How will driverless cars change our cities?

From Issi Romem:

  • Cities will greatly expand, again: Faster and more efficient transportation will convert locations that are currently too remote for most users into feasible alternatives, abundant with space. Like suburban rail in the early twentieth century and the mass consumer automobile that followed, driverless cars will generate a gradual, but dramatic expansion of cities.
  • Buildings and parking will be uncoupled, freeing up valuable land: After dropping off passengers, driverless cars will independently seek parking (or their next car-share customers) and they will show up for the return ride at the tap of an app. As soon as driverless cars are common enough, the demand for adjacent parking will dwindle and parking lots in areas where land is sufficiently valuable will be ripe for conversion to other land use. As parking in high-value areas is thinned out or altogether purged, the micro-structure of cities will change – you guessed it – dramatically!

For the pointer I thank Josh Hausman.


So why haven't taxis achieved much of the alleged promise of driverless cars already just for the much more modest goal of, say, only Manhattan?

Taxes have a government monopoly-ish.
Also they are insanely expensive. (A taxi licence in NYC is around USD $1,000,000.)

The NYC taxi market shows how a technology may be hobbled by regulation. Therefore, part of the success of driverless cars will depend on regulation. For example, legislation forcing internet car sales to go through a local dealer in each state killed the early potential of in the 90s. To this day, one's ability to go on the web, comparison shop and click to have a car ordered is totally hobbled by dealer preserving rules. I suspect viable driverless cars will see even more vicious fights.

Me, I wanna see how you hail a driverless car.

By looking at the taxi symbol in your Google Glass interface, obviously.

You get on/off the bus or train at the designated stop.

After all, when you take the train, you ride in a driverless car, and have for decades.

That's exactly right, so it's nothing about them being taxis rather than driverless cars per se that's the problem...

No, that's exactly wrong. Taxiis are far less efficient and create much more congestion. THey are also much more expensive because they require a human driver. Your question is sort of like "Why haven't super-human chauffers achieved these beefits already?"

HA. Boy you should talk to some cabbies, I think they'll be surprised to learn that they're the reason taxis are expensive.

Yes, also wrong because when you're carrying 6 bags of groceries or one screaming child, having a private ride to your doorstep is completely different than getting dropped 6 blocks from a 20 minute wait for a bus to take you to another bus that is 15 blocks from your doorstep. I never lived in the boroughs, but the train/bus penetration is horrible.

Most people don't need to drive within Manhattan. They need to drive to and from Manhattan.

Anyway, how far away is this supposedly 'uncoupled' parking? Driverless cars don't eliminate congestion; you think waiting 30 minutes or more for a car to arrive is going to fly with NY consumers? You've never met a New Yorker.

Taxis in Manhattan have largely achieved what driverless cars would do elsewhere. Why would we think they have not?

No they have not.

Taxis are a shared and continuously used resource. They don't need to park. Ever.

Why couldn't driverless cars be continuously used resources too?

I expect that (regulation permitting), driverless cars will mostly be taxis. The reason western countries are not taxi dominated is that taxis are expensive but with the driver out of the picture, I can imagine western commuting becoming more Asian. That means lots of small (three wheeled?) taxis for getting around a high density metro area with larger private cars (and some taxis) going further afield.

MOOC's and driverless cars have that in common: Before even early adoption steps in theorists spin hype about how they'll completely transform the world.

Why not wait a bit?

The tragedy of the futurist is other people, and vice versa.

I guess it's fun to talk about step 12 when you are on step 2 for some people. In reality, no one is going to allow a car without a driver on the road.

I agree in urban areas, where driverless cars just don't make sense. They would have to obey all traffic rules, while deferring to every rulebreaker.
How, for example, would a driverless car turn right in Midtown traffic? There's a constant stream of people walking across the light. The driverless car can't just plow into them.
However, I can see driverless trucks transporting goods in rural areas. I can also see automatic co-pilots for trucks, just like airplanes have. The auto-pilot handles rural Texas; the driver takes over in the outskirts of Dallas-Fort Worth. Overall, I suspect truck drivers will have a lot more to think about than taxi drivers.

This is a good insight. Will driverless cars eliminate long haul truck driving? Will Teamsters oppose driverless cars? Will it reduce shipping costs? A bonus for

Trains still need conductors and they are a million times safer than a long-haul truck on the road with smaller vehicles.


Not all trains need conductors.

Also, if you google "Driverless Trains" and in the wiki above there are quite a few articles suggesting driverless trains are not far off.

Teamsters would probably oppose, but they aren't nearly the force in trucking they once were. Less than 10% of long-haul truckers are Teamsters.

Trains need conductors by regulation.

I've always thought this would be the first real app for driverless vehicles: you may even still have a driver, but instead of operating the rig for 10 hours and stopping, he operates it for the "hard" 10 hours and sleeps another 10 while the autopilot drives the truck at night. You've effectively doubled the productivity of an expensive rig for probably a couple tens of thousands initailly.

Not sure how long it will take, 5 years, 10 years, 20 years. But, driverless cars will end up being much safer then humans at the wheel. Already true with airplanes (but, they have fewer requirements to meet.) Basically, computers don't get bored, tired, distracted, or forget their training for unusual situations. It's simply a matter of including case after case after case and testing each one. It takes a looong time to do this. But, it doesn't really require any breakthroughs in technology or anything.

In the early days their may still be a few safety issues due to unusual cases that were overlooked, but in the end the fight won't be about safety it will be about jobs. The luddites vs progress all over again. The article talks about driverless cars, but one likely early target will be driverless trucks because of the huge cost savings. With a driverless truck you don't have to pay a trucker, it doesn't have to stop to sleep or eat. In short, one driverless truck can get more work done maybe twice as much with less money. Don't expect the teamsters union to think this is a good idea.

Of course, the teamsters will scream safety long after this has been a real issue. When driverless trucks do start to be used, they will pounce on every accident. Not even a driverless truck can avoid being t-boned by a driver running a stop sign. The teamsters and their allies success is unpredictable since it is purely political, They may be able to stall wide spread use of driverless vehicles for decades. But just like the luddites, they will eventually lose and progress will win out.

In the end, not only will driverless cars be safer. They will end up being mandatory. No different from other car safety features like seat belts and airbags.

How will the driverless truck fill up its tank?

Find a full-serve gas station.

A robotically-controlled gas pump seems trivial compared to driving, but it doesn't seem worth the bother, until the full-serve gas station attendants threaten to unionize.

There are no full-service truckstops (aside from possibly NJ et al where full service is mandatory). I suppose they could arise in response to the trend though, so maybe that's where all the truckers are going to go.

You're much more optimistic about an automated gas pump though than I am...I can't imagine how that would work. If it were that easy I think we'd already have them.

Are you kidding? You don't think they could make automated gas pumps for automated CARS? Drive up to the spot, the gas door pops open, the arm extends...

>>> Not even a driverless truck can avoid being t-boned by a driver running a stop sign.<<<

It just might. Gas pedal full down might have a chance.

Simply automating a gas pump motion can be done now. The question is how to make it actually work in the real world. For instance, where does the arm extend to? Is the machine to somehow read the model of car and adjust accordingly?

>>> Not even a driverless truck can avoid being t-boned by a driver running a stop sign.<<<

In addition to Rahul's point, the computer-driver truck will be aware of the other dude approaching the stop sign too fast. It will see many weird stupid things that a human naturally wouldn't think to be looking for. It will still be possible for someone to t-bone the truck, but they wouldn't be able to get a running start.

In a world populated by driverless cars alone we may not need any stop signs......

There's a classic sci fi story where auto-driving cars are common, and people jaywalk everywhere, including freeways, because the auto-cars are guaranteed to stop for them.

If it gets bad enough, legislatures could set the rules so that the cars video recording of the intersection is usable to charge people with jaywalking.

ok, maybe t-boning wasn't the best example. My point is that a driverless truck that is safer than a human driving still won't have instant acceleration, braking. or the ability to levitate. I will give a simpler example. Rear ended, or even a head on, while sitting at an intersection too busy to enter.

They may also have problems with weather conditions just like humns. Hard to see glare ice, or hydro-planning when it first starts to rain. Undoubtedly they will be programmed to slow down but that takes time. Time during witch something can go wrong. I consider it unlikely that better than humans will ever mean perfect.

Not so much a reply to Dan but to the whole "self-driven cars are safer". No one seems to be thinking of the possibilities presented for cracking the security and operating system of these cars to make them hit jay-walkers, kill the occupants, help hijack the trucks or perform limited types of seek and destroy missions.

Yes I agree, malicious cracking would seem likely to occur. Unless it can be done remotely to whole fleets of cars like computers on the internet, I doubt it would be any more troubling or common then current spree killers. Every once in a while their would be horrific and scary reports about such a case on the news. But, in the end they would just be using a car instead of an AR-15.

Of course their is that remote cracking possibility, one sure fire solution is to have very little remote input. Maybe "come to me" or "go home" are your only choices. Hard to crack that. But, it all depends on how they are designed. Surely, their will be uses for more connectivity. We will have to wait and see.

There is a bit of a difference between "that last MOOC really worked for me" and "I can't wait for my driverless car." Do you notice that MOOCs are increasingly taught by central figures in a knowledge domain? Dan Ariely teaches a behavior MOOC, Martin Odersky teaches a Scala MOOC.

Central figures have been writing textbooks and giving talks for a long time.

Do we have a Professor whose Resume says "BA Coursera" or some such? Or even a lowly PhD student?

Not in our lifetimes, I fear.

It has a real "dioramas of the future" feel to it. And by that I mean more than a little artists' fantasy.

This is a particularly American perspective. If cities have much cleaner air and safer streets, with a lot of land freed up, it is likely that many people will be more greatly disposed to live within their limits. Europeans have been choosing to live in small apartments rather than in the sticks for decades - but then many of their cities are much more liveable.

I'm going to wait for the self-driving RV and just live in that splitting time between Wal-Mart and truck stop parking lots..

I've been to Europe, but lived in Asia for 10 years. I think Europe is actually less livable than either the U.S. or Asia. The problem is that population density in Europe is not high enough to make a car-less life style practical and convenient. In the major cities of Japan and Taiwan, the population density and access to trains and subways are high enough to make such living practical and convenient. In Kaoshiung, I lived in a high-rise condo. I could do my day to day shopping just by riding the lift down to the street and walking to the store across the street.

With few exceptions, you do need a car to live in the U.S. But the population density is low enough, and housing cheap enough, to make this life style practical and convenient. Where I live, every store I shop at, the gym, and several good restaurants, are all less than 1/2 mile from my house.

Europe seems to be stuck halfway between the U.S. and East Asian standard in population density. Its too high density and housing too expensive to make U.S. car life style practical. At the same time, most of Europe really does not have a high enough population density to make for lots of trains and subways and easy access to them. Also, most European cities do not allow for the construction of high rise condos that are common in much of East Asia and Vancouver, BC.

I live in Munich, and I find the carless lifestyle convenient (though not so much as it was in London). Munich is not a large city by US or Asian standards -- but it is large by German standards. Indeed Germany is a great car loving nation and I suspect most people need them.

Europeans have been choosing to live in small apartments rather than in the sticks for decades – but then many of their cities are much more liveable.

More myth than reality. Densities of European cities, like American cities, are far below their historic peaks. Major cities have freeway networks as extensive as those in the U.S. and those roads are clogged with commuters (try spending a couple of rush-hours on the Périphérique or the around Paris).

"Despite some of the most stringent anti-sprawl regulations in the world and high gas prices, the population of the City of Paris has declined by almost a third since 1921, while its suburbs have grown. Over the last 15 years, the city of Milan has lost about 600,000 people to its metropolitan fringes, while Barcelona, considered by many a model compact city, has developed extensive suburbs and has experienced the largest population loss of any European city in the last 25 years. Greater London, too, continues to sprawl, resulting in a population density of 12,000 persons per square mile, about half that of New York City."


"Polls consistently confirm that most Europeans, like most Americans, and indeed most people worldwide, would prefer to live in single-family houses on their own piece of land rather than in apartment buildings," Bruegmann writes. So strong is this preference that certain European countries such as Ireland and the United Kingdom now have higher single-family house occupancy rates than the United States, while others, such as Holland, Belgium, and Norway, are comparable. Half of all French households now live in houses."

What effect will driverless cars have? I see no reason that they would affect people's living space preferences. And -- unless all cars are legally required to be driverless -- there will be no real effect on total road capacity, so...I'd predict at most a modest effect in the direction of greater sprawl. It will become more feasible for office workers to have longer commutes, given that the trip can be made inside a truly functional, connected mobile office. But I also predict these commutes will not be done in driverless cabs, but rather in driverless cars that spend the night in their owners garages (consider -- if there's no driver, who's going to keep passengers from leaving a mess in the back of a driverless cab, or cleaning it up when they do?)

Better remote access technologies will hopefully reduce the need for driverless cars, at least in terms of long daily commutes. What percentage of the NY workforce physically must be at the workplace in order to perform their jobs? 60%? Less?

Anyone care to venture a WAG?

Current technology dictates you need to be physically near to kiss the boss's ass.

In the short-to-medium-run, this will only increase the number of cars on the road, because for the first time in history, there is no 'proportional to population' growth in the number of cars.
Remember, it is too bad already.

I think we will end up with airborne drone-like transport for commuting and local transport before driverless cars become popular. It's an easier computational challenge to move through the air than on the ground. I see the current small drone architecture of a many ducted fan lifting a small airframe, like a carbon fiber chair. Maybe a small gasoline engine driving a generator, then feeding power to a number of fans. Sit down, tell the computer, Siri like, where you want to go, and the computer plots the course and deals with the avionics.

On the demand for driverless cars though, it will depend on the cost. I am always surprised that more people don't have full time drivers, even when they can afford it and the cost benefit equation in terms of their own productivity would easily justify it. I have a full time driver and I get a lot of work done in the back of the car.

Totally agree. I find it shortsighted that at best 'we' can imagine automating a mode of transportation that has huge sunk and ongoing costs (roadways, signaling, etc.) versus a new technology that would both automate travel and reduce or - in a best of all possible worlds, completely eliminate - costs.

For most use cases, to get somewhere I need to go I don't need a personalized mini-plane. I need a jet pack. For long trips, family excursions, or cases where I need to lug something to or from somewhere, maybe then I would need something with storage and/or seating.

A one-man vehicle with a gross weight of 200 kg, supported by two one-meter ducted fans of 90% propulsive efficiency, would require approximately 150 kW, or 200 hp, of lift power. That's not exactly a "small gasoline engine". It would also produce a 100-dB noise signature at roughly twenty meters distance during takeoff and landing, and I'm pretty sure the downdraft would kick up small rocks and discarded bottles, etc, at dangerous velocity.

VTOL aircraft of any significant scale tend to work poorly in urban areas, for reasons that are not really amenable to technological solution because they depend on the basic physics of how much air has to move how fast to produce a given amount of lift. Either you use fans/jets/rotors of impractical scale, giving you a parking problem far worse than with mere automobiles, or your smaller discs consume impractical amounts of power while producing intolerable levels of noise, etc. Or both.

If you've heard that ducted fans are much more efficient than rotors, and thus ducted-fan vehicles will let us do all the fun things helicopters can't, there's a catch. Ducted fans are more efficient than rotors of the same size. The fundamental scaling of power (inversely proportional to disc area) and noise (disc area to the -1.5 power) more than makes up for that, unless you use ducted fans the size of helicopter rotors - and even then, you get only modest gains.

Driverless cars will increase congestion and commute times as well as increase gasoline consumption. People will go to work and send their cars home, only to come back, pick them up, and drive them home. Double the driving, double the miles, double the congestion. This will block the freeing up of cheaper land so our urban sprawl dynamic will look much like it does today with development riding the edge of reasonable commuting times.

Why would they send their cars home? Because they're replacing a second car needed at home?

No parking downtown. Or expensive downtown. This applies to cities rather than suburban office parks.

It would probably be cheaper to send it to a parking lot on the outskirts on downtown for little money rather than pay gas to go home and back.

Also, with driverless cars, it's more likely people will rent cars as needed by the hour or for the commute. Most of the time your car is doing nothing and instead of driving home , that car can pick someone else up nearby who needs a ride.

No. Everyone needs a car at the same time. This is why everyone owns a lawnmower instead of having on shared mower for the block that people rent. Plus people use cars for much more than commuting. Other than places like New York already served by zipcar I don't see ownership abating.

Right now, there are many commuters from NE Pennsylvania to the NY metro area, driving their own cars. If you convert them them to driverless cars, they can make use of the time in other ways, such as working. If you are going to spend 4-5 hours in the car working, why bother to come into the office most days, anyway? If your job requires physical presence (say in building maintenance) you could sleep, but then you have these disconnected periods of sleep.

An interesting speculation, but I suspect that telepresence will trump the driverless car.


No one in their right minds will spend 4-5 hours a day commuting - even if they can work in the car. Driverless cars will allow people to maintain travel times at today's levels, but to cover a much greater distance over that time. For more details see the full blog post, not just the snippet re-posted here.


I did read it. And people are currently spending 4-5 hours in their cars now. If they can travel further in the same time, they will. I'm not knocking your post, which was interesting, but I also think that a lot can happen while we are waiting for these cars to be common.

I spend 2 - 2 1/2 hours a day commuting into Manhattan by rail. I do think these cars have the potential for killing off a lot of the mass transit operations. A lightweight, barebones car would be excellent for commuting - perhaps they could convert the rail right of way and some of the highways into dedicated routes for these cars. Instead of discrete trains, there could be a continuous flow of driverless cars, with much more flexibility.

You might find a greater effect on air travel - if the speeds are high enough, the driverless cars might be better than flying for trips of less than 500 miles - maybe even 1,000 miles.

The Future with Driverless Cars: Massive traffic congestion from cars unwilling to take risks and driverless car inhabitants stuck in alleys Googlemaps identified as streets.

Just came back from Sicily, having traveled using an IPad, Google Maps and a GPS.

1. Italians will scoot in front of you, and dodge under falling railroad standards. The risk taker will dominate a driverless car programmed not to have an accident.

2. GPS and Ipad took us down an alley that got narrower, and narrower....until you could barely back out and certainly couldn't turn.

3. Imagine a politically correct driverless car that doesn't recognize crime ridden neighborhoods. GPS and Googlemaps said it was the shortest distance, you say to yourself, as people gather omininously in front of your car at the traffic light.

On this point, it is interesting that Google Maps would not take me through the South Bronx, but to get from Queens to Arthur Avenue, it insisted on routing me through Manhattan.

Sounds similar to the tale of NYT's Tesla car reviewer saga......

When people come to visit they invariably are routed around tolls. Doesn't matter if it is google or garmin. Takes them longer and they are often confused about why they are in the boonies. But I digress.

Well said.

Avoidance of high crime neighborhoods will become a standard once there is a high probability that 12 randomly selected individuals form any given community rely on GPS assisted navigation services.

"Italians will scoot in front of you, and dodge under falling railroad standards. "

I'm sorry,w as this supposed to be an argument *against* driverless cars?

Urso, I should have been clearer. It's an argument that the Italian driver will be willing to take risks, even if your driverless car does not. The car came from behind me and dodged under the standard.

Also, Urso, I did say: "The risk taker will dominate a driverless car programmed not to have an accident." By dominate, I meant that the risk taker would take more risks, possibly endgagering the non-risk taker because the non-risk takers type is known in this game.

A classic "change one variable and assume everything else stays the same" prediction. At least the probable accuracy of this prediction is highly predictable. Must be a slow blogging day.

Futurists generally overestimate the willingness of middle-class couples and families to live in high-rise and other small, dense, unlanded properties. Suburbanists, on the other hand, probably overestimate people's willingness to drive increasingly-long commutes. If you're not telecommuting into work, then you're probably doing something which relies in some way on being around people or machines at work. I think the "mobile office" concept of driverless cars is greatly overstated. So the "pull" effect of driverless cars is just as overhyped as the "push" effect of decreased parking requirements. I think the desire to live in a large-ish home with a yard, that's within 20 minutes of grocery stores, schools, malls, etc., and within an hour of work, will prove to be a very stable preference for the middle class.

I can't put down a timeline for widespread adoption but I think 15-20 years at the earliest is probably right, and at least 30 years for insurance premiums for non-robotic cars to start skyrocketing. At that latter point, when you have basically legislated and insured human-driven cars off the road, things can change much more dramatically:
-Cars can get smaller, cheaper and more energy-efficient because the chance of being plowed over drops a lot. As I understand it, much of the current weight of smaller economy cars is there as padding to prevent them being pancaked by idiots in SUVs. No more idiots-->smaller cars. Smaller cars can be cheaper, making them more affordable.
-The logistics industry changes again. If labour makes up 20% of transport costs (just speculating, could be higher or lower) then transport companies will benefit greatly from automating road haulage, and they can use smaller, human-less vehicles for greater efficiency. This change is enabled by a fully-automated transport system because it lets freight haulers experiment with smaller designs (say, a go-kart-sized Individual Delivery Vehicle) that could confuse human drivers but pose no problem to a computer system. Pizza delivery vehicles could go in breakdown lanes.
-Families will still probably need two cars, but one can be the Family Vehicle while the other is a cheap two-seater for chaffeuring Mom to work, then driving Billy from school to soccer practice, driving Mom home from work, and then getting Billy again. Not exactly dependent on fully-automatic system, but as with the first point it can become a lot lighter.

What we should remember is that technological breakthroughs tend to pave the way for cultural changes, not vice versa. We shouldn't expect everyone to adopt driverless taxis as their default transport solution, because our auto culture suggests that we're strongly attached to our own personal vehicles. But we can predict several cultural ramifications:
-"Drunk riding" becomes socially acceptable replacement for drunk driving (bad-idea hookups become even easier for 20- and 30-somethings!)
-Further isolation between parents and children as the role of parent as chaffeur diminishes. Latchkey children will barely need to see their parents at all.
-Cheaper taxis in cities should really happen but there are a lot of interest-group obstacles to this. That said, the threat to their livelihood from increased inner-city car ownership (if you can buy a decent new car for $7000, why not?) may prompt taxi companies to turn on their drivers and accept taxi liberalisation. Not sure on this though.

In general the question of what interest groups will support-oppose driverless cars is fascinating. Automakers and tech companies will develop the technology but will have to work to lobby Congress for liability reform so they're not screwed the first time one of their cars has an image processing error and kills its passenger. Insurance companies would be in a bind: they'd love to insure the first drivers, since you'd be immune to all but others' mistakes, but as adoption approaches 100% the need for mandatory insurance is going to come up for debate. Automakers might even be pushing it... mandatory insurance raises the cost of car ownership, and manufacturers will probably figure they're already liable for any accidents that come from their vehicles malfunctioning, so car insurance would be a bit of a relic. The public safety argument will become increasingly convincing, attracting the MADD types and allowing politicians to make driverless cars a moral issue.

As for the transition period, where driverless cars are legal but not ubiquitous, there could be special lanes where driverless cars have to operate. Eventually those lanes would be turned into "driverless car-only" lanes, and before too long there'd just be one or two human lanes on a four-lane highway. Seeing the cars to your left going 80 mph while staying bumper to bumper will help push the laggards to switch.


This is a very thoughtful reply. I am rethinking much of my own ideas about driverless cars.


I totally agree that "the desire to live in a large-ish home with a yard, that’s within 20 minutes of grocery stores, schools, malls, etc., and within an hour of work, will prove to be a very stable preference for the middle class."

The catch is that driverless cars will allow travelling much greater distances than our cars do today in just 20 minutes. For more details, see the full blog post - not just the snippet re-posted here.

My question is how much faster and efficient will cars really get under a driverless car system? I don't doubt that commutes will be significantly faster, but are we talking a 30% improvement, 50%, 100%, 200%? The way I see it, the suburban expansion of the second half of the 20th century was fueled by technological changes--faster cars and the interstate highway system--that allowed commute distances to multiply fourfold, fivefold, and more. That's the kind of dramatic change that I think might be necessary to fuel expansion of cities at a noticeable rate. Moreover, the time scales we're talking about are quite long-- suburbanization in the US took decades, and this suburban expansion would only really be able to start once all cars are driverless. That's a 30 year timeline, minimum, and commutes would have to get noticeably shorter (that 90-minute commute takes 30 mins now; the 45-minute drive is just 15 minutes, etc...). Moreover people's preferences for real estate are even less flexible than their car preferences, which many here suggest are pretty ingrained, so you'd see this unfold over a very long time. At this point, these suburbanization trends would be competing with urban renewal involving the repurposing of large parking areas for new developments. (There could be as many 8 parking spots for each car in America, which is way more than necessary under a driverless car system. So I accept that we may see some suburbanization, but it's not inevitable, it may not be large-scale, and it won't be rapid.

-- Families will still probably need two cars, but one can be the Family Vehicle while the other is a cheap two-seater for chaffeuring Mom to work, then driving Billy from school to soccer practice, driving Mom home from work, and then getting Billy again.--

This sounds a lot like what you were criticizing initially in terms of overestimating the willingness of persons to do certain things. Suburban families are free to buy cheap, small econoboxes to get around in now. But typically, they don't. A premium is put on issues beside efficiency and economy and I don't see any reason to assume that removing a driver from any vehicle will change those considerations in any substantial way.

-- As I understand it, much of the current weight of smaller economy cars is there as padding to prevent them being pancaked by idiots in SUVs. --

I'd be interested in a cite for that. Even if you remove SUVs completely, larger vehicles (work and hauling vehicles) are still present on the road and will be in any scenario. If there is evidence that auto manufacturers are incurring substantial expense in order to specifically protect drivers from SUVs I'd like to see how that is working and if those monies can really be saved with driverless cars. I can certainly see savings in safety terms with driverless cars in certain instances, but that scenario seems far fetched IMO.

-- This change is enabled by a fully-automated transport system because it lets freight haulers experiment with smaller designs (say, a go-kart-sized Individual Delivery Vehicle) that could confuse human drivers but pose no problem to a computer system. --

I don't quite follow this. Small vehicles like motorcycles and minis are already common on the road and drivers deal with them and haulers could utilize them. If haulers could save money with tiny vehicles, why are they not doing that now?

I appreciate your comment- and I hope I don't seem too critical- but it seems generally that some of the changes that people are envisioning are changes they want, not necessarily changes that would flow from this technology.

Hi Maurice,
Thanks for your comment; your points are all reasonable:

-This sounds a lot like what you were criticizing initially in terms of overestimating the willingness of persons to do certain things. Suburban families are free to buy cheap, small econoboxes to get around in now. But typically, they don’t. A premium is put on issues beside efficiency and economy and I don’t see any reason to assume that removing a driver from any vehicle will change those considerations in any substantial way.

My argument about culture vs. technology is technology can't usually *demand* a cultural change in order to win over customers: the first automated cab company shouldn't rely on people being willing to give up their preferences for personal cars. But here, I don't think there's an ingrained preference for large sedans. People buy them because "economy" cars are not wildly cheaper, because large cars are generally safer, because they want flexibility for carrying a bunch of kids, etc. Even if you're right and consumer preferences don't shift dramatically, this is something that happens *after* driverless cars achieve ubiquity. I'm confident on the ubiquity aspect, so if this prediction doesn't come true, I'm not that ruffled.

-"– As I understand it, much of the current weight of smaller economy cars is there as padding to prevent them being pancaked by idiots in SUVs. –I’d be interested in a cite for that"
I'm no car expert, but I'd seen this in articles/comments over the past couple months and threw it in. Just now I googled some articles and while nothing is fully conclusive, it seems like frames have gotten bulkier and are using more advanced materials in order to earn 5 star crash test safety ratings. Then there's airbags, etc (those will probably stay for a while). (Here's a sketchy overview of what's making cars so heavy: It seems a bit inconclusive: more computers and fancy systems will actually add to the weight, but miniaturization should be possible. Overall, I think the computers' and sensors' contribution to the weight will decrease as everything is miniaturized, and I'd expect a similar capacity car to be lighter in 2040 than it is now. But anyone with actual expertise is welcome to weigh in.

–" This change is enabled by a fully-automated transport system because it lets freight haulers experiment with smaller designs (say, a go-kart-sized Individual Delivery Vehicle) that could confuse human drivers but pose no problem to a computer system. –
I don’t quite follow this. Small vehicles like motorcycles and minis are already common on the road and drivers deal with them and haulers could utilize them. If haulers could save money with tiny vehicles, why are they not doing that now?"

Well, I was spitballing on this a bit. The idea is that without drivers and with faster transport systems, individualised delivery becomes a lot more economically feasible. There have been experiments with urban courier systems in big US cities and there have been cute things like the TacoCopter reported on MW. If the vehicle is stripped down to being essentially a motorized box on wheels, it might facilitate these businesses. Think of ordering from a store and having it delivered within an hour. The automated transport system could be necessary because it's so tiny that it falls into human drivers' blind spots. But here we're getting into starry-eyed futurism again, I agree. I hope my larger proposition--driverless cars revolutionize the logistics industry-- is acceptable.

I think the “mobile office” concept of driverless cars is greatly overstated.

Well, there's also the personal mobile Pullman Sleeper concept. The point is that, at the margin, the ability to work, relax, or sleep while traveling to work could not help but to increase tolerance for greater commuting times and distances. Of course, a continued transition to telecommuting would be much better (and a big flameout by Yahoo now would help advance that cause). And telecommuting is obviously practical now -- whereas driverless cars are at least years/decades out (if the concept really pans out at all).

owen wins the thread, i agree driverlesscars are gamechangers even in the limited use, slow adoption scenario (the ability to do other things while driving is huge, people already try at the risk of their lives, a one hour commute is much less onerous if you can use that time for all that time wasting (e.g., web surfing, calling the plumber back, etc.) you generally do elsewhere). if we're lucky, we'll also (eventually) get smarter traffic lights which, combined with driverless cars, could provide significant further efficiencies (far better handling of heavier loads, no waiting at a red when no one's around, etc.).

I think Issa's being a bit utopian here.

As someone who commutes nearly three hours a day, let me tell you, driverless cars are not going to get me to commute any more. Rather, distance will be a function of speed--and I think that's going to be hard to raise in areas like metro New York. In any event, there's only so much time you want to spend commuting. Believe me.

Driverless cars will greatly affect parking and pick up patterns. On the whole, however, we would expect more cars, I think, but distributed quite differently. Keep in mind that the allowable delay between summoning and pick up is only about 2 min. So we'll want those driverless taxis to be distributed. Some will be parked on your residential street (and a reason why electric is the way to go).

The real change will come to those without mobility today: the elderly, infirm and school kids--and the people who usually drive them. Adults have their own cars today. They won't be more mobile, just more productive and less tired. (I can't tell you how I have craved a self-driving car those late summer Fridays when I drive up from Manhattan to Cape Cod.)

So, here are a couple of lectures I could do for MR University:

Basics of Market Analysis
Economists do some ridiculous modeling sometimes because they don't know how to do basic market analysis for a product or service. For example, any reasonable market analysis would show that electric cars would fail commercially. The government could have saved a couple of billions dollars if it knew how to do this analysis.

By contrast, this same analysis suggests a huge market for self-driving electric cars. This presentation is really more of a business school topic, but you really need to know how to do this stuff even if you're an economist. Many of the issues discussed in the comments here are actually amenable to quantitative market analysis. Issa would benefit from this course.

Valuing Time in Transportation
Whenever you sit down to analyze a transportation model, you inevitably discover that what you're really selling is time. It's the value of time which determines the viability of proposed transportation solutions. If you use this model, for example, you realize that you can effectively cover the entire country with CNG stations if you just put one next to each Wal-Mart--the economic drivers are virtually the same. So you don't need 125,000 filling stations to offer CNG; you can do it with about 5,000 at a cost in the $5-10 billion range. Really not that big a deal.

In any event, valuing time is not rocket science, but it's critical to understand if you're talking about electric cars, buses, air travel or high speed rail. Your product is not transportation; it's time.

Thanks for the positive comments, guys. The way I see it, driverless cars face a litany of criticisms about unfeasibility, but these tend to be brought up in a haphazard, unsystematic way and presented as if it's a novel idea that completely sinks the idea once and for all. Most have answers, though, and many can be implemented without major bureaucratic/political/cultural adaptations, which are the real source of friction. People argue that cars can't make the tricky decisions required in modern traffic situations, and that's probably true at the moment, but the idea that instantaneous reaction times, precision turning and speed control, and 360 vision can't make up for this is a bit short sighted. Computers will probably be better than us at navigating ice and snow, they'll be much quicker at reacting to kids running into the street chasing balls. They'll be programmed to drive very defensively, because the liability will have shifted over to the manufacturer, so they may merge and turn very cautiously, which could be annoying. Clever manufacturers will hopefully devise some kind of local area network communication protocol (like WiFi Direct) to let cars figure out if other robocars are nearby. If the approaching car is a robocar, it becomes a lot easier to negotiate a merge.

The interesting question is how the technology will actually come about: will it be gradual adoption of automated futures by current automakers, or disruptive innovation by a company like Google? I'd guess a bit of both: today's carmakers will continue to be the front-end manufacturers, distributors and suppliers even in the robocar era, but Google will probably be selling or licensing its technology to them. Current and nearterm automated car tech is incremental: Lexus has collision avoidance and automated emergency braking, Volvo has pedestrian detection, Infiniti has blind-spot warnings, several companies have adaptive cruise control and lane departure warnings. But these are far off from total automation, and there's a quantum leap required to get to that point. You can't have someone controlling a car just half the time-- it's arguably more dangerous, since they'll be drifting in and out of distraction even more than they do. So someone's going to have to take the leap and test out a model that's 99% automated. Google will come in huge here: they have the rich map database and the computer vision algorithms which the automakers need, while Google itself has no real reason to jump into a market in which it has no expertise. I expect to hear about a partnership between Google and a major automaker within the next 3 years to move toward a commercial production model. That would be a long-term research and development process, but by the time the deal's signed, I think Google will have proved itself in the field and someone will want to lock them into an exclusive arrangement. If the Toyota Rover launches in 2020 with exclusive Google autonomous capabilities, and proves a commercial hit, it will be "the" robocar, just like the Prius is "the" hybrid car.

What sort of demonstration can be made of the technology that would make people accept the technology?

I think the best case for driverless cars would be in a gated suburbam community with golf-cart style cars that don't travel out of the neighborhood, but instead drive you to a parking garage or mass transit system at the edge of the neighboorhood. These could be electricity based and would be great for older people and kids. The neighborhood could have shopping, restaurants, schools and the like. The cars would be travelling slowly with low levels of traffic and well defined paths. The cars could drive to self-charging systems when they are done. If people talked about driverless cars this way I would be more convinced.

However, most people babble about long-haul trucking and high-speed mixed traffic areas. Really those should be replaced with mass transit and rail not with driverless cars.

> What sort of demonstration can be made of the technology that would make people accept the technology?

A driverless car should win Le Mans. A racing series would do wonders.

Yes, just like predicting that years of diesels winning Le Mans would lead to a flood of diesel-powered sports cars. It all depends on how the rulebook tilts the playing field. Nevertheless, it is useful publicity. The high-efficiency DeltaWing racer would have done much better if it hadn't been painted stealth black, making it difficult to avoid when the maneuvering gets tight and visibility is critical.

Endurance racing with a driverless car should be an advantage, since computers don't get tired and lose focus, and they avoid the need for pit stops to change drivers, leaving only fuel and tires to slow cars down from racing speed.

I don't think there was ever any doubt about the basic functionality of diesels, nor predictions that people would buy diesel 911s. There's plenty of doubt (and justifiably so) about driverless cars. A car that could win Le Mans would need to satisfactorily answer many of the questions.

I interpreted the question as "what would make a skeptical engineer accept the technology was reasonably robust?" I don't think toodling around California between 10am and 4pm qualifies. If albert meant "what marketing campaign would work best?", I don't have an answer for him.

I'd love to see more freight travel by rail. Unfortunately, it's much slower than freight that travels by truck. Moreover, the "last mile" problem still requires truck delivery. I suspect the costs of transferring freight from trains to trucks completely overwhelms any cost savings found via efficient rail transport.

As a frequent interstate driver, driverless long-haul trucks would be awesome. No more fucking around on a two lane road while two truck drivers play "I'd really like to drive one half of one mile per hour faster than you" while 30 cars are stacking up behind them.

Might not be as good as you think. I expect driverless trucks will always drive exactly the speedlimit when weather allows. Most drivers go faster. We can hope they will all line up in one lane though.

Wonder how Google managed to navigate the patent thicket without stepping on anyone's toes.

It's going to be a boon for urbanism *and* for sprawl. The issue of wait times for mass transit autos/mass car sharing will long favor cities with higher exchange volume. Just like taxis now are more attractive wherever it's dense enough support a large stock of cars constantly loading and offloading. Long Island-type suburbs will be the first to stop needing to own their own cars, and that dynamic will enable density growth. As has been mentioned, it can evolve into something like 4-person subway cars going everywhere the roads go, with another "train" coming every few seconds. Ironically, car ubiquity will enable "car-free" lifestyles, in the deeper sense of the term. The death of mass transit will be the life of mass transit lifestyles!

That's all happening at the same time as the preferences everyone has described for space and privacy begin to push distance and sprawl. Metropolitan clusters will grow denser even as they sprawl further. The balance of the effects will emerge from the balance of preferences for space+privacy vs. density+vitality. Tomorrow will bring more Matt Yglesias urbanism, but it will also enable more grumpy or private people to indulge their preferences to move further away from noise and bars and hipsters. We'll have to see how the dust settles--particularly since the dust has yet to be kicked up.

Anyone who thinks that driverless cars means that people will give up either driving or car ownership has never lived in America.

I suppose the same was said about horse riding and horse ownership, once.

Of course, the extant drivers will continue to have preference for driving and cars, as they are already used to.

But with gradual generation change, the cohort of youngsters who will think of cars as autonomous robots, will expand, and these youngsters will often have no compelling reason to undergo driver training and have a driver licence. For a beginner, learning how to drive is a nuisance. In this way, driver licences will go the way of dodo, telegraph and written letters in ~30 years.

The end of sprawl is coming due to demography, not anything else. I suppose it's possible that driverless cars, by reducing the inconveniences of parenting, would affect demograhic trends to smaller families at the margin, I have a hard time seeing it as significant. And it is families, and family formation, that drives the suburbs. (Much of the hatred for the suburbs is simply a result of an association with the nuclear family.)

My guess is that it will be better to be a teen, and better to be elderly, than it is today. Teens will suffer a reduction in their unsupervised discretion, but it isn't discretion they often exercise well.

I expect that alcohol producers are likely undervalued in today's market. Norms of alcohol consumption have changed in large part in response to laws on driving while intoxicated. When that becomes irrelevant, I see bars and restaurants becoming more popular, and alcohol significantly more popular. (And other substances as well.) It is possible that social capital formation will be improved as a result (though there would obviously be countervailing effects from the increased consumption as well).

There are those who claim that the dispersed neighborhoods made possible by automobile ownership have made people fat, because the distances make walking inconvenient. Does this mean people get even fatter?

I know that driverless cars are coming but something makes me doubt that they will be prevalent as people think they will be.

Comments for this post are closed