A vision of city transportation, circa 2043
A Sidewalk Talk Q&A with forward-looking transport scholar David Levinson.
It’s 2043. Few people in cities own cars anymore. It’s cheaper to rely on electric, self-driving taxis. Some vehicles are big enough to share; others are individually sized to make the most of limited street space. They have one button inside: Stop. Dynamic curbs — patrolled by enforcement droids — remain clear for deliveries, pick-ups, and drop-offs. Street parking no longer exists, and this space has been recaptured for better public uses.
That’s the future as seen by David Levinson, the University of Sydney transport professor who writes the popular Transportist blog and is co-author of the 2017 book The End of Traffic and the Future of Access. “Look back to the 1920s, and you have magazines that ask: What does the future look like?” he says. “Some of it is absurd. Why would we all be using blimps? But some of it’s still like: Why doesn’t the future look like that?”
The truth, he says, is that imagining tomorrow’s urban mobility raises far more questions than it answers. If we get used to the idea of using taxis, what other things will we no longer feel the need to own? What are the new things we now can do because robots can move around without supervision? What will we do with all the extra time we don’t have to spend driving? How do you allocate road space in a world with delivery drones?
“These things are unpredictable in how they play out,” says Levinson, who’s an advisor to the Sidewalk spinoff company Coord, which recently launched a data integration platform for urban mobility. Levinson spoke to Sidewalk Talk about the challenges facing cities today — and the innovations 25 years or so down the road.
You give a lot of thought to the future of transportation. How do you see the biggest challenges facing urban mobility at this moment but also in the short- and longer-term future?
There’s the litany of automobile evils we all know: lack of safety, pollution, congestion, and so on. Those are all here and have been here for decades and will remain here for at least a little while longer. Trying to actually solve those collective set of problems, which can be done (a) through technology and (b) on the demand side, is the project for the next couple of decades.
On the technology side, the rollout of electric vehicles is relatively straightforward. The rollout of autonomous vehicles, which is more complicated technically, will probably be a little bit slower. There’s simultaneously the rollout of the transformation from an ownership model to a mixed model of fleet-owned vehicles. And along with this transition toward fleet vehicles there’s also the opportunity to right-size the automobile itself, so we don’t have these large, oversized vehicles holding only one person in them. Moving towards the one-passenger vehicle has huge benefits, and that’s the biggest challenge we’re not recognizing.
On automation, people have an unreasonable expectation of how quickly we can deploy this kind of technology, but we’ve moved faster than I imagined we would. We’re getting to the point where we’re going to have passengers in cars where the only thing they have is a “Stop” button. And that’s great, but it’s going to take decades to fully deploy this, because such a big system has to be transformed. Remember a few decades after the mobile phone, and a full decade after the iPhone, just under half of homes still have a landline.
To get to a fleet of taxis with just a “Stop” button there’s so much data the car will need to have to make choices, or to offer you choices as a passenger.
There has to be a real-time map of the environment at different scales: of the infrastructure, of the presence of other vehicles on that infrastructure. Then there’s a services layer that Coord is doing, a real-time map of road prices, curbside regulation and availability, and parking regulations and availability.
Then you have the question of the user’s value of time. How much are you willing to pay to save a minute, because Road X is more expensive but faster, and Road Y is less expensive but slower? That’s if you imagine we will have some kind of spatial differentiation — I’m not convinced we will. It might just be going toward a universal time-of-day pricing, where it’s higher at 4 pm than 2 pm, but it’s not higher on Road X than on Road Y.
Say on the freeway you’re charging more than on a local road per mile of travel because it’s faster. Then more people will use local roads, and that’s not what you want. But if you want more people using the freeways, are you going to charge a discount on the freeways? That’s counterintuitive. It’s going to look something more like a mileage charge with a time-of-day discount than a differentiation by route. That’s my sense of where this goes.
Would you say road pricing is fundamental to a better future of transportation?
It’s fundamental to a less time-wasting form of transportation. I think there’s significant gains to be had from automation and from refactoring the automobile. That is, if we can convince 90 percent of the trips that they can use a one-passenger car, we can double the capacity of roads just from splitting lanes. Then with automation we can double it again, because vehicles can travel closer to each other. That solves a lot of the problem in most places. If you can double the capacity of the roadway, that alone buys you 40 years of population growth.
There’s a big question as to how curbs will be managed given the increased demand from new mobility services. Can that happen in the absence of road pricing?
I think curb management is very ad hoc right now. In big cities, this is a tension. Getting that data streamlined and leading to more rational policies, Coord can improve the situation. There is a lobby for people who would be against on-street parking; that would be the people who own off-street parking. And there is the transformation towards shared vehicle fleets in cities; many fewer vehicles need to be stored on the road at any given time because most are in motion, and there are fewer vehicles around because they’re used more efficiently. So that opportunity to eliminate on-street parking and transform that space into bike lanes, bus lanes, and loading and unloading spaces is ripe. The time is right.
Information technology is making it so that we can track and enforce use of lanes in real-time with cameras. There are many ways this could play out. Maybe enforcement needs to identify vehicles by the license plate, which means the camera angle has to be right, which means cities might need a robotic meter maid. Every block could have its own little enforcement droid to make sure no one is violating the rules about parking their car too long or loading and unloading for longer than needed. And you can do all that without road pricing.
If we get rid of street-parking, do you envision the curb needing the same types of definitions it has now? This is clearly a loading zone. This is clearly a bike lane. Or could it be more flexible? You’d probably need a real-time sense of regulations.
It could be more dynamic, for instance loading from 4 am to 6 am then it’s for movement from 6 am to 9 am then it goes back to loading. Something like a bike lane you’d want to make more permanent. And for a bus lane, when there are enough passengers to justify a bus lane, it should be a bus lane. You already have cities that have parking until 3 pm, then from 3–7 pm it’s a bus lane. And they have at 3 pm a vehicle and crew making sure the street is clear and ticketing the vehicles that are there. It works well enough, though it is a bit labor intensive.
You can imagine once existing rules are in place and well documented with a systematic way of describing and mapping them, people can think more rationally about which of these parameters they want to adjust. Then it’s just showing what the map looks like to someone on the local Curb Management Board, a new institution responsible for those regulations deciding how to maximize the value of curbspace for the community.
In this scenario, how would you envision pick-ups and drop-offs happening? Would they also be charged? Or they’d be directed to a certain place?
Certainly they’d be directed to a certain place. You’d want to avoid loading and unloading them at an intersection. Maybe some midblock taxi stand equivalent. As part of your taxi license you get to pick-up and drop-off in whatever district you’re in, or maybe you pay on a per-drop-off basis. It depends how you’re collecting revenue from your taxi operators. In London, with the congestion charge, they exempted taxis, arguing the price is embedded in the price of the license. That seems plausible if the license fee is large enough and you want people to use taxis instead of parking, then you want to encourage it and not put in another fee. And if you don’t unload at the designated place, that’s illegal and you get a fine, automatically assessed by camera or enforcement droid.
So how do we get to that place where we move away from the ownership model and toward the fleet model?
It’s a value proposition for the consumer. We already have taxi markets. But most residents say to themselves, it’s cheaper to own a car now than to get taxis every day. So if it costs $10,000 a year to own a car, that works out about $30 a day, and I’d pay more than that for my daily taxi, it’s cheaper for me to own a car. If the cost of taxis comes down to less than $30 a day, the value proposition says, I shouldn’t own.
Now there’s the out-of-pocket versus the fixed-cost question, but that can be dealt with through a subscription model. So just like my cell phone, and I have “unlimited” data up to some threshold per month, I might have “unlimited” rides up to some threshold per month, and then I go above that the taxi company charges me per ride.
The reason you should be able to get it under even $20 a day is that if you go toward automation and electric vehicles, the price of the vehicles should drop. Electric vehicles should be less expensive. To date they are not, but we’re moving in that direction pretty rapidly. Automation is the second thing. You’re saving on the cost of parking. That’s $30 a day alone in big cities. You’ve eliminated the driver so taxis are cheaper, so the marginal cost per ride is really low — cheaper than taxis are today.
But unless you can get that cost structure in place, people are not going to give up their car. And I don’t think public incentives are going to matter a lot here, because most cities won’t have the will or the money to subsidize shared rides just so there’s fewer private cars on the road.
So you see self-driving fleets that aren’t shared?
I see multiple models here. You have a taxi provider like Uber or Waymo providing what we would call a taxi service today, except it’s automated. You’ve got leased cars that are maintained centrally in some respects but you take them home with you, so you have them on-demand. Then you have privately owned cars, or less rigorous leases much more akin to today’s private cars.
In an urban area, I can have a car on-demand and some fleet manages it. In a remote area, a car on-demand is a 10-minute wait. Rural users are less likely to want a shared vehicle.
Do you have the same fear of the zombie or ghost vehicles — cars without any passengers in them?
There will be empty vehicles moving around in any case. There are passenger-less taxis moving around now. That’s probably on the order of 10–20 percent of distance. With a well-managed system, you get that down. With private cars, someone could say: I will drive into the city, then send it home to park, then have it pick me up in the afternoon. That would double miles traveled. That would be terrible. You’d need to have some sort of penalty for that. Road pricing becomes perfectly justified if that kind of behavior emerges.
What about the nightmare scenario where it circles the block for eight hours?
If people start doing that, road pricing is the obvious solution. But even without road pricing, you could make it a crime to circle the block more than once in a short period of time. You have road pricing by ticketing. If it’s automated with cameras, then it’s: we’ve identified your car on the same block three times in the past 15 minutes, that’s a $30 fine. People might complain a little bit, but cities will see that as a good way to disincentivize it, and they don’t have to go through the pain of implementing road pricing. Instead it’s a new crime, enforced using technology the way red-light running and speeding are now in some cities.
There’s going to be all sorts of new regulations. Teenagers will step in front of the automated vehicle to make it stop — we know this is going to happen. We will invent a set of laws and regulations once these problems begin to emerge. Cities and counties are pretty quick at copying the regulations of adjacent jurisdictions. It’s a diffusion of innovation process. One place writes the rule, they get the rule right, then all the counties and cities around it just copy the rule.
Are there aspects of the future of transportation we’re not focusing on enough?
I think the curb space question is more generally the road space reallocation question. How do we recapture capacity we no longer need to move automobiles, and what do we do about it?
Doing the same thing better is the obvious first thing that happens. But what are the new things we can now do? It’s not just cars moving people. It’s person-less vehicles moving goods — and they’re not cars anymore. That’s going to change a lot about how we shop: what is the retail experience, what does it mean to want something?
Most discussions of shared autonomous vehicles have an urban-centric viewpoint. How technology changes the world outside of cities is not well understood or much contemplated. It might be that the new transport’s impacts are less outside of cities.
The futurist’s job is to put trends together and paint scenarios, but in the end, we’re 25 years since the Mosaic web browser for the World Wide Web was released, and it has turned out different than was expected. Imagine in 1993 someone said: “Fake news being generated by Albanian teenagers for the purposes of getting ‘ad clicks’ on a social network called Facebook from 60-year-olds would shape the outcome of the 2016 election and elect noted casino owner Donald Trump.” This is not a scenario a futurist could have foreseen. 2043 could be very strange indeed.
This Sidewalk Talk has been edited for length and clarity.
Follow what Sidewalk Labs is thinking, doing, and reading with our weekly newsletter.