Transit Arteries and Autonomous Veins
How public transit can survive, and thrive, in the age of the autonomous taxi
Disclaimer: I’m not a certified city planner, I’m just a guy who cares a lot about public transit. I’ve linked to primary sources wherever I can.
2022 was a great year for my favorite genre of internet content: tech bros getting publicly humiliated for their terrible ideas. Whether they’re losing billions of dollars to “own the libs”, going to prison for running obvious scams, or failing to ruin Fiji, we can always count on the tech bros for a good laugh. Thanks, tech bros!
So imagine my horror then when, barely 24 hours into 2023, out comes a Reddit post where — god forbid — I actually agree with the tech bro.
Fleets of rentable self-driving cars (aka. Autonomous Taxis) with dedicated lanes have the potential to solve a lot of problems, including many of the problems intrinsic to trains. They can navigate around obstacles, deliver people directly to their destinations, and traverse tight curves and steep inclines. They can spontaneously form ‘platoons’ (basically, trains) for long-distance efficiency, then split apart into smaller units that serve communities too sparse or poor to support a conventional train line. They can even use existing infrastructure more efficiently than human drivers , increasing the availability of transportation without major capital expenditures.
But don’t get me wrong: trains are absolutely part of the future of transit. I’m a big fan of trains. I own these shoes:
Indeed, self-driving cars have nothing on trains when it comes to per-acre capacity. A 2022 study estimated that a dedicated lane for self-driving cars could move around 3600 vehicles per hour under normal driving conditions. Taking into account the average US car occupancy of 1.5 people, that means each lane could move 5,400 people per hour. Even the single relatively-modest subway line in my hometown of Seattle can move 3 times that many people — the equivalent of an entire highway dedicated to autonomous cars. Plus, unless the autonomous-only lanes lead directly into your garage, you’re still going to have to contend with normal traffic at some point.
If self-driving cars replace mass transit entirely they will make congestion worse, not better. Luckily, that’s not what I think will happen.
America, 2038
It’s the year 2038. You live in America’s trademark suburban sprawl.
You’ve had a car since you were a teenager but for the first time ever you’re considering going without one. Your old Camry is in need of expensive repairs, and for what? It’s obvious that your long commute downtown is making you sick, anxious, and tired.
The little town you settled in never had great bus service, but now self-driving taxis are rarely more than a few minutes away— maybe it’s time to ride the wave of the future.
Driving all the way downtown by autonomous car would only solve half of your problem: it’s still a long and tedious commute, not to mention expensive. Sure, you could try to sleep or work in the car, but you get motion sick. Plus, you still don’t entirely trust the AI driver — your hand rests on the emergency brake every time you ride. Luckily, there’s a train that stops in the next town over. All you have to do is make it to the station and within 10 minutes you’re flying past traffic in a cushy seat with free Wifi and a little fold-out desk where you can get some work done.
The train station downtown is still a few blocks from your office. You could summon another taxi (some days you do) but most of the time it’s more fun to just rent a e-scooter. The city has turned a lot of street parking into bike lanes — an obvious improvement now that most cars can autonomously find parking elsewhere. It’s a pleasant ride. You’d never really seen the city before, you were always too focused on driving. You find a cafe you like along the way. You talk to people more.
All-in-all your commute takes about the same amount of time as it did back when you drove yourself, but it feels so much faster! You can even leave the office a little earlier than you used to and work on the train, and you don’t have to plan your days around avoiding rush-hour traffic. Plus, if you want to grab a couple drinks after work, it’s no big deal — you’re not driving.
How do we get there?
This vision of the future assumes 3 things:
- Self-driving taxis become widely available and undercut the cost of traditional taxis (which: maybe, maybe not).
- Mass transit is able to remain competitive with self-driving car fleets on frequently-travelled routes.
- Cities respond to changing needs by replacing street parking with high-quality bike and pedestrian infrastructure.
Points 1 and 3 are fairly straightforward. Not easy by any stretch, but at least there’s a clear path forward. Point 2 is trickier.
Keeping Mass Transit competitive
I don’t think autonomous taxis will ever be strictly cheaper than taking the bus or the train, but they might get cheap enough that some current transit riders make the switch to taxis. This can lead to a disastrous feedback loop.
“Fare Box Recovery” refers to the percent of operating expenses covered by passengers in the form of tickets or monthly passes. Successful transit systems, like Hong Kong’s MTR or The London Underground, actually earn more in fares than they cost to run, putting their Fare Box Recovery above 100%! Unfortunately, most of the world’s public transit systems (including every single one in America) are nowhere near that point.
The reason, to use a tech-bro buzzword, is that most transit systems are sub-scale: they are not large enough to fully meet customers’ needs. A transit system that gets you half way to your job, or to 3 of your 10 favorite restaurants, is probably not going to be your primary mode of getting around — but it’s still going to cost your city a fortune to run.
Conservative politicians love to disparage public transit funding as government subsidies for city-folk, totally overlooking the fact that it’s a good investment. Not only does mass transit reduce traffic congestion, carbon emissions, and poverty — it can literally make money! Fresh investments in public transportation, if done well, can turn the negative feedback loop into a positive one (in business terms, a flywheel) and eventually create systems large enough to sustain themselves.
And yet, thanks to the looming prospect of autonomous cars, all new public transit projects have a shadow of uncertainty hanging over them. It’s hard for politicians to invest trillions into transit infrastructure when tech bros are whispering fearful thoughts about self-driving taxis into their well-funded ears. And, those tech bros aren’t entirely wrong! Given the current state of mass transit in countries like the US it’s almost inevitable that self-driving vehicles will accelerate the negative feedback loop of transit ridership, eventually leaving our existing systems in ruins.
So what do we do? Well, we stop trying to fight the autonomous car and instead embrace it as part of a balanced transportation diet.
Transit Arteries and Autonomous Veins
There’s a lot we don’t know about autonomous vehicles. When will they be available? How safe will they be? How much will they cost to buy and operate?
The only safe bet is that autonomous taxis will never rival the density of mass transit — the passengers per hour per precious square foot of downtown real estate.
So this is my blueprint:
Focus on excellent mass transit for dense well-travelled routes, and be ready to cede everything else to autonomous taxis.
Those taxis would have to be subsidized — the same way buses are today — in order to keep transit affordable. That might sound crazy (the city paying for you to ride around by yourself?!), but plenty of cities are already experimenting with this idea. Los Angeles has its Metro Micro vans, and dozens of other cities have piloted programs with ‘microtransit’ startups like Via. It turns out it’s not actually that hard to productively compete with bus service.
The cost of buses
In 2019, King County (which contains Seattle) spent 1.437 billion dollars operating bus service. Passengers made 123 million bus trips that year, meaning each trip cost $11.68 in total. The $2.75 fare I pay when I board the bus is somewhat of an illusion since, as an average King County taxpayer, the full $11.68 comes out of my wallet one way or another.
Is $11.68 too much? I can’t answer that quantitatively, it depends on how much you personally value the societal benefits afforded by public transit. For me, I’m happy to pay it. Still, if we could move more people for less money, I’d be even happier!
So why are buses so expensive? It varies city-to-city and year-to-year. The most recent numbers I could find for King County come from 2012:
This doesn’t include the cost of buying the buses in the first place. I couldn’t nail down actual numbers for King County but the hybrid diesel-electric articulated buses they use are probably in the ballpark of $750,000 and last for 12–15 years.
It’s not one thing making buses expensive, it’s a combination of everything. They’re large vehicles that need frequent maintenance, yet still the biggest expense isn’t the vehicle but the human driving it. Therefore it’s cheaper to run a small number of large buses than it would be to run a large number of small buses — even though the latter would provide more frequent and reliable service to riders. Naturally, if the driver were no longer necessary, this financial algebra would change.
The _real_ cost of buses
Discussing financials is all well and good but let’s not forget that money is an intangible fantasy we made up for our own amusement. It’s a nice thing to have, but it’s not food or water or clean air — all of which are actively under threat from climate change. So as much as money shapes public transit in most of the world, discussions around how public transit should look really aught to revolve around those more tangible costs: carbon emissions, the use of finite resources, and air pollution.
So how do buses stack up? Well, still not great!
King County operates about 1,500 buses, 80% of which are hybrid-diesel vehicles averaging just 4 miles per gallon. Given that a Toyota Prius can achieve well over 50 miles per gallon while carrying 5 people, a diesel bus would have to carry 71 people just to break even! Of course, that’s the absolute best-case scenario for cars. To get more realistic numbers, let’s imagine a city-owned autonomous taxi service consisting of a fleet of Priuses. These are the parameters:
- Each car averages 57 MPG (to be fair to hybrid-diesel buses I am specifically using a hybrid-gas car here, not an electric car)
- Each car spends fully 50% of its time traveling empty on its way to pick up a new passenger
- Each trip moves an average of 1.5 people
I also assumed that buses spend 5% of their time traveling empty to their operations base. The more realistic break-even point: 7.9 passengers. In other words: if you see a bus with fewer than 8 people on it, that trip would have been more carbon-efficient if operated by taxis.
That’s a low enough ridership target that, at least in King County, bus service is more energy efficient than the taxi alternative on balance. Importantly though, many less-travelled routes within King County are, considered individually, below the break-even point. I pulled ridership data from 2019, accounted for electrification, and found that of 130 routes:
- 76 routes (59%) are carbon efficient during peak commuting hours
- 48 routes (38%) are carbon efficient during ‘off-peak hours’ (9am-3pm)
- Only 20 routes (16%) are carbon efficient during night-time runs
To be fair, King County Metro is aiming to be fully zero-emissions by 2035 by transitioning to battery-electric buses. Still, zero-emissions isn’t the end of the climate story. Energy efficiency will always be important, either because some electricity is generated via fossil fuels or, optimistically barring that, because setting up renewable energy plants has environmental costs of its own. So I’d be remiss not to point out that King County’s new battery-electric buses can only travel 0.43 miles on a kilowatt-hour of electricity, while a Tesla Model 3 can travel 4 miles on the same charge.
Investing wisely
I’m not here to dunk on buses. Sure, I don’t have shoes with buses on them, but what are my beloved trains if not buses-on-a-stick? Trains generally have higher capacity and nicer ride quality, but they also cost a lot to build. There is a place for Bus Rapid Transit in the future — bus lines with high ridership and frequent service are doing their jobs beautifully and don’t need to be ‘disrupted’.
But, as we saw, 41% of all bus routes in King County never have enough ridership to justify their carbon cost — not even during rush hour. These routes also tend to be infrequent, running only once or twice an hour. If missing your bus means standing in the rain by a busy street for half an hour you’re probably going to spend at least some of that time seriously reconsidering your transportation choices.
These routes are the small veins and capillaries of the living city — necessary to keep all the appendages alive and healthy but too sparse to justify large vessels. On these routes, traffic and congestion are less of an issue. They do not require mass transit, only affordable transit. Replacing buses with autonomous taxis here could save carbon, save money, and provide better service to residents.
We often refer to downtown cores as the “heart” of the city, so it’s only natural to extend that biological analogy to transit. Your heart doesn’t have hundreds of veins leading into it from every corner of your body — it has only 6 arteries which branch off into smaller veins and even smaller capillaries. Instead of your local suburban transit system taking you directly downtown it should take you to an artery — a subway station or a frequent bus corridor. Sure, nobody likes having to transfer to another line, but if the wait times are only a few minutes and the stations are pleasant it’s really not a big deal.
This bit of biomimicry could drastically improve service, but isn’t really possible for transit agencies whose resources are spread thin across sprawling suburbs. Something needs to change, and autonomous taxis are (or rather, hopefully soon will be) poised to help.
Let’s get weird with it
There’s so much room for nuance and creativity here! Cities could run traditional bus service during rush hour and switch to autonomous taxis at night. They could operate both fully-accessible vans and tiny single-occupancy vehicles, dispatching each as needed. They could even get involved in food and package delivery using their existing autonomous fleets, further reducing people’s dependence on private cars. Or, why not go full “Florida retirement community” and design new neighborhoods around golf carts and e-scooters?
Moving away from one-size-fits-all solutions could be the key to growing public transit systems into growth engines that literally and figuratively drive our cities.
False Dichotomy
Discussing nuance is hard and boring and honestly just makes you sound like a pretentious snob. It’s much more fun to either be team “Fuck Cars” or team “Fuck yeah, Cars!”
But of course, that’s not really helpful. So let’s put on our “pretentious snob” hats (which I imagine to be some sort of bowler, or maybe a beret) and think through the details.
Transportation is the art of moving people from Point A to Point B. It sounds like the simplest thing in the world — but, where are Point A and Point B? How far apart are they? How many people need to make the trip? And, what do those people value?
It should be obvious that no one mode of transit will be a panacea that solves every problem. Self-driving cars cause congestion, buses are inefficient, trains are expensive to build, and all of them sink when you try to drive them across the ocean. The future is not about which single technology will “win”, but rather how all of the technologies we have can work together to create a whole greater than the sum of its parts.
Or, in the words of every improv teacher I’ve ever had, “Yes, and”!
Should we embrace the self-driving taxi? Yes! And, we should offer subsidy programs to ensure they’ll serve lower-income communities that currently depend on bus service.
Should we continue to invest in other modes of mass transit like subways and Bus Rapid Transit? Yes! And, we should focus on improving the reliability, frequency, and comfort of the transit systems we already have to ensure they stay competitive.
So, which is better: Self-driving cars or mass transit? Well…
