Dense cities don’t have room for everyone’s car. If too many people use cars, they take up all the available space and still get in each other’s way, which is what congestion is.
This was all obvious, and much discussed, when cars first appeared on the scene. So the prospect of making the car the dominant tool of urban transportation — as opposed to, say, something you might rent to make a trip into the countryside — should have been easy to recognize as a scam.
Historian Peter Norton’s first book, Fighting Traffic, chronicles how this scam took over the United States to create the way of life that most Americans now see as normal. Exploiting understandable frustrations with the for-profit transit of the time, the nascent car and petroleum industries “partnered” with government to build a sense of inevitability around car-based travel. This campaign had all of the disastrous results that were in fact predicted at the time — road carnage, pollution, and congestion.
Why did people fall for it? In part, because “innovation” was going to fix those problems soon, leading us to a new utopia where we could take our cars wherever we wanted, safely, cleanly, and without delay. Norton’s new book, Autonorama: The Illusory Promise of High-Tech Driving, fills in more detail on this critical element of the scam, and shows how it operates in the driverless car narratives of today.
Obviously, actual technological improvements to make driving safer are to be welcomed. The danger lies in the impossible visions of the congestion-free autonomous-car-dependent city, which is then cited as a reason not to invest in proven methods of urban transportation, such as public transit. The claim that autonomous driving can fix congestion is no longer as loudly proclaimed as it was a few years ago, but it’s still out there. The only basis of this claim is that because a computer’s reaction time is faster than a human’s, autonomous cars could drive closer together at high speed, taking less space. This, of course, is a minor improvement compared to the countervailing force of induced demand: Eliminating the hassle of driving will cause a lot more driving. We have seen this before.
In the century-long history of high-tech car boosterism, Norton detects cycles of peak hype roughly 25-30 years long, peaking in the 1930s, 1960s, 1990s, and now. At the peak of each cycle, a burst of technical innovation, fused with intense funding and public relations efforts, seems to bring the dazzling future almost within reach. When the vision fails to deliver, there’s an inevitable pause of 20 years or so. Memories fade, and perhaps more important, a generation reaches their 20s who don’t remember the last cycle, and whose sincerity and energy give the effort new life.
Norton calls the newest of these cycles Autonorama (a portmanteau of Futurama and autonomous), but his description of it captures what all four cycles have had in common:
Autonorama is the place where old-fashioned car-dependency is lent new credibility through the application of a fresh gloss of high-tech novelty, where simple possibilities are neglected not because of their inferiority but because of their simplicity, and where implausible promises of perfection divert attention from practical possibilities of actual improvement. In Autonorama transportation research looks like public relations (and vice versa), theoretically possible performance is equated with actual performance, and technology is less a human means to human-chosen ends than a mysteriously willful entity that inevitably delivers ever-better solutions …
None of this is a secret, really. If you read business journalism you can find corporate gurus explaining their methods with pride:
In 1929 [Charles] Kettering distilled his advice into an article, written for Nation’s Business, called “Keep the Consumer Dissatisfied.” “If everywhere were satisfied,” he explained, “no one would buy the new thing.” To Kettering, transport sufficiency was a threat to motordom’s future. He advocated perpetual insufficiency, propelled by an ever-receding promise of future perfection.
In the book’s first four chapters, Norton explores the four cycles that we’ve been through so far, ending with the current moment of autonomous-car boosterism. But the most powerful chapter is the fifth, “Data Don’t Drive,” which will train you to recoil when you hear the term data-driven. Norton explores how invocations of data as the ultimate authority invite us to surrender to interests and goals that may not be ours.
Part of the problem is that data is a valuable commodity. “Data is the new oil,” as they say. Norton even turns up a McKinsey report arguing that the real importance of driverless cars is that it will allow us to spend more time interacting with screens, generating data about ourselves that can be used to target and manipulate us.
But the real issue is that data is a tool, not a goal, and only humans can specify the goal. As Norton puts it, “data can tell people which efforts are serving their goals and which are not, but the goals must be chosen first, and by people.” In my own career, I’ve seen countless studies that sought to overwhelm the reader with data and analysis, not to illuminate the real choices (as our firm‘s work does) but to make them surrender to the goals (sometimes not clearly stated) of the proponents. Traffic engineering is full of this kind of talk (“the data show that we need to widen the road”) and you’ll sometimes hear it in transit planning too.
I heartily recommend this book. It will remind you, once again, of why historians are as urgently needed as scientists in our brave new technological future.
‘The only basis of this claim is that because a computer’s reaction time is faster than a human’s’
In the transitional phase with both autonomous and human-driven cars this is mostly true (it might be fairer to say the computer reacts near-instantaneously in milliseconds instead of seconds). But the claim was that when all cars were autonomous (within a certain area which might deliberately exclude human drivers) movements could be planned rather than reactionary.
Which isn’t to say I disagree with the broader point!
I’ve seen online demos of the so-called high-tech intersection before. The big thing they neglect to mention is that, in order to work, not only does every car need to be autonomous, but everything that crosses a street needs to be a car, so what they’re proposing is essentially an outright ban on bike and pedestrian travel in the middle of the city. To which people reply that you can have overpasses and underpasses, and that ordering an autonomous Uber to take you one block across the street is hardly a big deal. To which I respond that pedestrian overpasses and underpasses are too costly to build on more than a very tiny number of streets, and having to wait several minutes and pay several dollars just to get across one street feels very hellish, not to mention people without money would be essentially trapped on whatever block they happen to be on, unable to move at all without getting hit.
Of course, it is also possible to implement a middle ground solution, where you have a traffic signal that alternates every minute or so between a phase where the pedestrians move in all directions simultaneously while the cars wait and an autonomous vehicle phase where all the cars move simultaneously in all directions (under control of the algorithms) while the pedestrians wait. But, now the vehicular capacity of the intersection is drastically reduced to the point where you haven’t actually improved anything over the traditional traffic signal.
Similarly, since autonomous intersections don’t work in improving vehicular capacity on city streets, even autonomous driving on freeways doesn’t actually help either. It may result in better throughput when three lanes merge down to two, but assuming that everyone eventually needs to exist the highway to drive anywhere useful, the AI tech simply shifts the traffic bottlenecks from the freeway merge points to the exit ramps, where everybody has to wait in line for the traffic light where they meet city streets.
My guess is we’ll land on the ‘signalised pedestrian crossing’ outcome, and we’ll get a small but measurable improvement mainly from vehicles accelerating and braking together.
In any case, my point was just to identify that ‘reaction time’ claim isn’t really where autonomous car supporters think a big improvement will come from (rather, they think a small improvement will come from this without requiring any physical or policy changes).
Just to push you on this a bit, do your feelings change if we’re in the situation where cars are excluded and all the vehicles are buses? (See also, Oxford Street Pedestrianisation scheme).
I think full autonomous driving will decrease freeway capacity! Unlike human drivers, computers will actually maintain a proper margin of safety – while it is rare things like wheels falling off of cars happen, computers will stay far enough apart that they can actually stop in time or maneuver around. Humans are forced to hit something because they are not leaving enough margin.
If human driers actually backed off to the recommended following distance for human drivers, then autonomous with faster reaction times could improve things. However highway safety experts and drivers ed has been telling people to do that for decades to no change.
Something like that is already implemented… in high-tech warehouses.
HOWEVER, only one type of vehicle is allowed there, and nothing else, especially, no humans. And the vehicle density is low enough that they can be controlled from one center without overloading the communication network.
And that’s where it should stay…
“Norton even turns up a McKinsey report arguing that the real importance of driverless cars is that it will allow us to spend more time interacting with screens, generating data about ourselves that can be used to target and manipulate us.”
Sounds a lot like Brave New Word. Yuck. I fight not to be tied to my phone but other people’s priorities (often companies’) just make it easier for me to acquiese to the phone’s solution more and more. Sure I have the choice to not use a phone, but that often involves extra time and hassle or just missing out on things.
Considering how much of a premium is placed on data, management, professional services, finance and mindless ‘innovation’ we seem to be hitting some real economic and technical walls. How many start ups are every going to produce a product that is both useful and humane? Despite the breathless reporting on innovation and technological change, are we really making progress or just tweaking a few things here and there. Too much treasure and human potential is wasted on busy work, marketing, and fake innovation. Meanwhile real problems and human needs go unmet. Bless the ‘free’ market.
This is a timely post given that San Francisco officials recently sent a letter to the state of California opposing Waymo’s request to operate an unlimited number of robotaxis anywhere in the city. The letter documents the issues, including AVs driving through construction sites when and where people were actively working.
Twitter thread summarizing letter – https://twitter.com/DavidZipper/status/1664683301435473925
Letter – https://drive.google.com/file/d/1Rg2UrLqjF5t7A_wWxec_SWX5N-qtqOfu/view
“Dense cities don’t have room for everyone’s car.
* * *
So the prospect of making the car the dominant tool of urban transportation — as opposed to, say, something you might rent to make a trip into the countryside — should have been easy to recognize as a scam.”
I don’t think “scam” is the right word, rather than Norton’s characterization of it as a revolution or battle to actually redefine the urban form. *Dense* cities don’t have room for everyone’s car, but *less dense* cities do (for the most part). The rise of Big Auto in the U.S. wasn’t auto supporters convincing people (falsely) that autos could meet the transportation needs of places like Manhattan, but rather convincing people that the future of urban construction should involve building more decentralized places like Los Angeles or Houston that *could* accommodate autos for most users:
https://pedestrianobservations.com/2023/05/21/the-origins-of-los-angeless-car-culture-and-weak-center/
IOW, it was a claim that we should hereafter stop building places dominated by strong city centers, but instead build places that autos *did* function well in – not a claim that autos would work well in the central business districts of historic cities like Boston or NYC.
If I’ve understood Jarrett’s writings correctly, a single small dense urban area is not a good candidate for public transit. To operate cost-effectively, public transit would ideally have passengers at all times along every route.
People sloshing into a city centre in the morning and back out again in the afternoon is bad, because in the morning the transit run empty out of the centre (and vice-versa), and during the day fewer people are moving around so the transit vehicles are underused.
It would be interesting to see what the mathematically opportunity-maximising urban form would be – i.e. Jarrett’s isochromes would cover as many destinations as possible. Intuitively, I think it ought to be completely flat density with destinations of different types completely mixed. But I might be wrong about that?
I think your intuition is correct – but you still need a threshold amount of density (describing any uses, not just residential) before Jarrett’s geometric argument takes hold. A low density city doesn’t *need* transit to function, because at lower densities you don’t have the spatial limitations that necessitate shared transport. That doesn’t mean, of course, that there don’t remain benefits to transit that might lead a lower density city to adopt it. It just means they’re not in a situation where it is *impossible* to move people around solely through private vehicles.
Most urbanized areas in the U.S. are pretty low density. And outside of a handful of metros, the few areas that have high density only have it in the CBD’s. Borrowing Alon’s argument, I think that’s because most of our urban areas (again, outside that handful) co-developed along with the automobile. That’s why an outsized portion of our transit use takes place in NYC, and most of the rest is in that handful of cities – and most other cities, even very large ones, don’t have much transit mode share at all. The rest rely almost entirely on private vehicles. And not because there’s been a “scam” when people are deluded into thinking that cars can work for dense cities – but because we stopped building dense cities, and mostly built lower density urban areas, where cars work.
I think the most transit-friendly cities have what I call a “spider” form. They ideally have a fairly large urban core and sprawling development concentrated to the “legs”. This is common in mountainous areas where urban development is concentrated to the valleys, this is one of the reasons why Switzerland has so great transit, but it occurs in some flat areas too, such as in Copenhagen, where urban development has been deliberately concentrated to the rail corridors. The idea is that most travel will take place along the “legs” which are naturally suited to fast high-frequency transit. For this to work, most workplaces must be in the core, otherwise you will have much travel between the legs where high-quality transit never will be viable. Even though vehicles will be underused in the “reverse” peak direction, this disadvantage is highly overshadowed by the advantage with concentrating travel to a few corridors.
A great example of a city with a “flat” density is Los Angeles. Contrary to popular belief, Los Angeles has a fairly high density, but is evenly spread out with a very small core. Most travel occurs between areas outside the core where there are too little demand to make high-quality transit viable.