This is basically Arlington, VA, where I live (or at least a lot of it). And it's one of the safest, most affluent, most diverse, and physically healthiest places in the country. I like this future!
Yeah, it's nothing new really... just a return to the natural pattern of development before euclidian zoning outlawed it. The future just sounds like any college town or streetcar suburb built before WWII. Sign me up!
I think people have this misperception that euclidian zoning is a recent thing. Which I think you intend to mean 'grid like' but Euclidian grids have been used to layout cities and towns for centuries. The 'Garden City' movement sought to break with the grid and introduce a variation of order but with the meandering streets. Laying out development patterns on a grid increases efficiencies vs. developing along cow paths.
The worst example of the meandering development and road layout is the Seven Corners in VA. What a shit-show to navigate and the built environment is terrible.
I lived a few blocks away from the Braddock Metro in Alexandria for several years and really enjoyed the walkability and access to the metro. Parts of Arlington are okay but it's still very much car-centric. I couldn't stand Potomac Yards, but there are other parts of Arlington that are nice. A lot of the DC/VA area has changed drastically since I moved away in 2009. Some good and some bad.
Parts can be for sure. Crystal City is a bit plastic. North Arlington has a lot of character though! I live near the old but revitalizing area near Langston Blvd (formerly Lee Hwy), and it's exciting. So many local small businesses--restaurants, retail, and services--and the community input about the new development plans all reflects the stuff Noah's talking about here.
One element you left out: I think we're a few years out from the beginning of pressure for the state to enable Accessory Commercial Units in the way that it has enabled ADUs. For many of the R1 zones, 80+% of the people living in them are an unreasonable walking distance from any commercial hub. The only way you turn those into 15-minute neighborhoods is to allow corner stores and cafés to be opened by residents on formerly-exclusively-residential parcels, where it is presently illegal to operate a business.
When I was a kid, I remember my family got our hair cut in the basement studio of a walking-distance neighbor. I think we're going to see a lot more of that in the future.
Wow! Moderate optimism, what an unusual stance for a blog. One quibble. You suggest that the newly densified inner suburbs will have shopping areas with mixed retail. I'm not sure bricks and mortar retail is likely to expand in any conceivable future. Lately, I've been confronting some age related incapacities. My shopping has become almost exclusively on-line. This is not because I want to make Jeff Bezos the first trillionaire. Trying to walk to and through a retail store is the equivalent of torture these days, hence my aversion to physical shopping. Of course, I also share many economists incomprehension regarding the notion that shopping is entertainment.
It probably won't be retail that sells physical goods, other than food and drink. Rather it'll be small business that cater experiences - fitness studios, bars and restaurants, co-working spaces, and perhaps a growing array of commercial spaces for hobbies, like board game stores and drunk-painting classes and axe-throwing bars.
1. Growth needs to be, and will be, organic and incremental. I've read some anti-NIMBY screeds (often East Coasters) promoting some nutty ideas; e.g., single-family city neighborhoods (like mine in L.A.) need to be replaced with 40-story complexes. No! For many reasons, it cannot, and will not, work that way. A mix of single-family homes, some with ADUs, and low-rise apartments is a much, much better way to go.
2. Not sure cost of car ownership in future goes up. Looking at TCO, EVs are more affordable than ICE cars. Not only fuel costs, but maintenance (no engine, transmission, etc.). Our EV expenses: <$20/month fuel, $0 maintenance.
3. Political ramifications of denser cities/suburbs may become more challenging that they already are today. We're about 80/20 urban/rural now, and political power of the less-dense areas are able to thwart the will of the majority. What if the split goes to 90/10? Will that flip some red states and make progress more possible, or will the stalemate continue, raising frustrations?
post, Noah. I’ve seen you mention micro-mobility as real game changer for how we will get around in this denser America. I’m an electric scooter owner myself in the suburbs of Boston and I think you are a bit too optimistic. Electric scooters still can’t ride in the rain and are basically unrideable in sub 40 degree weather. In Boston that means I can only ride mine 7 months of the year and when you subtract days that it rains it’s closer to 5 months out of 12. It’s still not practical for most people. Only reason I do it is because I don’t have my license yet; once I get that in November I’ll have a car.
In Los Angeles there is already a major difference in traffic congestion on rainy days vs dry days (in this case, largely because streets that don't see much rain actually build up a lot of oil and dust and become very slick when a mild rain falls on them). But if walking/biking/scooters become a bigger part of everyday travel, then I think the effects of rain on traffic will become a lot larger.
I don't know what an appropriate response is - convert some existing streets to bikeways, that open back up to cars on rainy days?
1. The issue with suburbs isn’t just zoning that allows denser housing, of equal importance is zoning that prevents businesses and commercial properties. The reason that some suburbs are unwalkable is because there is no place to walk to.
I lived in Europe for 12-years. Most of the time in smaller villages or cities. The great thing about Europe was there was always a little shop, bakery or pub within walking distance. Pittsburg suburbs and surrounding cities are like this. At least the older established ones.
If you really want to change the landscape, then this zoning issue needs to be addressed as well.
2. Commuter trains just aren’t going to be a thing across the United States. Sure they are established in some cities in the North East and a few big west coast cities, but across the US they are just non-existent. At best bus service will fill in some of the gaps, but I doubt it.
3. I’m surprised you didn’t mention self-driving cars. Now… let me say up front… self-driving taxis will never be a thing. People are nasty. They would quickly become disgusting. I’ve seen some people say that they could have automated cleaning, but that would mean they would have to have hard surfaces which would basically be uncomfortable. Think about your average bus, and they have a bus driver present.
4. I think the scooters and the bikes or an interesting proposition/idea. To tell the truth I never really thought of them as a commuter item. It’s partially limited by the weather, especially in winter. But it actually has potential, but only if you can at the same time increase bike paths which I assume they will use as well.
5. I’ve traveled the world, except for Asia. I need to get to Japan to check out how these Japanese cities you talk about really are. In someways what you described it sounds like parts of South America. I’m in Argentina right now. I know ironically, a lot of cities in urgent Tina and other parts of South America, high-rise apartments are more upper class, where the poor working class people tend to live in smaller single-family houses, even if they are densely packed together. I asked one of my engineers why this is, and he says it’s because of security. A high-rise apartment usually has a gate and enclosed area and a full-time security guard to watch.
6. Anyway. Great post. I wish you had some more photos, but aerial shots showing a few blocks. It would help me visualize better.
Addendum. I forgot to add about self driving cars. They could change the landscape in one way without self driving taxis. Imagine a family that has three or four cars. Like my house. We could probably replace four of our cars with one self-driving car. It could drop me off at the airport at 6 AM, returned to take my wife to work, return to take my daughter to work, then pick my wife, etc. This is were self driving cars will become very useful. Hell, you wouldn’t even need to park them at your house. You can keep it in a parking lot several miles away, and then just call it to come to you when you needed. Especially in the denser cities were people might like to have cars, but they’re parked everywhere, such as New York City.
Unfortunately I think the self driving future is still going to be decades away unless infrastructure is built specifically for them and them only. It's likely a problem that requires close to human level machine intelligence to safely solve the way our current road systems are set up, and I won't be surprised if we don't see that in our lifetimes.
I don't know what the non-train transit solution would be. I love cycling, and would prefer to ride my bike around a city to get around and for lite errands, but as others have pointed out, that's tough in Fall/Winter in a lot of areas.
I do share your skepticism about whether most towns in the west would be willing to put in lite rail infrastructure but the irony is that trains built the west. Done right, I think they could be well-received. Seattle has put one in and it's been expanded. Think Boise would bite?
No way Boise bites. It would be cool, but the cost is just so prohibitive. Apparently we did use to have a trolley system. There are plans for a transportation corridor along State Street, with raised bus stops and rapid transit going into downtown, but Boise just doesn't have the centralized jobs downtown to justify it. So many of our new residents are moving to Boise and working remote. Lots of apartments and townhouses going up downtown, but its a lifestyle thing for access to restaurants and bars and the University. Not so much for jobs. Plus, they plan on putting a huge homeless center right on the main bus route between outskirts of downtown and downtown, so that will probably end up negatively effecting what little ridership their is.
I agree with you about self-driving cars. I'm not sure "decades" but at least a decade away.
Personally I'm a fan of the 10-minute city block idea (cant remember the official name), but Noah has written about it before. Sort of a lot of self contained mini-downtowns.
To clarify, I wish we had a train/trolley system, but only if it worked until the late late evenings. And all day on weekends. I just know we wouldn't be able to justify it
Yeah, I'm extremely skeptical that safe "full self driving" is going to arrive in the next five years. I feel confident that _eventually_ a computer will drive better than almost any human, the same way computers now fly better than humans. But the environment on a busy urban street is wildly more complicated than anything an airplane has to deal with. I think we're more like 20-25 years out from being able to cope with the unexpected.
I think the greatest challenge for self-driving cars is that while we shrug at ~40,000 traffic deaths per year for human-driven cars, 40 deaths from self-driving cars might drive the industry into regulatory oblivion. It's the shock of the new and the strange.
I could see a time where we shrug our shoulders at, say, 4000 deaths/year from self-driven vehicles, but I'm not sure how we get from now to then.
I think the challenge with street traffic vs air traffic is greater than that. A lot of what we do when we navigate street traffic relies upon our background understanding of peoples' intentions. I think these are the majority of the cases too, not just edge cases. The problem with understanding intention is that to really understand it, _as such_ it takes an agent with at least the cognitive capacities of a 3-year old - though our laws suggest 16-year olds.
There's a view in philosophy of mind and cogsci called "mental holism." It's the idea that you can't really get a foothold on a working understanding of other minds and mental states without understanding _almost all of it_. If it's true it will present a whole lot of problems for AI use cases that depend upon understanding mental states. I happen to think it's true, so that's why I'm skeptical. All the traffic data in the world would still massively underdetermine the data you need to understand even basic mental states that other road users have, like the desire to not die, the desire to get where they're going (ideally efficiently), the desire to cooperate for safety, and a near-infinite set of other desires, beliefs, etc. that go into our understanding of what it is to be operating a vehicle on the road.
Heh, you think I'm under-stating the problem, I think you're over-stating it. My undergrad degrees in are cog sci (with concentrations in philosophy of mind and lingustics) and computer science. Basically I studied Natural Language Processing, and worked on various natural language products at IBM, Microsoft, and Motorola, for quite a few years. I studied in the lab of Fred Jelinek, who famously joked that every time he fired a linguist, the accuracy of his language model got better. You seem to be asking the driving model to "think" about driving the way a person does. It won't, and that won't stop it from doing the job better, eventually, any more than it stops the best chess AIs from playing chess better than a person, even though it's obvious that they're going about the task very differently.
You really _don't_ have to know _why_ people are doing what they're doing. You just need a broad enough training set to accurately predict _what_ they're doing, regardless of _why_. The problem is that the set of possible things that can happen on the ground is so much broader that it's going to take _vastly_ more training data and processing power to generalize from a training set, so that novel situations are at least _enough like_ something from the training set that we can match a pattern and respond appropriately. We're _already_ at the point where for highway driving, the computer is probably better than a person at least 99% of the time; it's only really weird edge cases that fail to parse. But surface streets are still much worse than that, and even for highways, a 1% error rate is still way too high, that has to get improved by maybe three orders of magnitude before it's safe to deploy truly driverless vehicles.
I sort of think "what" an agent is doing in a bounded game like chess or go is waaaay easier to understand than what they're doing in the less bounded games of life. There's a view that you can't really understand what an agent is doing without understanding their mental states. E.g. you might be able to predict that the human-shaped object will move from one side of the street to the other at a predictable rate, but you won't know that it's crossing the road, to get to the other side, to get ice cream. It is very plausible that understanding action _as such_ and not just as physically indistinguishable brute physical movement requires understanding intention, and that understanding intentions can't be done piecemeal in an atomistic or even molecularist kind of way.
That's why I happen to think it's going to be really hard. FWIW Wittgenstein had a big influence on my thinking here. On the idea of "thinking like us" being a necessary condition for recognizing some activity as thinking, I guess I just sort of take that as trivially true. Insofar as we can justifiably call some activity "thinking" surely it is in virtue of its similarity to something we do; doesn't have to be exactly the same, just sufficient enough for us to even recognize it as such.
The other thing with self driving safety that I think people really care about is the intelligibility of incidents where they do occur. It's perfectly predictable the kinds of human conditions that make incidents more likely (DWIs, speeding, unkept equipment, distracted driving, etc.). We can monitor, control, and attempt to prevent those things to reduce our risk. But driving full speed under a semi truck in full view because it was crossing in front of the setting sun is really an unintelligible error from a human perspective and makes it hard for me to personally trust these technologies (what other stupid, unintelligible, unpredictable mistakes are they prone to that I might not know about?)
Maybe not under that label, but sure, I'd expect we'll see some kind of self-driving taxi fleet. Elon has been quite clear that a big part of why he's pouring money into self-driving research is that he sees "you can flip a switch in the app and send your car off to earn money as a taxi" as part of a future value proposition that will make the total cost of ownership of a Tesla more competitive against other brands. And the fact that the auto division _is_ turning a profit on its core business (as compared to Uber and Lyft which have blown through billions and still can't seem to explain how they'll actually become profitable) means even if it takes decades, Tesla can deal with that. The only obvious competitor I can name off the top of my head is Waymo, which is similarly subsidized out of a separate profitable business. And Google doesn't have Tesla's data set. Google has GMaps, sure, but every Tesla vehicle is feeding back basically its entire driving record into the training set for Autopilot. Waymo execs would trade their firstborn for that kind of data.
No way is it viable... I've driven ubers before... people are nasty. The only possible way it would work would be if you were super super selective about users, and it would inevitably have all sorts of disparate impact issues.
Brussels might be an existing prototype of this! Europeans think it is quite American in that the desirable neighborhoods are suburban, but they're fairly dense and full of e scooters and streetcars.
I was wondering why you didn't mention the changing climate, need for lower energy intensive housing etc? It is as if the recent IPCC report "Code Red for humanity" was never written....
Most all of these types of articles seem to concentrate, look at, what might happen in the densely populated areas of the coasts etc. What about the vast majority of the counties and cities in the middle of the country with low density and plenty of land? Do you see those changing? If not, do you see the differences in lifestyle exacerbating political differences? As in why should someone in Iowa etc want to spend billions on Amtrak or Bart improvements, entities that will always run at huge losses and need to be forever funded by the federal government?
Do the housing bills like SB9 etc just mandate density or do they look at construction codes also- that is do they also facilitate modular and off site and more mechanized forms of home/housing construction?
The population is pretty much leaving anything that isn't a significant metropolitan area at this point. So few are moving to areas with lots of land and there isn't going to be much construction there of any type.
None of the current YIMBY laws mandate density. They just remove existing maximum, or cut red tape restricting new development.
I think it was Alon Levy who made the argument that this kind of missing middle housing is a bad equilibrium since it is dense enough to cause bad traffic but not dense enough to support frequent transit. I gather the counter argument here is that battery based e-bikes, scooters, etc. change that calculus?
I think that's correct. The bet Noah is making (and which I mostly agree with) is that people are willing to deal with worse traffic and transit in exchange for lower density, more greenery, cheaper housing, and a guaranteed parking space at your house. Which isn't unprecedented—this model of "dense suburbia" is basically present-day Los Angeles
Need to see some more thought on carless grocery shopping. How does this work in big cities now? My family goes shopping for once a week, and I have a hard time believing most people are going to carry a week's worth of food on a scooter. Do you fill up your backpack with canned goods or are people doing their shopping more often than once a week?
What percentage of people are going to keep going into grocery stores themselves for years to come? During the pandemic, a large number switched to getting most of their groceries delivered, and I think this will continue to grow. Especially as households cut down on car use.
In denser cities, the grocery stores are closer to home, so it's easier to walk anyways. Also, in cities where it's traditional to go more than once a week, there's a combination of more eating out (particularly in France) and just getting whatever you need to make whatever you wanted to have that particular evening/couple days.
I do it all the time, and my town's not all that dense.
I think in general, you just go to the store more often and only buy what you can reasonably carry home. Which isn't much of an inconvenience when the store is only a couple blocks away.
For bigger shopping (or longer trips) people just get it delivered these days. Or if you're on a budget, by granny cart, cargo bike or the bus.
Yeah... living in an European city now, I go shopping multiple times a week, and carry only as much as is possible in my backpack... maybe 10 items (less than 5 kg) at most...
For Boston at least: I don't know anyone who take a scooter to go shopping. I know some people have a foldable "shopping cart" that is sized to carry a couple of bags and could be taken on the subway or a bus. Most people I know though walk and take the bus/subway and then carry a couple bags back with them. I'm frankly not sure what elderly people do though - drive?
Probably more of the future looks like grocery delivery, especially in these environments, but that's hard to predict.
In terms of transit, it seems to me that we are missing a real use case for existing AI technologies. I live in Philly which has a large commuter train system that is relatively little used. The trains don’t run very often and if you end up waiting a long time for a ride. But the tracks are there. It seems to me that it would be pretty simple to add small bus sized trains that run with no crew at frequent intervals. If we can have cars that “nearly” drive themselves…driving on a track should be pretty easy. This is AI driving that is achievable today…not an ever receding “few years”.
BART in the San Francisco Bay Area was originally going to be automated, but they couldn’t make it work and ended up adding in train drivers. This was decades ago though.
I think you have a couple of plain mistakes here, but I'll also add some editorial comments.
1. "Around the commuter rail stations, larger shopping complexes will be common — big-box stores and malls. This is because the rail stations will be centrally located, zoning rules will likely be relaxed, and high-density housing will be located nearby." I don't think this is in the cards at all. Big-box stores and malls are not compatible with public transit of any kind, people go to big-box stores and malls to load up their pickup trucks and SUV's with large and voluminous products that are purchased rarely. Add to that that commuter rail stations are almost exclusively located in older, already built-out areas, making it near-impossible to meet the parking needs of big-box store or mall. There is another model of the urban big-box store or mall on top of a heavy rail station in the downtown zone that seems to sometimes work, but there are still and will continue to always be 100 of the suburban-type to every one of those for the simple reason that people go there to buy large and voluminous items. Even if people don't own cars, they'll rent them to do those trips, which are generally once a week at most, usually way more occasional than that.
2. The "suburban" neighborhoods in all of your pictures are not suburban neighborhoods at all, they are moderate (by American standards) West Coast urban neighborhoods. Putting aside San Francisco, the other big pre-war cities in the West (basically, Seattle, Portland, Oakland, Los Angeles, Long Beach, San Deigo) all have neighborhoods that have a mix of two/three-story detached single family homes, similary-sized duplexes with the occasional 3-4-5 story apartment building over ground floor retail mixed in, usually along the arterial streets. They're usually pretty nice, sometimes gentrifying, sometimes they've always been nice, occasionally tattered, and they offer that mix of a bit of space to live but also easy to walk, bike, ride transit for daily needs. But these are not suburbs. They "were" suburbs in the 1920's and 1930's., but not now. Maybe it's because these new laws are actually "new" (!) and so we haven't really seen what it looks like when a single-family, cul-de-sac neighborhood densifies, and the only photos are from neighborhoods that have gone through this kind of transition decades ago.
But densifying a true, one story, sprawling, single family suburb, where everyone has a 2-3 car garage, ample space in the front and back yard, etc., etc., is different. I've lived most of my life in these above-mentioned big city medium density neighborhoods and now live on a street full of single family homes and so I have some perspectives maybe to offer. For nearly everyone who moves to "the 'burbs", there is really one and only one reason: kids. The kids on my street and throughout my neighborhood have endless opportunities for outdoor play: they take over the street itself all day every day, but everyone also has backyard play structures and trampolines. Nearly every weekend someone is setting up a jumper, or a pool, or a slip n slide in the front yard. The nearby schoolyard is not locked outside of school hours, so kids are constantly over there playing as well, not to mention that it hosts the local Little League and AYSO. And wherever they go, it's all the same group of kids who all know each other and whose parents all know each other.
Thinking back to the medium-density urban neighborhoods I've lived in, it's not like that at all. There's always traffic on even the side streets, so kids cannot commandeer those streets. Very few have enough yard space to really do much with, and the local schools were gated, fenced and locked outside of school hours. Parks can be popular, but few parents are willing to let their under 10's go to a city park unsupervised, so kids can only go there to the extent that they have an adult to take them there, and city parks are also more anonymous than frontyards, backyards, streets and schoolyards, so it's not obvious that all their friends will be there any given time.
Density has long been proved to inversely correlated with "getting to know your neighbors". The fact that you might live 100 feet instead of 200 feet away from someone doesn't really increase the likelihood that you'll know that as much as the greater volume of people in a given space increases the likelihood that the standard of interaction will be more anonymous. This is a basic reality of human interaction that is seen in everything from small colleges to small towns, if you limit the universe of people, you're more likely to get to know those who are there.
Again, however, the logic of social interaction is highly variable between young singles and families. If you're single and "looking" obviously, there's an interest in running into as many people as possible in your local environment, increasing the odds of finding a mate. If you have a spouse and kids, your interest is much more in the quality of the people, in trust built over time, in the kind of deep multifaceted connections that involved when one is allowing others to influence the growth and development of your progeny.
But it's undeniable that density is coming to those kinds of single-family neighorhoods, and I think you're generally right that those neighborhoods will start to look like the pre-war medium density neighborhoods pictured. But I don't think that will really be a change in the general trend, all that will happen is that more territories within these big coastal metro areas will be dense and urban, but the population growth in the United States will continue to be shift to the Sunbelt, where new family-oriented exclusively single family neighborhoods will continue to be constructed and protected. The shift to remote work will intensify this trend, as fewer people are required to live in the big metros in order find work. It will look like the flight to the suburbs always has, except now increasingly at the scale of the nation, rather than region.
I think that a problem in the US and Canada is that generally speaking, relatively few multi-story apartment buildings get build within a "park area"...
e.g In Germany, most apartment buildings have at least their own courtyard (inside the "block" of apartment buildings), and many newer apartment buildings are built inside a "park area", with retail on the street sides, but a big courtyard/park in the middle, which is usually where additional apartment buildings are found...so more green space for children etc. to play around, but still urban...
Good point. I lived for 13 years in a place in Los Angeles where there were apartments and townhouses surrounding interior green spaces. It was a pretty good place for kids but nowhere near as good as the suburban cul de sac that I now live on.
This is basically Arlington, VA, where I live (or at least a lot of it). And it's one of the safest, most affluent, most diverse, and physically healthiest places in the country. I like this future!
Yeah, it's nothing new really... just a return to the natural pattern of development before euclidian zoning outlawed it. The future just sounds like any college town or streetcar suburb built before WWII. Sign me up!
I think people have this misperception that euclidian zoning is a recent thing. Which I think you intend to mean 'grid like' but Euclidian grids have been used to layout cities and towns for centuries. The 'Garden City' movement sought to break with the grid and introduce a variation of order but with the meandering streets. Laying out development patterns on a grid increases efficiencies vs. developing along cow paths.
The worst example of the meandering development and road layout is the Seven Corners in VA. What a shit-show to navigate and the built environment is terrible.
I lived a few blocks away from the Braddock Metro in Alexandria for several years and really enjoyed the walkability and access to the metro. Parts of Arlington are okay but it's still very much car-centric. I couldn't stand Potomac Yards, but there are other parts of Arlington that are nice. A lot of the DC/VA area has changed drastically since I moved away in 2009. Some good and some bad.
A bit sterile but still nice
Parts can be for sure. Crystal City is a bit plastic. North Arlington has a lot of character though! I live near the old but revitalizing area near Langston Blvd (formerly Lee Hwy), and it's exciting. So many local small businesses--restaurants, retail, and services--and the community input about the new development plans all reflects the stuff Noah's talking about here.
One element you left out: I think we're a few years out from the beginning of pressure for the state to enable Accessory Commercial Units in the way that it has enabled ADUs. For many of the R1 zones, 80+% of the people living in them are an unreasonable walking distance from any commercial hub. The only way you turn those into 15-minute neighborhoods is to allow corner stores and cafés to be opened by residents on formerly-exclusively-residential parcels, where it is presently illegal to operate a business.
https://www.strongtowns.org/journal/2020/8/15/accessory-commercial-units
When I was a kid, I remember my family got our hair cut in the basement studio of a walking-distance neighbor. I think we're going to see a lot more of that in the future.
The house next to my building had a basement barbershop! It's being torn down in favor of a 12-studio apartment building.
Wow! Moderate optimism, what an unusual stance for a blog. One quibble. You suggest that the newly densified inner suburbs will have shopping areas with mixed retail. I'm not sure bricks and mortar retail is likely to expand in any conceivable future. Lately, I've been confronting some age related incapacities. My shopping has become almost exclusively on-line. This is not because I want to make Jeff Bezos the first trillionaire. Trying to walk to and through a retail store is the equivalent of torture these days, hence my aversion to physical shopping. Of course, I also share many economists incomprehension regarding the notion that shopping is entertainment.
It probably won't be retail that sells physical goods, other than food and drink. Rather it'll be small business that cater experiences - fitness studios, bars and restaurants, co-working spaces, and perhaps a growing array of commercial spaces for hobbies, like board game stores and drunk-painting classes and axe-throwing bars.
If anything I think Noah is more optimistic than most political-economic commentators I read, but we've probably got different samples
The future: I like it.
1. Growth needs to be, and will be, organic and incremental. I've read some anti-NIMBY screeds (often East Coasters) promoting some nutty ideas; e.g., single-family city neighborhoods (like mine in L.A.) need to be replaced with 40-story complexes. No! For many reasons, it cannot, and will not, work that way. A mix of single-family homes, some with ADUs, and low-rise apartments is a much, much better way to go.
2. Not sure cost of car ownership in future goes up. Looking at TCO, EVs are more affordable than ICE cars. Not only fuel costs, but maintenance (no engine, transmission, etc.). Our EV expenses: <$20/month fuel, $0 maintenance.
3. Political ramifications of denser cities/suburbs may become more challenging that they already are today. We're about 80/20 urban/rural now, and political power of the less-dense areas are able to thwart the will of the majority. What if the split goes to 90/10? Will that flip some red states and make progress more possible, or will the stalemate continue, raising frustrations?
post, Noah. I’ve seen you mention micro-mobility as real game changer for how we will get around in this denser America. I’m an electric scooter owner myself in the suburbs of Boston and I think you are a bit too optimistic. Electric scooters still can’t ride in the rain and are basically unrideable in sub 40 degree weather. In Boston that means I can only ride mine 7 months of the year and when you subtract days that it rains it’s closer to 5 months out of 12. It’s still not practical for most people. Only reason I do it is because I don’t have my license yet; once I get that in November I’ll have a car.
In Los Angeles there is already a major difference in traffic congestion on rainy days vs dry days (in this case, largely because streets that don't see much rain actually build up a lot of oil and dust and become very slick when a mild rain falls on them). But if walking/biking/scooters become a bigger part of everyday travel, then I think the effects of rain on traffic will become a lot larger.
I don't know what an appropriate response is - convert some existing streets to bikeways, that open back up to cars on rainy days?
1. The issue with suburbs isn’t just zoning that allows denser housing, of equal importance is zoning that prevents businesses and commercial properties. The reason that some suburbs are unwalkable is because there is no place to walk to.
I lived in Europe for 12-years. Most of the time in smaller villages or cities. The great thing about Europe was there was always a little shop, bakery or pub within walking distance. Pittsburg suburbs and surrounding cities are like this. At least the older established ones.
If you really want to change the landscape, then this zoning issue needs to be addressed as well.
2. Commuter trains just aren’t going to be a thing across the United States. Sure they are established in some cities in the North East and a few big west coast cities, but across the US they are just non-existent. At best bus service will fill in some of the gaps, but I doubt it.
3. I’m surprised you didn’t mention self-driving cars. Now… let me say up front… self-driving taxis will never be a thing. People are nasty. They would quickly become disgusting. I’ve seen some people say that they could have automated cleaning, but that would mean they would have to have hard surfaces which would basically be uncomfortable. Think about your average bus, and they have a bus driver present.
4. I think the scooters and the bikes or an interesting proposition/idea. To tell the truth I never really thought of them as a commuter item. It’s partially limited by the weather, especially in winter. But it actually has potential, but only if you can at the same time increase bike paths which I assume they will use as well.
5. I’ve traveled the world, except for Asia. I need to get to Japan to check out how these Japanese cities you talk about really are. In someways what you described it sounds like parts of South America. I’m in Argentina right now. I know ironically, a lot of cities in urgent Tina and other parts of South America, high-rise apartments are more upper class, where the poor working class people tend to live in smaller single-family houses, even if they are densely packed together. I asked one of my engineers why this is, and he says it’s because of security. A high-rise apartment usually has a gate and enclosed area and a full-time security guard to watch.
6. Anyway. Great post. I wish you had some more photos, but aerial shots showing a few blocks. It would help me visualize better.
Addendum. I forgot to add about self driving cars. They could change the landscape in one way without self driving taxis. Imagine a family that has three or four cars. Like my house. We could probably replace four of our cars with one self-driving car. It could drop me off at the airport at 6 AM, returned to take my wife to work, return to take my daughter to work, then pick my wife, etc. This is were self driving cars will become very useful. Hell, you wouldn’t even need to park them at your house. You can keep it in a parking lot several miles away, and then just call it to come to you when you needed. Especially in the denser cities were people might like to have cars, but they’re parked everywhere, such as New York City.
Unfortunately I think the self driving future is still going to be decades away unless infrastructure is built specifically for them and them only. It's likely a problem that requires close to human level machine intelligence to safely solve the way our current road systems are set up, and I won't be surprised if we don't see that in our lifetimes.
I don't know what the non-train transit solution would be. I love cycling, and would prefer to ride my bike around a city to get around and for lite errands, but as others have pointed out, that's tough in Fall/Winter in a lot of areas.
I do share your skepticism about whether most towns in the west would be willing to put in lite rail infrastructure but the irony is that trains built the west. Done right, I think they could be well-received. Seattle has put one in and it's been expanded. Think Boise would bite?
No way Boise bites. It would be cool, but the cost is just so prohibitive. Apparently we did use to have a trolley system. There are plans for a transportation corridor along State Street, with raised bus stops and rapid transit going into downtown, but Boise just doesn't have the centralized jobs downtown to justify it. So many of our new residents are moving to Boise and working remote. Lots of apartments and townhouses going up downtown, but its a lifestyle thing for access to restaurants and bars and the University. Not so much for jobs. Plus, they plan on putting a huge homeless center right on the main bus route between outskirts of downtown and downtown, so that will probably end up negatively effecting what little ridership their is.
I agree with you about self-driving cars. I'm not sure "decades" but at least a decade away.
Personally I'm a fan of the 10-minute city block idea (cant remember the official name), but Noah has written about it before. Sort of a lot of self contained mini-downtowns.
To clarify, I wish we had a train/trolley system, but only if it worked until the late late evenings. And all day on weekends. I just know we wouldn't be able to justify it
Yeah, I'm extremely skeptical that safe "full self driving" is going to arrive in the next five years. I feel confident that _eventually_ a computer will drive better than almost any human, the same way computers now fly better than humans. But the environment on a busy urban street is wildly more complicated than anything an airplane has to deal with. I think we're more like 20-25 years out from being able to cope with the unexpected.
I think the greatest challenge for self-driving cars is that while we shrug at ~40,000 traffic deaths per year for human-driven cars, 40 deaths from self-driving cars might drive the industry into regulatory oblivion. It's the shock of the new and the strange.
I could see a time where we shrug our shoulders at, say, 4000 deaths/year from self-driven vehicles, but I'm not sure how we get from now to then.
Yep... we see that with Covid as well.
I think the challenge with street traffic vs air traffic is greater than that. A lot of what we do when we navigate street traffic relies upon our background understanding of peoples' intentions. I think these are the majority of the cases too, not just edge cases. The problem with understanding intention is that to really understand it, _as such_ it takes an agent with at least the cognitive capacities of a 3-year old - though our laws suggest 16-year olds.
There's a view in philosophy of mind and cogsci called "mental holism." It's the idea that you can't really get a foothold on a working understanding of other minds and mental states without understanding _almost all of it_. If it's true it will present a whole lot of problems for AI use cases that depend upon understanding mental states. I happen to think it's true, so that's why I'm skeptical. All the traffic data in the world would still massively underdetermine the data you need to understand even basic mental states that other road users have, like the desire to not die, the desire to get where they're going (ideally efficiently), the desire to cooperate for safety, and a near-infinite set of other desires, beliefs, etc. that go into our understanding of what it is to be operating a vehicle on the road.
Heh, you think I'm under-stating the problem, I think you're over-stating it. My undergrad degrees in are cog sci (with concentrations in philosophy of mind and lingustics) and computer science. Basically I studied Natural Language Processing, and worked on various natural language products at IBM, Microsoft, and Motorola, for quite a few years. I studied in the lab of Fred Jelinek, who famously joked that every time he fired a linguist, the accuracy of his language model got better. You seem to be asking the driving model to "think" about driving the way a person does. It won't, and that won't stop it from doing the job better, eventually, any more than it stops the best chess AIs from playing chess better than a person, even though it's obvious that they're going about the task very differently.
You really _don't_ have to know _why_ people are doing what they're doing. You just need a broad enough training set to accurately predict _what_ they're doing, regardless of _why_. The problem is that the set of possible things that can happen on the ground is so much broader that it's going to take _vastly_ more training data and processing power to generalize from a training set, so that novel situations are at least _enough like_ something from the training set that we can match a pattern and respond appropriately. We're _already_ at the point where for highway driving, the computer is probably better than a person at least 99% of the time; it's only really weird edge cases that fail to parse. But surface streets are still much worse than that, and even for highways, a 1% error rate is still way too high, that has to get improved by maybe three orders of magnitude before it's safe to deploy truly driverless vehicles.
Guess we'll just have to wait and see!
I sort of think "what" an agent is doing in a bounded game like chess or go is waaaay easier to understand than what they're doing in the less bounded games of life. There's a view that you can't really understand what an agent is doing without understanding their mental states. E.g. you might be able to predict that the human-shaped object will move from one side of the street to the other at a predictable rate, but you won't know that it's crossing the road, to get to the other side, to get ice cream. It is very plausible that understanding action _as such_ and not just as physically indistinguishable brute physical movement requires understanding intention, and that understanding intentions can't be done piecemeal in an atomistic or even molecularist kind of way.
That's why I happen to think it's going to be really hard. FWIW Wittgenstein had a big influence on my thinking here. On the idea of "thinking like us" being a necessary condition for recognizing some activity as thinking, I guess I just sort of take that as trivially true. Insofar as we can justifiably call some activity "thinking" surely it is in virtue of its similarity to something we do; doesn't have to be exactly the same, just sufficient enough for us to even recognize it as such.
The other thing with self driving safety that I think people really care about is the intelligibility of incidents where they do occur. It's perfectly predictable the kinds of human conditions that make incidents more likely (DWIs, speeding, unkept equipment, distracted driving, etc.). We can monitor, control, and attempt to prevent those things to reduce our risk. But driving full speed under a semi truck in full view because it was crossing in front of the setting sun is really an unintelligible error from a human perspective and makes it hard for me to personally trust these technologies (what other stupid, unintelligible, unpredictable mistakes are they prone to that I might not know about?)
I wouldn't bet against you. So question though.... when it is a thing, will there be self-driving ubers?
Maybe not under that label, but sure, I'd expect we'll see some kind of self-driving taxi fleet. Elon has been quite clear that a big part of why he's pouring money into self-driving research is that he sees "you can flip a switch in the app and send your car off to earn money as a taxi" as part of a future value proposition that will make the total cost of ownership of a Tesla more competitive against other brands. And the fact that the auto division _is_ turning a profit on its core business (as compared to Uber and Lyft which have blown through billions and still can't seem to explain how they'll actually become profitable) means even if it takes decades, Tesla can deal with that. The only obvious competitor I can name off the top of my head is Waymo, which is similarly subsidized out of a separate profitable business. And Google doesn't have Tesla's data set. Google has GMaps, sure, but every Tesla vehicle is feeding back basically its entire driving record into the training set for Autopilot. Waymo execs would trade their firstborn for that kind of data.
No way is it viable... I've driven ubers before... people are nasty. The only possible way it would work would be if you were super super selective about users, and it would inevitably have all sorts of disparate impact issues.
I could see self-driving commercial carpools.
Brussels might be an existing prototype of this! Europeans think it is quite American in that the desirable neighborhoods are suburban, but they're fairly dense and full of e scooters and streetcars.
That's not the case anymore. Being close to a "superstar" city and how easy it is to get to the city center is more important now.
I was wondering why you didn't mention the changing climate, need for lower energy intensive housing etc? It is as if the recent IPCC report "Code Red for humanity" was never written....
https://news.un.org/en/story/2021/08/1097362
Oh that's a good topic, but I have another big post about that coming soon...!
Yup, this is why the Sierra Club has guidance that development should focus on infill -- a world where we've resolved the climate crisis is a world with a significantly denser pattern of residences. https://www.sierraclub.org/redwood/blog/2019/09/sierra-club-updates-urban-infill-policy
Sounds like America will be where Melbourne is now. Hopefully by then Melbourne will be further along in the fight against nimbyism
Most all of these types of articles seem to concentrate, look at, what might happen in the densely populated areas of the coasts etc. What about the vast majority of the counties and cities in the middle of the country with low density and plenty of land? Do you see those changing? If not, do you see the differences in lifestyle exacerbating political differences? As in why should someone in Iowa etc want to spend billions on Amtrak or Bart improvements, entities that will always run at huge losses and need to be forever funded by the federal government?
Do the housing bills like SB9 etc just mandate density or do they look at construction codes also- that is do they also facilitate modular and off site and more mechanized forms of home/housing construction?
The population is pretty much leaving anything that isn't a significant metropolitan area at this point. So few are moving to areas with lots of land and there isn't going to be much construction there of any type.
None of the current YIMBY laws mandate density. They just remove existing maximum, or cut red tape restricting new development.
Excellent questions. Being in Wisconsin I was wondering the same thing.
I think it was Alon Levy who made the argument that this kind of missing middle housing is a bad equilibrium since it is dense enough to cause bad traffic but not dense enough to support frequent transit. I gather the counter argument here is that battery based e-bikes, scooters, etc. change that calculus?
I think that's correct. The bet Noah is making (and which I mostly agree with) is that people are willing to deal with worse traffic and transit in exchange for lower density, more greenery, cheaper housing, and a guaranteed parking space at your house. Which isn't unprecedented—this model of "dense suburbia" is basically present-day Los Angeles
Example post: https://pedestrianobservations.com/2018/09/10/is-missing-middle-really-missing/
Need to see some more thought on carless grocery shopping. How does this work in big cities now? My family goes shopping for once a week, and I have a hard time believing most people are going to carry a week's worth of food on a scooter. Do you fill up your backpack with canned goods or are people doing their shopping more often than once a week?
What percentage of people are going to keep going into grocery stores themselves for years to come? During the pandemic, a large number switched to getting most of their groceries delivered, and I think this will continue to grow. Especially as households cut down on car use.
In denser cities, the grocery stores are closer to home, so it's easier to walk anyways. Also, in cities where it's traditional to go more than once a week, there's a combination of more eating out (particularly in France) and just getting whatever you need to make whatever you wanted to have that particular evening/couple days.
I do it all the time, and my town's not all that dense.
I think in general, you just go to the store more often and only buy what you can reasonably carry home. Which isn't much of an inconvenience when the store is only a couple blocks away.
For bigger shopping (or longer trips) people just get it delivered these days. Or if you're on a budget, by granny cart, cargo bike or the bus.
Yeah... living in an European city now, I go shopping multiple times a week, and carry only as much as is possible in my backpack... maybe 10 items (less than 5 kg) at most...
For Boston at least: I don't know anyone who take a scooter to go shopping. I know some people have a foldable "shopping cart" that is sized to carry a couple of bags and could be taken on the subway or a bus. Most people I know though walk and take the bus/subway and then carry a couple bags back with them. I'm frankly not sure what elderly people do though - drive?
Probably more of the future looks like grocery delivery, especially in these environments, but that's hard to predict.
Noah,
In terms of transit, it seems to me that we are missing a real use case for existing AI technologies. I live in Philly which has a large commuter train system that is relatively little used. The trains don’t run very often and if you end up waiting a long time for a ride. But the tracks are there. It seems to me that it would be pretty simple to add small bus sized trains that run with no crew at frequent intervals. If we can have cars that “nearly” drive themselves…driving on a track should be pretty easy. This is AI driving that is achievable today…not an ever receding “few years”.
BART in the San Francisco Bay Area was originally going to be automated, but they couldn’t make it work and ended up adding in train drivers. This was decades ago though.
I think you have a couple of plain mistakes here, but I'll also add some editorial comments.
1. "Around the commuter rail stations, larger shopping complexes will be common — big-box stores and malls. This is because the rail stations will be centrally located, zoning rules will likely be relaxed, and high-density housing will be located nearby." I don't think this is in the cards at all. Big-box stores and malls are not compatible with public transit of any kind, people go to big-box stores and malls to load up their pickup trucks and SUV's with large and voluminous products that are purchased rarely. Add to that that commuter rail stations are almost exclusively located in older, already built-out areas, making it near-impossible to meet the parking needs of big-box store or mall. There is another model of the urban big-box store or mall on top of a heavy rail station in the downtown zone that seems to sometimes work, but there are still and will continue to always be 100 of the suburban-type to every one of those for the simple reason that people go there to buy large and voluminous items. Even if people don't own cars, they'll rent them to do those trips, which are generally once a week at most, usually way more occasional than that.
2. The "suburban" neighborhoods in all of your pictures are not suburban neighborhoods at all, they are moderate (by American standards) West Coast urban neighborhoods. Putting aside San Francisco, the other big pre-war cities in the West (basically, Seattle, Portland, Oakland, Los Angeles, Long Beach, San Deigo) all have neighborhoods that have a mix of two/three-story detached single family homes, similary-sized duplexes with the occasional 3-4-5 story apartment building over ground floor retail mixed in, usually along the arterial streets. They're usually pretty nice, sometimes gentrifying, sometimes they've always been nice, occasionally tattered, and they offer that mix of a bit of space to live but also easy to walk, bike, ride transit for daily needs. But these are not suburbs. They "were" suburbs in the 1920's and 1930's., but not now. Maybe it's because these new laws are actually "new" (!) and so we haven't really seen what it looks like when a single-family, cul-de-sac neighborhood densifies, and the only photos are from neighborhoods that have gone through this kind of transition decades ago.
But densifying a true, one story, sprawling, single family suburb, where everyone has a 2-3 car garage, ample space in the front and back yard, etc., etc., is different. I've lived most of my life in these above-mentioned big city medium density neighborhoods and now live on a street full of single family homes and so I have some perspectives maybe to offer. For nearly everyone who moves to "the 'burbs", there is really one and only one reason: kids. The kids on my street and throughout my neighborhood have endless opportunities for outdoor play: they take over the street itself all day every day, but everyone also has backyard play structures and trampolines. Nearly every weekend someone is setting up a jumper, or a pool, or a slip n slide in the front yard. The nearby schoolyard is not locked outside of school hours, so kids are constantly over there playing as well, not to mention that it hosts the local Little League and AYSO. And wherever they go, it's all the same group of kids who all know each other and whose parents all know each other.
Thinking back to the medium-density urban neighborhoods I've lived in, it's not like that at all. There's always traffic on even the side streets, so kids cannot commandeer those streets. Very few have enough yard space to really do much with, and the local schools were gated, fenced and locked outside of school hours. Parks can be popular, but few parents are willing to let their under 10's go to a city park unsupervised, so kids can only go there to the extent that they have an adult to take them there, and city parks are also more anonymous than frontyards, backyards, streets and schoolyards, so it's not obvious that all their friends will be there any given time.
Density has long been proved to inversely correlated with "getting to know your neighbors". The fact that you might live 100 feet instead of 200 feet away from someone doesn't really increase the likelihood that you'll know that as much as the greater volume of people in a given space increases the likelihood that the standard of interaction will be more anonymous. This is a basic reality of human interaction that is seen in everything from small colleges to small towns, if you limit the universe of people, you're more likely to get to know those who are there.
Again, however, the logic of social interaction is highly variable between young singles and families. If you're single and "looking" obviously, there's an interest in running into as many people as possible in your local environment, increasing the odds of finding a mate. If you have a spouse and kids, your interest is much more in the quality of the people, in trust built over time, in the kind of deep multifaceted connections that involved when one is allowing others to influence the growth and development of your progeny.
But it's undeniable that density is coming to those kinds of single-family neighorhoods, and I think you're generally right that those neighborhoods will start to look like the pre-war medium density neighborhoods pictured. But I don't think that will really be a change in the general trend, all that will happen is that more territories within these big coastal metro areas will be dense and urban, but the population growth in the United States will continue to be shift to the Sunbelt, where new family-oriented exclusively single family neighborhoods will continue to be constructed and protected. The shift to remote work will intensify this trend, as fewer people are required to live in the big metros in order find work. It will look like the flight to the suburbs always has, except now increasingly at the scale of the nation, rather than region.
I think that a problem in the US and Canada is that generally speaking, relatively few multi-story apartment buildings get build within a "park area"...
e.g In Germany, most apartment buildings have at least their own courtyard (inside the "block" of apartment buildings), and many newer apartment buildings are built inside a "park area", with retail on the street sides, but a big courtyard/park in the middle, which is usually where additional apartment buildings are found...so more green space for children etc. to play around, but still urban...
Good point. I lived for 13 years in a place in Los Angeles where there were apartments and townhouses surrounding interior green spaces. It was a pretty good place for kids but nowhere near as good as the suburban cul de sac that I now live on.
This is basically Hillsboro, OR and quickly becoming Beaverton, OR, as well. This isn't far from my house:
https://en.wikipedia.org/wiki/Orenco_Station,_Hillsboro,_Oregon#/media/File:Orenco_Station_Plaza_and_Rowlock_Apartments_(2016).jpg
The future vision for the US seems very similar to what Australia is now