The biggest promise of self-driving cars is that traffic will finally be democratized. All vehicles will flow smoothly through the city using optimized routes, getting everyone where they need to go safely and efficiently. But there’s a very good possibly that’s not how it will work at all.
In a piece called “When All Is Optimized,” Alexandros Washburn, the former chief urban designer for New York City, acknowledges that all the smart transportation technologies we’re layering onto our streets can make life great. But he also points to a frightening way it all could go very awry:
Imagine that roads have reached capacity in a mega-city. Our Waze app has hit the asymptote of optimization. You won’t go any faster from A to B whether you go this way or that. Traffic has become a zero-sum game. To go faster in a zero-sum city, each of us needs a WazeMe app to compete against other drivers. If I arrive sooner, it will mean someone else arrives later. Imagine if driving across town was a contest between competing algorithms. These algorithms would use existing data but they would also create synthetic data to spoof the data of a competitor. When we have optimized the collective, someone begins to win and someone begins to lose. We have crossed the line from tactics to strategy.
Think about this for a second. Plenty of apps already have these tiered levels of service. What if, in addition to Waze, there was WazePro that specifically routed some cars along the swiftest routes. What if, instead of just a nicer vehicle, the premium rates of UberBLACK might buy you a faster travel time to your destination. Suddenly, the idea of elevated, “congestion pricing” becomes exactly that: A way to pay more money during busy times to “beat” the traffic.
Scarier yet: In your app’s quest to best the other navigation apps around it, your car might flick out some fake data (watch out, road debris!) to throw the other algorithms off-course.
This isn’t necessarily a concern that’s native to urban transportation, Washburn argues. You can already see precedent for it in the financial world. Stock market trades are increasingly made by algorithms, not humans, that are specifically designed to sabotage other trades. Washburn writes:
Their algorithms are sensing data. They’re also creating data to spoof the sensors of other algorithms. They place trades and cancel them in nanoseconds, leaving the responder’s price naked.
In the short term, before the autonomous revolution, this algorithmic control of our streets points at yet another danger of the digital divide: the richest, most tech-equipped residents are promised the best urban experience. And it’s another good argument for making all self-driving vehicles part of a car-sharing system that functions more like public transit.
But it also points to another potentially hopeful trend. Maybe the future won’t even be about shelling out tens of thousands of dollars to buy the most tricked-out new self-driving car on the market. It will be about buying the best algorithm so you’re not sitting in that car all the time.