One of the least understood aspects of self-driving cars is the extent and capacity to which they rely on remote assistance operators (RAOs).
When a self-driving car finds itself confused in an uncertain or tricky situation (like when one rolled into an active shootout), there are typically two safety valves. The first is a manual override, where someone like a first responder might jump into the front seat and take control of the steering wheel. And the second is assistance from a remote operator.
If the car gets confused, a human can tell it, "Hey, you, follow this path." But how often is remote assistance being called upon? And who is actually responding on the other end? Apparently, the answer is, "I don't know."
According to a recent report from Senator Ed Markey, every major AV company refuses to disclose how often they rely on an RAO. And in the case of Waymo, they rely on overseas operators in places like the Philippines.
This has led to new proposed legislation that would, among other things, limit the number of vehicles that a single RAO can oversee, mandate that the RAO be located in the US, and require the humans to hold a local driver's license. You know, so they're sure to know the rules of the road.
There's a lot to figure out, and it seems a bit messy. But that's what it takes. As one would expect, this is par for the course when you're trying to rewrite urban mobility.
Cover photo by Leo_Visions on Unsplash
We’ve been talking about the merits of congestion pricing for as long as I’ve been writing this blog. But it remains politically unpopular, despite the overwhelming evidence that it consistently does what it’s supposed to do: it reduces congestion, shortens commute times, improves air quality, and raises money for alternative modes of transport, among other things.
The status quo bias is strong, but right now we have an opportunity. Self-driving cars are in the midst of shifting the mobility landscape, and there’s a growing belief that (1) roads are going to need to be more accurately priced to deal with the surge in demand, and (2) this is a moment in time that grants us the opportunity to do it. Here’s a recent tweet by Chris Spoke of Toronto Standard that makes this point and that I agree with.
The basic idea behind point number two is that many voters don’t like the idea of a congestion charge for themselves, but will probably mind a charge on robot cars a lot less — both because they are robot cars and because there are relatively few of them on the road today. However, at some point, robot cars will form the majority of vehicles on the road, so now would be a good time to establish new practices.
What do you think?
Cover photo by Minku Kang on Unsplash
Fred Wilson chose the perfect quote by William Gibson, here, to describe the current status of self-driving cars: "The future is already here — it's just not very evenly distributed." That's how it feels right now.
Waymo isn't in Toronto yet, but they are expanding rapidly throughout the US and elsewhere. Last week they announced fully autonomous driving in five new cities: Miami, Dallas, Houston, San Antonio, and Orlando. Autonomy is here, as we have talked about many times. There's no longer a question.
But what's interesting is that we're at the point in the hype cycle where expectations are not as inflated as they were a number of years ago (at least that's the way it appears to me). Years ago, everyone in real estate was talking about how it would disrupt parking requirements and reshape the landscape of our cities.
So when does this happen?
Fred ended his post by saying that "the downstream effects of this technology and behavior change are going to be profound." But he doesn't get into what these changes might be. Let's do a reminder of that now. Some of the most commonly believed consequences are as follows:
Cars consume a vast amount of real estate and also spend the vast majority of their lives just sitting around idle. Switching to a "mobility-as-a-service" model will require dramatically less parking. This is going to force landlords to repurpose the parking they already have and it's going to encourage developers to build new buildings with reduced parking, or no parking at all. That will be good for housing affordability.
However, the autonomous vehicles will need to park and corral somewhere at some point. My guess is that we will see something akin to rail yards today. This would be a good use for some of our excess parking, though this use won't require nearly as much. I would also imagine that many of the cars will leave the most valuable and dense parts of a city during off-peak periods.
At the same time, it's not clear what the winning business model for AVs will be. Will it be a Waymo-like model where the ride-hailing company owns and operates all of the cars? Will it be a Tesla Robotaxi model where individuals own the cars and put them out to work? In this case, maybe the Robotaxis just go back to people's individual garages. Or will Uber remain the dominant platform? Meaning, an asset-light model that aggregates customer demand remains the highest-value component of the stack. Personally, I can't see Tesla's Robotaxi model being very lucrative for individual owners, so I'm inclined to look toward Waymo and Uber.
Street parking will be replaced by a proliferation of pick-up/drop-off zones. This urban design problem will need to be solved as we dramatically increase the number of people getting in and out of AVs on busy urban streets.
In the mid-1990s, Italian physicist Cesare Marchetti remarked that, all throughout history, humans have tended to cap their commute times at about 60 minutes per day. Something like a half hour each way. This became known as Marchetti's Constant. What this has meant is that as new technologies (streetcars, cars, and so on) allowed us to move faster within that 60 minutes, humans have tended to sprawl further outward. Will AVs do the same, and could they actually break Marchetti's Constant?
As we all know, the key difference with AVs is that we will no longer need to pay attention to our commute. We could sit in an AV and sleep, work, watch a movie, or do whatever else we'd like. One can think of it like a mobile office or mobile living room. This should, in theory, make commuting long distances a lot more enjoyable and encourage even greater "super sprawl."
The counterforce to this phenomenon is that if more people are willing to commute long distances in an AV, we will see demand greatly outstrip supply on our roads. In other words, traffic congestion in large cities will get even worse. I think this will force more/most cities to adopt congestion pricing. Politically, it will finally become acceptable, because now we'll be able to use "the machines" as our scapegoat. They're overrunning our cities! Ironically, this means that we won't adopt the thing that makes driving a lot better until we all stop driving.
So where do these opposing forces ultimately net out? Well, my view (and bias) is that human-scaled walkable communities will always have value. We are social animals. I also think that the experience within our cities will improve dramatically. Pedestrian safety will increase (the data already supports this) and far less space will be dedicated to cars. Good.
At the same time, I think that reducing commute friction will encourage an exurban explosion. Like the technologies that came before AVs, it's going to empower humans to further decentralize. What this will do is exacerbate the divide between our urban cores and our suburban and exurban fringes.
Of course, this is just me surmising. I don't really know. But AVs are here, and I think it's time we get back to discussing and planning for the second and third-order effects of this technology.