» Do we have the political will to build a different type of transportation system? Or will the rise of autonomous vehicles simply reinforce existing norms?

There is now no debate over whether automated technology will revolutionize the transportation system—the question is just when it will do so. Industry watchers who four years ago predicted it would take a decade and a half to realize fully driverless cars now suggest they’ll be online, everywhere, by 2020.

Indeed, just this week, Tesla announced that all of the vehicles it is producing, including its newest, cheapest Model 3, will have “full self-driving capability at a safety level substantially greater than that of a human driver.” Its cars will feature eight cameras and 12 ultrasonic sensors, designed to evaluate the world around the cars and respond immediately. The technology won’t be activated immediately; it will collect data as drivers use the cars to improve its calibration before release.

Tesla’s announcement comes in the midst of a public debate about how such driverless technology should be rolled out. Should it be simply integrated into the existing, privately owned personal automobile system, such that people buy autonomous cars to keep for themselves, or should it be rolled out in connection with driverless fleets designed for shared use?

Robin Chase, the founder of Zipcar, has laid out an intuitive way of understanding this issue using a binary “heaven or hell” construction (note: I’ve interviewed her in the past on how autonomous cars will impact the transit system). According to this formulation, we could have “heaven” if we had fleets of shared, electric, driverless cars powered by renewable energy, plus a redistributive economy that ensures that people who once had jobs in the transportation sector have access to a minimum income. On the other hand, we could have “hell” if everyone owns his or her own driverless car that does our errands, parks our cars, and circles the neighborhood waiting for us to need it again.

Tesla seems to be resolving this issue for us.

In the video the company released announcing its new technology, a man enters a car outside of his bucolic suburban home, from which it whisks him away—without him touching the steering wheel—to the (also suburban) Tesla factory. Once there, he gets out of the car, at which point it goes off to find a parking space, where it will presumably sit all day until he’s ready to go back home.

One promotional video is hardly enough to make an assessment about the future, but the imagery Tesla is projecting, which is of an anti-urban, individualized nature, certainly aligns closely with Chase’s “hell” scenario. After all, multiplied across the millions of people living in a metropolitan area, Tesla’s independently owned self-driving cars would simply replicate much of the existing transportation system, except this time, unlike for current drivers, they’ll have no incentive to minimize driving time—since automated cars can go driving off, circling the block or finding some distant parking space, without inconveniencing the driver.*

And why wouldn’t this be Tesla’s vision? Its hope for future sales is dependent on being able to sell tens of thousands of cars, so it has an interest in making the market for car consumption as large as possible. Keeping automobiles underutilized may not make much sense for a city, or for the society, but it certainly makes sense with respect to automaker profits.

Some argue that fleet cars will simply come naturally; after all, it will cost less to order a shared autonomous vehicle via an app than it will to own your own autonomous vehicle. Yet it is worth noting, as Joe Cortright has written, that the price of such ordered, shared autonomous vehicles is only cheaper than owned cars when all costs are accounted for. Just the cost of fuel (or, in the future, electricity)—the primary operating expense that most drivers care about and notice—is now and will continue to be cheaper than the cost of hiring a shared vehicle.

This suggests that, for many people, waiting for a shared car—which may be dirty, or uncomfortable, or smelly—simply won’t be worth it compared to owning one’s own autonomous vehicles.

That’s particularly true given the fact that driving overall (no matter who owns the vehicle) is likely to decrease in cost significantly compared to today because of the fact that people are no longer paying for parking, they’re able to take advantage of the time in the car to do other things, and the cars themselves may be cheaper, since manufacturing costs will decline because the cars won’t have to be designed to withstand crashes they won’t get into. Why not, given those conditions, spend the money on one’s own autonomous vehicle?

I should say that in urban contexts where people are reliant on walking, biking, and transit, shared-car fleets are likely to be more prominent than individually owned cars. In dense cities, the availability of frequent car service and the limited distance of most trips will minimize the value of ownership. Yet those contexts represent just a small portion of the living environments of today’s America, and people there already don’t own many cars. People taking advantage of shared fleets likely will be doing so to replace what are now walking, biking, and transit trips. Moreover, most of the country lives in suburban environments where trips are longer (thus people want their rides to be more comfortable) and where densities are low enough to make waiting for a shared vehicle more of an inconvenience than it’s worth given the limited additional cost of owning a car.

The case for political agency

Where, then, does that leave us? Do we have no choice but to accept the reality that the arrival of autonomous vehicles will do little more than replicate our existing transportation system?

Many have suggested that the arrival of autonomous vehicles—and the reduced transportation costs they can be expected to bring—will eliminate the need for much of the existing transit network. This argument suggests that we should stop investing in transit that we’re not sure can compete with autonomous cars, that we should simply pause planning for the future of our cities’ transportation systems because we need to wait for autonomous vehicle implementation to know what to do.

Payton Chung, who is more skeptical than me of the ease with which the transportation system will evolve into universalized autonomous vehicles, notes thoughtfully that “It’s important to remember that American suburbia is a political and social construct, not a fact of life, and that policies put into place immense structural supports for American suburbs.”

Indeed, the same can be said about our transportation system as a whole, and its relationship to the cities it serves; it isn’t a random coincidence that people commute in very different ways in New York and Dallas. We do not have to accept the “hell” scenario of Tesla’s creation—but working to produce “heaven” requires more than resting our hopes on the economic benefits of sharing vehicles versus owning our own. Advancing positive change for our cities means recognizing the trouble with simply accepting whatever is most appealing on the market, or whatever the market leaders are promoting.

In the course of my work developing Transport Databook (which, if you haven’t checked it out yet, is a resource for up-to-date transportation data), I pulled together information on changes in transportation mode shares in U.S. cities over time. I was sadly unsurprised to find that the share of people commuting by car in the country’s ten biggest cities is little changed from what it was in 1970. The level of car reliance is a fact of life, one might say. Given public interest in autonomous vehicles, cities should give up on bus lanes, abandon pedestrianized streets—just give people what they want.

Yet examining similar data for Paris and London suggests that, in fact, policy can matter. In those wealthy cities, the level of automobile travel has declined quite substantially since the early 1990s. That decline is not an accident. It is the product of clear-headed policies that implemented a vision of the city where travel by walk, bike, and transit is prioritized; the recent pedestrianization of the Seine highway and the plans for new cross-Thames pedestrian bridges, for example, reinforce those policies.

How will American cities respond? Will they allow the automobile to continue to define the way our transportation system works? Or will city residents, and their representatives, fight back?


Submission to what, or to whom? Perhaps I am fear-mongering. Am I falling prey to unreasonable concerns, just as Michel Houellebecq does in his racist predictions for a future France? Perhaps the evolution of our transportation system toward driverless cars will simply mean an evolution of current conditions which, while not in any way ideal, nonetheless are, at minimum, functional.

But perhaps I am right that the increasing complication of the transportation system, and the speed at which changes are being made, really do mean that we as a public have no choice but to accept the reality that our technological masters provide us. Or, at least, we may not have the will or the means to produce a viable alternative in time to prevent the realization of that future.

* It is worth noting that this vision is not shared by all private actors. Lyft has made clear that its vision for the future is of shared vehicles, not private cars. If the public sector fails to intervene against a car-ownership society, maybe another part of the private sector will.

19 replies on “Submission”

The one word answer is no.

Because all that Tesla and similar entities are suggesting is what I call “next generation asphalt nation.” be aFor the most part, they are promoting single occupant vehicle trips. SOV trips are not optimal ways to utilize the transportation system.

While it is true that automated vehicles can be a useful addition to transit networks and sustainable mobility, and enable intra-district forms of transport (shuttles, taxi collectif, jitneys, etc.) often referred to as “first mile/last mile” trips, but also subtly different from that in what I call tertiary transit subnetworks, that aren’t cost effective to provide now because of the cost of labor, it appears as if, at least now, firms are focusing on automated vehicles as the means to transport individual riders.

Not that this can and won’t change, and that certain firms will aim to fill these kinds of segments.

At the same time, the public transport field will have to come together to support innovation and delivery of services that can serve these needs. Elected officials have a difficult time looking at Transportation System Management rather than individual types of conveyances and modes.

E.g., with the things that Uber, Lyft and others are doing with public transit organizations now, frankly, these services and initiatives seem no different to me than Montreal STM’s taxi collectif program. Nothing prevented transit authorities from trying to develop such services already, they just didn’t take the initiative.

Insightful article. I follow Tesla very carefully and have a few comments with additional info:

1.) Tesla is launching a car-sharing service called the Tesla Network next year. People that cannot afford a car or don’t want a car, will be able to call a Tesla on demand to wherever they are. Tesla owners will be able to make money on the side when their car is not needed (on vacation, at work, sleeping) that will offset the cost of car ownernership and make Tesla’s affordable for most. In cities where demand is greater than supply (owners willing to share their car), Tesla will supply their own cars to make sure anyone can access a car when they need it.

2.) Tesla is also launching an autonomous point-to-point minibus sometime in the next few years. The bus will be able to pick up multiple passengers with different destinations and reduce the cost of transit even further. It will likely replace most bus systems operating today and can have a significantly disruptive effect on public transit.

1. I think I like Tesla Network, but rest assured, Tesla rides will not be as inexpensive as Uber-X and Lyft equivalent. That means Tesla Network will be high-end ride sharing.

2. I like autonomous minibus in theme parks and other private restricted environments. But rest assured, the first autonomous minibus accident on public roads that causes death or maiming will attract a $$$,$$$,$$$-lawsuit. Even if settlements are 1/10th that amount, they will add up.

3. Millions of people will vote against any spending to convert freeway lanes to autonomous vehicle lanes.

With this thread there are a couple issues. First, I always ask ‘how is the market for autonomous vehicles different from the market for car sharing now?’ Granted there is a difference between one way and two way car sharing, but that is a developing market that only has so much oomph (see the discussion of innovation diffusion in works by Everett Rogers, and work that built upon that by Geoffrey Moore).

Note that car sharing is more successful in denser places like center cities and urbanized suburban centers and college campuses.

Second, I guess the difference in consideration of the value of autonomous vehicles depends on whether the technology is applied primarily to deliver SOV trips vs. shuttle/taxi trips. The value with capturing more SOV trips is limited but still significant, a reduction in the number of owned vehicles and the need to store them. This is an already demonstrated value of car sharing, where various studies find that each car supports between 10 and 20 households and reduces the number of owned vehicles and demand for on-street parking in congested areas (e.g., Hoboken, DC, Arlington County, VA, etc.).

The value in the latter is to enable cheaper means to serve “first mile/last mile” and intra-district mobility needs, which is an underdeveloped segment of the transit network.

The thing about capturing SOV trips as a fractional use as opposed to someone owning their own car is that this value of reducing the number of cars varies between cities and suburbs and there are plenty of other steps cities can take to encourage take up of sustainable modes besides building an ITS that enables autonomous vehicles.

Furthermore, as long as the US is a major oil producer and automobile manufacturer, policy is going to preference SOV trips. Therefore that is the focus of a lot of the discussion and policy development for autonomous vehicles.

On the contrary, Tesla rides will likely be MUCH cheaper than Uber/Lynx rides, because you don’t need to pay for a human to drive you. The human is the dominant cost.

It costs 41 cents/minute to use a one way Car2Go, plus there is a $1 charge (plus most local jurisdictions charge sales tax). A SmartCar costs less to buy than a Tesla.

More likely a cost framework is what BMW does for their ReachNow program, which is more costly than Car2Go. They dropped the service in SF. Now it’s in Portland and Seattle where because they compete with Car2Go they are using the same Car2Go pricing, at least initially. They have a mix of minis and BMWs. After the promotional period, it seems pretty likely that the BMWs will cost more to use than the Minis.

The info on the SF service is no longer available online. It was $12 for the first 30 minutes. While that is on track for the Car2Go service (which has a flat rate of about $15/hour), most of the intra-city one-way car share trips are for trips shorter than 20 minutes. (At least, that’s how we use it.) The price is comparable to the cost of two transit fares (DC has very high pricing for subway use compared to most cities).

… again, this is why I argue that the market demand for SOV autonomous vehicle trips is likely to be analogous/substitute for the car sharing market from the standpoint of own vs. fractional use. Again, Rogers and Moore’s works on innovation diffusion are relevant.

For a variety of reasons, labor in the taxi realm is pretty discounted. They don’t earn a whole lot more than minimum wage when you figure out all the costs.

So it’s not likely costs will be significantly lower, just that the pricing framework will be different.

The costs consumers will pay should be around the same on the Tesla service as current car-sharing services, but you will get increased convenience in a much nicer car that will likely be prepopulated with your favorites settings, music stations, etc.

As for the raw cost per mile, it will be less since electricity is less (sometimes far less) than gas even with today’s prices and Tesla’s require less maintenance than an ICE car. Remove the $15-20/hr for the driver on top of that and you get a service that could be much less. However, after Tesla and the car’s owner take their cut I think it will even out.

1.) Think the cost will be similar after the cut for Tesla and the car’s owners. The electric powertrain, low maintenance, and lack of a need for a driver all push the cost per mile down.

2.) There has been one fatal autopilot accident out of 300 million miles of use where the system was truly at fault, and that driver was using the system on a road where Autopilot should not have been used (he was not following directions) on top of watching a DVD while driving. Already the edge case that happened there has been patched and will not happen again, and that is with the old hardware. A self-driving bus would be at least on 3rd generation hardware with highly refined software. The chance of an accident will be dramatically less than if a person was driving.

3.) There is no need for autonomous-only lanes right now. You can easily have manual and automatic driving intermixed. In the future it might be a good idea so that the size of lanes can be reduced and speed limits can be increased.

Road use charging would go a long way to discourage the ‘hell’ scenario. Your car can’t circle indefinitely if it’s racking up peak hour congestion rates on top of the base use charge, or a logarithmic fee based on cumulative annual usage, or whatever.
This will be necessary given electric cars don’t use gas/petrol. While a flat fee is the simple substitute, I think, and hope, we’re well past that.

I’m a frequent skeptic on this, but I don’t see—in the US, at least—how we go anywhere but the “hell” scenario. Many commentators (not you) fall in to the trap of talking about how cars sit idle 96% of the time and therefore we can decrease the fleet size by 96%, right? Except demand for mobility is not exactly flat. I was biking and running around this morning at 4 a.m. and there were not exactly a lot of cars out. Yet by 7:00 the roads were full. Demand is concentrated in certain areas and at certain times, which is pretty obvious.

Then there’s how people use cars. You astutely point out that much of the market for cars is not in dense, urban areas, and that demand for vehicles is ever so slightly different there. If everyone on transit, foot and bicycle used a car instead cities wouldn’t work (road space, curb space, parking, etc; Jarrett Walker has made this point). We could price SOVs more than shared rides or transit, although if and when Uber has to make money, its fares will have to rise, unless it can figure out how autonomous vehicles work in dense urban environments, which I still think is orders of magnitude harder than teaching a car to drive around wide streets in the suburbs devoid of pedestrians or cyclists.

Luckily (?) this isn’t necessarily a lost cause. AVs may make it easier to live in the suburbs and work in the suburbs, but that won’t make it much more desirable: people are voting with their feet and moving towards denser environments, and I don’t believe that it’s solely because of suburban traffic (on the contrary, I believe that the trends would be faster were it not for half a century of suburban job sprawl). The scenario posited last week of 100 mph AV travel from the country to the city fails on several levels even before you figure out how to get a car in to the city (100 mph is quite dangerous in an open system or any weather, roads are designed for 70 mph, rarely faster, 100 mph travel is very energy-intensive, etc). So people will likely keep moving towards cities, since I doubt that Millennials are really that enticed by exurban living and a weekly trip to Applebee’s.

As for how people use cars: I live in the city. I have a car. I did the math a few years ago (it helps to have six years of personal travel data) and with the type of travel I generally use a vehicle for, it would cost more for me to use car sharing. Beyond that, there’s the fact that I keep and travel with a lot of stuff in my car. For three months of the winter, there are half a dozen pairs of skis, a blow torch, a towel, a change of clothes, a box of wax and other Nordic ski accessories in my car. I could take them in and out of my house every time I ski, but the car would have to be a lot cheaper for that to happen. (I’m not alone; my neighbor coaches youth soccer and keeps 100 soccer balls in his car.) Then there are how AVs may and may not work. If I take a trip to hike or ski on a weekend, the destination is often beyond the pale of cell phone coverage. If I end an AV trip in the end of a valley with no cell coverage, the car has to sit there until I return hours or days later. Demand in rural areas, where cars are really quite necessary, does not lend itself to a fleet of vehicles; if the wait is 10 or 15 minutes or more the market will be very small indeed. And teaching AVs to drive on washboarded dirt roads out in the sticks is probably a couple steps down the line, too.

So the long and short of it is that outside of major cities, private car ownership isn’t going anywhere. In major cities, we’ve been able to drive it down somewhat (if I lived in Oslo, I wouldn’t need a car to go hiking and skiing, because there’s good transit and a huge forest abutting the city), but that’s by providing a range of transportation options. If we wed ourselves further to the car—autonomous or not—we’re not going to make any progress towards a more sustainable future. The automobile should be seen as a tool (and a very useful one, at that) in the transportation system. But what Tesla et al want is to sell a lot of cars.

You’re right about energy costs. It takes 4 times more electricity to go 100 mph than to go 50 mph. As a society, it only makes sense burning such high levels of electric energy for public transit vehicles that transport at least 30-40 patrons, if not more.

Regarding the number of cars needed to accommodate peak demand, the best analysis can be found in this ITF study modeled using real passenger, origin-desitinaiton, timing matches for the city of Lisbon:

The bottom line:
Need only 30% of the cars if people do CAR sharing
Need only 10% of the cars if people do RIDE sharing (very reasonable wait time expectations)
Need only 3% of the cars if we use 6 passenger vehicles (this number is really shocking).

This is exactly why has published a version 1.0 of a National Strategic Framework – see my recent conversation with Ashley Halsey, transportation reporter from the Washington Post discussing ATI21’s National Strategic Framework to help the next US President advance self-driving:

Not having your august position but have discussed this issue in my blog, in the comment section of this entry.

The issue is that where “self-driving’ is easiest to implement is on freeways. It will take a couple decades or longer before street networks in cities can rebuilt to accommodate autonomous cars.

But in the freeway situation, just like with the “last mile” issue with transit, once a car moves off of the freeway network, it will have to self-operate.

But like how e-bikes are promoted as a mode for center cities, when really they are more appropriate for 5 to 15 mile trips which are more likely to originate in the suburbs, autonomous vehicles are promoted as a center city solution when they are not, except for tertiary network (taxis, shuttles, etc.) applications.

The point of a city from the standpoint of exchange and the creation of a rich sustainable mobility platform is to overall reduce the “need” for a car, or to shift people from ownership to fractional use. But the complexity of the urban road network and the semi-unreliability of the digital IT and telecommunications network–even if it is reliable 99.999999 % of the time, one system breakdown will paralyze the system, comparable to China’s multi-day hundreds of miles long traffic congestion/road breakdowns.

Analog can be more reliable than digital (e.g., the difference between air compressors and electric compressors as water pumps for the NYC Subway system in the wake of Superstorm Sandy. The air compressors continued to work despite the power outages and this was a significant contributor to the speed of the system’s ability to recover in certain sections).

The way I look at this issue is the huge amount of $ being invested in driverless technology (and the vehicles themselves) while we continue to under-invest in public transport and bike/ped infrastructure. Where I live in Boston the governor and his suburban supporters would love to kill the Green Line (light rail) extension and won’t consider completing the Community Path as an alternative to the busy street cycling corridor where 2 cyclists were killed in the last 4 months. Why? Insufficient funds since the same suburbanites have only allowed a 3 cent increase in the last 20+ years.

We hear about Tesla because that’s accessible and understandable to the broad audience. What we don’t hear about are the automated solutions for larger vehicles like buses and trucks. The operators of these vehicles would benefit even more than than the general public from AV technology because it breaks the connection between labor and cost.

By automating buses, more buses could run a fixed transit route for the same operational cost. This could make people more likely to take transit because the bus is always coming. As autonomous vehicles increase in market share we can also expect more efficient road usage because autonomous vehicles don’t get distracted by text messages or phone calls, drowsy, or slow down/speed up for no reason. This means traffic moves more efficiently. That, in turn, makes buses more reliable.

If we then consider how much society would have to accept and trust automated transportation, more transportation jobs (i.e. rail motorman) could be replaced with computers further decreasing the labor costs of transit allowing for increased frequency. In fact, many of the same video and ultrasonic data processing algorithms for cars and trucks could make it easier to adapt older rail systems to fully autonomous operations at less expense than traditional driverless systems.

With so many moving pieces, can we conclude that AVs will be a heaven or hell scenario? Can we then reliably determine the best policies to implement now, to address future problems that haven’t proven to be problems?

Legislation concentrates benefits and socializes costs, which creates special interests that defend those benefits, sometimes when the costs far outweigh the benefits. So, before we create more special interests, we need to see what problems arise. We have time to see what happens, because once AVs are priced affordably, it will still take time for consumers (both individuals and businesses) to replace non-AVs. We’ll see the problems grow, and then we’d have evidence that something is wrong, and we’d have starting point for figuring out how to fix it.

I join Ari Ofsevit’s and Jarett Walker’s argument on still highly consuming public space by individual motorised traffic; and similiar to comments on Jarett Walker’s blog article, Elon Musk just made another step to merge PRT (personal rapid transit) and road traffic by simply using publicly financed infrastructure (road) in an inefficient way (by individual other than mass transit).

For quite a lot of politicians anywhere in the world, solving the noise and pollution issues is all what we might need to do. But comparing North American settlement structures even in metropolitan areas with those of Europe or Southeast Asia, for example, there would still be a huge gap in efficient use of urban space as only efficiently organised mass public transit can cope with huge flows of passenger transport. And comparing North American smaller cities with such examples as Amsterdam (Netherlands) or Münster (Germany), there’s still a lot to do on how to make smaller cities move liveable and usable for walking and cycling…

By simple “cancelling” huge parking lots in order to replace them by either highly used buildings -or- by parks, cities could additionally being adopted more towards a mass transit oriented structure -or- at least improved air purity.

Therefore, you always need “PUSH” (see above) as well as “PULL” actions: A basic “PULL” action is a citywide and attractive public transit network 24h/day every day in the year with different levels of services depending on demand, network structure, day and time.

So, Elon Musk’s “PRT” can play a good additional role when concentrating on reducing individual car traffic, but cannot realistically replace mass public transit in any urban areas.

Driverless cars don’t solve the space-usage or congestion problems — at all. Great for solving the reckless-driver problem. But try to use one for a commute, you’ll be stuck in traffic.

And that’s going to continue to cause a strong push for MASS transportation which can move hundreds of people at once — in other words, trains.

Leave a Reply