THE SECOND COMING OF CARS: Will Self-Driving Cars End Congestion, Improve Safety, and Save the Environment?

Self-driving cars, a.k.a. autonomous vehicles (AVs), are all the buzz these days.  They are already being rolled out for real-life testing; within a very few years, sooner than anyone believed possible a short while ago, they will soon be nationwide.  Whether we want them or not; whether we are ready or not, they are moving from “Level 3” autonomy, where a human must be available to retake control, to "Level 5," cars that go on their own.  The breathless headlines announcing their arrival amplify our society’s techno-utopian impulse, with enthusiasts (and marketers) describing the countless ways AVs will revolutionize and improve nearly every aspect of our lives and society.  We are being told that autonomous vehicles will come to the rescue of our increasingly dysfunctional transportation system.  Car crashes won’t happen.  Pollution will decrease.  Congestion will go away.  Parking lanes will be turned into parks or bike lanes.  Access disparities will decrease for low-income and rural areas.

At the risk of being labeled a Luddite, I don’t believe it. 

Change: Yes.  And lots of it; much of it disruptive.  But improvement?  No more likely – in fact, probably less likely – than damage.  The only thing that has a chance of creating a more positive outcome is proactive regulation of the product and its use.  As ZipCar founder, Robin Chase, has been pointing out, we are faced with a Heaven or Hell choice.  Without successful strategies to steer us towards positive outcomes, AVs will not eliminate traffic congestion, reduce aggregate vehicle miles travelled (AVT), injuries, air pollution, or the need for parking, and may actually make it all worse.

The good news is that so much of what we need to do to maximize the benefits, and avoid the catastrophes, of the seemingly inevitable onslaught of driverless cars are the common-sense things we have already started doing because they are worthwhile under any circumstances -- prioritize modes that move the most people, cause the least environmental damage, and equalize access; build more transit, bike lanes and sidewalks; price highway access, curb access and parking; etc.  However, the coming of AVs means we have to do a lot more of it, sooner.



Autonomous vehicles (AV) may reduce traffic crashes, at least after a potential demolition derby transition period when driven, human supervised, and self-driving vehicles share the road, or in areas (such as urban downtowns) where driven cars are forbidden to go.  (The tech firms behind AVs have already begun protecting themselves: “California Limits Crash Liability For Self-Driving Cars”)  AVs will only reduce pollution, at least in cities, if they are electric – but there is no current reason to assume that they won’t use fossil-fuels.   If electric, they do open the possibility of drawing on renewable energy, but our national grid will continue to draw on coal and gas for years to come; and the total energy needed to move a vehicle depends on its weight, with electric cars not inherently lighter than gas-powered ones because of the bulky batteries – weight mostly goes down in tandem with size, and American’s seem to have an unending appetite for bigger vehicles.  

Congestion may go down if people are willing to share AV trips, as with a shared Lyft/Uber ride, but will increase if the AVs convenience and low cost are enough to pull people out of mass transit, or if individually-owned AVs are instructed to keep circling the block until their owners return from shopping or work – a “zombie car” apocalypse probably only avoidable through congestion pricing!  AVs will be affordable if they are scaled into fleets, with users only paying for the distance they travel, but will remain out of reach of the bottom 40% of our population if the Tesla model of individual/family ownership prevails.   Even if we regulate AVs to require electric fleets, keep vehicle weight and size low, and incentivize group trips, it is probable that road use (aggregate Vehicle Miles Traveled -- VMT) and parking will only go down if there is no need for the car to drive around (or park) between letting one group off and picking up the next and if lower commuting costs (and continuing increases of in-town housing prices) convince people to move even further out of town requiring even longer commutes. 

In any case, rather than replace mass transit, AVs will not eliminate the need for a vast upgrade and expansion of urban-core mass transit systems.  There is no possible way to have enough cars on our roads to simultaneously carry the hundreds of thousands of people who go to work and back home every day at roughly the same time.  And if there were enough cars, none of them would be able to move, even if their AV-brains allowed them to be mere inches apart – which is exactly how they’d end up.



Of course, there are some socially valuable functions for properly regulated AVs to play, as there is for almost all technologies.  First, they are an effective adjunct to mass transit. Transit is irreplaceable because as Jarrett Walker noted in CityLab, “an efficient mass transit network is designed to do most of what thousands of people want. It is by being mostly useful to so many people that buses and trains attract lots of riders. And carrying lots of riders is how buses and trains deliver so much liberty and opportunity to citizens while taking so little space.” In lightly settled area, AV vans holding 6 to 12 people might be a flexible and low-cost way to expand the transit system assuming they are convenient and cheap enough to attract users.  In denser areas they are an excellent method for handling the last mile problem of getting people from dispersed homes to train, trolley, and bus stations – most of which is along city roads rather than highways.  (However, without congestion pricing or other disincentive, people who hire an AV to get to the train station are likely to skip the train and stay in the car all the way downtown.)  Some of the AVs used for this can later serve to bring off-peak commuters to work at times when mass transit isn’t scheduled to run, preferably as a group/shared ride.  Other AVs can remain in the community, both within city neighborhoods and in suburban towns, for local trips.   All these uses fit into the Uber/Lyft and taxi model of short rides. 

AVs might even be useful for long, family trips to places not well served by train or bus lines – fitting into today’s car-rental business model of full- or multi-day use.  However, one-way trips to rural areas, where it is unlikely that a second user will want to hire the vehicle within a short time after the end of the first person’s trip, will probably invoke a surcharge.  And the long time it would take for a car to get from its main activity area to a rural user may discourage that kind of use – unless rural agencies such as Regional Transportation Authorities (RTAs) own a small fleet based in their area.

Looking beyond cars, AV buses might be another way to merge the technology into mass transit, particularly if cities establish bus priority lanes and intersections.  This might free up the bus (or trolley or subway) staff to focus on customer service – we should not allow the MBTA to use AV as an excuse to leave bus riders without any staff support during their trip. 



Whether future AV fleets are owned by public agencies or private firms is not ultimately the issue, although public benefit must shape the context and market for profit-seeking.  A broad coalition is beginning to coalesce around “Shared Mobility Principles for Livable Cities”. In addition to proactively requiring that all AV’s be electric and creating a variety of incentives for them to be fleet-based rather than individually owned (or putting prohibitively high excise and registration fees for non-electric or individually-owned AVs), there are other steps that government should take to lay the foundation for the successful introduction of this new technology such as raising the gas tax and imposing a carbon-emissions fee.  We need to remember that 37% of Boston’s households do not currently own a car.  The mobility needs of these people need to be a central concern for policy-makers.

Creating bus-priority lanes and intersections has already been mentioned, but expanding the network of High Occupancy Vehicle (HOV  -- preferably requiring 3+ passangers) lanes for cars with more than one person is another reform that will be both helpful today and set a good direction for the future.  Imposing a VMT tax on non-electric or individually-owned AVs will help reduce their wasteful use.  Congestion pricing has already proven to reduce down-town congestion and will help keep individually-owned AV’s from undercutting transit, if combined with a “time-in-the-area”  fee it could also encourage owners to send their empty cars out of the downtown to park.  This should be complemented with demand-responsive (market-rate) parking fees in commercial areas that will reduce today’s circling-the-block problems while keeping future AV vehicles from stuffing up spots as well. 

More fundamentally, roads need to become much more people-oriented rather than vehicle-focused.  Pedestrian space needs to be expanded along with safe spots for sitting, playing, and socializing.  Intersection’s need longer Walk signal times.  Bike facilities, with protected bike lanes on streets with high car volumes or speeds, need to be an automatic part of every street repair.

No technology will, by itself, solve our society’s growing inequalities.  But that’s no reason to allow it to get worse; we need policies that will ensure equal access to AV fleets in low-income and minority areas and that improves their affordability for people whose budgets are already stretched thin.

Cities, rather than the state, own most of the roads and all of the curb-space.  They need to use that ownership to impose controls on the rapidly approaching AV era.  We need to rethink – probably expand -- the number and location of truck and shared-vehicle unloading spots.  We need to “price the curb” to generate the revenues needed to monitor the evolving situation and implement improvements.  But we should also more boldly start putting a price on use of the travel lane through congestion pricing, permit fees, and other mechanisms.

Finally, we need to keep shaping the land-use context that shapes our transportation choices by doubling down on smart growth zoning, encouraging more mixed use developments where people can live and work nearby,  steering development towards denser transit-accessible locations, and requiring all firms with more than 20 employees and all retail locations with more than a half-million dollars in gross sales to implement strong Transportation Demand Management plans to reduce single-occupancy-vehicle use by staff and customers.



Frequently, the introduction of a radically new technology triggers a similar cycle.  First, it becomes fuel for the techno-utopian impulse, with enthusiasts (and marketers) describing the countless ways it will revolutionize and improve nearly every aspect of our lives and society.  Then comes the business and power-protection filtering, where the technology gets shaped to fit into the most profitable uses that do not upset the larger hierarchies of our social system.  Third is the roll out, when the transformative impact either manifests or (at least temporarily) disappears while its use becomes increasingly integrated into daily life. If the technology does take off, does become a commodified normal part of everyday activity, people start pushing its boundaries, using it for purposes its commercial pushers never contemplated – and perhaps turning it into a tool for greater equality and general well-being after all.

Let’s hope that with Autonomous Vehicles we can run through the cycle quickly.


Thanks to Robin Chase and Tony Dutzik for comments on previous drafts.


You may also find these previous, related blog postings interesting. 

> Smart Cities, Power Politics, and Quality of Life:  Technology and what it’s used for

> Redefining Transportation:  From Moving Vehicles to Place-Making   

> Digital World Creates New Road Functions & Requires New Design Standards

> The Interacting Ingredients of Livable Cities:  Complete Streets to Interior Design, Transit to City Planning, Art to Education

Showing 7 reactions

Please check your e-mail for a link to activate your account.
  • Steven Miller
    commented 2018-02-09 13:21:06 -0500
    Again, good points bringing out some of the nuances and hidden pitfalls.
  • John Allen
    commented 2018-02-09 10:33:14 -0500
    Steve said — “Even worse, I think the reduction of crashes is only attainable when/if there are ONLY computer-controlled cars on the road — my own experience in high tech makes me uncertain that the interaction of humans and machines will go well because humans are more unpredictable than any program can anticipate.”

    Yes, indeed, but — if there are only computer-controlled cars on the road, then no bicyclists, and no pedestrians!

    There is a give and take in interaction among humans, whether behind the steering wheel or the handlebars, or on foot. This involves at times, stretching the rules. Example: motorist inches out from a driveway where parked vehicles in the street make it impossible to see approaching traffic. The driver entering from the driveway is supposed to yield right of way under the law, but as a practical matter, drivers in the street have to yield. There are many other examples, ranging from understandable and necessary like this one, to others which amount to outright or generalized intimidation (e.g., failure of motorists to yield to pedestrians in a crosswalk, risky close passes of bicyclists).

    I don’t think that automated vehicles are going to be very good at this game, because they must err on the side of caution to achieve the greater safety which is promised for them, and to protect deep-pocket corporate entities from lawsuits. Consider another scenario: it’s trash collection day. Gust of wind blows an empty trash can out into the road, blocking it. Human driver stops car, turns on 4-way blinkers, carries the trash can to the curb, gets back into the car and proceeds. Driverless cars can’t do this. Pedestrians also could throw trash cans into the street and block traffic, but this sort of action is rare, seen only in political demonstrations. With driverless cars, the usual game of motorist intimidation (or courtesy) and pedestrians looking for gaps and calculating risk fails. Blocking traffic becomes more practical. Preventing it becomes a law-enforcement and civil-rights nightmare, requiring infringement of privacy, going as far as to require people to carry electronics which can be queried to identify them and their location.

    Getting back to your point — yes, automated vehicles, even in platoons like railroad trains, could work on limited-access highways, increasing capacity and lowering crash ratesI even think that they can work when there are also human drivers on thsoe highways. We already have platoons of human-operated trucks But I don’t think that driverless cars are going to solve transportation problems in urban areas. You just can’t get rid of people , or trash cans, there.

    On the other hand, crash-avoidance technology can be very beneficial. Sensors which are coming into use today can prevent rear-end collisions and right-hook collisions.
  • Steven Miller
    commented 2018-02-08 23:00:22 -0500
    Hi Steve, nice job.
    I don’t think of you as a Luddite for wanting to control a technology with such potential, some of it for ill. But then again, I always thought that the Luddites have been unfairly treated. Given the social disruption the technology change in industrial looms caused, there was lots of reason to control that new technology. The problem is always whether it is possible for government to control a middle ground strategy to take advantage of a new technology for social benefit without being captured by the political pressure to either be total abolitionists who refuse to use the technology at all, or profit maximizing boosters who don’t care about the social damage at all, and prefer to neither mitigate it nor offset it with programs to compensate the losses caused by the new technology.
    The idea of seeing the city as the level of government that might have more motivation to regulate in the public interest, and which controls the streets where some of the issues will play out, seems sound. But the technology enthusiasts are already ahead of the game in seeking to establish the national level as the place for minimalist regulation to maximize production industry benefits. So, it is important to not ignore the federal level.

    The potential to introduce pay to drive insurance, road charging, and vehicle-based certificate of entitlement charges for AVs, while they don’t yet exist as a political force is very sound, and urgent. Once any significant number of people individually own AVs there will be the same opposition to change that causes inaction on these issues for the traditional vehicle fleet.

    I still think that the very long period when there will be a mix of traditional vehicles and AVs on the road is not enough considered, as you mention. One way to deal with the problem of the risk that national action crowds out public interest regulation at the city level could be to focus on safety as the primary issue to justify intervention, and require that close tailgating, and excessive speeding, that are celebrated by the enthusiasts, should be prohibited and regulated to reduce accidents. If there were a strong discussion of safety issues, especially for the inevitable period of mixed operation of AV s and traditional vehicles, there might be a better chance to not be stampeded by industry enthusiasts, and get a serious multi-jurisdictional regulatory schema in place.

    A slight disagreement that I have with your article is the underlying presumption that congestion pricing will solve many of the potential problematic aspects of the technology. Pricing systems are inherently biased towards the interests of those with high income, and much of the potential for harm (like routing vehicles through neighborhoods where congestion may be lower than on main streets) ought not to be permitted at any price. So I would caution against the belief that it will be politically feasible to control road pricing to consider social concerns beyond revenue maximization.

    Anyway, thank you for a readable and thoughtful article.


    I’m glad you found the blog interesting. I don’t think of myself as a Luddite (a group that, as you point out, is very unfairly maligned) except in the sense of deeply mistrusting the techno-utopian claims that some new technology will “solve everything.” As a former high-tech person, I am deeply respectful of the importance of good tools and the disruptive power of new business relationships. But I am skeptical that either of them will solve the most important issues of our (and probably every) time relating to justice, equality, freedom, order, meaning, and love.

    I also share your concern about federal pre-emption. The Republicans are already using it on the state level as well to keep cities (often run by Democrats) from passing minimum wage bills and other positive measures. As you suggest, safety may be the only handle that we can use to get around federal limits, although I fear that safety is itself too limited a scope to allow us to shape the way AVs eventually fit into our society. (I have to confess that the Republican idea that they are engaged in a prolonged war with the existing dynamics of our economy and government totally explains - and perhaps even justifies- their willingness to break every existing rule and use every mechanism of power available to them. Were a truly radical left movement to take power, as Trump and his corporate-reactionaries have, I would only hope that they would be as bold and successful in pushing past the boundaries of the past.)

    As for congestion pricing: yes; it is like all pricing-focused forms of rationing inherently biased towards the rich. But in my mind there are two legitimate criteria for accepting that fact in particular situations. First, if you are dealing with something that is generally socially negative — smoking for example: I have no qualms about raising the price of cigarettes. Second, if there are more equitably available alternatives or mitigations — making car driving more expensive is ok if cushioned by robust and affordable transit as well as tax refunds for low-income families (which economically offset but still maintain the congestion-pricing disincentives).

    Finally — would it be ok with you if I copied your comments to the Comments section under the blog on the web when I get back to town a week from now? I think this discussion raises additional points worth sharing.

    Hi Steve,
    Thanks for your response, and of course you should feel free to resend it to anyone you want.

    On the pricing issue. I certainly agree with you on pricing tobacco products to reduce consumption, without worrying about being discriminatory towards the rights of lower income people to poison themselves as freely as higher income people. There are huge external costs imposed on society by tobacco, beginning with second hand smoke victims in the household.

    But when talking about the imposition of road pricing to encourage reduced driving. There are other considerations. Most driving, other than joyriding, is not about driving, but about access to opportunity, and unfortunately no American cities, with the possible exception of New York, are equally accessible by public transportation as by auto. Refunding the money collected to low income people doesn’t restore the accessibility deprivation imposed by road pricing. Only substantially improved public transportation can do that.

    When "Red Ken " Livingston introduced congestion pricing in central London, the cost of parking was so high in central London that he couldn’t significantly hurt the auto accessibility of moderate income people, because they had no auto accessibility to the center of London to begin with. He also put all the proceeds from road pricing into improving bus services, which were already very good, and used some of the freed-up space to provide wider sidewalks and provide better bike safety, so he improved the accessibility of all by non-automotive means.
    But there is a dangerous theory among some anti-auto advocates who believe that all we need to do to improve our accessibility is to restrain auto demand WITHOUT IMPROVING TRANSIT CAPACITY. (They must never have tried getting onto the Green or Red or Orange Line near rush hour). I am fully in favor of imposing pay to drive insurance on AV s, and Uber et al, because that is just a reasonable way to make more visible the cost they are imposing on the safety of everyone, in a better targeted way. But there is unlikely to be enough money from road pricing to invest in serious capacity expansion of transit.

    In London, Crossrail is being funded by a combination of an extra business property tax in most of Central London, a national government contribution that is probably about the amount that the national government will collect in added income taxes from construction workers who are employed building Crossrail, and by a contribution from higher fares to reflect the increased value of transit access to Crossrail will produce. Road pricing is not a significant part of the funding, and the fight from auto interests would almost certainly insist that all road pricing proceeds “should” be spent on road expansion, not transit.

    If instead we focus on gasoline taxes and carbon taxes, as you suggest, it is clear we are recognizing the negative externality of added fuel consumption, and shifting auto purchase demand towards ultra fuel-efficient cars, and funding transit is a much more clear mitigation of the externality of excessive fuel use, and a means to increase the transit accessibility of the city. As in the case of London, and nearly every European city you may visit, clear regulation of street space to provide wider sidewalks or pedestrian zones, reserved lanes for bus or LRV services, and improve bike safety is a much more straight-forward way to improve non-auto accessibility, than engaging in the political fight for road pricing. The fact that road pricing advocates actually propose to LOWER gas taxes to offset the impact of road pricing on drivers as a political strategy to make road pricing politically feasible only makes it more clear that what the road pricing advocates believe is politically feasible is actually substantively useless, providing NO added revenue for expanding transit accessibility.

    Finally, serious levels of carbon pricing or gasoline tax increases would actually not be fully borne by the driver, but would be partially absorbed by the petroleum producers. This is seen as a disadvantage by those who prioritize increased pain on drivers as the solution. But for those of us who think that the primary purpose of a tax is to collect money to spend on better public services, having some of the money actually come out of petroleum producers instead of drivers is a good thing, that we should calculate and publicize. The Koch brothers and petroleum interests understand this, that is why they oppose carbon taxes and gasoline tax increases. But the driving public does not understand that the petroleum producers would bear part of this cost, so there is no political support generated from what should be a major selling point for carbon taxes.

    To reiterate and summarize , I share your distrust of the technology enthusiasts who see the A V as some sort of magic solution, I just also mistrust the road pricing enthusiasts.

  • Steven Miller
    commented 2018-02-08 22:49:17 -0500
    QUINTON: I am less worried about the technology than about the markets that will shape its implementation. There is no such thing as a pure, “free” market — every financial (and other) interaction is shaped by the surrounding context’s differing empowering of the participants. Someone is always stronger, more informed, with greater resources to buffer risk, with more support from society. That is why we need to keep pushing for democratic regulation of our economy, not centralized command and control, but proactive shaping to at least increase the odds that things will move in a sustainable, equity-enhancing, and quality-of-life-improving direction. Of course there will be unintended consequences of this kind of intervention, some good some bad. But allowing profit-seeking to be the sole driver will produce even worse consequences for ourselves and our world.
  • Steven Miller
    commented 2018-02-08 22:42:35 -0500
    John: I have to confess that I never thought of the problem of smelly previous occupants or scary shared users. Even worse, I think the reduction of crashes is only attainable when/if there are ONLY computer-controlled cars on the road — my own experience in high tech makes me uncertain that the interaction of humans and machines will go well because humans are more unpredictable than any program can anticipate.
  • John Allen
    commented 2018-02-06 12:18:02 -0500
    Good thoughts, Steve. Interaction with humans is central to whether autonomous cars improve or worsen conditions. There is a supposed reduction in congestion through car sharing, but who will want to get into a car where the previous user may have been a drunk who (thankfully) wasn’t driving, but puked all over the seat? Or carpool with strangers and no driver? I know that techno solutions have been advanced for these problems, but I’m not convinced. I do see the ability to avoid crashes as an advantage — and it comes with technology that is already being implemented in human-driven cars.
  • Quinton Zondervan
    commented 2018-02-01 13:52:30 -0500
    Thanks Steve for another great blog post. I’m less pessimistic than you about the technology; I’ve been watching it closely for over 20 years and as a software engineer I’m certain AVs will be far better drivers than humans :-) That said, I agree with the need to regulate the technology; Uber/Lyft and Airbnb are recent examples of what happens when we fail to do so effectively: we get chaos and unintended consequences. And as you said, most of the regulations should be done anyway, like congestion pricing and a mileage tax, which we’ll need for EVs anyway. How do we move the ball forward in Cambridge?