Self-Driving Cars: A Coming Congestion Disaster?

 

We're starting to see professional reports echoing long-standing concerns about how driverless cars will affect our cities. This new one from KPMG, in particular, is getting a lot of press.  It's actually a focus group study about the transport desires of different generations, but it confirms the thought experiments that many of us have already been laying out for a while.  

Much depends on whether these cars are owned or spontaneously hired like taxis, Uber, and Lyft.  A taxi model is definitely better in its congestion impacts, but that doesn't mean it will happen.  The ownership model is closer to the status quo, and the status quo always has enormous power.  Driverless taxis will not always be available on demand, especially in suburban and rural areas, so a legitimate fear of being stranded will make people in those areas prefer the security of having a car just for them. And of course, that's just the effect of rational concerns about relying on taxis.  Less rational desires for car ownership, as an expression of identity or symbol of liberty, will also not vanish overnight.

This leads to a nightmare scenario that University of Washington's Mark Hollenbeck laid out in our recent Seattle Times panel.  Paraphrasing Mark:  A suburban father rides his driverless car to work, maybe dropping his daughter off at school.  But rather than park the car downtown, he simply tells it to drive back home to his house in the suburbs.  During the day, it runs some other errands for his family.  At 3 pm, it goes to the school to bring his daughter home or chauffeur her to after-school activities.  Then it's time for it to drive back into the city to pick up Dad from work.  But then, on a lark, Dad decides to go shopping at a downtown department store after work, so he tells his car to just circle the block for an hour while he shops, before finally hailing it to go home.

This is really easy and obvious behavior for a driverless car owner.  It reduces the number of cars someone needs to own, and reduces pressure on inner city parking, but would cause an explosive growth in vehicle trips, and thus in congestion (not to mention emissions and other impacts).  Just the commute behavior doubles car volumes, because the car now makes a two-way trip for each direction of the commute, instead of just one.  And if everyone shopping downtown has a car circling the block waiting for them, well, that level of congestion will far exceed what's generated by cars circling for parking today.  It could pretty well shut down the city.

This is the good old problem of induced demand, which is what happens when you make a resource available at an artificially low price – as we do with most urban roads today.  If you don't pay the true cost of something in money, you will pay it in time, and that's what congestion is.  (It's also why in the old Soviet Union, people spent hours waiting to buy bread: Soviet price controls made the price too low to compensate the suppliers, so there wasn't enough bread, so everyone waited in line.  Congestion — waiting in line to use an underpriced road — works the same way.)  

Pricing of some kind will be the solution, but we tend to do this only when things get really bad.  Notice how bad congestion has to be today before solutions like toll lanes and transit lanes are finally accepted as necessary.  

As always, the very worst scenario won't happen, but some really bad ones still can.  If the economic functioning of downtown is too badly impaired by driverless cars circling the block waiting for their owners, the government will intervene to save the economy, as it always does, probably with some kind of downtown street pricing on the London or Singapore model.  But this only happens when congestion threatens the economy.  That's a very high bar.  Long before that point, congestion will be bad enough to be ruining people's lives, wrecking the urban environment, strangling public transit, worsening climate change, and so on.  

As always, the scary thing about congestion is how bad people (and therefore governments) allow it to get before they start making different choices to avoid it.  The level of congestion we (justifiably) complain about is much lower than the level that we choose to tolerate, and this is the real reason for pessimism about how bad congestion could potentially get, if driverless car ownership — like cars today — are so massively underpriced even in the context of high urban demand.

44 Responses to Self-Driving Cars: A Coming Congestion Disaster?

  1. AvgeekJoe November 25, 2015 at 2:46 pm #

    I’m happy Jarrett you’re stepping into this debate. Driverless cars powered by cheap energy are just going to compound congestion problems. Originally as a disabled person who can’t drive, I thought driverless cars would even the playing field. Per your thoughtful blog, guess not.

  2. Paul Kidd November 25, 2015 at 4:05 pm #

    A friend of mine wrote a paper with some more detailed modelling suggesting there would be huge increases in congestion associated with autonomous vehicle take up, but that things would get much better if the fleet were to fully switch over. Interesting reading:
    http://transposition.com.au/research/AutonomousVehicles.pdf

  3. Carl Youngblood November 25, 2015 at 7:01 pm #

    I think there are strong economic incentives to prevent this from happening. Far fewer people will own cars outright when then can instead rent access to an on-demand fleet.

  4. Rob Fellows November 25, 2015 at 7:06 pm #

    Perhaps people are doing deeper thinking than I’m aware of, but my guess is that most of the modeling done has no behavioral basis, and is just modeling the expected vehicle movements. It reminds me of the day when urban freeway planners calculated capacity needs by drawing desire lines from homes to central business district worksites and failed to factor in all the shifts in land use, business location and travel behavior that occurred as people maximized the value of personal mobility provided by cars and urban freeways – with disruptive impacts the next generation has spent trying to repair.
    I wonder whether autonomous vehicles will have a similar scale of unexpected impact. I think we need to think more deeply about how commuters, teenagers, shippers, retailers will change behavior to avail themselves of new value from autonomous vehicles … and the ripple effects that might take place as the dynamic land use and transportation systems get rebalanced. Until we’ve got a behavioral model, I think all the modeling will give us wrong answers.

  5. Paul Godsmark November 25, 2015 at 7:44 pm #

    It’s all about policy. If a city desires to reduce congestion then it will put in place policies to promote ride-sharing of fleets of shared autonomous vehicles, and discourage private ownership. The transit community, more than anyone, knows how powerful policy can be in steering transportation towards a city’s desired goals. My own calculations suggest that if a city can achieve an average vehicle occupancy of 1.8 during peak hour flows, as opposed to the current 1.1-1.2 in most cities, then even allowing for a considerable degree of induced demand, congestion reduces. Combine this with new modes, such as pods that combine into road trains, and the situation will improve further.
    But if a city sets no policy, then we may see continued high levels of private vehicle ownership, which as noted in Jarrett’s post, is not such a great idea.
    Interestingly enough, how a city deals with the Uber issue may well indicate how successful it may be in the future when negotiating the autonomous vehicle policy minefield.

    • Jerry Wheeler March 18, 2016 at 11:10 pm #

      Should ride-sharing in a large city become the norm, shall we say mandatory, would not there be massive problems which occur such as: 1.fellow passenger sick an infects all companion commuters? 2. fellow passenger is a criminal packing a gun and robs all companion passengers? 3. fellow passenger is insane or rapist and maims, kills or rapes other riders? 4 to infinity. write your own horror possibility.

      • dan June 1, 2016 at 8:44 am #

        “Ride Sharing” does not mean what you think it means.

  6. EJ November 25, 2015 at 8:30 pm #

    The answer is to bake more bread.

  7. Michael Abrahams November 25, 2015 at 8:39 pm #

    Disaster? Ridiculous.
    These problems are engineering questions. They have such easy fixes, a disaster could never happen.
    I find your view of policymakers extremely pessimistic. “Affecting the economy” doesn’t mean “directly affects tax revenues.” Governments intervene when people complain asking for things to be done differently. If the traffic engineers make stupid decisions when they design the system, people will ask them to fix it.

  8. sasha November 25, 2015 at 10:20 pm #

    Even if we assume the taxi model prevails over the private ownership model, to achieve the kind of response time to hailing a vehicle people will want/expect, is going to require lots of empty cars circulating around, ready to pick someone up somewhere at a moment’s notice. I don’t see any way persons per vehicle trip can rise above what it is currently with the amount of deadheading driverless cars will generate.

  9. el_slapper November 26, 2015 at 12:51 am #

    @EJ : if the answer is to make more bread, bread, in that case, is road space. Which is horribly costly in current urban environments.

  10. Gl November 26, 2015 at 1:32 am #

    Perhaps instead of instructing his car to circle the block while he is shopping he can make some extra money by having it accept rides from the public. Perhaps that can even be enforced by traffic laws and software: any self driving vehicle traveling without any passengers must be automatically set in “taxi mode”.

  11. Jeff Wegerson November 26, 2015 at 2:06 am #

    Two major traps to predicting the future are expecting some things to work the same and the others that work differently get expected to work rationally.
    I expect that all automatic-automobiles (auto-automobiles) of the future will be electric, so we can cross off extra pollution to the induced congestion of self-driven cars.
    Electric cars also have the issue of paying for infra-structure since they do not pay gas taxes (as if that covered it all anyway.) That brings us to the under priced resource of highways and streets built for cars. In budget challenged Illinois the one part of government that runs a surplus is the tollway system entirely in the suburbs. Now everyone who drives in the suburbs has a RFID transponder. You see where that is going right.
    It is crazy to expect a free (well in a way) roadway system as well as free parking. While we just got a 90 miles of additional lanes one one of our tollways, there was no issue paying for it. Since the new lanes are in the ex-urbs I am pretty much blah about it. The exception is that the shoulders were upgraded for bus usage when congestion slows traffic below 35mph. I am looking forward to good utilization of that potential resource.
    Where AA cars come in is that we have to assume that every AA car will have not only a transponder but also regular network connection. This will allow for congestion charging anywhere on the one hand and intelligent parking on the other. For non-private or at least corporate-owned AA cars after dropping off their passenger(s) they will proceed to as close as possible to where they will be needed next. That might be to a stadium where the game is about to be over or downtown where rush hour is beginning or out to some suburban street to wait the morning rush. When I say needed next I do not mean some place they have been called to. Rather I mean to a place where statistics predict they will be needed.
    Speaking of transponders, my guess is that the will become personal rather than vehicular. That way a car with more than one person can be split or charged less even.
    Of course a lot of the above falls into the “rational expectations” future. In many places rationality might be trumped by privately owned rental reality.

  12. twitter.com/bradtem November 26, 2015 at 2:09 am #

    Having a car make a long return trip to the Suburbs is not practical if the trip is long. Shared cars which get heavy usage (which this is on the cusp of, though it’s not shared as much as a taxi) wear out by the mile, not by the year, and you think of their cost more as a per-mile cost. If you double the commute distance, you double the cost of the commute. The commute travel is a significant fraction of the travel a typical car will do.
    Generally, it will be cheaper to own two cars than to send your car back home if the commute is moderate to long. Or cheaper to make use of robotic taxi services at the home rather than send the car home. The robotic taxi services will be cheaper than sending your car back in almost all cases — remember they don’t have to pay drivers. You just pay for relatively inexpensive parking (the vehicle does not need to park precisely where you work, unlike a human driven car, and so can shop the spot market) and use the taxi service at home if available.
    Of course, if you do still want to spend the extra money to send your car home to do service there, you will be doing it in the anticommute direction, where road capacity is plentiful and congestion is minimally increased.
    So what’s the issue?

  13. Richard Gadsden November 26, 2015 at 3:38 am #

    Something that a lot of car owners use their cars for is as a portable storage facility. They have a quantity of their “stuff” in the car that it’s helpful to have available sometimes – but the need is difficult to predict.
    Thinking through examples from people I know, there’s a change of clothes (in case you get dirty/wet), driving shoes (very few women can drive in heels) and gloves (some people prefer to drive in them), umbrellas, warm clothes/coats (for if the weather changes), and even things like overnight equipment (tent, food, water, blankets, sleeping bag) for people who may get trapped away from home in the car (e.g. if it breaks down).
    The amount of convenience lost by not being able to leave stuff permanently in the car is not zero, and at least some people will pay for that privilege.

    • Jeff March 11, 2016 at 11:18 am #

      The convenience is the big thing. If you have kids you keep extra clothes, diapers, toys, car seats, etc in your cars at all times. If you do not own a car you will need to have all of that ready at all times? Lets say your kids play sports. Every time you go somewhere you have to repack up an AA? That seems absolutely ridiculous to me. I think every family will need to have their own car. Another big thing will be the cost in the long run. People can buy a used car for as little as 3 grand. If you commute 20 miles to work everyday, 40 miles round trip, at $0.50 per mile, it would cost you 5 grand a year to go to work. I assume that will go down a bit in price, but how much? At that price it is still better to own a car in the long run.

      Also cleanliness of a ride-sharing autonomous vehicle is critical. You will have people leaving used Kleenex, water bottles, pop cans etc. Plus what if the person before you had the flu or some other kind of contagious disease. It will increase the spread of illnesses. Will they be able to self clean? I know this isn’t much different from public transportation, but I know plenty of people who refuse prefer not to use public transportation for this reason.

      For all these reasons I just do not see the practicality of the ride-sharing vehicles. I can see it being a niche business but not an everyone usage.

  14. Jackrhoo November 26, 2015 at 5:23 am #

    So the argument is that congestion will increase to the max, because driverless cars will be very cheap to run?
    I disagree that they will be. ELECTRIC cars will be cheap to run, and of course the saved driving time will be a huge economic benefit. Also parking can be automated, and can even be located some distance away form need, so can be much cheaper than present.
    But the scarce resource is what will be charged for, and that means the road space itself, and suggests that toll roads are almost inevitable, unfortunately.
    In that case, everyone will end up paying per mile, regardless of owning a car or not, so the incentive to own one for cost reasons will disappear – it will be a luxury choice, although possibly a popular one.
    The real expense will be actually using the roads. And to sell that I imagine the marketing and pricing will be similar to mobile phones, where you get packages of free miles in a certain level of car, then pay per mile after that, etc. Possibly there will be multiple levels of tolls too, where some routes are congested and others not.
    I guess it depends on how far different countries fall for the free-markets-are-king mantra.

    • Jerry Wheeler March 18, 2016 at 11:21 pm #

      How do you think driverless cars would work out in bad weather – hurricane, tornado, flooded roadway, crossing a railroad, avoiding a person walking across a street, elementary students commuting to and from school. I personally view driverless vehicles as insane in the not so long haul.

      • Mike August 26, 2017 at 5:49 am #

        I share your views there, Jerry. It is insane for all the reasons outlined here and then some. It’s amazing how lazy we’re becoming as humans as it’s now to the point where people can’t even be bothered to drive themselves around anymore and are begging for cars that will do that for them.

        On top of everything already mentioned here, these cars will be run entirely by computers and ALL computers on the face of the planet are hackable. From PC’s to phones, tablets, laptops, servers, TV’s, A/V receivers and the list goes on; all hackable. People want to put their lives completely in the hands of a computer on the road? This alone has the possibility to cause accidents several times bigger than we’ve ever even imagined before and certainly ever seen before. These cars have the potential to cause far more deaths on the roads than we see currently and it’s not even really close and all this because people feel life isn’t convenient enough for them. It’s pathetic IMO.

        There are always unforeseeable problems with things. We still have big problems with traffic and roads now and how long have we had to fix all that? Still not fixed and we’re about to introduce more variables to the equation? Besides, there are plenty of people that love to drive their cars and I don’t see that going away any time soon either.

        I could see self-driving cars as being a viable taxi service especially for the elderly and people who can’t safely drive themselves around. For them, these cars would be a godsend. Going down the road where people will ultimately be forced to use these cars is not viable IMO. These cars will have a demand, but trying to force it further than needed isn’t a good idea at all.

        There’s also the privacy aspect. Doesn’t the government already spy on us enough already? Do they or anyone really need to know where I am at all times? With the way the world is going, having these cars connected to the internet poses even more concerns. With Facebook now using AI to police content on their site to seek out “extremist” content, what if something you posted on there is wrongfully flagged as extremist or terrorist content? That car you’re driving in that you have no control over could be locked down and forced to drive to the nearest police station to be questioned. That’s just one possibility in the future.

        The price for convenience is likely to be far, far higher than anyone really anticipates. It’s pathetic. Pretty soon, we’ll have little vehicles to drive us around the house automatically too because we can’t be bothered to walk. It truly is pathetic what humans are becoming.

  15. Eric November 26, 2015 at 5:26 am #

    I think this discussion parallels the discussion of through-running for rail. If most of the LIRR’s morning passengers want to go to Manhattan not from Manhattan, is it worthwhile continuing the trains to New Jersey or else parking them in Manhattan? The competent professional opinion seems to be that continuing to New Jersey is preferable. Similarly, it would seem, it’s preferable for cars to drive themselves out of downtown rather than needing downtown parking lots or garages.
    This seems especially true in North American cities with street grids. Since there are so many parallel streets in a grid, technically capacity to/from downtown is well over demand. Human drivers prefer a faster and smoother freeway ride, so the freeways get congested, but there is no reason why a self-driving car couldn’t slowly make its way out of downtown on local streets.

  16. Kevinklink November 26, 2015 at 5:45 am #

    It’s very hard to take that KPMG study seriously. They have a very small sample size, admit themselves it’s not scientific, and then they top it off by using “generational” labels, which is about as meaningless for real data as you can get.
    It’s also very hard to understand why we think that usage of autonomous vehicles will really be that much different than Uber. The big tech breakthrough has been on-demand vehicles via smart phone. Whether there is a human or not behind the wheel seems to really be a marginal difference, other than that it will allow on-demand companies to likely have more vehicles operating and thus mor coverage.

  17. Jack November 26, 2015 at 7:04 am #

    Topical! For pundits to see disaster in anything new. Journalist get attention by feeding the fear of the unknown. The savings of life, limb, time and property will be of enormous value. While the “old generation” will be slow to relinquish control (ownership), it is silly to suggest that transportation service will not ultimately beat the ownership model. It is equally silly to suggest that the service model will lead to more congestion than the status quo. We can reduce the number of vehicles by 90% and still provide a better experience. V

  18. Tim Kramer November 26, 2015 at 8:54 am #

    I see this article as (maybe) just a bit above click-bait as it vocalizes a “foreseen” problem but doesn’t address possible answers. I’m not a public transit professional but I can think of a couple solutions. (I don’t claim that these are actual solutions.)
    One obvious solution is already in use: set up (public or commercial) parking lots where driver-less cars can go to “rest” for a small fee. Another would be to set up monitoring and ticketing to prevent use of the highway as a moving parking lot, to address the “circle the block” issue. If a vehicle is moving, but unoccupied, it should be coming from or going to a defined location. Otherwise, issue a ticket to the vehicle’s owner.
    I’m disliking this article because the author is self-defined as a professional, is griping about a yet-to-be-realized problem, and doesn’t offer discussion of multiple, possibly valid solutions. “Pricing” is a short-term stop-gap solution that is subject to politicking and impacts the people who are in most need of self-driving cars: the blind, the elderly, and the physically disabled. “Pricing” pushes those people back into the margins.

  19. Eric November 26, 2015 at 9:01 am #

    Pricing is the important solution, but if it is done, then the regions that pull it off will certainly start looking better. So the question becomes, what will those locations look like? What does the solution look like, if pricing signals are allowed to provide feedback and impact decisions?
    It is impossible predict what will happen, but I would suggest the solution would take advantage of transit. Consider that with pricing signals taxi fleets will be locked in a race to provide the most efficient baseline service they can, which means ever lighter and more energy efficient solutions that meet the demand “just enough” in the mobility mix. It is all about what becomes the most convenient of the cheaper alternatives you can scout in real time from your mobile.
    Part of the solution obviously will really depend on where you live. We could have some places within cities completely leapfrog automated taxi predominance with services that utilize high frequency transit as the backbone and that interface fluidly with something readily available in the mix like the MIT Mobility Lab’s aptly named “Persuasive Electric Vehicles“. How behavior can adapt in time is what I’m thinking most about. I believe the factor of transit is one that is very important, because where it exists and is effective, it could pull the “just enough” services into the mix. All these have to do is to provide a really convenient mobility mix in good locations. Transit will already be there for upstarts to exploit in their quest to quickly win customers from taxi fleets.

  20. Davidseattle727 November 26, 2015 at 9:09 am #

    It’s terrifically important for us to contemplate scenarios, both negative & positive.
    But bottom line, we haven’t any real idea about this technology will emerge, how it will be offered and how it consumers will use it.
    For example, it’s quite conceivable that at the outset (say) Google will ONLY lease to fleet operators so that self-drivers are ONLY used as part of taxi fleets. (For security reasons, Google might decide that it wants to keep as much control as possible so won’t sell to individuals but insists on fleets where it can physically inspect the cars on a daily basis.) That sorta changes your scenario, Jarrett.
    So while it’s important and useful for us to spin possibilities, there are NO certainties at the moment.

  21. Alex Bailey November 26, 2015 at 10:53 am #

    While I acknowledge that Hollenbeck’s scenario is a worse case political outcome it seems very unlikely to occur for several reasons.
    The biggest reason is lobbying power, the ultimate antidote to the people’s will. Because of economics, it is extremely likely that Uber, Lyft etc. oligarchy will dominate city auto markets. With that percentage of trips operated by effectively the same entity, the negative externalities of congestion will become internalized. Thus, Uber will miss out on potential profits if their vehicles are stuck sitting in traffic.
    As a result, these companies will lobby city, state and federal officials hard for congestion pricing. I also suspect that they will lobby for a GPS-based pricing system, which would both maximize the efficiency of the system and be fit most comfortably around their systems. It is not certain that they will win the fight, but it seems likely to be the nudge needed to get economic rationality imposed on road usage.
    The dynamics of the political economy will change in other ways too. If people start seeing lots of empty vehicles around them while they sit in traffic, I suspect there will be outrage because “those” guys are taking up their road space. This “us” vs. “them” phenomena (referring to two different types of auto users) is notable as it doesn’t really exist in the current political environment. I suspect this new found outrage will lead to additional pressure to prohibit or price the sort of behavior that makes Hollenbeck’s hypothetical so awful.
    In both these instances, parties will have an obvious interest in modifying the status quo. That is far different from today where it is broadly speaking unclear to the typical driver (for perfectly justifiable reasons) whether they benefit from a congestion pricing regime, even if the societal economic benefits of such a policy are pretty obvious from even a rudimentary analysis of the issue. As a result, I suspect this sort of perpetuation of the status quo is an unlikely political outcome.

  22. kyle November 26, 2015 at 11:22 am #

    If shared-use automated shuttle buses can offset SOVs the congestion game will change drastically in the other direction. Big changes coming in the hybrid bus/taxi world. I absolutely don’t want a personal self driving vehicle, but I can’t wait to use on demand services.

  23. Jarrett November 26, 2015 at 11:46 am #

    Everyone.
    I appreciate the many comments of optimists here. As for Tim Kramer’s demand that I should lay out solutions, I did lay out the only solution that works with the facts of math and the law of supply and demand, and that’s more accurate pricing. Almost everyone who looks at these problems — including many self-described conservatives — ends up identifying the real problem as a pricing failure. Road pricing, plus more public transit for those who don’t want to pay the prices, is the only solution that is something other than a majoritarian handout for people who like cars.
    But transport experts across the political spectrum have been saying that for decades, and it just doesn’t seem to affect policymaking or even infrastructure planning in a big way. As long as most voters are car-owners, it will be hard to get any government to treat such people as one interest group among many. In my work I rarely even try.
    I agree with many commenters about what should happen. But my 30 years of watching these issues, and my reading in psychology and evolution, make me wonder if what should happen will happen, when the market and the culture are already deeply set on a different set of habits. If, as Alex Bailey suggests, we have to rely on Uber to out-lobby General Motors, well, that doesn’t make me feel hopeful. In any case, the feeling of entitlement to driving and car ownership is so strong that we many need a couple of generational cycles to really let go of it. That’s why it’s reasonable to suspect that we may create Hollenbeck’s Dystopia before we figure out why it’s a disaster.
    As always, when contemplating a new technology, we should be thinking about outcomes other than the best ones. And that means we need to think beyond the hype and understand how this will really affect things we care about.
    If, in the 1920s, more people had been thinking this way about the whizbang new technology called the car — if they has asked not just “how cool will this be?” but also “what will our world be like when everyone has one?” we would be in a much better place today. That’s why I feel like someone needs to be thinking this way now.

  24. The_ararar November 26, 2015 at 2:02 pm #

    @Jarret
    We are still vastly better off than any other decade in the past despite of living in the car dystopia.
    So while it’s not the optimal outcome, and we should avoid such things in the future, this is not a disaster, and I don’t expect self-driving cars to cause one either. It’s a missed opportunity.
    Also while states might have problems introducing mobility pricing and there is widespread opposition to any such proposal, I’ve actually seen some progress over the years in my car-dominated corner of Europe, and it has accelerated as the traffic got worse.

  25. SingleOccupant November 27, 2015 at 12:25 am #

    The analysis assumes self-driving cars designed like current cars. As Seth Porges points out in his Forbes’ article “Why Google’s New Self-Driving Cars Could Change Auto Design As We Know It”, When you move to a fully electric vehicle, you have no need for the trappings of car design that have been present since the Model T days.
    With the invention of the highway-capable four-wheeled fully-electric lane-splitting capable Tango from the US company Commuter Cars, it’s proven that safe and fast single-width cars are absolutely viable on current roads and, indeed, far more efficient regarding road width. If single-occupant drivers drove a right-sized single-width Tango during their commutes, road space would more than double as the space between current lanes could be used for driving.
    As Google demonstrates with their self-driving car design, the shape of cars will not necessarily resemble the shape of current side-by-side seated cars. In the future, it’s most likely that self-driving cars will be narrow for parking advantages alone, and, in turn, vans and buses may become single-width vehicles, too.
    Just as side-by-side seated bicycles are available but not often ridden, narrow cars may become the standard shape of cars over side-by-side seated cars which will greatly challenge current congestion modelling predictions.

  26. Bolwerk November 27, 2015 at 6:58 am #

    Optimistically, driverless cars convert cars to transit. It is not practical or affordable to share cars right now, which is why most sharing is carpooling, which in turn remains at best a marginal activity. But sharing a vehicle becomes much more practical when you can timeshare. For instance, a couple working different hours could own one car and handle both commutes. Especially in our increasingly low-wage society, this has got to introduce potential for cost savings of hundreds of dollars per month to many strained families. The same technology can also change the economics of surface transit; on a per-seat basis, buses’ costs would be reduced to be closer to or even lower than that of rail.
    On the more pessimistic flip side, there is precious little reason to believe self-driving cars will reduce pollution very much. Cars have become relatively minor contributors to global emissions as is, directly anyway, and the environmental and ecological impacts of cars would be the same: driverless car culture would make living in energy-inefficient suburban housing even cheaper and would consume similar amounts of land as is consumed now. So, in the end, well-populated urban cores with rich multimodal transit would remain as desirable as ever.

  27. Rick Robinson November 27, 2015 at 2:26 pm #

    Great question! But I somewhat lean toward the optimists here:
    Self-driving cars are also self-parking; ‘circle the block for an hour’ can easily be reset to ‘circle till a parking place opens up’ – likely the wait-for-me default built into the software. Parking doesn’t need to be walkable from wherever your car lets you out, just drivably close. And parking lots can be used more efficiently, valet-style, cars moving to let others out. (And with electric cars there isn’t the big energy/emission cost of starting a car in order to move it a few feet.)
    I am skeptical of how many car owners will choose to set their cars in taxi mode for strangers to ride in sight-unseen, but robo-cabs in some form will surely be widely used, being much cheaper than cabbie-driven ones are today.
    But in the bigger picture, does all of this really change the geometry issues that Jarrett has talked about? When we lived in San Francisco, what really got us on the bus was the congestion and parking hassle, and in large cities that will remain, whether it gets expressed as inconvenience (time cost) or road/parking pricing cost. There just isn’t room for a zillion vehicles, parked or moving, and smaller/narrower cars, like automated ones, can only have a marginal impact. A bus (streetcar/subway/monorail/whatever) carries a lot more people a lot more compactly. And robo-bus service will likely be cheaper for the transit agency, thus more frequent, and faster because transit preference can be baked into the control software.

    • Scott D February 22, 2016 at 9:12 pm #

      This is one of the few comments I have seen about autonomous shared cars needing to be cleaned. Given how often a taxi driver has to clean the inside of their car, despite having been present to say “HEY! don’t do that in my cab!”, I hesitate to contemplate the thought of calling an autonomous vehicle that has had other passengers since it was last taken out of service to be cleaned. I don’t think I’d enjoy riding in a sealed box that is all surfaces that can be auto-cleaned like the inside of my dishwasher between uses, either, which is a logical extension of not wanting a potentially dirty ride.

  28. Rick Robinson November 27, 2015 at 2:38 pm #

    Or, in short form: Traffic congestion is Euclid’s way of telling you to take the bus next time.

  29. Steve Sidner November 28, 2015 at 12:14 pm #

    The transition to driverless self-driving cars would be to first have self-driving cars that still have a licensed driver in the driver’s seat. Thinking through the scenario you posit in your article much if not all of the increased congestion doesn’t occur. Communities may elect to never go to driverless, just as many have elected not to implement open- or concealed-carry gun laws for safety reasons. In other words we may eventually trust cars to drive themselves, but not without a human being in the driver’s seat. I think that is big leap for ordinary people who are not enamored of the shiny tech. 30K+ people a year die in car accidents (most not wearing seatbelts, but still…) Death by automobile is still a real hazard in the US.

    • Scott D February 22, 2016 at 9:22 pm #

      What could the “driver” be licensed to do? There are no manual controls connected to the drivetrain or steering. A licensed “owner” could make some sense, in terms of knowing what it is legal to tell your vehicle to do, but those rules could be coded into the car Operating System and broadcast to the car by the road itself, so that the car cannot break the rules, so the owner is not required to know them.

      I agree that it is likely that the evolution from current automobiles to auto-automobiles likely includes a period where the cars can drive itself but still has some level of hands-on controls as well, so might require a licensed human driver, or at least require that if there is a human in seat 1, they must be licensed to drive, but there are good reasons to remove the requirement for someone to be in seat 1 at all – starting with an automated parking garage for automated cars can be much smaller as there does not need to be space to open the doors or walk between vehicles. It can look more like a current warehouse than a current parking garage.

  30. Peter Muller November 28, 2015 at 12:28 pm #

    I believe ride-sharing will be more important than car-sharing in reducing VMT and therefore, congestion. Ride-sharing has been touted for decades and never really caught on – even with expensive driven taxis – so why it would with cheaper driver-less taxis I am not sure.
    Note that shorter headways (time between vehicles) rely more on the laws of physics (that remain unchanged) than human reaction times that computers can change.
    Whether or not driver-less cars will lead to reduced capacity and increased congestion before, perhaps in many decades, leading to increased capacity, remains, in my opinion, very wide open.

  31. Andre Lot November 29, 2015 at 4:42 am #

    @Steve Sidner: I think that, much on the contrary, as soon as it becomes evident that deaths and serious injures plummet when the proportion of autonomous vehicles rise, there will be an enormous push for adoption of self-driven cars on safety grounds alone.
    These self-driven cars can easily avoid more than 95% of all road-related deaths that currently occur.

  32. The_ararar November 29, 2015 at 2:44 pm #

    the transition will surely be very slow and initially a driver will still be required and these will just be safety features.
    Although due to prices I’m not really sure on how long it will take to transition completely, considering that most cars on the road are low-end.
    So regardless of what happens, there will be decades to observe the effects (certain localities with lots of money and new cars will surely transition almost completely before most other places, thus providing statistics about everything) and adapt the legislation to avoid the negative effects of the massive induced demand of the cheap taxi-like service (regardless of whether it’s your car or the transit agency’s) that these cars will offer.
    Mobility pricing through congestion charges will surely spread to medium-sized cities as well.

  33. Jim November 30, 2015 at 9:23 am #

    I’ve yet to experience government fixing a problem they didn’t first create, and traffic congestion is no exception. However, a preemptive strike against hypotheticals is both wrong and counterproductive. Whenever movie makers create a future Earth, they get a few things right, but there are often gaping holes. In Back to the Future, where is Facebook mentioned, the greatest single media revolution in history since TV? We assume that driverless cars will circle the block just like a regular driver and cause congestion. Does it ever occur to you that a driverless car will use intelligent algorithms or be part of an intelligent traffic network coordinator? Well before that car gets to downtown to pick up the guy, it would pull over to the side of the road and wait instead of wait to get downtown. Even if it’s downtown, it would find unused streets to double-park. When a spot becomes available or that street gets busy, it moves. There are so many scenarios. It could leave downtown and return.
    You forget that once we have driverless cars, the relative cost of car ownership goes up. It will only truly be for the rich. Why own a car just for the occasional convenience of having a car guaranteed? This doesn’t make sense. If you really want a driverless car to pick you up, you would pay a higher rate like surge pricing to get it. It will all be about demand and supply. Why buy a car when you can easily afford to pay more when there is high demand for cars to pick you up? On top of all this you forget the drop in trips when drones can deliver us groceries, food, and other items we often use cars to pick up.

  34. Peter Chiller November 30, 2015 at 7:35 pm #

    I’m definitely in favour!

  35. Bill Cappel December 1, 2015 at 3:50 pm #

    If the car is smart enough to perform all those tasks on its own, what kind of job does the father have downtown where his human labor is still necessary?
    The advances that will enable fully autonomous vehicles will cause disruption in many other fields.

  36. Marc Clifton December 6, 2015 at 6:59 pm #

    Ownership of driverless vehicles is ridiculous. You simply tell the system (not the car) where and when you need a vehicle, and a car shows up. The car you used to get you to work would be in use by other people, or parked somewhere waiting, especially in off peak hours, like work hours. Furthermore, a smart system would be able to automatically and dynamically create carpooling routes, reducing the number of vehicles but also ensuring that everyone gets where they need to get to at the right time. Ironically, there would actually be “less” congestion. The model described in the article is based on the (hopefully) soon to be archaic concept of “ownership” and “owner-vehicle” single relationship. If think of vehicles as a system (much like a subway or bus) then you start to see the real benefits of a driverless vehicle.

  37. Wanderer January 15, 2016 at 9:03 pm #

    The people who I see advocating driverless cars are not sustainability advocates. Their agenda is not to cut the number of cars and move from an ownership model to a service model. Instead I see suburban leadership that wants to be able to tell itself that individual car based sprawl is fine, as long as the cars are electric and driverless. The agenda is getting through changes in the energy system without anybody having to change their lives. I’d be very cautious about predicting that the cars will become shared and properly priced because they’re driverless. Instead, I think we’ll see the argument that driverless cars can (presumably) drive closer together, congestion is no problem.