Why Autonomous Cars Won't Save Us

We hear a lot that autonomous cars are the future. And there are distinct ways in which autonomous cars would improve the current car situation: in theory they won't violate traffic laws (like speeding), they can safely follow each other more closely than human-driven cars can, and they can have reaction times to things like pedestrians that are significantly better than humans, for example. These aren't true yet, but you can see the potential.

Yet autonomous cars have real disadvantages as well, and in light of their growing deployment in US cities, I wanted to touch on some of them.

1. No Fix for Congestion

The trigger for this post was reporting that companies like Uber and Waymo are projecting autonomous vehicles to increase, not decrease, the number of cars on the road. That means that autonomous cars won't fix the problems with traffic, because they won't be reducing car trips. Sure, they might go faster or more smoothly, but there's still a throughput issue, especially on city streets that have slow speed limits for safety reasons. And of course, stoplights still cause backups for any car.

These cars are stopped, whether they're autonomous or not.

There might be fewer crashes and cars might potentially move less jerkily through bad traffic, but cars on the road are still cars on the road--and the projections of increasing vehicle hours would make that worse, not better.

That's because autonomous cars, unlike regular cars, are not usually conceived of as stopping in a parking space (or parking lot) when they're empty. Instead, Uber and Waymo conceive of them as constantly cycling, looking for more people to drive. And that means that a not-insignificant quantity of those autonomous trips will be empty of people--a problem that unsurprisingly does not exist for human-driven cars, and which can create a massive uptick in traffic.

2. Not Scalable

A related issue is that autonomous cars, just like regular cars, don't really scale as population grows, especially if (as Uber and Waymo typically project) we're still looking at a model that is car-like, rather than bus-like, with individual people calling their own car for their own destination rather than carpooling or driving fixed routes. 

Making this autonomous does not increase its capacity. The driver is just...not driving now. 

It's true that it might have a major benefit for people who cannot drive themselves (youth, the disabled, the elderly, etc.) who can now get a car. But that would run us into problem #1, as it just increases trips and thus congestion. There's no efficiency gain here. 

Note that this does not apply to all autonomous vehicles: autonomous buses, and especially autonomous trains, do provide an efficiency gain, as bus and train operation is often bottlenecked by the ability to find trained operators, and autonomous operation allows more trains per hour on tracks or even the expansion of tracks to increase a system itself. 

But of course, that's older technology, since Vancouver SkyTrain has been autonomous for longer than I've been alive.

No driver, no problem, but also nothing new.

Should we expand driverless train operations? Of course. But that's not the sort of autonomous vehicles that people are excited about right now, and it's a different conversation. 

The car version simply doesn't scale the same way--or it if does, it basically turns into inefficient versions of the trains by putting a separate engine/tires/doors/drivetrain/etc. into every 4 or so seats, compared to having a vehicle like a train with a lot more seats per unit volume.

3. Pollution

And both of the above points contribute to the problem that cars are a polluting form of transportation, especially as compared to other options for moving lots of people.

In theory, I suppose, there's a big benefit to switching to autonomous cars, which is that they'd be newer cars and often electric cars and thus much more fuel-efficient and less polluting than the average of the actual car fleet, which is older and less efficient.

But...this car is an efficient car, but it's not a non-polluting car.

My Nissan Leaf may not produce tailpipe emissions, and in some places it may even not produce much emissions to produce its electricity (say, if I live near a hydroelectric dam). But cars produce other pollution too.

Tires are bad for the environment, and all cars run on tires. 

Sure, there are rubber-tired subways and of course buses also run on tires. But see above about scale: those move a lot more people per tire.

And of course not all autonomous cars are electric or even all that fuel-efficient. 

So increasing the number of cars on the roads is going to massively increase pollution.

And that's physical pollution; I haven't even mentioned other kinds like noise and light pollution.

So AVs may have benefits; I'm not pretending they don't. But they also won't actually save us from some of the bigger issues that cars produce, and they might actually increase some of the problems (like increasing traffic). So maybe we should stop treating them like an urban panacea and consider investing a bit more in some older technology--like our good friend the train.