Transit-Oriented Development and Rail Capacity
Hayden Clarkin, inspired by the ongoing YIMBYTown conference in New Haven, asks me about rail capacity on transit-oriented development, in a way that reminds me of Donald Shoup’s critique of trip generation tables from the 2000s, before he became an urbanist superstar. The prompt was,
Is it possible to measure or estimate the train capacity of a transit line? Ie: How do I find the capacity of the New Haven line based on daily train trips, etc? Trying to see how much housing can be built on existing rail lines without the need for adding more trains
To be clear, Hayden was not talking about the capacity of the line but about that of trains. So adding peak service beyond what exists and is programmed (with projects like Penn Station Access) is not part of the prompt. The answer is that,
- There isn’t really a single number (this is a trip generation question).
- Moreover, under the assumption of status quo service on commuter rail, development near stations would not be transit-oriented.
Trip generation refers to the formula connecting the expected car trips generated by new development. It, and its sibling parking generation, is used in transportation planning and zoning throughout the United States, to limit development based on what existing and planned highway capacity can carry. Shoup’s paper explains how the trip and parking generation formulas are fictional, fitting a linear curve between the size of new development and the induced number of car trips and parked cars out of extremely low correlations, sometimes with an R^2 of less than 0.1, in one case with a negative correlation between trip generation and development size.
I encourage urbanists and transportation advocates and analysts to read Shoup’s original paper. It’s this insight that led him to examine parking requirements in zoning codes more carefully, leading to his book The High Cost of Free Parking and then many years of advocacy for looser parking requirements.
I bring all of this up because Hayden is essentially asking a trip generation question but on trains, and the answer there cannot be any more definitive than for cars. It’s not really possible to control what proportion of residents of new housing in a suburb near a New York commuter rail stop will be taking the train. Under current commuter rail service, we should expect the overwhelming majority of new residents who work in Manhattan to take the train, and the overwhelming majority of new residents who work anywhere else to drive (essentially the only exception is short trips on commuter rail, for example people taking the train from suburbs past Stamford to Stamford; those are free from the point of view of train capacity). This is comparable mode choice to that in the trip and parking generation tables, driven by an assumption of no alternative to driving, which is correct in nearly all of the United States. However, figuring out the proportion of new residents who would be commuting to Manhattan and thus taking the train is a hard exercise, for all of the following reasons:
- The great majority of suburbanites do not work in the city. For example, in the Western Connecticut and Greater Bridgeport Planning Regions, more or less coterminous with Fairfield County, 59.5% of residents work within one of these two regions, and only 7.4% work in Manhattan as of 2022 (and far fewer work in the Outer Boroughs – the highest number, in Queens, is 0.7%). This means that every new housing unit in the suburbs, even if it is guaranteed the occupant works in Manhattan, generates demand for more destinations within the suburb, such as retail and schools.
- The decision of a city commuter to move to the suburbs is not driven by high city housing prices. The suburbs of New York are collectively more expensive to live in than the city, and usually the ones with good commuter rail service are more expensive than other suburbs. Rather, the decision is driven by preference for the suburbs. This means that it’s hard to control where the occupant of new suburban housing will work purely through TOD design characteristics such as proximity to the station, streets with sidewalks, or multifamily housing.
- Among public transportation users, what time of day they go to work isn’t controllable. Most likely they’d commute at rush hour, because commuter rail is marginally usable off-peak, but it’s not guaranteed, and just figuring the proportion of new users who’d be working in Manhattan at rush hour is another complication.
All of the above factors also conspire to ensure that, under the status quo commuter rail service assumption, TOD in the suburbs is impossible except perhaps ones adjacent to the city. In a suburb like Westport, everyone is rich enough to afford one car per adult, and adding more housing near the station won’t lower prices by enough to change that. The quality of service for any trip other than a rush hour trip to Manhattan ranges from low to unusable, and so the new residents would be driving everywhere except their Manhattan job, even if they got housing in a multifamily building within walking distance of the train station.
This is a frustrating answer, so perhaps it’s better to ask what could be modified to ensure that TOD in the suburbs of New York became possible. For this, I believe two changes are required:
- Improvements in commuter rail scheduling to appeal to the growing majority of off-peak commuters as well as to non-commute trips. I’ve written about this repeatedly as part of ETA but also the high-speed rail project for the Transit Costs Project.
- Town center development near the train station to colocate local service functions there, including retail, a doctor’s office and similar services, a library, and a school, with the residential TOD located behind these functions.
The point of commercial and local service TOD is to concentrate destinations near the train station. This permits trip chaining by transit, where today it is only viable by car in those suburbs. This also encourages running more connecting bus service to the train station, initially on the strength of low-income retail workers who can’t afford a car, but then as bus-rail connections improve also for bus-rail commuters. The average income of a bus rider would remain well below that of a driver, but better service with timed connections to the train would mean the ridership would comprise a broader section of the working class rather than just the poor. Similarly, people who don’t drive on ideological or personal disability grounds could live in a certain degree of comfort in the residential TOD and walk, and this would improve service quality so that others who can drive but sometimes choose not to could live a similar lifestyle.
But even in this scenario of stronger TOD, it’s not really possible to control train capacity through zoning. We should expect this scenario to lead to much higher ridership without straining capacity, since capacity is determined by the peak and the above outline leads to a community with much higher off-peak rail usage for work and non-work trips, with a much lower share of its ridership occurring at rush hour (New York commuter rail is 67-69%, the SNCF part of the RER and Transilien are about 46%, due to frequency and TOD quality). But we still have no good way of controlling the modal choice, which is driven by personal decisions depending on local conditions of the suburb, and by office growth in the city versus in the suburbs.
Pedestrian Observations 
From figure 6 and 7 on https://www.gov.uk/government/statistics/rail-factsheet-2024/rail-factsheet-2024 in table view London is 45% peak arrivals and the regional cities are 27% peak arrivals.
Isn’t this question interesting but inherently quite academic and (in a well run world) quite detached from real world decision making?
You don’t need to know the capacity of the current train service, or even the capacity of the rail line with hypothetical improved service. You should just build more buildings without worrying about capacity of rail. Crowded trains is a better problem to have than empty trains.
Actually this brings me a question. I read from Jarret Walker that half empty buses aren’t an actual issue, because the cost of running buses is mostly the cost of the driver salary, therefore you should just give him the biggest vehicle you have and avoid the risk of having too small buses: https://humantransit.org/2020/04/whats-wrong-with-an-empty-bus.html
Does this hold true in trains as well? Or train vehicle maintenance is expensive enough that running shorter trains represents a significant saving? Can we say “the cost of trains is mostly the infrastructure, therefore let’s just put an over-demand capacity of trains per hour to maximize the amount of return for the fixed investment”?
Maybe the discussion is not about capacity, but about how do you calculate the benefit of people changing from can to mass transit (super hard to estimate I guess), and how this links with the function of ridership depending on frequency?
Longer trains is somewhat more expensive because you have to expand platforms – all of them, which are already longer than even a very large bus (in most cases). This often means expanding stations and moving the switches before the station, and in turn buying a lot more real estate. Over the next 100 years of the expanding platforms the cost of operations is probably going to be more than that cost, but we don’t know for sure – lines have been abandoned, ridership may never develop, or something else not even thought of might go wrong.
If money is no object you would run short buses only until you are running a bus every 5 minutes and still are running out of seats, only then you increase the size. However in the real world money is a limit and so you follow Jarret’s large bus everywhere plan where possible (in many cases the roads don’t allow a large bus to take the corners thus forcing shorter buses), since a large bus isn’t must more and you don’t have the money to increase service until the large bus is near full anyway – if a short bus is full you can’t afford to run two short buses in general.
In most cases trains should be looking to upgrade to fully automated systems (most cases – trams often run in situations where you need a human to hit the brakes, but most trains couldn’t stop in time anyway and so rely on guard rails, grade separation and otherwise have absolute priority over anything else and so it is easy for a computer to automate and do better than humans). As such you shouldn’t have driver costs, and this means you should aim to run as many trains as possible before looking at making trains longer. This is different from a bus where we still don’t have automated drivers and so cost means we need to run longer buses instead of more frequent. (as soon as automated buses come I will look at data and likely change my recommendations – this is a place to keep watching, though I will admit to having once predicted they would be ready about 5 years ago which obviously didn’t happen)
Partial door opening is also an option that helps.
Most of the places we have discussed are already doing this at peak – only really are the high speed lines often not being run at capacity at peak.
Even planes which have had ‘autopilot’ for half a century or more still have a driver on board.
In this context, partial door opening hurts more than it helps. If a station has nonnegligible alighting passenger traffic, it will very frequently be the case that at least one passenger has to do substantial internal circulation before they can alight, meaningfully extending dwell times, which are the bottleneck on train frequency (unless you Swiss up the station, but if you can do that, you almost certainly can just extend the damn platform). Frequent riders will also curse you and your ancestors for having to listen to the partial-door-opening announcement every single time.
Train drivers generally speaking tend to carry one-ish order of magnitude more passengers than bus drivers, so their costs are correspondingly less interesting.
They do selective door opening at quite a few stations in London. It’s fine and services aren’t typically delayed – https://clondoner92.blogspot.com/2024/09/list-of-london-underground-overground.html?m=1 has a full list.
Also as per https://tfl.gov.uk/corporate/transparency/freedom-of-information/foi-request-detail?referenceId=FOI-1627-2324 there were only 14 complaints about selective door opening. One per 10 million customers or something. g
My degree is in human / machine interaction. (even though I never use it). I would love to figure out how to make partial doors opening work in the real world without the downsides. It sounds like a fun problem. However in the end just expand the platforms it still a better answer even if you can make it work.
Platform length is not the only issue with partial opening though. You still need the switches far enough away that the stopped train isn’t blocking other traffic. The large complex “throats” that allow any train to use any platform are a bad idea, but you still need every track to have a couple platforms options so you can do maintenance without shutting the whole system down and at times you will need to operate a switch while a train is at a platform and that can’t happen if the train is on the switch. In many cases the switches are too close and so you need to rebuild a lot of track before you can do partial opening. Of course you might not have this problem at the station in question, but it is still something you need to consider before you think about partial opening.
@Henry, I think fixing the switches etc is cheaper than extending the platforms – as that is why they have chosen that approach in Britain so much.
And some of the stops with short platforms like Cutty Sark and Camden Town are very touristy.
Vancouver skytrain has been driverless (proper driverless with no staff on the train) for 40 years
Moving switching are very much it depends. If you own the land it is cheap to build more track and move the switch elsewhere. However if the station doesn’t own the land you have to buy it and that gets tricky you may as well just expand the whole station while you are at it – you at least get something to show the public for your taking.
In most cases trains should be looking to upgrade to fully automated systems (most cases – trams often run in situations where you need a human to hit the brakes, … this is different from a bus where we still don’t have automated drivers…
Even planes which have had ‘autopilot’ for half a century or more still have a driver on board.
Vancouver skytrain has been driverless (proper driverless with no staff on the train) for 40 years
Waymo and Zoox are real things, that right now as I write this or you read this have fully automated cars picking up passengers in mixed traffic. Straight and level autopilot has existed for over a century; the first instance of an aircraft taking off, flying across the Atlantic, and landing fully on autopilot was over 75 years ago. Modern planes do not require a pilot anymore than an elevator does. London’s Victoria line was fully automatic 57 years ago when it opened (still with “drivers” though), and Kobe’s Port Island Line was built fully driverless more than 45 years ago.
As Jordi points out below, driverless busses and trams, even in mixed traffic on city streets, is an easier problem than what Waymo and Zoox are doing. Driverless metros/mainline trains shouldn’t even be a question, the technology was mature over a generation ago. Transit, at least in the US, is hitting a crisis in many areas due to the the shift to work from home as a result of Covid. The loss in ridership/revenue is pushing them over the “fiscal cliff”. The fact that almost all of these systems are still using drivers for fixed route or even grade separated transit when driverless cars exist is going to push them to irrelevance.
As everyone notes, drivers are the major cost driver for most service. Waymo is very expensive right now because it is a niche product in a limited area that is overcharging tourists for the novelty because they can, but driverless services are going to eat transit agencies for breakfast, lunch and dinner in most areas. Even if it isn’t exactly as cheap as a bus/train ride, the convenience of a vehicle that comes to you with little waiting and takes you directly to your destination will devastate fixed route transit. Only in areas of very high density/demand where car/van sized vehicles don’t scale will traditional transit be viable.
But there is potential for transit services. Also as everyone notes, driverless means you can run service much more frequently, with smaller vehicles. With smaller vehicles you can build through lanes/tracks and turnouts instead of stations. Demand assignment technology for elevators has been widespread for about 20 years; instead of getting on an elevator and choosing your destination, you select your destination first and are told what elevator to get on for efficient operations. Apply this to a driverless transit system, and when arriving at a stop/turnout you select your destination, and you watch several vehicles go by until one stops to pick you up, with the exchange that the vehicle goes non-stop to your destination (or makes a few stops to pick up others with the same destination). In this way a very inexpensive street only right of way can provide service as fast as most subways, with the frequency of a vehicle every minute or so to minimize waiting (even if some pass you by) combined with close to non-stop service speeding travel.
Such a system can also carry as many or more passengers than many train systems. The entry to Manhattan that carries (carried?) the most people at rush hour isn’t a subway tube but the Lincoln Tunnel due to the Exclusive Bus Lane (XBL). Trains have space between them, while the XBL at peak islike a single long train with busses bumper to bumper. This works because the busses pour into the Port Authority Terminal and don’t have to stop at a station en route (if they did stop it would introduce gaps). A driverless system with turnouts, however, can have near continuous flow as vehicles in transit bypass the stop and vehicles dropping off or picking up. Most light rail/pre-metro/medium capacity transit services should be made obsolete by very frequent driverless bus/van lines.
At the highest traffic levels a system like this won’t scale. A bus every 30 seconds won’t carry as many people as a full subway train every 90. The larger you make the vehicles, the more people you carry, but the harder it is to offer demand assignment and few stops along the way (more people on the bus means more possible destinations, even if the computer is grouping them efficiently). The busiest, densest metro systems are not in danger of replacement (but should still be driverless). In general though, most average bus lines must be replaced by driverless vans operating on fixed routes, and most rail lines must be replaced by driverless busses operating on fixed routes, to include busses in tunnels or elevated ROW. If transit systems don’t make this shift, they will collapse.
If Waymo was actually cheaper to operate than a taxi it would be worldwide news like Google Deepmind’s programming breakthrough yesterday that was genuinely impressive (of course we also have to ask how many millions of dollars that cost to pull off – because if it wasn’t that expensive they would have said what it cost publicly).
@Onux
This is a very interesting comment and you make a tightly argued case.
Respectfully, however, I see some issues.
The first is that modern passenger aircraft absolutely do require a pilot. (I’m an engineer in the aerospace industry.) Autopilots do not have the authority to navigate traffic, especially in congested airspace or in relatively bad weather, because not all aircraft are equipped with the same level of compliant navigation radios or traffic separation devices. UAS are restricted to small aircraft and are not generally fully autonomous. Autopilots excel at routine maneuvers such as – as you yourself pointed out – straight and level flight, or climbing and descending, or landings in good weather with no crosswind, but still require steering commands from a pilot. Autopilots cannot manage emergencies. For instance, loss of airspeed indication (one consequence of an iced-up pitot-static system) will cause an A330 to revert from Normal Law (its typical operating regime) to Alternate Law, with far fewer protections.
As a result, we are a long way from fully autonomous unmanned passenger aircraft, and not because of some parochial concern like pilots’ unions trying to save jobs. There are real safety issues at stake here, and real engineering challenges to overcome, before the aerospace industry will fly 250+ passengers on a vehicle without a human in the cockpit.
So, too, do I feel that self-driving vehicles in mixed traffic will have a very slow rollout. I’m also not convinced that Waymo and Zoox vehicles are in fact fully autonomous at all times or capable of navigating the entirety of the cities where they’ve been rolled out. Amazon and Google may claim that’s the case; they aren’t trustworthy. Either all cars have to communicate with each other and roadway infrastructure using radio datalinks for navigation, traffic flow and separation, and deconfliction, or a self-driving car will need to extract that data from its environment using cameras and pattern recognition; and given the infinite permutations of the latter case, I believe we’re a long way off before widespread adoption. That’s before pedestrians enter the mix.
And this is under an ideal scenario where Amazon and Google actually apply capital to solve this problem, which is not the case now – their profits come from extracting rent and enshittification, and that’s not going to change anytime soon. Their operating philosophy is opposite to what the aerospace industry refers to as Design Assurance Level (https://en.wikipedia.org/wiki/DO-178C) – where software related to flight criticality must be rigorously designed and tested such that failure rates are minuscule; Big Tech routinely rolls out buggy software and treats regulations as legal obstacles rather than safety constraints. Then there’s Tesla, run by a drugged-up madman staring down a Chinese EV juggernaut. Tesla has no future.
But you do have a great point that autonomous vehicles in non-mixed traffic are a fantastic use case. A fixed guideway removes the majority of the permutations previously mentioned, and by limiting excursions from the guideway, you have removed the need to solve a lot of the navigation and deconfliction problems. You’re absolutely correct that this represents an opportunity for transit providers.
Great comment, very thought-provoking.
J.G. the driver on the Victoria line is basically for emergencies too. That said there is a principle with the railways where either you have a guard or you staff the stations. All of the metro lines where they are automating the trains have platform edge doors (with an handful of grandfathered exceptions like the Vancouver skytrain) and staffed stations. So the station staff will be handling any emergencies.
And anyway on a metro (or a metro-like mainline like the Yamanote line) you wouldn’t stop between stations for an emergency anyway as it barely saves any time and access is much more difficult.
There was never staff on the skytrain or at most of the stations when I was there a few years ago. Just the automated ticket machines. I’m sure some stations have full time staff, but some of them were small enough that I’m confident in saying nobody was there. Of course I rode 2pm on a sunday or other such times when I wouldn’t trouble. I’d expect security guards of some sort after a big game or party – but I wouldn’t be going to them and so I wouldn’t see it.
I suspect you are right about Vancouver.
But all the more recent ones have platform edge doors throughout and almost certainly the stations are also staffed.
The other thing is the Spanish ones we all praise is being low cost also have not really done any automation.
No system is 100% automatic. There’s always people in the background operating, monitoring, taking decisions (even if they don’t need to be real time). I’m sure Waymo has a team of people in the background telling the car how to take decisions when they’re stuck, or maybe the car is disabled and somebody needs to be sent to tow it. One cost gets cut but not everything changes at the same time. In mainline rail, as Alon says that the cost is driven by the infrastructure and vehicles, because the ratio passenger to operator is already huge, probably it’s worth to keep drivers minimizing the impact of rare situations. In subways we seem to need so much infrastructure to automate because we’re making them work with so much stronger security conditions than we do for cars.
Driverless cars remove one hurdle of taxis, but then this just pushes the system to the next hurdle. Traffic is already a problem, imagine if driving was easier and cheaper. Bus bunching is already a problem, will this force cities into more transfer-based systems? I’m skeptical of “more smaller buses”, because the higher frequency will make buses more useful and there isn’t much saving in having them smaller, so maybe it will be “more same size buses?”
“The other thing is the Spanish ones we all praise is being low cost also have not really done any automation.” – Let me correct wrong information here. Barcelona lines 9, 10 and 11 are driverless, line 11 adapted in 2004 to make tests so all three operated automatic since 2009. Actually, one of the first experiments with metro automation was done in Barcelona in the 1960s! https://ireneu.blogspot.com/2018/07/ato-linea2-bcn.html?m=1 Meanwhile, Madrid is retrofitting self driving in an almost 50 years old line.
@Matthew Hutton same with the DC Metro. They also indirectly manage dwell time by opening and closing doors, although I’m sure there’s guidelines for how long they can leave doors open. There’s one person in each station outside of transient janitorial staff. No platform edge doors, though. I think it’ll be a while before those are widespread in the US.
I wonder if WMATA has considered getting rid of the train “drivers.” Station masters kind of make sense to help clueless tourists buy tickets or deal with fare gate issues. Source: was a clueless tourist. In my defense, wrangling small children is distracting.
@J.G. “UAS are restricted to small aircraft and are not generally fully autonomous.”
The RQ-4 long range surveillance drone is fully autonomous and has a wingspan equal to a 737. You are confusing people people controlling a UAS to get a particular result (“go do surveillance over here”) with people being required to fly the UAS and keep it in the air. There are drones that can operate fully independently (if built/designed for it, obviously many small/cheap drones are actively flown).
“For instance, loss of airspeed indication (one consequence of an iced-up pitot-static system) will cause an A330 to revert from Normal Law (its typical operating regime) to Alternate Law, with far fewer protections.”
This is obviously a reference to the Air France 447 crash, but it doesn’t help your case. That aircraft did have pilots, but they did not perceive/understand the changes that came with Alternate Law and ‘flew’ the aircraft right into the ocean. The initial cause of the accident after the pitot ice over was the aircraft starting to move due to turbulence that the autopilot had smoothed out before it was disconnected, which the pilot then overcorrected – the computer was a better pilot than the human pilot! An computer flown aircraft would not have been confused by the switch to Alternate Law (provided it was programmed correctly; separate from the Alternate Law issue was the fact that the stall system was programmed as an IF…WHILE loop that only alarmed when it received valid airspeed input instead of an IF…UNTIL loop that kept alarming the stall until it had confirmation that aircraft was flying correctly – when the pilot was overcorrected and had the aircraft pointed too high he received no stall warning even though the aircraft was plunging out of the sky, when he dipped the nose to gain speed and recover (the proper move) he got a stall warning as a perverse form of negative feedback).
“So, too, do I feel that self-driving vehicles in mixed traffic will have a very slow rollout.
The fact of companies rolling out to new cities all of the time right now disproves your feeling.
“or a self-driving car will need to extract that data from its environment using cameras and pattern recognition;”
That is exactly what is happening. The jump in AI/pattern recognition the past few years has been extraordinary. You don’t have to believe it, but it is true.
@Jordi
“No system is 100% automatic. There’s always people in the background operating, monitoring, taking decisions (even if they don’t need to be real time).”
Yes, but one person can respond to more than one vehicle, probably many more, which provides the huge savings that would make transit so much more frequent and useful.
“because the higher frequency will make buses more useful and there isn’t much saving in having them smaller, so maybe it will be “more same size buses?””
It may very well be. There is a range of passenger loads to deal with. At the very densest/busiest (Manhattan, London, etc.) traditional mass transit is still probably required, a train every few minutes will still carry more people than a bus or two every minute. I can see many other cities replacing plans for subways, medium capacity transit or light rail with regular size busses every minute or two for similar capacity but better frequency and no infrastructure cost except for the bus lane paint. Articulated busses will probably go away. For less dense areas where no bus is viable now high frequency vans could be competitive.
@Onux
No UAS is as fully autonomous as a purported self-driving automobile, performing missions completely independently, whether being actively flown or commanded to do a mission and leaving the details to a flight director. Even the RQ-4, which is limited to relative autonomy only for a limited set of missions and not in highly trafficked airspace. And it is a very, very far leap from an unpiloted military UAS designed to go into harm’s way to a civilian aircraft with passengers aboard. Military UAS do not have to comply with crew or passenger safety requirements. Military aircraft as a rule are not FAA or EASA certified either, except for a limited number of aircraft adapted from civilian designs.
It was not specifically a reference to AF447, but since you brought it up, let’s talk about that.
AF447’s autopilot disengaged when the pitot-static system iced over. The computer wasn’t better than a human pilot, because the computer wasn’t flying, and was designed to be incapable of exercising that level of control authority in that regime, due to the lack of suitable aircraft state data. It is precisely those types of instances that require a human being in the cockpit. The fact that AF447 crashed doesn’t mean pilots are obsolete. It means pilots need to be trained better.
Incorrect. The event chain was:
Pitot-static ice-over–>loss of airspeed–>Normal to Alternate Law–>autopilot disengage–>Altimeter blip–>Pilot panics and climbs, reaches service ceiling–>stall–>CRM breakdown, pilots begin fighting each other’s control inputs–>stall, throttle up, roll, stall, roll, stall, stall–>crash.
There were plenty of stall warnings and airframe telltales. They were all ignored.
With all due respect, a “correctly programmed” computer-flown aircraft is a handwave. There is no “correctly programmed” commercial passenger aircraft DAL A autopilot that exists or is likely to exist or is even being worked on, and get FAA or EASA certified in the near future. I can go on about this, but it’s beside the point.
I didn’t say no rollout. I said slow. This isn’t like Uber or AWS. This is not their core business. This is a vanity project and a money loser that doesn’t get profitable with scale. Remember who we’re talking about here. Google and Amazon have no interest in solving real problems that real people have. They are rentier enterprises dependent upon monopolistic market domination that continually market products and services which get worse over time.
AI pattern recognition is not particularly repeatable or reliable for complex problems, certainly not enough for applications like road safety; and Waymo or Zoox robotaxis aren’t risking the latency to employ AI tools in daily ride usage. In other words, the processing power is restricted to that which is present on the vehicle.
In any event, the vast majority of the AI industry is a bubble and doomed to failure. Some niche academic and scientific uses may survive, where they will add tremendous value for those deep-pocketed institutions willing to pay the non-VC-subsidized costs of the tools. The rest of us will shake our heads at the trillion dollars set ablaze in this mad orgy of brain rot economics. And when (not if) Google and Amazon feel the fallout of OpenAI and Anthropic and others imploding, they will abandon this robotaxi nonsense so fast it’ll make your head spin. I would absolutely short them if they were independent companies.
But despite all this, I return to the point where I think we agree: autonomous road vehicles have a huge opportunity to thrive in relatively fixed, exclusive guideways, removing labor costs and increasing efficiency and passenger throughput. I think you’re absolutely in the right there. Cheers and appreciate the good discussion.
This is a vanity project and a money loser that doesn’t get profitable with scale.
In any event, the vast majority of the AI industry is a bubble and doomed to failure.
“No one will pay good money to get from Berlin to Potsdam in one hour when he can ride his horse there in one day for free.” -King Wilhelm I of Prussia, 1864
“This ‘telephone’ has too many shortcomings to be seriously considered as a means of communication.” -Western Union Internal Memo, 1876
“data processing is a fad that won’t last out the year.” -Editor, Prentice Hall, 1957
“There is no reason for any individual to have a computer in his home.” -Ken Olsen, founder of DEC, 1977
“…the idea of a wireless personal communicator in every pocket is ‘a pipe dream driven by greed’.” -Andrew Grove, CEO of Intel, 1992
By 2005 or so, it will become clear that the Internet’s impact on the economy has been no greater than the fax machine’s.” -Paul Krugman, Nobel Prize Winner, 1998
I guess time will tell which of us was right on autonomous vehicles.
AI pattern recognition is not particularly repeatable or reliable for complex problems, certainly not enough for applications like road safety; and Waymo or Zoox robotaxis aren’t risking the latency to employ AI tools in daily ride usage. In other words, the processing power is restricted to that which is present on the vehicle.
I never said AI as in accessing a webpage. Yes, the vehicles only use the processing power they carry, but that is more than enough processing power to autonomously guide them through repeated complex problems, namely navigation on streets in mixed traffic. Mark my words, autonomous vehicles will be MORE safe than cars with drivers, because computers do not get tired, or drunk, or angry, or distracted by the kids in the back. The technology is real. It is in use today. It is not just Waymo and taxis: Auroa does Class 8 self driving trucks while Gatik is doing mid size vehicles like vans. Avrride has autonomous vehicles driving around the Gangnam district of Seoul, one of the most crowded and complex urban environments on earth. This revolution is coming, transit agencies that ignore it are in trouble.
I saw something in the Economist yesterday saying that they were 20-40% more expensive than a taxi in San Francisco and that their primary advantage was a choice of music for the passenger, but with the disadvantage that they are slower as de-facto human drivers always have priority in a driving conflict so they are slower.
That doesn’t seem like a winning technology in the medium term unless they can get costs down a lot.
And he was basically right, the economic win by 2005 was that sort of size. And because of the negatives of social media the wins from the internet since 2010 have also been pretty small.
What Krugman said as a followup is that you can see the Internet everywhere except in productivity statistics. American TFP growth in the Internet Age has been pretty shit (link). There was a 10-year period, 1995-2005, in which growth almost matched the average before 1973, and then it reverted to ever lower, averaging 0.35% in 2005-19.
@Onux, human driving is also significantly safer outside the United States because the driving tests are more robust and you don’t have the traffic doing 70mph through central Las Vegas where for some reason there are no speed cameras to keep speeds at something more sensible like 50mph.
I took the economist data above and a search for taxi costs from San Francisco airport and it looks like Waymo has running costs of something like $4-6/km. That compares to something like $1/km here.
I saw something in the Economist yesterday saying that they were 20-40% more expensive than a taxi in San Francisco and that their primary advantage was a choice of music for the passenger, but with the disadvantage that they are slower as de-facto human drivers always have priority in a driving conflict so they are slower.
That doesn’t seem like a winning technology in the medium term unless they can get costs down a lot.
Waymo is able to charge a lot right now because it is a novelty and exploiting supply and demand. In SF tourists are riding it just so they can post the “invisible man driving” clip to tik tok, and paying accordingly. Also lots of things start out as a luxury item before they are commoditized (Tesla exploited this expertly when it moved from the Roadster, to the Model S, to nipping at the heels of Renault group for total vehicles sold).
But in any event, I am not arguing for driverless taxis as the future, but driverless technology for transit vehicles. Any personal taxi runs into Jarrett Walker’s law of geometry, where single occupancy vehicles take up too much room in crowded areas. You don’t have to get costs down a lot if you can spread that cost over 8-10 people in a vehicle (i.e. a van or mini-bus, not 8-10 people crammed in a current Waymo) even if Waymo’s prices are not inflated by supply/demand. If you have a demand response system then skipping the unneeded stops makes your service much faster than current busses, even if slower than a human driving a taxi (almost all public transit is slower than a taxi now).
What Krugman said as a followup is that you can see the Internet everywhere except in productivity statistics.
That is an pathetic pivot and cope to try and recover from such a terrible prediction. Productivity growth is only one kind of impact, and the economy exists outside of the US. Why is Amazon the second largest retailer in the world? Why is Tencent the largest Chinese company by market cap and Alibaba second largest (both companies were founded after Krugman made that quote!!!) How many people work in a shipping warehouse fulfilling online orders versus a brick and mortar retailer? How many brick and mortar retailers don’t exist anymore as a result? Centralization is some areas (Uber and Lyft as national/global brands, vs regional holders of taxi medallions) decentralization in others (who needs a major label record contract when you can self publish or just get famous/rich from clicks on social media?). For a person whose entire economic livelihood stems from the internet (this blog leading to your shift from mathematician to transit researcher, and the ability to coordinate with people around the globe on your work) to try and defend Krugman on this is a bit rich.
@Onux
If Waymo was making an operating profit in San Francisco even with high fares for the novelty and even with San Francisco’s extremely high wages then we would know about it because it would be an international headline.
That they haven’t announced this means they are not making such an operating profit.
@Onux, in terms of buses, perhaps theoretically in San Francisco it would be profitable to run a driverless bus.
However in a bus you have to be more assertive than a car driver to eventually get somewhere, and there is also the problem in San Francisco and other big cities with security if there are no staff on board.
@Matthew Hutton
I don’t know if Waymo is making an operating profit just on service in SF. Even if they were, I would guess the overall company is not profitable yet due to R&D and expansion, which is normal. Also Waymo is currently using Jaguars as their base vehicle, which is expensive upfront and to maintain. Watch them switch to a more reasonable base vehicle (from other manufacturers or an in house design like Zoox) and their all in cost drops quite a bit.
But this is all kind of irrelevant to the point, because no public bus service in America makes a profit on fares! (That I am aware of, maybe there is an outlier?) To implement what I am suggesting Waymo does not have to make money as a taxi, they just need to provide a system that runs a bus cheaper than a driver does. This is a virtual guarantee.
perhaps theoretically in San Francisco it would be profitable to run a driverless bus.
If it is possible then there is even less excuse for transit agencies not to adopt it. If SF covered all of its bus operating costs with fares then it would have a ton of money to expand service (an thus get even more fares – the growth cycle not the death spiral) or invest in other transportation needs (DTX tunnel for Caltrain to get downtown, bike lanes, you name it). For what its worth I think it will profitable to run driverless busses/vans most places there is some bus service now. After a decade of people predicting Uber’s collapse because “they are not profitable” Uber has posted an annual profit for the last two years; Lyft did the same last year. Transportation is a service that can be sold. If it is high quality (fast, comes quickly) that is what people will pay for. When a private company figures this out with fixed route service not taxis, public transit is in trouble.
However in a bus you have to be more assertive than a car driver to eventually get somewhere,
Says who? I know of no problems with Waymo being able to pull out of parking or navigate in city traffic because it is not ‘assertive’ enough. Plus very frequent smaller autonomous vehicles that use turnouts for stops and a demand response system are perfect fits for bus-only lanes. You can get the capacity and speed of underground light rail for basically just the cost of paint. Then there is no issue with needing to be assertive because you have dedicated right of way.
there is also the problem in San Francisco and other big cities with security if there are no staff on board.
Try again. SF is among the many cities that have placed their drivers in pods with clear plastic dividers between them and passengers (for driver ‘security’) so there will be no staff intervening in problems. This is also a non issue. Do people avoid riding Vancouver Skytrain because there are no staff on board? Do people have a problem with riding at the end of BART train in SF hundreds of feet from the driver who can’t see or hear them?
@Onux
If they were they would publicly announce that they were. Any remotely competent business would even if they are losing money on expansion. Uber made it clear when they were profitable in some of their initial markets – London was an early one.
In terms of buses lets look at London. In London there were 453 million bus kilometres in 2023/24 as per https://www.gov.uk/government/statistical-data-sets/bus-statistics-data-tables#local-bus-vehicle-distance-travelled-bus02 regional data. As per page 74 of https://content.tfl.gov.uk/tfl-annual-report-and-statement-of-accounts-2024-25.pdf the London bus network cost £2269m in the same year to run. That is £5.00/km or $7.15/km PPP.
As Waymo costs $4-6/km to run a taxi in San Francisco I think it is pretty unlikely they could get remotely close to the $7.15/km it costs in total to run the buses in London – even if they covered for free the upfront development costs of a bus product and made them sufficiently assertive to handle driving in London.
The Economist said this in their recent article on Waymo.
@Onux
Yes, this is a company I trust to make good decisions
@Matthew Hutton
As Waymo costs $4-6/km to run a taxi in San Francisco I think it is pretty unlikely they could get remotely close to the $7.15/km it costs in total to run the buses in London
Rule of thumb is that half of the cost of a transit vehicle is salary for the driver. That would mean London is spending $3.58/km for the bus. If Waymo is running their taxis at $4/km then if the cost to run their cars is $0.43/km or more, then it is more than likely they can run busses at $7.15, it is certain. A quick google search says that cost to own an I-Pace is $1/km, or that owning an average vehicle is $0.51/km. Cost to own is highly variable depending on assumptions; I presume London’s bus cost is direct operational costs with capital procurement covered elsewhere, while Waymo’s cost per km includes depreciation for owning the vehicle. That would make Waymo look even more favorable. And remember that Waymo’s vehicles have a fixed cost at all times, while drivers on late/overnight shifts or drivers working overtime get paid more per hour. This makes it cheaper with autonomous vehicles to run more service at all hours of the day.
More broadly, are you seriously arguing that automation is more expensive than using drivers? Have you visited a high rise building in the past few decades? When you did and got into an elevator, was there an elevator operator there because it was cheaper for the building owner than buying all those expensive computers that allow people to press a button? Did Toyota become the most efficient and largest car manufacturer in the world by banishing assembly line robots from their factories?
the upfront development costs of a bus product
They already have a product that can drive a vehicle autonomously in mixed traffic on city streets. Adapting it to a different vehicle would not be terribly costly compared to the past decade(s) of R&D to get to this point. And as I have said before, programming a bus to run a fixed route is simpler than a taxi that can go anywhere.
and made them sufficiently assertive to handle driving in London.
Unlike human-piloted taxis, autonomous ones rarely break traffic rules; you cannot tell a robot to “step on it!”
This, as the saying goes, is a feature not a bug. Are you seriously arguing that obeying traffic rules and not running over people is bad thing?!?!?! The article complains that Waymo is slower than a taxi, but busses are already much slower than taxis (because busses stop along the way) but people still ride them. But if you have more autonomous vehicles on the road and can program them to only make certain stops and group destinations (the demand response operation I have referred to) then you can speed up busses a lot – limited stop busses have up to double the speed of local/all stop busses.
The very first line of the Economist article:
AUTONOMOUS TAXIS are taking over San Francisco. Waymo, which launched only in 2023, may now have over a fifth of the city’s ride-share market. Waymo, which launched only in 2023, may now have over a fifth of the city’s ride-share market.
Yes, a sure sign of failure is that fact that people use the service a lot. Why are all of those people taking Waymo rides instead of listening to the pundit who wrote the article explaining how bad it is for them!?!?
On automatic trams and buses, I’m an IT professional, and I just don’t understand how we’ve started with self-driving taxis, when self-driving trams and buses should be a much much easier technology. Software system complexity is a factor of the amount of edge cases it has to cover, and with a bus route you can manage the environment. You can define that your self-driving bus only works if the route is like this and that and that. And modern trams make this even much much simpler, with routes almost 100% dedicated that only cross with traffic in intersections controlled by traffic lights. Well, I guess I know why we started with cars, it’s much easier to make a business case to sell thousands of millions of cars than a few thousands buses. And this will reflect on the investment on lobbying as well…
We do have self driving metro systems in a number of countries, but typically they cost a lot more to build.
Self driving trains are easy if the trains don’t mix with traffic. The skytrain was automated more than 20 years ago. Since nobody can accidentally get on the track we can assume nobody is there and so don’t need safety systems. Trams are harder because they often operate where somebody could walk in front of the train and expect the train to stop. Buses are harder because they operate in mixed traffic, but if you can do a taxi you can do a bus.
Which is to say if you are a bus operator you should be demanding driverless buses. If you can’t get them the reason needs to be some technical point that I’m not aware of, not political.
You asked,
Couple of data points for you:
https://www.mta.info/document/176336
YTD (May) financials for Metro-North report $483M of labor expenses out of total expenses of $682M. (71%)
https://www.amtrak.com/content/dam/projects/dotcom/english/public/documents/corporate/monthlyperformancereports/2025/Amtrak-Monthly-Performance-Report-July-2025.pdf
YTD (July) financials for Amtrak report $2.5B of labor expenses out of total expenses of $3.8B. (66%)
At a glance, these are similar to other reporting periods’ financials.
One can likely conclude that salaries, wages, and benefits represent a clear majority, between two-thirds and three-quarters of operating expenses for passenger rail transportation providers.
A larger data set should be looked at to further validate this conclusion. In addition, the data should be partitioned between services of similar type.
I hypothesize that Amtrak’s Long Distance service line has far higher labor expenses than NEC and Acela services due to: overnight duty; crew layovers; higher maintainer costs because of aging rolling stock, more extreme environments, and diesel road power; and higher crew to passenger ratios because of more luxurious food and beverage service and sleeping car attendants. For electric commuter/regional services these are not true as only conductors and drivers make up operating crews; and on the maintenance side, mean distance between failures for electric multiple units are over 10x those of diesel-electric locomotives (e.g. >800,000 miles for M8 vs <70,000 miles for GP40s), requiring less routine maintenance and shop overhaul labor.
Fun accounting problem.
This isn’t the case for other countries.
Very fair point, perhaps I should have restricted my conclusions to domestic passenger rail services providers.
I notice that in some cases you note total expenses and others operating expenses. Was that deliberate, in that in some cases the labor and expenses represent both capital and operating?
Trains have much higher fixed costs and much lower variable costs than buses. Bus costs can be modeled as the driver’s wage plus some maintenance wages that scale linearly with service-hours plus a little bit extra. Rail costs are different: maintenance is based on regular intervals more than on intensity of usage, the driver’s wage is a secondary element, station and equipment acquisition is a large fraction of the cost, track construction and renewal are the primary cost if a new line has to be built. The marginal cost of raising off-peak service until it matches peak frequency is very low, unlike on buses; American transit agencies err in running too high of a peak-to-base ratio, especially on commuter rail.
Is that true even for systems that have reasonably good rolling stock utilization?
The same type trains working Tokaido Shinkansen seem to be serviced more frequently and definitely get replaced more frequently than the ones that stay on Sanyo/Kyushu Shinkansen.
The shinkansen maintenance schedules are typically based on a time interval/service miles/km run system, whichever comes first, i.e. on the Tokaido Shinkansen the most intensive “general inspection/overhaul”, is performed every 40 months or 1.6 million km, whichever comes first.
Do you have a source for train inspection intervals in Japan? I only know the inspection regime in Germany, which is not the most reliable system right now.
You are still basically saying that the New York commuter railways could increase their passenger numbers by 40% by running more service in the daytime and evenings for driver, guard and some cleaning/train/track maintenance costs.
Commercially it undoubtedly makes sense.
Often when rail maintenance is done at regular intervals without regards to intensity of service, that comes at the detriment of maintenance, and so you get trains breaking down more often. The cost ends up being paid one way or another, and it might be through higher capital costs of more equipment as its worn out faster and insufficiently maintained. Case in point: every fleet that runs out of Jamaica Yard.
I guess you’re not a big fan of activity-based transport models, because technically, they claim to answer all the questions you’ve posited…
Because that is the way they work. If everybody commuted to Poughkeepsie it would be part of the Poughkeepsie region. If they all commuted to Schenectady it would be part of the Albany region. It’s how most people have commutes that are under a half hour. They take jobs near home.
I realize the view across Ninth Ave. can be dim. They already have that. Because they developed in the 19th Century. Before there were cheap automobiles. Which are a 20th Century innovation. They all can’t look like Morristown or Stamford. It’s okay if a few of them have ….. elevator buildings… and most of them have a clump of non-residential and some of them have a post office and cab stand because retail would be so so declasse.
The view across Ninth Ave. is very dim.
People who have automobiles are not going to walk to the station to go to the chain store three stops down the line when there are chain stores near their station. Or someplace they can drive to faster than the walk to the station.
They don’t already have that, is the point. The schools are pushed away from city centers due to grants for new greenfield school buildings, the retail is in malls accessible only by car, and what’s left in the town center is pretty sad.
Schools are surrounded by athletic field deserts and parking. Not a good use of downtown. Suburban students get there by bus, mostly because of stranger-danger hysteria but they get there by bus. The bus can go anywhere.
There isn’t going to be a Sears, Woolworths and A&P at every train station. And there definitely isn’t going to be a Lord and Taylor at every one. There will busier downtowns like Morristown or White Plains. Or Bridgeport or Paterson. Some of them like Millburn will have banks and medical professionals and a hardware store and….. People in Short Hills, a separate zip code in Millburn will have a post office. They will go down the hill for medical care. Or up the hill to Summit. Or a bit farther down the hill for the big box stores that are in Vauxhall. Or up the hill to Short HIlls Mall. Unless they are west of the Gap. Without consulting transit dweebs about how they should live.
The retail at 110th and Broadway isn’t Lord and Taylor and an A&P either. Things aren’t going be Herald Square everywhere.
While school parking lots should shrink into irrelevance given good foot/bike/transit accessibility, the athletic fields are a good point. While I understand the regulations to be somewhat flexible, they are very …aspirational. IIRC someone figured that with walkup-midrise residential and a “families” number of children per adults, half the “built” area would be school grounds. For this reason, I consider schools to be “light”, they should “float” away from the town center (as opposed to retail&c which take much less floorspace per capita of population than residential, thus “sink” to the center of a monocentric model too small for Hotelling’s law to operate).
Separately but acting in the same direction, some American delusions would create a political force to eject the schools from the town center, particularly the bus->train connection. Maybe the school could exist a few stops up the bus route, because there there wouldn’t be all these (poor, and also non-local a.k.a. transient) people (menacingly) standing around, conspicuously doing nothing.
Do they use atomic weapons to clear the suburbs or more conventional means? So they can be reconfigured to look like places where people are poor and fuel is heavily taxed?
Schools tend to be poor fits for bus routes unless they are at a major station. You get 1000 students who want to arrive and leave at the same time and overwhelm the system for 10 minutes, the rest of the day that account for zero riders. High schools are worse as they have to handle 10,000 people leaving the homecoming game, but that is a once a year event. (sport teams in general have this problem, as to concert halls).
Yet the majority of students in the U.S. use a bus to get to school. Many districts require them to use the bus. And a bus to get back home.
Is this “schools are outside the town because they need huge athletic tracks and kids need to be hauled by train or car” a common thing outside America? Looks like I come from a different planet where schools are in town, kids mostly walk to them, the school has a patio where they can play futsal/basketball, the entrance time is 99% of people at the same time, but the exit time is all over the place because there’s all kind of activities happening around the school… Only the super-posh international schools resemble the space needy design.
In the US, whether the place is affluent or not, competitive sport is largely school based. Schools have a large budget for various sports and stadiums where attendance can run in the thousands. The kids who excel at high school sports get a scholarship to college and from college teams may be picked by professional franchises. The young adults who don’t become professional have nowhere to go. The pyramid promotion and relegation structure of independent sport clubs found everywhere in Europe doesn’t exist.
Each kid uses a bus to get to school, but 10-20 different buses running different routes arrive at school at the same time. A transit route wants to run 1 vehicle every X minutes all day, with the a reasonable amount of riders on each. If your school is also the transit transfer point it would be fine, but it doesn’t work to have the school farther out on a route since you need so many extra vehicles to arrive at one time.
You don’t want the school to be the transfer point though because (see discussion elsewhere) this is expensive real estate and should be used for those willing to pay for it. Schools with captive users should be farther out.
dralaindumas wrote,
Someone’s been watching too much Friday Night Lights.
Most schools, if suburban in location, have enough spectator facilities for a few hundred, not thousands, and they’re bleachers, not stadiums. Depending on the district athletic facilities may be shared between schools or centralized with town parks and recreation facilities (e.g. soccer fields) and the children are either driven, bused or walked to those facilities.
I’ve lived almost my whole life in suburbs but when I did live in a city, my school used city facilities or public parks.
The kids who excel at high school sports do not all get scholarships. Nowhere close to 100% of Division I athletes are on scholarship, let alone D2 or D3. A vanishingly minuscule but notable fraction of individuals go straight from high school to professional.
The college students who don’t become professional athletes go into the workforce with the degrees they earned from the colleges they attended, which includes the overwhelming majority (>90%) of scholarship athletes. They don’t “have nowhere to go.” I hope this is a language barrier thing, because this is a breathtakingly dismissive assessment of most college students who play sports. They’re there to earn a degree, not to go pro.
@ J.G. It isn’t a language barrier thing. I am not discussing the scholarly achievements of college athletes and I have no idea what Friday Night Lights is. I am writing in a blog about transit-oriented development, about where pieces of urban architecture are located.
The stadium and the soccer teams were the beating heart of the place where I grew up. They took over this role from the church or the train station before I was born. The entire village could hear the roars coming from the stadium when the senior team was playing. I was responding to Jordi who may have had the same experience in Catalunya.
In the US, competitive sports follow the suburban high schools or colleges in distant green fields surrounded by large parkings. The architecture of the competition is also different. The European leagues are a pyramid with a large number of teams at lower levels. American sports is a ladder you climb as you grow up. Once you fall out of it you have nowhere to go as a competitive player. In Europe if you don’t make it as a professional or are too old to remain one, you will return to your original club and be a local hero long into adulthood. I was a mediocre player. Overcoming the stress of starting a game against stronger or more skillful opponents and at times finishing on top are my proudest achievements.
@dralaindumas
Let’s try this again.
You wrote “The kids who excel at high school sports get a scholarship to college.” This is incorrect. Most children who excel at high school sports do not get athletic scholarships, and most college athletes are not scholarship students.
You wrote “Schools have a large budget for various sports and stadiums where attendance can run in the thousands.” This is incorrect.
You wrote “The young adults who don’t become professional have nowhere to go.” This is incorrect. College athletes have somewhere to go: into the workforce. If you meant “have nowhere to go as a competitive player” it’s still incorrect. What you may be unfamiliar with is farm systems, minor leagues, and free agency.
But sure, let’s talk about TOD. Why not talk about it in the context of sports stadiums?
Well, to pick DC, New York, Los Angeles, Chicago, and Baltimore, all of those stadiums have robust transit connections. In fact, if the Mets and Yankees make the World Series, it’s called a Subway Series, because they can take the subway to each others’ stadiums. I’m aware that all seven hundred or whatever London-area Premier League teams can take the Tube or Overground to each others’ facilities and it’s not special; let us Americans have this one thing.
I think there’s an opportunity for TOD and transit stops centered around minor league stadiums, though. Recently, ticket prices for major league games have become incredibly inflated, and the middle class is getting priced out. I remember as a college student 20+ years ago getting $5 tickets for upper deck at Camden Yards; third base line at Yankee Stadium on a weeknight now crosses $150. That leaves minor league teams as attractive entertainment options for urban and suburban families.
Here in Hartford, Connecticut, the Hartford Yard Goats minor league baseball team and its stadium, Dunkin Park, is reinvigorating development in the poorest city in a very wealthy state. It is the centerpiece of a new neighborhood called Downtown North. To the east, near the Connecticut River, Trinity Health Stadium hosts the Hartford Athletic of the United Soccer League. I’ve been to games at both, and parking is a royal pain the ass. I would absolutely take the train to the game, if there was one.
Trivia: a yard goat is a switcher locomotive. The team is so named after a small rail yard nearby, and the logo recalls the visual design of the New York, New Haven & Hartford Railroad.
PS. I know what a promotion and relegation system is. I’ve seen Ted Lasso.
You are accusing me of watching too much TV or making up stuff but my daughter grew up in Connecticut. Her high school stadium had a 3500 seat capacity. She was interested in cheerleading but the 49 week a year commitment was incompatible with our schedule. Another high school she decided not to join seemed to have a bigger emphasis on football while the public high school has a 12000 seat stadium. This contrasted with my own upbringing in Europe where academic and athletics developments are separated. My friends and I were soccer mad but our schools or universities did not have stadiums or teams.
@dralaindumas
You made a blanket statement that is nowhere close to universally true despite your anecdotal experience. Move past it, we’re done.
I think we made some progress. First you argued that what I wrote was false, now that it is anecdotal. It is indeed anecdotal because my intention was never to discuss the size of audiences to sporting events. For the same reasons that when Transit-Oriented Development is discussed here, people take about life-style and not about investment opportunities, I am more interested in the life-style effects of school-based and independent sports leagues. Adults in my soccer club practiced competitive sports for years after leaving school, whether they were highly educated (a dentist and a pharmacist were on the roster) or not. Amateur adult competitions are less common in the US. Ten teams compete in the Adult Baseball League in CT. A French soccer district with CT’s 3.675 million population would have about 1000 adult teams.
After living in multiple US locations where I saw large gyms, accompanied by sizable parking lots when located in the suburbs, I don’t think Americans are lazy. Maybe as with transportation, Americans lean more towards the individual practice while Europeans are still more communal.
If we move past his we have to move past yours. There have been some real knee slappers in this thread.
That depends. If the train station there for the purpose of a bus stop (presumably because we are so dense that a bus can’t meet demand on the line) then it won’t have that much. Even then the dense parts will have some retail at the station because there is enough density to support walk to the local grocery/restaurant/gym – it may or may not be mega department store, but there will be density to support some retail in that area and they will want to be by the station anyway. Less dense parts of the lines may not have that retail, but most stations should be that dense – because why would you pay for the cost of a train if you won’t have that many riders?
Most train stations should be larger meeting points where all the local buses (and local trams) meet for a transfer to faster transit. (there might not even be trains, bus only cities operate a hub system like this all the time) There should be more than enough demand to support a lot of retail at these major transfer points everyone is going to.
Of course if you have a car you can go anywere, but even cars have place retail at busy places. There is a reason every freeway exit in the US has a gas station and a fast food place – people are going there anyway and so those exits function as meeting places that retail wants to be. It works slightly differently with transit, but only slightly, retail still wants to be where people are going.
Yes it does depend. On the people who make careers out of deciding where to put a store since Frank W. Woolworth himself was running the company. Again, I don’t know how things work on your planet but here on Earth there will be more retail where there are more people. It’s one of the things people who make careers out of picking retail space look at. Since people aren’t spread evenly around there will be differences in how much retail there is.
Those retail people can’t put their stores where the people are because transit agencies put large parking lots around their big transfer stations. These should be the place they are paying the most for, but it isn’t an option so they are putting in stores elsewhere, and then putting in parking lots because the need to transfer means “normal” people are going to drive just because the trip with all the transfers is not reasonable.
It’s just awful the way people from the city are allowed to move to the suburbs. If they didn’t live in the suburbs there wouldn’t be a need for a bus to the train or a train, would there? Do you want to use bulldozers and wrecking balls to reconfigure things or do it all at once with an atomic bomb or two?
I want to bulldoze the parking lots around train stations – those are owned by the transit agency and so this should be easy. Replace them with dense retail because that is where the people are and so that in turn is where retail wants to be if they want to get transit riders as customers as opposed to drivers. Well I’d replace with mixed use, but the goal is to get ground floor retail paying high rent because that is where the people are going anyway, and then upper floors of residential because people should want to live near the train. (at the far reaches of the parking lot maybe just residential because there isn’t that much retail demand).
If you think people are going to take a very long walk to the bus, from their cul de sac… I want your dealer’s number because he’s got good stuff. Really good stuff.
The parking lots are frequently owned by the town. Which means the town can restrict access to residents. Good luck with redeveloping that. Because people aren’t going to take a really long walk from their cul de sac to get to the bus. Unless you think people are going to abandon their cul de sac to let it go back to nature. Which means your dealer has truly extraordinary stuff.
The people down at the end of the cul-de-sac won’t walk. Their kids will when they realize that the long walk means they can visit their friends. (though if they are locked on devices they won’t – I hear all kinds of scary stories about this and then someone else says it isn’t true because… Let us just settle for some kids will go out and before their get their license good transit will attract them to walk to it). People half way down the cul-de-sac don’t have such a long walk though and will talk to good transit. This is a group worth getting. If anyone has any practical idea of how to get transit to those cul-de-sacs I want to know.
Clap harder. Maybe what you need is ruby slippers.
People who own cars will use them when it’s faster than taking transit. They take transit when it’s faster than driving. Even people who live at the bus stops at the new apartment in Ronkonkoma want parking so they can do things that would take longer by bus. Even though they live right at a bus stop.
Suburban kids know how to use the bus. They take the bus to and from school. They also own bicycles and use those until they get a driver’s license. Perhaps Glinda and Wizard is what you need.
@Henry, part of why Culham railway station has a tenth of the ridership of say Haddenham and Thame Parkway or Oxford Parkway is that it has a smaller catchment, but part also is the fact that it doesn’t have a car park.
Rural/suburban train stations (especially in the US where car ownership is particularly high) need to have a decent sized car park.
With TOD details matter and most attempts get obvious details wrong. I still want to see TOD around stations not parking lots, but they need to be done right or the whole will fail.
Every suburbs has a commercial district because the locals don’t want to go very far for the basic stables. Stations should be valuable places where stores want to be – getting the people from other parts of the city to them via transit. This value should be higher than the value of the parking lot.
The biggest thing to note above is unlike most (in the US) I want transit to be a way to get places, not just a way to get to your 9-5 job downtown. Most transit is designed around the 9-5 downtown jobs, and that means you must have a car for all other trips. If you must have a car for other trips it means all those apartments still need some form of “free parking” for everyone (3 spaces per unit!), and that quickly means you will prefer the cheaper apartments farther from the station where you can just drive.
Note too that I’m not against parking. However it shouldn’t be walking distance from the station. Put is 1km away, a suburban bus can pick you up. Sure it make driving even slower, but once somebody gets in their car we have already lost – they are always going to be looking for cheap parking closer to where they want to be (nobody over 4 wants to be on a train every day), and less traffic. They will be working against you anyway trying to get more money for roads to solve their car problems since the train is a compromise. (One could imagine a high speed train from the outer suburbs to Manhattan which is enough faster as to attract people even if there was no traffic – I’m not aware of any high speed commuter rail in the world so I’m not sure how it would work)
I would expect if you follow my plan that stations would be less busy in the short run – but the density it enables (hopefully at lower prices now that it isn’t limited spaces downtown) would over 20 years result in busier stations. Of course this assumes a dedicated effort to build a useful transit network over those 20 years, we haven’t seen much progress yet in the US.
Have you ever been in a suburb?
What compelling destination is there three stops away? Is the Walgreens so much better than the CVS right at the station the suburbanite is going to walk to? Is the bank branch two stops away so much better than the branch at the station they will walk to? People who have cars aren’t going to a supermarket, it would have to be a truly compelling. Perhaps, maybe, if the only car in the one car household was off doing something until tomorrow, would they go to a specialty food store halfway to the city. Or just put it off until the car is back in town. They probably considered that when the car wandered off for a day or two or three.
Ah yes, car owners are going to drive to bus stop to get to the train station so they can take the train a few stops away to go to Dollar General instead of Family Dollar that is on the street where the bus runs. Instead of driving there. Or going to Walmart. Target and Kohl’s have a lot of the same stuff. Or maybe … Have you ever been in a suburb?
My insurance doesn’t cover CVS so use the Walgreens is needed. Though drug stores, banks, and grocery stores are bad examples because there will be several options close and so you won’t go elsewhere. However there are a lot of other stores that won’t be there. I travel past a couple “guitar” music stores to get to the local violin store – my city can only support one violin store. Hobby stores likewise tend to specialize and there will be only one with the exact hobby you need. People looking for deals will check every goodwill (and other thrift / second hand stores). There are also many clothing stores that people travel to (even chains won’t be in every mall). There are a lot more restaurant options if you are willing to travel a stop down the line.
People who have a car don’t want to take the train at all. Even if traffic or parking fees make them use transit, they are going to want to fix traffic or parking so they can just drive all the way to where they are going. So we should aim to make travel by transit easy enough that people don’t get in the car in the first place – that means most of them have a quick bus ride to the local station and from there plenty of things to do and if that fails a quick train trip to other things. Maybe you keep the truck to pull the boat on weekends, but for most trips you don’t use the car at all and so you don’t have a second car.
@Matthew Hutton Denver built their train system around parking at the station and their system is under performing. I don’t know all the factors at Culham railway station, but it is clear that parking alone is not enough to explain it.
That is what is in the other suburbs up or down the line. You are the one imagining people driving to a parking lot to take a bus to the train station so they can go a few stops up or down the line.
Buses aren’t a way to fix lousy health insurance. There are many solutions to your quandary other than taking the train to the bus that ferries you to the parking lot, going all the way home, driving back to the parking lot, waiting for a bus taking to the train to get a few stops down the line and picking up your prescription. Maybe your coworkers can give you some advice. They might even suggest using the Walgreens near work. Or getting off the train on your way home.
It’s too bad the unicorn farrier is in the city, isn’t it? It’s good you don’t have to have it’s horn sharpened often. You could drop off the violin while it’s getting ground down.
Maybe you drive to Walgreens to pick up the prescriptions too. Because it’s next to the supermarket. That you don’t take the bus to because it would be difficult to haul it all home on a bus. Does the unicorn go in the truck bed or does the neighbor kid feed and water it on the weekend?
@henrymiller74, for Denver the mean commuter rail station has 750,000 passengers a year, that compares to 1.25 million passengers a year for the average British station. I would be surprised if you excluded the London commuter belt that the British average per station in the rest of the country wasn’t lower than 750,000 – personally on a vibes basis I consider 750,000 riders a year to be strong for commuter rail.
For the Denver light rail system, yes 11 million rides with 57 stations is weak. The Manchester Metrolink which people in general think underperforms gets 42 million passengers a year with 99 stations, which works out at 850k riders a station vs 390k in Denver.
Partly because you would have to tear down the existing mixed use TOD-y multi story development that would make a New Urbanist plotz. It’s okay that the only thing near the Green Farms station, which is also in the incorporated municipality of Westport, is the parking lots and Post Office. In the wasteland between the tracks and the Turnpike. Be a great place for the ginormous park-n-ride garages with dedicated ramps to the Turnpike. Which would keep them out of Westport or Southport.
You wrote,
I must confess I am surprised by this statement. May I respectfully ask, what data drove this conclusion? Was it an examination of median home sale prices and median rents in these suburbs? Was there other data in the mix such as surveys of city-to-suburb individuals, i.e. social data of preferences, rather than financial-data-to-causality?
Regardless. Very interesting post. I was recently in Ronkonkoma, where high rise mixed use development has sprung up around the LIRR station. It’s called Station Yards. But it’s not “transit oriented” to the extent of having all essential services walkable. There is a grocery store and urgent care, but the other commercial tenants are drinks and dining. There’s also an event space. So while schools can provide bus routes for students, any other destinations in very suburban Long Island require a car. So Station Yards has a big garage for residents, visitors, and commuters.
I do agree that this type of development isn’t truly walkable, but I think of it as a gateway drug. It’s hard enough overcoming zoning and NIMBY for development even of this kind, because while it might not eliminate certain car trips entirely, I hope it reduces them. So I’ll take this win. And it does hold out for the possibility of more future multifamily housing, services, and transit.
I’m basing this on BLS/BEA numbers giving relative prices in each region of the US; New York is large enough that it’s broken into city vs. suburbs, and both are very expensive, and the suburbs were more expensive last I checked (which was in the 2010s). So the housing cost would be owner-equivalent rent, because that’s what goes into the CPI.
I’m also basing this on much less granular income information that I have for city center commuters (link); for CBD workers, incomes peak around 50 km from the center. Long working-class commutes aren’t so much drive until you qualify as complex multi-bus trips from a residence in the Bronx to a retail job in outer Brooklyn.
As for the gateway drug point: it doesn’t work if the only TOD is residential. There are nationally-recognized TOD sites, like Mission Valley in San Diego, that nonetheless have high car ownership and very high car use for non-commute trips, because there’s no attempt at producing any commercial TOD. American urbanism unfortunately is stuck with the options of traditional Main Street retail and an auto-oriented suburban mall, and the former is at a scale that large discount chains can’t use. This is why I talk about co-location and about urban shopping malls, of the kind that’s normal even in France and Germany, let alone Canada or rich and middle-income Asia.
The one J.G. cited has retail. That doesn’t meet your epicurean tastes. It also has a bus route to the Smith Haven Mall and a second bus route to the Islandia Mall. That offend your epicurean tastes. That’s too bad.
Fascinating. I’m in a deep-dive now on FRED looking for whether this still holds true after Covid and ~10 years worth of city housing starts lagging population growth since your last check (I haven’t found anything yet), purely for my own edification, not because I think you’re wrong.
This is of particular interest to me because my home state of Connecticut’s a very funky and unequal place, and as recently as the 2010s had extreme difficulty balancing a state budget whose tax revenues would closely track its highest earners’ lumpy cash flow. The weirdness of the Gold Coast hoi polloi sitting atop a large working class minority population in the same county. A low-level but consistent population decline. But then, since Covid, the population decline reversed, and the budget went into surplus (because of either prudent management by a new governor and legislature or massive cash infusions from the feds or both or neither? I haven’t seen a satisfactory answer). Was the population decline a result of wealthy New Yorkers decamping for suburban living? A lot of anecdotes in the New York Times seemed to think so. But anecdotes aren’t data, and true to form the Times focused more on the schadenfreude of these hitherto apartment dwellers reliant on supers contending with managing single family detached paradise, with all the termites and leaky roofs that came with it, over a statistical reality. But I digress.
With respect to TOD — I was forced to confront my own biases. At first I got kind of defensive about my gateway-drug point, but then I thought about my own life. I do, in fact, do a lot of driving to places other than work — sometimes more than work, especially with a hybrid work arrangement. It would be difficult if not impossible, even living in a residential TOD, to conveniently visit all the locations I need using transit. Even if I lived in a place like Station Yards, you’re absolutely right — urgent care isn’t a primary care physician or specialist or dentist; an upscale Whole Foods wannabe isn’t Aldi or Stop & Shop; there’s no CVS or Rite Aid or Walgreens; a bus that runs every half hour to a mall isn’t practical for families with children, which constitutes about 40% of the US population–a demographic that absolutely needs to be considered if a TOD wants to remain viable. I shudder to think of manhandling a double stroller and two feisty toddlers onto a bus. Obviously people do it if they have no choice, but if I had a choice? Heck no.
I’m curious if there’s anywhere in the US with an urban mall and/or plenty of retail, schools, and services in close proximity in a walkable, residential-commercial TOD. The places I can think of, again from personal experience in DC, include Crystal City-Pentagon City, King St-Old Town Alexandria, Bethesda, and Silver Spring. Speaking of which, what are your thoughts on the Purple Line? It’s busy cutting a swath through my alma mater as we speak.
Do you actually think people are that stupid or are you that stupid?
There are plenty of low paying jobs in jobs in the Bronx. And plenty of similar jobs, that pay better, in Manhattan. Which makes me lean towards you being that stupid.
Go to the US Census On the Map tool (onthemap.ces.census.gov) Type “Bronx County” or “Kings County” into the search bar on the left side. Select the appropriate county in New York. After it pops up on the map, click “Perform Analysis on Selection Area”. Select the button for “Home” if doing the Bronx, for “Work” if doing Brooklyn/Kings County. Select the button for “Direction/Distance” and then the button for “All Jobs”. Click “Go” and what pops up is a map filled with dots for every location where someone who lives in the Bronx works, or where everyone who works in Brooklyn lives.
The map of Bronx residents has dots for employment covering basically all of Brooklyn, denser around downtown, but with pockets concentration all of the way out to the ocean. 169k people live in the Bronx but work more than 10 miles away from what the Census considers the center, which means out of the Bronx, and most of them are not in Manhattan because midtown is also within 10 miles of that spot. Similarly the map of Brooklyn workers has dots all over the Bronx.
The only thing that might be inaccurate about Alon’s comment is that most of these commuters are probably using the subway for at least part of their trip, instead of bus to bus, because of how well it covers even the outer areas of both boroughs. The highest concentration of people who live in the Bronx but work in Brooklyn is around 167th St, where the B/D and the 3 offer direct service to both DT Brooklyn and most of the rest of Brooklyn.
Your suggestion that people in the Bronx do not take jobs in Brooklyn, however, is, like almost everything else you say, clearly false based on the data.
I am aware that the Census Bureau attempts to track commuter flows. I do hope this is not a revelation, it’s how they determine metro areas. They have the employer’s reporting address, not the actual work location. It’s the same for many employers but not all.
They allow people in the Bronx to own automobiles. If Alon can imagine some particularly stupid cashier passing by cashier’s jobs in the Bronx and higher paying ones in Manhattan to schlep for two hours to Gateway Center in East New York. I can imagine a doctor in Riverdale driving to Coney Island Hospital until the family moves to Mill Basin. They allow people in Brooklyn to own automobiles too. A doctor in Mill Basin who took a position at Jacobi in the Bronx and drives until they find something in Pelham Manor.
None of them are going to take the train to New Jersey to shop in Walgreens because there are Walgreens all over New York City. Or a bus, there are lots of buses from Manhattan to New Jersey. From two different bus terminals. I’m almost sure there are no buses from the Bronx to Brooklyn. Though there might be ethnic markets being served by private buses. I’m digressing.
Though I suppose it could be that Alon cannot imagine there being high paying jobs in Brooklyn. Or the Bronx. The view across Ninth Ave can be very hazy.
Boy, this game gets really tough to follow when the goalposts just fly around the field.
That’s the way it works. It’s too bad the goalposts aren’t where you imagine they should be.
I don’t know why I bother, you sad, strange, hateful little troll.
The goalposts are where the Census Bureau plants them. I’m sorry you are disappointed but I can’t do anything about the Census Bureau. I’m sorry reality annoys you. I can’t do anything about that either.
You can pretend that no one travels from the Bronx to Brooklyn for work or vice versa, but the data is unequivocal that it happens. There are lots in the jobs in the world that are not as fungible as cashiers. Jobs are also not evenly spread. If the business in the Bronx has the cashier’s it needs then it won’t be hiring, and if one in Brooklyn is you will take that job so that you can have money to eat. Or you will stay with your employer when they relocate so that you don’t have to lose your job and go on unemployment, even if it means a longer commute. Same if your employer downsizes and offers to keep you if you switch to the location staying open. Or you move to the Bronx when you get married because you have a good paying job in Brooklyn while your spouse has one in New Rochelle.
The fact that the Census Bureau uses the formal reporting address is a known issue that only rarely causes problems. For instance, I believe that every teacher or parks worker or something in NYC is labelled as working in DT Brooklyn, because that is the reporting address for their HR department. But if you look at the On The Map Data (which I am rather certain you have not) you will see thousands of dots with values from 1 to 15. There are not thousands of businesses with 9 or 12 workers that maintain a Brooklyn address even though they do all of their work in the Bronx.
No matter what you say, tens of thousands of people commute from the Bronx to Brooklyn everyday.
I didn’t.
Alon is the one imagining people filling ordinary jobs paying minimum or near minimum wage taking hours long commutes. They aren’t that stupid. Or stupid enough to do it by bus. Because there are equally ordinary jobs closer to home. Or ones in Manhattan that likely pay better.
Alon is the one imagining people filling ordinary jobs paying minimum or near minimum wage taking hours long commutes.
Alon didn’t say hours long commutes, he said complex multi-bus commutes. I grant that multi-bus is unlikely in NYC, at least one leg would be subway, however, hours long commutes do exist for low income workers. Google Walter Carr, Trenton Lewis, or Patrick Edmund for the extreme examples that became famous.
Because there are equally ordinary jobs closer to home. Or ones in Manhattan that likely pay better.
I’ve given lots of examples why people can’t/don’t take jobs close to home. Then there are the outlying examples above. If people could get the better paying jobs in Manhattan they wouldn’t be working for or near minimum wage, would they. It’s people lower on the income ladder who have fewer options and have to make sacrifices like longer commutes for the jobs they can get.
I gave examples of people not taking multi-bus trips too. You want to imagine someone taking a “retail job” in “outer Brooklyn” who is also too stupid to take the train across Manhattan, instead of a “retail job” in the inner Bronx, go right ahead.
American transit executives have a hard time imagining using transit for anything other than going to work. That is why they advertise how much time you could save using transit instead of driving (often using very unrealistic routes instead of real world trips to work – there is good reason the bus doesn’t take an efficient route to the freeway, but in my car I’m going the direct route) . They often don’t run very reduced off peak service or even none at all – going someplace downtown after work: service ends before your event is over – when I was a kid growing up in Minnesota we often said “New York is the city that never sleeps, and Minneapolis goes to bed early” – there was nothing going on downtown after 5pm (I’m told things have improved?). Likewise if you attend church on sunday like 40% of the US (last I checked) you have to drive because they don’t run half the bus routes and the rest don’t run schedules that are useful for church (either arrive before the church is unlocked or arrive late). Sometimes they allow the poor to use transit for this stuff, but they make it clear “normal people” should be driving.
This is a self perpetuating cycle. Transit isn’t useful for this, so those building retail don’t build such that transit could be useful. And more and more jobs move to the suburbs because transit doesn’t have enough money to run good service where people live because they don’t capture the money from running all those other trips (not just fares, it is also mindshare of I use transit I wish it was better and am willing to pay taxes vs I use transit only until they fix traffic and parking so I can drive.
Step one is making transit more useful for getting into big cities where people don’t really want to drive. That’s where a lot of the train use in Britain comes from.
Instead of looking for a mathematical formula, Hayden could spend a few hours at peak times on Grand Central Terminal platforms to look at trains’ occupancy. For what it is worth, I took one of these trains on a Friday afternoon September 5 and there were plenty unoccupied seats.
This was consistent with MTA’s Metronorth’s current ridership which is only 80% of its 2008 peak. Most Connecticut stations on the New Haven line serve communities with tens of thousand inhabitants. Unless Hayden’s TOD is very large, it shouldn’t make a significant difference on board the trains.
It’s not going to be large because those pesky pesky 19th Century commuters went and built suburbs around the station in the 19th century. And the people living there now, who paid extra because the real estate ad said “walk to station” aren’t going to just wander off.