of refreshing. But it also seems to kind of miss the point of how driverless technology will actually be implemented.

We almost always prefer to re-invent than to retrofit. M-City seems visionary in the way futurism often seems visionary: It’s a re-imagining of our world, made from scratch.

But the future isn’t a fresh start—it’s built on top of the present, which is full of fallible, human-driven cars that are only getting older. The average age of a car on the road in the United States is now 11.5 years, and has been climbing since the recession. Even if self-driving vehicles eventually debut at a modest price point, Americans just don’t seem to want, need, or have the means to buy new cars. At this rate, it would be decades before self-driving vehicles were a significant force on our roads—the scenario M-City was supposedly built to test.

Driverless cars full of extra-productive workers packed together on highways at rush hour is an exceptionally retro vision of the future. It also might be the best we can do after decades of carving America into highways and suburbs—this may be the futurism we deserve. But it will be decades more before robot vehicles are a dominant force on the road. Automated vehicles will roll onto roads the same way other new cars do—slowly, and over a long period of time, during which they’ll mostly be interacting with older cars driven by dumb humans. Earlier this month, a Google self-driving car made national news when it was rear-ended by one of those dumb humans. Since 2009, according to Google, its self-driving cars have been hit by human drivers 14 times, about once for every 70,000 miles traveled.

Boosters claim that automated cars will boost worker productivity and unshackle us from the burdens of personal vehicle operation and even ownership, via legions of unmanned Uber cars. Boosters claim a lot of things. What we really need smarter cars to do is save us from our own dangerous driving. And that means we need to test them on real roads, where they’re still likely to be the small minority for a couple generations—barring an extremely expensive transit takeover by Uber.

A few years ago, automated cars driving among us inspired fear and loathing. But it seems scarier, and more dangerous, to plan primarily for the fully automated future. Google’s cars don’t appear to be moving terribly fast or breaking too many things, but we wouldn’t even know that if Google were testing its cars only in M-City with other self-driving cars and no dumb humans.

If M-City is meant as a public relations stunt to re-assert Michigan as the American home of the car industry, it may have some public relations value. If it’s meant to signal a more concerted effort on behalf of U.S. inventors and manufacturers to make more automated cars, it’s not terribly convincing. It will certainly be useful to test all-robot driving in some controlled capacity. After all, they aren’t exactly omniscient—at least not yet. Unlike humans, Google’s cars are still not capable of discerning and maneuvering around potholes, snow, and other obstacles. But a driverless fender bender on a public street—however ethically complex—will tell us a lot more about the future we’ll actually live to see.

The Crooked Valley is an illustrated series exploring the systems of privilege and inequality that perpetuate tech's culture of bad ideas.