Basics: Technology Never Changes Geometry

Like many fashionable tech folks, Elon Musk wants to replace city buses with little vehicles that protect you from having to share space with strangers.

With the advent of autonomy, it will probably make sense to shrink the size of buses and transition the role of bus driver to that of fleet manager. …  [This technology] would also take people all the way to their destination. …  (Emphasis added)

Musk assumes that transit is an engineering problem, about vehicle design and technology.   In fact, providing cost-effective and liberating transportation in cities requires solving a geometry problem.  This confusion is very common in transport technology circles.

In dense cities, where big transit vehicles (including buses) are carrying significant ridership, any “small vehicles replacing big vehicles” solution increases the total number of vehicles on the road at any time.  The technical measure of this is Vehicle Miles (or Km) Travelled (VMT).

Today, increasing VMT would mean increased emissions and increased road carnage, but let’s say technology has solved those problems, with electric vehicles and automation.  Those are engineering problems.  Inventors can work on those.

But there is still, and aways, the problem of space.  Increasing VMT means that you are taking more space to move the same number of people.  This may be fine in low-density and rural areas, where there’s lots of space per person.  But a city, by definition, has little space per person, so the efficient use of space is the core problem of urban transportation.

When we are talking about space, we are talking about geometry, not engineering, and technology never changes geometry.  You must solve a problem spatially before you have really solved it.

The reigning fantasy of Musk’s argument is that we must always “take people all the way to their destination.”   To do this we must abolish the need to ever change vehicles — from a train to a bus, from a car to a train, from a bus to a bike — and of course we also abolish walking.  This implies a vision in which buses are shrunk into something like taxis, because a vehicle going directly from your exact origin to your exact destination at your chosen time won’t be useful to many people other than you.

So a bus with 4o people on it today is blown apart into, what, little driverless vans with an average of two each, a 20-fold increase in the number of vehicles?   It doesn’t matter if they’re electric or driverless.  Where will they all fit in the urban street?  And when they take over, what room will be left for wider sidewalks, bike lanes, pocket parks, or indeed anything but a vast river of vehicles?

There are audiences for which Musk’s vision makes perfect sense, people for whom useful high-ridership transit isn’t an option anyway.  There are two big categories of these people:

  • People who live in outer-suburban and rural areas, where space is abundant and high-ridership transit isn’t viable.
  • The wealthiest 20% or so of urban residents, who can afford to use relatively expensive services that would never scale to the entire population of the city.

If you are in one of these categories, your most urgent task is to remember that most people aren’t like you, and that cities are impossible if everyone lives according to your personal tastes.  As Edward Glaser said, “one’s own tastes are rarely a sound basis for public policy.”

That issue, right there, is the great disconnect between tech marketing and genuine urban problem-solving.

Tech marketing is all about appealing to elite personal tastes.  It runs on the assumption that whatever we sell to the wealthy today we can sell to the masses tomorrow.  But some things stop working when everybody buys them.  Cars in dense cities, for example, are not a problem when only the top 20% are using them; it’s mass adoption of cars that makes them ruinous to a dense city and to the liberty of its citizens.  Ask anyone in a fast-growing developing world city about that.

Here is the harm that this all this elite chatter about abolishing the bus is doing:  It’s introducing fatal confusion into the discussion of urban development.

Dense cities that want to live in the real world of space and time, and that do not want to become dystopias that are functional only for the rich, need to use urban space efficiently.  There is some simple and well-proven math about how to do this, which is also the math of how transit systems achieve high ridership.

These cities need to organize themselves around frequent transit corridors, where big-vehicle frequent transit, bus or rail, can prosper, allowing the city to grow dense without growing vehicle trips.

Someday some of these corridors will be rail or Bus Rapid Transit.  But the only way to grow enough corridors quickly, so that you cover much of the city with frequent service that can succeed in ridership terms, is to take frequent fixed-route bus service seriously.  If you don’t do that in your land use planning, you’re going to end up building a city where fixed transit is geometrically impossible, and then you’ll have no alternatives when congestion begins to strangle growth.

Smart cities aren’t just the ones that chase the latest technology fads.  They’re the ones that think carefully about the spatial, geometric problem that a dense city is.   Because if it doesn’t work geometrically, it doesn’t work.

99 Responses to Basics: Technology Never Changes Geometry

  1. JJJJ July 21, 2016 at 7:41 am #

    Jarrett , please, he’s invented PRT! A prime recipe for success. The man is a true visionary.

    • David Vessel July 21, 2016 at 10:35 am #

      EXACTLY! I was thinking PRT.

    • Erwin van den Brink July 29, 2016 at 1:33 am #

      I think this whole Tesla ting is a big phoney. It is so very old school, old technology in a new fashionable package namely: a car, more over a ‘wannahave car’. First: If we are going to replace 1,2 billion combustion engine cars (yearly out put of 90 million cars a year) by electric cars, we create a environmental problem of far bigger magnitude than we have today with the fossil industry. Any idea of the amount of copper it takes, for a start? Lithium? (the ‘new oil”?)
      But more important is: new technology, I mean really new technology, like the 3D-printer and all kind of other micro-manufacturing technology that fits in abandoned offices, all kind of cramped back yards in (European) inner cities. This industrial renaissance is powered by young people who also wanna live in inner cities. A private cars is the lowest thing in their priority list. Working in the new economy means hanging out in inner cities, meeting en greeting, networking. Assets: a bike, a smarthone, a laptop of tablet, a wifi hotspot. Not a car. It is a financial pain in the ass, parking space is unafforable of juist is not available. Families in Amsterdam massively use cargo bikes to ferry their kids to school, we have thanks to informationtechnology concepts like car sharing, instant rental cars. But yuppies, they don’t buy and own cars anymore.

      To use instant rental cars, which takes a subscription to for example a mobilitycard of the dutch railroads, one never has to walk further than 100 m. (You would only use them for appointments at remote locations out side the city.)

      This whole car thing really is very 20th century.

      • Alan Peery August 21, 2016 at 2:40 am #

        * We’re already replacing those cars with new cars, so your thought that having electric cars requires a faster replacement cycle is not a given. You should also check out the overall environmental consequences of the electric car:
        * Lithium is plentiful mineral, is used many times, and will be recycled after the battery wears out. The lithium is an element, and it won’t be worn out.
        * Read the car sharing section of Musk’s latest plan:

        • Anri December 23, 2017 at 5:50 am #

          Please travel to south america and africa and visit yourself mines there and the real damage lithium is doing to the environment.

      • Hcat December 25, 2017 at 7:42 pm #

        Yuppies? The present day equivalent of yuppies, if they are successful, will find a way to get cars. If not, they’ll do without or join Zipcar. Even in cities with great transit, like New York and London, people take to cars until the streets get clogged and transit is more convenient. And non BRT buses, because they share traffic and stop frequently, will always be for those who can’t afgord cars.

        • Cole January 3, 2018 at 6:56 am #

          “And non BRT buses, because they share traffic and stop frequently, will always be for those who can’t afgord cars.” – not necessarily if you take into account the time it takes to park in a dense downtown area. During afternoon peak times, it might take 15-20 minutes to get out of a garage, making buses more convenient in some cases – even if you can afford to own a car and pay for parking.

    • Luca December 15, 2017 at 1:41 pm #

      Elon Musk has not invented PRT

  2. Sean Gillis July 21, 2016 at 7:50 am #

    Jarrett, a question: can we quantify the density level where a settlement goes from being able to operate with something looking likes Musk’s vision for many trips, and a settlement dense enough to need fixed routes and big vehicles? Low density depends on one’s vantage point.

    The ‘on-demand’ sounds great, until someone tries to explain what that would look like for an individual trip. I’m at a stop, I hail a bus with my phone or the button. Mini-bus shows up: maybe in two minutes? maybe in seven? I’m on my way to the destination, but where am I going next if someone rings for the bus near my route? How will the algorithm determine whose pick-ups and destinations get priority?

  3. Tom West July 21, 2016 at 8:12 am #

    There’s a ridership intensity threshold between demand-responsive and fixed-route transit. That threshold is dictated by how many boardings+alightings per hour you can do with demand-responsive transit (it’s about six, btw). That has nothing to with the *cost* of operating transit vehicles. Driverless transit vehicles change the cost, but don’t change the logistics of demand-responsive transit. So, there won’t be a shift in the types of areas served by one vs. the other.

    Driverless buses’ big advantage is lower cost. That means you can provide more service for the same subsidy and fare levels. Now, drivers are about half the marginal cost of operating buses, and fares rarely cover more than half the marginal cost of operating buses. That means that switching to driverless buses could let you abolish fares *and* have more service *and* still reduce subsidy!

    • Sean Gillis July 21, 2016 at 10:14 am #

      Tom, just for clarity: demand-responsive transit can only handle about 6 boardings per hour?! Is that a theoretical average, or a maximum, or from real-world examples. Regardless, pretty low intensity.

      • MikeP July 22, 2016 at 4:46 pm #

        Is that average number for something like an Uber or Lyft? Pick someone up, take them to their destination, drop them off = 10 minutes. I would believe that, but it implies an awful lot of robot cars to replace one urban bus!

        • Brian July 24, 2016 at 10:12 am #

          As an Uber driver I can atest that exceeding 4 rides an hour is very rare even in very high demand periods. It looks good on paper, but in reality many variables come into play. Passenger delays loading and/or boarding efficiently, traffic, destination changes enroute, ride request cancellations, drunks dropping pick up locations that aren’t even close, etc.

  4. Mike Lindblom July 21, 2016 at 8:42 am #

    Before the geometric argument, which is powerful, I think we have to overcome the more basic fallacy of people who insist on an either-or future, or preach a single solution for mobility. Otherwise, advocates will continue to debate high technology and multibillion dollar infrastructure at a level of a six-year-old.

    To be resilient, in a time of population growth and climate change, a smart society needs to innovate in more than one kind of transportation. Not to mention the obvious point that different users will have different needs, as they do now. Every day in 2016, I make a snap decision between bus, car, bike or a train-bus combo depending on the morning traffic conditions, no different than wearing a rainjacket or sandals. To depend on one mode to rule the transport system in a fragile area like Seattle is a recipe for disaster.

    • Dave July 22, 2016 at 12:42 pm #

      “Every day in 2016, I make a snap decision between bus, car, bike or a train-bus combo depending on the morning traffic conditions, no different than wearing a rainjacket or sandals.”

      I would venture a guess that most people don’t do this, at least not on a daily basis. Most people just do whatever they did yesterday – it’s why most people call it their “morning routine.” They keep doing this until either (1) their commute gets really awful for whatever reason and they consider switching to something else or (2) they add another destination/trip purpose between home and work (e.g. “I’m going to happy hour after work”) that forces them to reassess whether what they did yesterday is still appropriate for today (e.g. “I don’t want to be driving drunk so I should take the train today”).

      Truly being open to a snap decision on which mode(s) you’re going to take each and every day requires allotting a lot more travel time into your schedule, since almost always, you have to wake up earlier than you otherwise would if you’re driving to really make transit or biking or walking get you to work by 8:00am. Constantly resetting one’s alarm clock for different wake-up times is a pain in the A.

      (Not to mention that once you’ve decided to not buy a car or to buy a monthly transit pass or whatever, you’ve pretty much locked yourself into a mode for a while…)

      • MikeP July 22, 2016 at 4:51 pm #

        I agree with Mike. In dense cities, there are lots of options and they have trade offs. If the weather is nice and I don’t have to dress up, I bike. If I’m not in a hurry, I walk. If I’m feeling lazy, I take one of three bus routes, depending on which one is coming sooner.

    • corner soul July 25, 2016 at 12:41 pm #

      Problem is someone like Musk doesn’t really care about cities or mobility. He only wants to “disrupt” transit because he sees it as easy prey. Most of these tech entrepreneurs and venture capitalists are just 21st century robber barons. Monopoly is the name of the game.

      • Chris July 26, 2016 at 7:58 am #

        Musk is a techno-utopian. He might be misguided, or uniformed but if he was just interested in making money, there a lot of ways easier than starting a rocket company, an electric car company and an solar power company. Also, he relinquished all his electric car patents. Not the obvious move for someone trying to build a monopoly.

    • Lux August 6, 2016 at 1:55 pm #

      Hi Mike, what you said is exactly right. The worse is top down rule. Let there be private bus companies or share rice companies which make the decision what an area needs the lost. If a service is not needed, let it fade away.

  5. Andres Salomon July 21, 2016 at 9:18 am #

    Taking people all the way to their destination has another unspoken problem: one of health. Walking is great for people, both physically and mentally. Study after study shows that people who rely on public transit instead of driving end up getting more exercise and are healthier. Having lots of autonomous door-to-door van pools or taxis would be an absolute disaster for public health. On the other hand, creating communities where people can actually (safely, comfortably) walk to their destinations (or bike, or walk/bike to nearby transit stops) is cheap, doesn’t rely on potential futuristic technology, and creates a huge boon for public health.

    • Chris July 26, 2016 at 8:04 am #

      Perhaps we should force people to carry a 10lbs pound weight around. Then they’d really be healthy. I agree that transit promotes walking and walking is awesome (i walk to work) but a constant refrain of HT is that transit should be useful. Noting the autonomous taxis would be too convenient (useful) for people’s own good sounds a lot like the holistic benefit of building slow transit.

  6. Adam Old July 21, 2016 at 9:36 am #

    In Miami we have private jitney services that follow a few cherrypicked trunk routes (overlapped with buses). They are cheap, hot and crowded, but they come at higher frequency than buses and seem to move more passengers on the same routes—I could be wrong about that.

    I would love some automated drivers for those things because they are really not 100% excellent.

  7. P July 21, 2016 at 10:03 am #

    I think we already have this kind of transit, it is called “taxi”.

    (boom, tish!)

  8. Jonathan Rabelo July 21, 2016 at 10:13 am #

    It may seem Musk wants to replace all forms of mass transit for small eletrics cars, but i think operating variable sizes of buses along-side with big mass transit rails or BRTs, will be very benefical, since small cars can be more efficient to operate many routes while keeping a high frequence.

  9. Sky Collins July 21, 2016 at 10:14 am #

    I think at his core, Elon Musk has made one core miscalculation any time he’s talked about transportation beyond personal automobiles. He assumes that he’s got ‘the answer’. His ‘answers’ are wonderful in a way: big, dreamy hypotheticals that get our imaginations spinning and our endorphins rushing. Make no mistake, there is a very important place in the transit discussion for big, dreamy hypotheticals but only when they’re labeled as such. Hyperloop is the most baldfaced example of this fallacy and I’ll deconstruct it here.

    In Hyperloop’s original concept form you have a maglev capsule in a near-vacuum pipeline being whisked along at shocking speeds.
    First let’s start with the basic premise of needing to traverse such long distances in such short periods of time. Maybe 40 years ago, when the Jet Age was in full swing, this would be reason enough alone to build such a thing. After all, it was the pre-internet age that gave birth to the Concorde and the post-internet age (and costs) that killed it in the end. Beyond a certain efficiency, people just aren’t as concerned with getting from A to B as quickly as possible anymore. Most business meetings are via vid-conference now and only rarely do people need to actually assemble in a single place from across the globe. The concorde, for its failings, lasted as long as it did because it wasn’t targeted at the masses in the first place. Even when everyone DID have to travel, the Concorde was only for the upper-level travelers who could afford the astronomical costs.
    Next, the technology. It’s not hard to explain how this works. With traditional maglev, where the carriage hovers over the guideway on magnets, the only real frictional concern is air resistance. Remove that and your top speed potential shoots up like a rocket. But to do that you must place passengers in the same level of risk as astronauts. One minor structural failure in the capsule and everyone’s blood boils in an instant. Its not a pleasant thought but it’s a real concern AND its not something you could ‘be lucky’ with like a traditional railway crash. That capsule pops and so do you. So you have a potentially lethal method of transportation on one side of the court and an army of lawyers and insurance companies on the other with the Federal Government as referee and the rest of the nation politic in the bleachers. How long do you think it’ll take to sort that out? In that amount of time any number of alternatives could be well underway before the first approval comes down for Hyperloop.
    Finally we get to alternatives. For high-speed ground transportation you have two choices: Traditional high-speed rail (HSR) and Maglev. They each have advantages and disadvantages but one thing they both have in common is that they’re both market-ready. HSR is what we all think of when you say “High-speed rail”: a high-speed passenger train operating on specially designed but otherwise conventional steel rails and steel wheels. For this reason, HSR’s primary advantage is in ‘last mile’ compatibility. In areas where there isn’t sufficient room to build another at-grade railbed, HSR trains can simply operate at low speed on conventional tracks, simplifying their integration with existing urban rail infrastructure. Aside from that advantage, most of the advantages in the comparison fall to Maglev, which is the more cost-effective solution despite a higher initial cost to build. For one thing, there is little to no physical wear on either the train or the guideway in a maglev system since one never touches the other and there are virtually no moving parts save for the emergency support systems (wheels that carry the carriage in the event of a catastrophic power loss). This ultra-low maintenance eliminates a large chunk of operating expenses, meaning that the initial cost is recouped more quickly. Additionally, the electrical systems have regenerative braking built-in as this is how they actually stop. A big deal was recently made about regen braking being integrated into HSR but there are mechanical losses in these systems whereas regen in maglev is absolute. Speaking of stopping, that’s another advantage to Maglev’s name. Acceleration and deceleration rates are significantly faster than in HSR which means that minimum effective distance between stations is less for Maglev than for HSR. Maglev is also unaffected by weather conditions since friction isn’t part of the operation. A Transrapid maglev train will accelerate and decelerate at exactly the same rate in sleet as it would on a sunny spring day and can tolerate almost 10cm of snow buildup on the guideway.

    Sorry to ramble a bit but it annoys me that Musk can’t be satisfied with his ludicrous Hyperloop concept, and has to start pushing his nose into urban intracity transportation.

    • TomTX July 23, 2016 at 4:29 am #

      Wow, with such a fundamental misunderstanding of the original Hyperloop proposal – why should I read the rest of your crap? It was never maglev. It was supported on air, like an air hockey table. This was key to the method. Low pressure so that you can use cheap vacuum pumps, no expensive magnets, use the problem of remaining air to support the capsule. Magnets to accelerate and decelerate, but not for levitation.

      • Greg March 7, 2017 at 11:10 am #

        Tom, he seems right on this one. The recent MIT work has included passive maglev. The compressor has been removed from the design. Links are in the Wikipedia article.

        Calling each other “crap” doesn’t help. I’d much rather read each of your ideas than have you insult each other just because it is the internet.

  10. Nate Simpson July 21, 2016 at 10:22 am #

    I think we should wait and see what the designs are before making a ridiculous assumption that Musk wants to reduce the size of buses to three people each; he explicitly talks about increasing passenger density in a given area, which is fundamental recognition that geometry matters. There would be zero point in a plan to produce 3-person “buses” when a Tesla sedan could seat more people (seven, I believe) than that and he does intend to allow people to contribute their Tesla automobiles to a ride-sharing fleet. Why talk about buses at all in that case instead of explicitly saying “replace buses with ride-sharing” if that was his intention?

    I think it’s more likely that he has a plan for a physically smaller bus that contains the same (or nearly the same) amount of people as current buses, putting more of them on the roads which will each replace tens or dozens of personal vehicles, thus providing a net reduction in VMT. Until there are more details, this theory makes more sense than the author’s alarmist one.

  11. Justin July 21, 2016 at 10:37 am #

    Did you actually read Musk’s piece? Honest question here, because nothing about this criticism makes sense.

    Part Deux literally refers to one of their new product categories as “high passenger-density urban transport.” He describes _redesigning_ buses, not getting rid of them. _Increasing_ their passenger density, not lowering them. Nothing about what Musk is describing ignores geometry. Quite the opposite in fact, as automated vehicles that can adapt distances to each other perfectly is a better use of space than regular buses. An automated bus system would be more space efficient than humans doing the same thing.

    This narrative of the replacing mass public transport wither personal AI taxis and wasting space just doesn’t exist here. You made it up.

    • Jarrett July 21, 2016 at 10:57 am #

      Justin. Musk says “It will also take passengers all the way to their destination.” That’s what triggers all these geometric outcomes. Jarrett

      • Andrew Cone July 21, 2016 at 3:03 pm #

        Right, but the antecedent of “it” is “automation,” not “every vehicle that may be summoned to these newfangled bus stops”

        Musk worded this section clumsily, but I think if you read for intent, not for techie-hating call-outs, it’s clear that he means that it would be possible for an automated system to provide such a service, not that we will never need higher capacity busses or transit on fixed routes ever again.

        Musk is clearly not proposing getting rid of buses—he explicitly states that some will need to have higher capacity—and a single offhand remark about taking you to your destination does not prove that he is as geometrically naïve as you suggest.

        Look, I like busses too. But we need to do better than infrequent, sparse, bumpy service that can’t even fill its seats even when the government pays more than half the operating costs. Busses can be a great thing. But they have huge defects, which the improvements Musk suggests could ameliorate. As a transit consultant, I would hope you devoted more energy to trying to figure out what Musk and his ilk might be right about than trying to poo-poo them on the basis of one phrase of sloppy writing.

        • BK Bed-Stuy July 25, 2016 at 1:43 pm #

          I love innovation, and technology. We should incorporate more into our transit systems, and most systems should be doing it yesterday.

          But, the techno-uptopianism in these responses are over the top. Where is the energy coming from, and how is it possible that it’ll be so cheap as to power all the implied computing needs? How will these massive networks be reliable, sustainable, and resilient? How will the manufacture of this massive network be financed when capital is becoming increasingly scarce? If the capital does exist for this at some point, is this vision of transportation the best use of it? Does it enhance the prosperity and well-being of communities of diverse socio-economic and racial backgrounds which are largely geographically segregated? Doesn’t this vision encourage unsustainable (in every sense of the word) suburban/exurban development patterns?

          If, somehow, the techno-utopians can answer these questions, I’ll *begin* to take them seriously. Otherwise, our innovations in transportation technology should be focused where it can make a tangible, sustainable, equitable, socially-positive, cost-effective, and immediate difference. Keep dreaming big, but remember to be grounded in reality.

      • Jarenot December 24, 2017 at 4:32 am #

        I don’t understand how you keep repeating that he isn’t solving the problem about “geometry”.

        Anyway I don’t agree with your meaningless dialogue “technology never changes geometry” but that’s besides the point.

        What do you not get about the “Boring” Company digging “tunnels” to solve the spatial problem. That is literally THE reason he has formed the company. To. Solve. The. Geometry.

    • Tyler July 21, 2016 at 9:09 pm #

      He describes removing the aisle as a means to “increase density,” which would suggest a small van (that somehow accommodates for wheelchairs). When the trend is to reduce seating capacity on urban short distance lines since it opens up more aisle space where standing people are much denser. We’re talking going from 140 people on a bus with a single motor, AC system, effect on traffic congestion, etc to 24 vans of six people.

  12. engineer July 21, 2016 at 11:20 am #

    Wow. Smaller vehicles operating ad-hoc dynamically generated routes will spend a lot less time on the main roads that are currently congested and more time on roads that don’t have enough ridership to justify a bus stop now.

    EG: much of the “Additional” space is the un-utilized space now, which reduces demand on the more utilized space. Not all space is the same, most space isn’t even roads. But you’re pretending like it is. Not even all roads are the same.

    You’re applying a rule of thumb based on certain assumptions to a situation where those assumptions don’t hold.

    Further we have an industry (urban transit) which has wasted billions of dollars of taxpayer money to provide terrible service … Mostly because it is lost to unions and other forms of corruption.

    Your vested interest is in being a “consultant” to that situation– not in seeing superior technologies make this government stranglehold obsolete.

    So you are incentivized to not read what he wrote too closely and to knock down the strawman you did.

    • anonymouse July 22, 2016 at 7:27 am #

      We already have “smaller vehicles operating ad-hoc dynamically generated routes”, in the form of UberPool and friends. I’ve taken it. It’s annoying: you want to go directly from point A to point B, not take a 15 minute detour to point C to pick up a second passenger who will in turn be annoyed at the 15 minute detour to point B on his way to point D. Try it sometime, or its low-tech cousin the airport shuttle van, I think you’ll find the experience enlightening. Also right now, it’s wasting billions of dollars of Saudi oil money to provide this service.

      • Brian July 24, 2016 at 10:28 am #

        ‘Also right now, it’s wasting billions of dollars of Saudi oil money to provide this service.”

        …in China. That sentence was incomplete, he’s spending the saudi money to open China, not to subsidize rides in established markets.

    • Gabe July 22, 2016 at 5:37 pm #

      That’s completely wrong – a vehicle sticking to ad-hoc routes will only spend more time off main roads if its travel is localised to one subnetwork which is off a main road. The moment you have a passenger who wants to travel from one subnetwork to another, then you have to use a main road, which is exactly the problem, especially if the service is trying to compete with high capacity fixed-route services. I.E. basically what Jarrett said: low capacity, demand responsive service works for last mile but not for main routes, and only when it can be run commercially.

  13. Joshua Arbury July 21, 2016 at 11:57 am #

    Hi Jarrett,

    Good article. I wonder whether what really sits behind this ongoing misunderstanding of transit is because people keep thinking the capacity constraint is the number of people per metre you can carry while moving, whereas it’s actually about how many people per second you can get on and off the vehicles, and how much space this loading and unloading requires.

    Take a bus vs 50 (driverless?) cars or taxis. Travelling along a lane the space requirements aren’t hugely different, especially if the vehicles can travel closely together.

    The problem really comes when they all park and people get on and off. That’s where they all get in each other’s way or require huge infrastructure. Same with buses vs higher capacity PT options. The real value in LRT or heavy rail is the loading and unloading efficiencies rather than being able to carry so many more people while moving. Pretty sure one of those Hudson River tunnels in NYC carries huge passenger volumes. Of course it requires a monumentally large bus terminal to do so.

    Thinking about geometric capacity this way might actually break through these misunderstandings and highlight why more, smaller vehicles can never be a solution when demand is high and space is limited. No matter how close they can travel to each other when moving.

    • Eric Orozco July 22, 2016 at 8:32 pm #

      Elon’s point-to-point fallacy also calls me to question my carcentric assumption that an autonomous vehicle will always be available for us magically whenever we need it. That’s of course the case when you have it parked near you always. But with ridehiring, that certainly can’t be the case. We will certainly be waiting and queuing for use and that inconvenience seems would make quite a difference.

      That fact means that folks will not only need to tolerate the need to share rides, if they want to preserve commerce, but learn to avoid places where people concentrate too much and share events too much. Not only avoid those places but avoid ROUTES THROUGH THEM. This situation calls to question our ability to make people share as easily as we think, to make them prefer urbanist ideals of places, and even to remove parking as much as we think since fleets will want to be able to stack reserves everywhere. Yes, one can algorithmically predict demand and chain trips, but unless the world has parking lots full of reserves everywhere for point to point, you will be hankering for a predictable ride you might only find in a bus. Or in a bike you can turn.

  14. Andrew Cone July 21, 2016 at 1:38 pm #

    This is a good explanation of why busses are necessary in high density areas, but it kind of misses Musk’s point.

    As you surely know, in the vast majority of the US, busses are infrequent, sparse, and nowhere near full. In all but a few cities, bus trips account for a negligible fraction of passenger miles travelled. The summon-button-enabled, straight-to-destination bus stops Musk is describing are not, I think, imagined to replace heavily used stops in places like San Francisco or downtown Denver. They are imagined to *add* a bus-like service in somewhere like metro Milwaukee, which currently has no fixed-route transit to speak of, and given entrenched settlement patterns, is not about to.

    The geometry problem you describe applies in places that *already have* transit. For the vast majority of Americans, however, whose communities look more like Overland Park, KS than Queens, NY, the geometry problem is that fixed-route, larger busses will never be a practical way to get around.

    If Musk was saying “Soon we’ll abolish the MTA and the T and BART!” I’d call him a fool. But I read him as saying “Maybe people will come to use something like transit in Overland Park.” I find that both interesting and compelling, so I don’t find this article that helpful.

  15. Nick July 21, 2016 at 2:39 pm #

    Does the author of this text understand English? I’m in doubt to be honest, because he simply made up his wall of text out of one phrase. Shame on you.

  16. Theodore Maniatis July 21, 2016 at 5:30 pm #

    This article forgets that Elon’s fleet will not be replacing today’s buses. It will be replacing the cars with single occupant. A Tesla would pick me up and drop me off at my destination then it would pick up the next passenger and drop them off at the their destination. Geometry problem solved. That is one car on the road for what would have been two cars and only one car that needs to be parked therefore parking lots and parking spaces on roads would disappear thus more space for lanes on streets. GEOMETRY SOLVED! It’s not rocket science or maybe it is because only Elon Musk is understanding this solution and everybody else does not have his vision because if they did then they would have started SpaceX and Tesla.

    Small buses work best because they can pick up a few people at a time and drop them off at a locations that are next to each other.

  17. Jarrett July 21, 2016 at 6:30 pm #

    On the last three comments:

    Yes, the geometry problem is solved as long as you’re in a suburban place where there’s lots of space. And I agree, as I said, that the ‘direct to destination’ goal is probably possible with automation in lower density places, where you have room for all the small vehicles.

    And I would not react to Musk’s “direct to destination” language this way if I were not hearing the same confusion over and over in my work as a transit consultant, and seeing how actual useful transit systems are dying from the apathy — if not hostility — that this kind of tech PR is engendering.

    The larger problem on display here, in Musk’s text and in the last few comments, is the assumption that the suburban or small-town experience — “Overland Park, Kansas” in one commenter’s phrase — is the normal point of view. Because the “direct to destination” fantasy is doing tremendous harm to the ability of DENSE communities — places that are already dense or want to be denser — to have a clear conversation about what kind of transit they will need, and what kind of urban planning they need to do to get there.

    Musk may have exaggerated in this passage. Surely the clumsy middle sentence in the paragraph I quote suggests he doesn’t subject his work to editors. But it’s still fair game, because lots of people are exaggerating in the same direction, and doing the damage I describe.

    • Sean Gillis July 22, 2016 at 6:34 am #

      Jarrett, I just did a radio interview on rapid transit’s potential in Halifax, and a phone-in question was whether we should bother investing in fixed route service because of potential technology like automated vanpools. So even as a part-time advocate, I’ve come upon arguments against potential transit investment – in places that are underserved but could support good quality fixed routes.

    • Andrew Cone July 22, 2016 at 6:45 am #

      You mention “seeing how actual useful transit systems are dying from the apathy — if not hostility — that this kind of tech PR is engendering.” Can you give some specific examples?

      I live in the Bay Area and from what I can tell, there is overwhelming support for multiple BART extensions, Caltrain electrification, a new muni line, and to some extent BRT throughout the bay (though people never seem to get excited about busses). I hear people praise Uber all the time and even hope for a self-driving golden age, but it seems orthogonal to people’s support for transit improvement.

      • Ted K. July 22, 2016 at 5:28 pm #

        Improved mass transit for the S.F. Bay area ? Yes, rail and bus upgrades and extensions are welcome.

        BART extensions ? HELL NO !!!! BART’s technology comes at too high a price, they have starved other worthy projects of funding, and their designs are often seriously flawed (e.g. Colma Stn., Millbrae Stn., SFO extension). Their attitude is one of toxic arrogance (e.g. Oakland Airport Connector). The planning for a Silicon Valley extension includes a redundant station (Santa Clara) that is a veiled attack on both Caltrain and the CHSRA. That station may be a killer issue when locals vote in November on a Santa Clara Co. tax measure that has a transit component and a BART bond in neighboring counties.

        I live in San Francisco and have used BART for a long time (student, reverse commuter, and fun-runner). I reluctantly support an extension to the Diridon Stn. because that’s an interchange point for many services. But the route being followed by BART to Diridon is over long in both distance and decades. There are transit geeks in the South Bay who talk of a DMU network that would have been an expansion of Caltrain or a sister agency. It could have been built in the 1990’s for tens, maybe hundreds, of millions of dollars ($ 10**7’s, $ 10**8’s). But BART beat its not-so-little drum and got their way at a cost of billions ($ 10**9’s) and many years.

        So, please, be carefull what you ask for in the way of mass transit. There may be a hellacious poison pill buried in the fine print.

      • Jarrett July 22, 2016 at 7:57 pm #

        Andrew. Outside of very transit-savvy places like the inner Bay Area (mostly San Francisco and Oakland-Berkeley) I’m always noticing how few elected officials seem to care about their local bus systems. It’s a minority on most city councils I deal with, and that reflects the lack of engagement with issues among most citizens. My perception is that this problem has gotten much worse with the advent of tech industry PR about “disrupting transit”.

      • Amy Butterfield May 29, 2017 at 8:46 am #

        Andrew C, oh my. Not to divert the very interesting conversation going on here but… Here’s an example of dying transit, and/or anti-transit: I live in Milwaukee WI, and there is ENORMOUS hostility to any expansion of public transit.
        The most recent screams of rage related to the city’s plan to lay an electric streetcar route downtown, with the route ultimately expanding area by area to link different parts of the city. The project squeaked by, and rails are just being laid now — but the project is still getting scathing comments from all corners as a huge boondoggle we’ll regret later. Detractors point out that the money would be better spent on more police officers or what-have-you (go ahead, name any social/urban problem that requires $ and pit it against $ for transit).
        Also some years back WI Governor Walker rejected already-allotted federal funds for expanding rapid train transit, to hoots of encouragement and cheers from the ‘burbs and rural areas about how all the urban babies in MKE must be weeping and soiling their trendy diapers in disappointment.
        The overwhelming vibe in any transit-related conversation here in WI, is that urban Milwaukeeans who wish for non-auto transit options are self-indulgent whiny “liberal babies”, and that the good sensible people who live in surburban and rural areas don’t want to bleed their hard-earned tax money to fulfill the Urban Infants’ ridiculous whims about riding a Choo Choo Train to work.
        Is this just because MKE is the only large urban area in our state? Or does it reflect the vibe in all smaller cities across the US? IS IT stupid from an urban planning viewpoint for smaller cities to shoot for the kind of transit options larger cities have? All us small fry are inevitably growing from small to medium, arent’ we? And then on to who knows what in the future… is it dumb to plan ahead because we’re can’t be sure how big we’ll become? Or does the lack of planning ensure we won’t grow?!

    • Bfdonnelly October 31, 2016 at 3:28 pm #

      Musk doesn’t understand the most elemental thing: first vehicles take you to your destinations, and then *they move the destinations.*

      Interurbans moved the destinations out. Cars and freeways still further. Obesity, loneliness, and depression ensued. The problem isn’t engineering, and it’s only slightly to do with geometry. It’s to do with valorizing consumerism and individuality at the expense of community life.

      Personally, I’d like to see auto-jitneys like WePods and the Olli close the last mile (or two) to main-line transit. Make the transit more efficient, and keep the jitneys relatively local, with local flavor.

    • Hansen Lu December 18, 2017 at 4:20 am #

      “Musk may have exaggerated in this passage. Surely the clumsy middle sentence in the paragraph I quote suggests he doesn’t subject his work to editors. But it’s still fair game, because lots of people are exaggerating in the same direction, and doing the damage I describe.”

      Er, no it’s not fair game at all. There is razor thin evidence to suggest that Musk wants to do what you suggest in this article is his intent/opinion. So you have criticised someone based on a pet peeve that you hold about others, not based on Musks actual established intent / opinion. You should change the title to “Lots of other people don’t understand geometry. I don’t know whether Elon Musk does, but I’m going to criticise his views on that basis anyway.”

      This article IS stupid.

  18. Don Miller July 22, 2016 at 7:54 am #

    Is there any reason why the 20 tiny 2 passenger pods could not link up (spaced only a few cm apart, but not physically connected) and form a pod train and carry people in and out of the denser areas? They would only take up slightly more space than a current bus. Pods could join and leave the tight pod train as they fan in/out to reach the desired destination. I think you’ve missed the other half of autonomous transport which is vehicle to vehicle cooperation and communications at a granular level to meet the needs of the passengers and the goals urban planners.

    • Greg March 7, 2017 at 11:25 am #

      Pod cars that link into trains are not new. They were the original proposal which turned into the MPRT system at West Virginia University. Look up Starrcar. From the 1960s.

  19. Frederik July 22, 2016 at 8:24 am #

    “That’s because it takes lots of people out of big transit vehicles and puts them into small ones”

    Doesn’t he do the opposite thing? Getting people out from their own cars into larger busses?

  20. Luke Mellor July 22, 2016 at 9:02 am #

    Not sure why there is so much hostility from transit wonks about these on-demand transportation ideas. I see basically two options for long term transit planning for large metro areas. Either find a way to solve commuters’ first and last mile problem, or let car drivers continue to flood downtown areas and thoroughfares every morning and evening.

    People in the suburbs are not going to take fixed route buses en mass, let’s be honest, the geography won’t allow it. They are going to need some form of smaller four wheeled vehicle to make it out of their neighborhoods.

    The opportunity here is to combined and coordinate the two means of transport into one system. Use the on-demand vehicles (either autonomous or human) to pick up people from their McMansions and bring them to the dense transit corridors so they can get to work/shopping/whatever. You can use technology to make these transfers run smoothly and as long as both components (fixed and on-demand) are properly resourced and the operators are working together not competing, then they will compliment each other. Then everyone gets to where they need to go and we use less cars.

  21. grimace July 22, 2016 at 1:49 pm #

    Most of your argument seems to base around the assumption that getting taken to your destination implies never changing vehicles, but that’s just not true. It’s obviously impractical for an inner city setting.

    Instead, imagine Manhattan traffic, but in such a way that you could just hop on every 10th vehicle you see. The flow of traffic would easily take you to your destination. The aptly named Manhattan distance also tells you that in many cases you will need to only switch vehicles once. Granted, not every city is laid out like this, but certainly many American cities are, to a large degree. In such a setting, there is also no difference in distance traveled compared to a bus, it’s just that the exact stops are variable.

    • Eric Orozco July 24, 2016 at 9:23 am #

      Grimace, now imagine Manhattan traffic and asking it to perform without the subway system. All that efficiency in moving large groups of people who embark and disembark vehicles en mass in a minute, allowing a massive scale of space savings, suddenly is removed. All that trip demand now transferred to city streets and curbsides. All those swarms of vehicles suddenly attaching the journeys that once were performed on foot to the transit system and now conflicted in travel against one another at the intersections.

      • Brian July 24, 2016 at 10:50 am #

        Now imagine that the tunnels and tracks are converted to SDC roadways where people can go to a subway station, call a car and take it to a specific stop non-stop and at much higher speeds and lower costs along with reduced wait times. Imagine these subway cars act as both the subway car and last mile with exits at less congested areas that allow this.

        One car pops out for last mile, another pops in to become a subway car.

        • Eric Orozco July 25, 2016 at 1:12 pm #

          Urban space is just no issue for you? So now, we turn Broadway into some version of a limited highway, or cap it for this distended subway made of little cars. Nothing tells me more about the serious disconnect of technologists Jarrett is talking than this scary comment.

        • Cynthia mae Curran July 26, 2016 at 3:14 pm #

          Well, Musk wants to moved the population out of the city to country area where people can travel at fast speeds and worked in New York and live out in the country. New York might be nicer if it less million again.

  22. Brad Templeton July 23, 2016 at 5:01 pm #

    You may feel he’s ignoring geometry but you’re ignoring what engineering can do with the existing geometries. It will not surprise you to see me agreeing with Elon, since I’ve been saying the same thing for years, and recently posted some very similar visions in my articles at and which you may find of interest.

    It’s extremely unlikely, in my view, that the typical 40 passenger bus class is the optimum group vehicle size once you don’t need to pay drivers. (With drivers as the biggest cost, all transit systems are pushed to using larger than optimal vehicle sizes.) Smaller vehicles can provide more frequent service and do vastly less road damage, and offer superior load factors and efficiency in places where load factors are low (which is almost everywhere in the USA.)

    Smaller vehicles also have a huge benefit for the geometry question, because they take up less road space when they travel. Today’s US city streets are still mostly occupied by cars, and will be for the future we can see, but big bus-sized vehicles and street cars become obstructions on the road, worsening traffic with their huge size and slower accelerations.

    The robotic transport world of the future is door-to-door, even when group vehicles are used. In that case, what you get are small, ultra-efficient single person vehicles that take people from their doors to a common spot where they board and fully occupy a van selected for their group, which then travels to near where they are all going and they take single-person pods the last mile. The transfers are instant and the services are all competing private companies, not bloated and bureaucratic transit agencies, and they do it all without subsidy.

    And all of it is going to use a great deal less energy and road space than today’s systems. Transit lovers are going to love it — as long as they don’t imagine that the only way to do group transportation is the way we did it in the 20th century.

  23. Brian July 24, 2016 at 11:35 am #

    This article reads just like all the Uber driver posts I see claiming self driving is decades away or impossible. They have a vested interest in the status quo so they look for roadblocks instead of answers.

    There are possible answers not being addressed in this article:

    Convert the subway system to SDC expressways allowing point to point non-stop transportation at much higher speeds. Place the entrances and exits in less congested areas. These tunnels could also be used for SDC parking off the roadway in low demand periods.

    Ban human driven cars, or restrict them to limited roadways. SDCs could use fewer roads far more efficiently if you make them exclusive because traffic lights can be removed allowing non-stop free flowing cross traffic that only stops for pedestrians. Even that may not even be necessary as they can just adjust speed based on pedestrian behavior communicated from cars ahead.

    Use VERY small single person electric cars. Cars can easily be thin enough to be two or even three wide in a single lane given that the distance between cars could be much smaller and lane markers would be meaningless, opposite direction traffic could pass a few inches apart, so SDCs could absorb most of that space.

    The efficiency of scenarios like this are amazing. Speed could greatly increase meaning less cars on a road at a given moment. Roadway lanes that change direction automatically based on need with all cars responding immediately. Better yet, no lanes at all. Center “lanes” could become even higher speeds with outer being for those preparing to stop or turn and, unlike on today’s roadways, this would be respected 100% of the time.

    In a disaster, this could be huge for moving people away while being able to send the car back for more people without endangering a driver. Remember katrina? Millions of cars could have responded to an evacuation before it hit. Instead of spending hundreds of millions on rescue and hospitalization, we spend it on extremely efficient evacuation with lane direction and routing changing on the fly.

    In addition, with no need for human cars, and a great reduction in car ownership, SDC designated roadways won’t need parking lanes so they can be converted to driving and/or pick up/drop off lanes.

    You also won’t need turn lanes as turns are coordinated V2V well ahead of time and no stopping or waiting would be required.

    TLDR: This article assumes integration of SDCs into an existing and terribly inefficient system rather than replacing it utilizing the efficiencies inherent in SDCs, V2V communication, and by repurposing current infrastructure.

    • LuisG July 25, 2016 at 5:26 pm #

      And where are the pedestrians in this “all roads are like freeways” scenario?

      What about traffic congestion from the constant drop-offs and pick-ups?

  24. Brian July 24, 2016 at 11:46 am #

    Also, more infrastructure funds will be available because getting rid of buses will greatly reduce road damage. Delivery trucks could also become much lighter single delivery reducing damage even more. The value of SDCs won’t be in passenger density, but in SDC’s small size and efficiency.

    • Jarrett July 24, 2016 at 7:36 pm #


      You are still spending a fortune to keep walls between every pair of travelers, and to prevent any traveller from having to make an unwanted stop, or to walk, or to meet anyone they don’t want to meet.

      What % of people do you think really need those things? .Are you sure this isn’t the old problem of elites wrongly assuming that everyone wants what they want?

      • Alan July 25, 2016 at 7:54 am #

        I don’t think it reflects an assumption of elites. My understanding is that as a *very* broad generality, in most metros the average commute time for those using transit is longer than those traveling by private car. That travel-time difference would be even larger outside the peaks. It’s hardly an “elite” assumption to expect that commuters would want a shorter commute – to say nothing of the possibility of accessing parts of the metro that are not served or are ill-served by transit.

        As for the geographic issue, autonomous cars are hoped to be more efficient at using a given area of space than a human-controlled car. If that were to happen, you *would* have an instance where technology changes geometry, since the throughput of a physical space (ie. a lane) can be altered by improving the capabilities of the vehicles that use it.

  25. cjl July 24, 2016 at 9:27 pm #

    It seems to me that the problem that Musk is trying to solve is not how to move people through cities but rather how to sell lots of driverless electric cars and batteries.

    • Paul July 26, 2016 at 12:28 pm #


  26. Akahadaka July 25, 2016 at 8:00 am #

    I haven’t managed to read all the comments to see if someone had addressed this already, but my understanding of part of the grand plan is to get most private vehicles off the roads. Also, a side-effect of autonamous driving is also increased efficiency.

    Yes, transferring hundreds from buses to individual vehicles under our current traffic conditions would be a nightmare, but Elon HAS brought this back to an engineering solution whereby the traffic conditions with shared autonamous electric vehicles will be completely different. Main streets, for example, could be bumper-to-bumper, but still moving swiftly along as if you’re the only car on the road.

    Not to mention the advanced routing strategies that will be at their disposal. Having a constant information feed on traffic conditions on any route from other vehicles can allow an individual ride to reroute pre-emptively and intelligently.

    Additionally, future city planners can start getting really creative with roads, because they won’t need the old constraints human drivers ensured.

    Now we still factor in the other benefits of no transfers and no central congestive hub.

    I’m really failing to see the geometry argument here. Then again, it’s not my field, so perhaps I’m the one suffering from short-sightedness. Nevertheless, I don’t think Elon plans to solve this alone. Judging by his companies’ accomplishments so far, I think he’s grasped the concept of working with a team of experts in each field.

    • Jarrett July 25, 2016 at 8:07 am #


      These benefits arise mostly from near 100% uptake of autonomous vehicles, which is very far in the future. For the next few decades fleets will be mixed and driverless cars will need the usual stopping distance when dealing with human driven cars.

      In any case, space benefits of driverless cars are small compared to massive increase in spatial demands that arise from moving people from big vehicles to smaller ones.

      • Akahadaka July 25, 2016 at 8:22 am #

        But what % of commuters are moving through central hubs unnecessarily because that is the only option for transfers?

        And yes, you’re right, fully autonomous roads will not happen overnight, but neither will replacing current public transit systems.

        • Jason July 26, 2016 at 3:57 pm #

          Not as much as you think?

          Lots of people work in the downtown core which by definition is dense. There I don’t think you can create that much more capacity by spreading cars throughout side streets – it might even worsen traffic because of a lot more stopping at intersections. A lot of traffic also goes through highways. Often there aren’t that many different reasonable routes into downtown if you travel on highway.

          As for your previous points, if you have every car tailgating behind one another then you maybe increase throughput by 2-3x. A 40-feet bus is about 3 cars in length and fits 50 people. 3 autonomous sedans won’t be carrying anything close to 50 people any time soon.

          • Anie July 15, 2017 at 6:38 am #

            I would be curious to know how many routes and times a day/week/month each bus is *actually* carrying 50 people? As commented above, the size of the bus is demanded by the need for a driver (which is interesting in and of itself) so by reducing the size of the vehicle used to transport x number of people, there is added efficiency on many levels.
            As a bus rider under many conditions, I could see the benefits of having 3 autonomous sedans or mini buses do the work of one 40 foot bus. Route lengths could be more precise to address needs and could be done more efficiently, ridership would benefit: all the folks to the financial district get on jitney A, all the high schoolers get on jitney B and the folks going a mile to the grocery store/pharmacy/pet store get on jitney C. They each have different needs of their transportation and yet will have different habits on boarding and unboarding times, precise location of pick up drop off and timing of their actual need of a ride.

      • Sam Crutsinger July 26, 2016 at 12:52 am #

        Next few decades? I think you underestimate how fast this is going to take over. This is going to be faster than Internet adoption because you don’t have to buy the car to participate. You only pay for the miles you ride. Once the tech is switched on, states will be falling over themselves to pass legislation as soon as possible. Taxi monopolies will be thrown under the proverbial bus. Young people won’t even get a license any more. If you say that you drove your own car, people are going to look at you funny.

        The only thing that will slow down the process will be how fast they can crank out the cars.

        • Jason July 26, 2016 at 4:07 pm #

          Someone somewhere still has to buy the car and you are still paying for it in some way. Yes for regular trips you can ride share so you need fewer cards and you only pay for your rides. However, you also have to consider all the OTHER trips that require full time use of a car. It will take time for the trucking industry to replace the entire fleet, for every trades worker to buy a new van, for every pizza delivery guy to buy a new car. What about specialized vehicles like police cars, fire trucks, and ambulances? There isn’t enough capital lying around to replace all that “faster than Internet adoption” and unless we have near 100% autonomous vehicles no way can you produce all of fantastical those benefits.

      • benschon August 1, 2016 at 2:09 pm #

        The presumption is that autonomous vehicles will cannibalize transit. How does it look different if they are predominantly cannibalizing private vehicle trips?

        The most radical change I see with robo-cars is a complete collapse of the market for parking, at both ends of the trip. Vehicles either keep circulating to carry other passengers, or during off-peak periods park themselves at the urban fringe.

  27. Sam Crutsinger July 26, 2016 at 12:40 am #

    This article assumes that people will cling to their existing one person per car lifestyle in this future scenario. If carpooling is comfortable, door-to-door, and affordable, then fewer cars will be on the road, because they’ll be using the service, with 3-5 other people who also aren’t sitting in a car alone, taking up a wasteful amount of space.

    If you can get a personal chauffeur for less than the price it costs to drive yourself and pay for parking, you go for it. The more people use the service, the FEWER cars are on the road. The goal is to make shared transportation enticing enough to get people to let go of their gas cars and kick back in a zero emissions Tesla. The geometry works out. You’re just stuck on existing public transit patrons and not visualizing the value of expanding the user base.

    • Jarrett July 26, 2016 at 2:24 pm #

      Sam Crutsinger. Again, I am responding to “take people all the way to their destination.” Door-to-door service is always incredibly unproductive.

      Look at the numbers, Sam. A big urban frequent bus routinely does at least 30 boardings per hour that the bus is operating. The best are over 60. There’s no way you can do that while taking everyone to their door. Demand-responsive service (picking up and delivering everyone to their door) maxes out around 6 boardings/hour, because there are only so many doors you can go to in an hour.

      • Alan July 27, 2016 at 5:28 am #


        I think your analysis ignores something you highlighted in your original comments – that the single large vehicle will be replaced with many more *smaller* vehicles. You can’t do 30 boardings per hour taking everyone door-to-door *in one vehicle*. But if you replace a single bus with 10 small autonomous vehicles, each AV may only hit 4-6 boardings per hour – but collectively, they can equal the boardings per hour of the big frequent urban bus.

        • Jason July 29, 2016 at 8:42 pm #

          Yes, but that is also incredibly inefficient both space wise and money wise. In fact it would cause a huge traffic jam if lots of people happen to want to go to the same place at the same time and we’d run out of places to drop people off. No sane transit agency would buy 500 autonomous cars if the same market can be served by 50 big buses running fixed routes. I don’t believe door-to-door works in dense urban areas. It will probably work in many low density developments and may actually integrate into the transit system pretty well though.

          • Alan August 1, 2016 at 3:44 pm #

            “No sane transit agency would buy 500 autonomous cars if the same market can be served by 50 big buses running fixed routes.”

            Why not? That would depend on a lot of factors, including costs and the relative efficiency of the two types of service – *and* whether autonomous cars are able to improve the efficiency of traffic more generally. There are very few U.S. cities where most commuters don’t *already* go to work in private cars, and that modal share is even lower outside of the CBD and peak use times. Even in “dense urban areas,” most of us are already using door-to-door.

            If giving low-income residents vouchers for autonomous Lyft/Uber rides ever ends up being cheaper than subsidizing their travel through maintaining a transit system, the latter is going to be under a *lot* of pressure.

  28. Cynthia mae Curran July 26, 2016 at 3:13 pm #

    The problem with everyone is going to ride a bike, well the population is aging not getting younger why go to bike riding so much since the 65 and over population is increasing, This is what Elon Musk is interested in and most cities are below 200,000 per people. I think Elon Musk has a great idea make Los Angeles less populated by 500,000 and people can make it green again and developed places like Needles Ca and so forth.

  29. Nathan July 26, 2016 at 5:39 pm #

    Why does this article ignore that more convenient methods of public transportation would also reduce the number of private vehicles on the road? Seems like an obvious flaw in your logic…

  30. Scorp July 26, 2016 at 10:38 pm #

    I get your door to door arguement, but I don’t think Elon really believes that the solution is entirely door to door. As far as buses go, in most large cities, simply removing the driver, running off electricity, shrinking capacity to around 20-30 and having a self assigned stop selected by passengers so the bus doesn’t have to make every single stop, would close the business case for many passengers. It’s basically commuter rail with less infrastructure capital and higher frequency.

    On the very seperate topic of self driving taxis:
    (Don’t be mistaken, I’m a laymen who only toys with these ideas during my commute and even I could tell it was a seperate strategy)

    I’ve been waiting for someone to champion the electric self driving taxi for years.The market is already heading in this direction. Here in Denver we have a small pool of 2 seater smart cars ran by a company named Car2Go who’s main advantage is parking availability and who’s main competitor is uber. A small self driving car pool would literally combine the advantages of both companies.

    In the end, passenger density is really going to be a variable controlled by profit margin, target audience, and market conditions. These technologies are not moonshots or engineering for the sake of engineering. They are inevitable, and they will all allow increased passenger density. The question is how much density the market, and the people, will want.

  31. Dan July 27, 2016 at 11:37 am #

    I’m not sure that I agree with Elon Musk’s plan but the author of this piece, Mr. Walker, leaves out a very important variable in his calculations. He mentions geometry, which in a city is not really a 3 dimensional but a 2 dimensional problem since we don’t have flying cars yet. But really this is a 3 dimensional (2 physical and one temporal) problem. It is true that increasing VMK increases the usage of space but he’s completely left out time from the function. I’m not talking wait time for the bus, I’m talking how long does the bus occupy the space. If Musk’s plan were to increase VMK but drastically reduce the time it takes to go from point A -> point B then the net effect may be a reduction of emissions, wait time, or other variables that lead to a positive effect. Take the function to it’s extreme. Assume that Musk’s vehicle can travel at near the speed of light. Assuming objects don’t crash into each other going that fast I could theoretically make VMK infinitely large because the objects traveling the road occupy that space for an incredibly small amount of time.

    Musk being right or wrong is one thing but I’d suggest not going on the attack about the ‘maths’ and then sort of bungling the math.

  32. Bill Snebold July 27, 2016 at 3:21 pm #

    Sounds like the author has a bias towards liking fixed route transit. Musk’s plans sound like a godsend to me, especially in LA where there has been quite a bit of activity in the area of fixed route transit. But there’s a whole lot more cars. Way too many. There’s a reason for this. The fixed route options are too slow and inconvenient for most people. The problem with a car is that once you get to your destination you have to park it. That costs money and, if you’re parking at work it generally means that your car sits taking up valuable space (geometry) for about 8 hours.

  33. Hermes July 27, 2016 at 8:50 pm #

    I get the geometry argument, but simply relying on VMT is an oversimplification that ends up being counterproductive. Yes, the dynamics of urban transportation are different from rural and suburban, but the assertion that technology can’t address a geometric challenge is absurd, as is linking Musk’s approach to any political notions of elitism or social stratification. In fact I would say that Musk’s idea could, under certain circumstances, offer an even greater degree of transportation efficiency in urban environments than transit corridors could ever hope to. An objective engineering approach would necessarily measure the effectiveness of any method of transportation by the overall efficiency of the resources used to effect that transportation. Say the goal, for instance, is for 50 people to get from one side of town to the other. On the one hand, you have a transport corridor – say, dedicated bus rapid transit lanes, and on the other, you have Musk’s approach. When you look at the roadway itself as a limited resource, you see how Musk’s approach is both more efficient and more easily scalable, even though the BRT/corridor approach yields a greater VMT. When you factor in the “roadway consumption cost”, the maths will clearly show this in relation to the geometric challenges inherent to transportation. A bus is a large moving structure that takes up “X” amount of square footage of a roadway at any given moment along its route. If the 50 passengers onboard were all headed toward the exact same destination, then it would be much more efficient by far – but this is not the case, and that is what the corridor approach fails to consider. Musk leverages a technological approach based on the notion that the same total area of roadway consumption being taken up by a bus would be much better served by being divided up into smaller chunks, where those 50 riders are still, in their aggregate, consuming the SAME amount of roadway as the single bus. The idea, therefore, is to more efficiently allocate the limited resource – roadway use – by spreading out the single area taken by a large vehicle into smaller chunks taken up by many smaller vehicles which add up to the same roadway cost for transporting the same 50 people that were riding the bus. Because of this, Musk’s idea is lends itself much more easily to scaling in urban environments, but also provides an inherent load-balancing to the urban transportation grid as a whole: by pairing small groups of people in smaller vehicles that will take them directly to final destinations in close proximity to each other, the roadway demand is spread from a single corridor roadway to a much larger area, easing congestion in major thoroughfares through dilution. Simultaneously, the ride-sharing principle in effect maximizes the collective utility of this method by ensuring the largest amount of people, those heading to points in “neighborhood X”, get there while utilising 1. the smallest necessary amount in of roadway they need to get there, and 2. do so without adding any undue traffic congestion load to other riders bound for points in “neigborhood Y”, and 3. will consume that roadway less, or more, based on demand at any given time. By comparison, the collective utility of a Bus will fluctuate: highly efficient when full, highly inefficient when it consumes the same roadway area while carrying only one or two riders. Add to this the fact that buses run on schedules, not on demand, by which their roadway consumption will remain constant regardless of ridership at different times of day, and the collective payoff to the urban grid is even more diminished despite unchanged, high VMT numbers. A technological approach, then, is able to maximize the collective utility of the urban roadway, precisely by leveraging geometric principles but applying them to a smaller ridership-per-vehicle scale. By comparison, the VMT measurement approach seems myopic, a tunnel-vision methodology that judges efficiency merely by the vehicle’s capacity without regard to the area of roadway it consumes.

    When the added effects of roadway consumption inherent to corridor-style initiatives are factored in, the difference is even more stark. Dedicated bus rapid transit lanes constitute roadway land area that sits unutilised in between buses, a clear waste of resources. In traffic congestion, the problem is entropy – a waste in and of itself, but you’re not going to solve the problem by trading one type of waste for another! Instead, the effect on the grid is merely
    psychological, whereby there is now an incentive to embrace public transportation as favourable to dealing with increased levels of stress as a driver when the available road surface area is reduced for the sake of dedicated bus lanes, compressing all other traffic into less space. Lastly, from a public works perspective, the reduction of heavy, large vehicles will incurr much less wear and tear on roadway surfaces, leading to lower maintenance costs, which is another collective benefit that Musk’s approach offers. Far from any notion of making public policy based on individual preferences, Musk blends the best aspects of both private and public transportation approaches and leverages the geometry of the urban roadway environment to achieve a much more efficient, sustainable, transportation method.

    • Jarrett August 14, 2016 at 8:00 am #

      Hermes. This article is about dense cities, where it’s very easy to get 50 people on a bus going in the same direction. They don’t need to be going to the same destination, but just in the same direction, to places within walking distance of the route or to connection points to other services.

      I don’t follow your claims about buses and VMT. Bus lines designed for high ridership are always the lower-VMT solution compared to small vehicles. The confusion arises because there are also bus services designed for low ridership

    • Anie July 15, 2017 at 6:43 am #


  34. echoman2 July 28, 2016 at 6:35 am #

    We are missing some important data which is what are the shortest paths between people’s homes in people’s work. For example, I live in a small town just inside of 128 by route 93. In the past year, I’ve worked in south Boston, Allston, and Watertown. Each one of these places takes between 40 and 50 minutes by car. If I go by public transit, it’s closer to an hour and a half to 2 hours with at least 3 transit changes plus mile long walks at each end. Sucks when it’s hot, cold, raining, snowing which describes three quarters of the year New England. It double sucks if you have to take a pee.

    It would be nice if we had a crowdsourcing project gathering work and home end points and then we could see just how badly or how well transit matches shortest path.

    Note I’m sure someone will say why don’t I move closer to work. That’s a frequent response to a lot of “transit takes too long” objections. And I don’t move for 2 important reasons. Cost and leases. These 2 points are also where Elon makes the biggest difference in transit.

    Fixed route/access transit systems radically distort real estate prices. If you are trapped into living within a certain area because that’s where transportation is, landlords have free reign at raising rents especially if we have a high demand situation like we do today. For examples of this rent seeking behavior, I point you to articles about the Green line extension in Somerville Massachusetts. Apartment rents climbed anywhere from $500 to $1000 a month just on the suggestion that the greenline was going to have stops in certain neighborhoods.

    Now this is where autonomous vehicles can make a huge difference. By freeing you to live “anywhere” landlords have less power to raise rents. Yes, they will still raise rents around transit stops because that’s what they do but now you can make a better choice you’re not forced to live near a transit stop and pay those rents.

    Addressing the other half of my problems with living close to work is leases. If I find an apartment I like and I signed a lease, I’m trapped in that place for year. I cannot move. I’m not sure I want to move if I like where I’m living. At the same time, I can’t control where I have a job and I will take a job 50 miles away if it’s a good job.

    This is where the value of variable sized autonomous sized autonomous vehicles kick in. If I travel on normal times, there is a high probably other people the travel on a similar path and the mini-bus can have a more or less standard pattern taking a few people from point A to point B. If, as is usual for my life, I’m working late. I don’t have to choose between sleeping overnight in the office, walking 15 miles in the dark, or spending $$$ bucks to get home in a taxi. An small autonomous vehicle can get me home at a much lower cost when I want to go home.

    Elon may not understand this writers version of urban geometry but he does understand the way people really live. Like the automobile, his transportation vision enables people to choose where they live by their values of population density, perceived safety, urban versus rural and doesn’t force them into a one-size-fits-all shrunken world of modern urban transit.

    • Jarrett August 14, 2016 at 7:57 am #

      Echoman … If you live in outer suburbs, including anywhere outside the Boston area’s 128 beltline, then the core point of this article doesn’t apply to your personal experience. The problem I’m describing is mostly in dense cities.

      • Anie July 15, 2017 at 6:45 am #

        Jarrett~on your maths then, specifically as they pertain to this article, where/which are these dense cities?

  35. Eclipse May 22, 2017 at 12:33 am #

    Now Musk has his ‘boring’ company he’s going to charge cars to get onto a lift that takes them down to the tunnel, where the ‘lift’ turns out to be a large steel ‘sled’ that shoots the car along at 200kmh, maybe even recharging the car while it rests on the sled. In other words, he wants a vast tunnel network underground to solve these problems of geometry, but at what expense?

  36. Mark December 17, 2017 at 3:34 am #

    Ever tried first principle thinking? That means you reduce a problem (like urban transport in dense spaces) to its core variables (like geometry, passenger demand, time and space, frequency and modes of transports possible,..) and from there you reason the most viable options.
    You can not do this if your thinking ia based on status quo. Thats what seems to make it very hard to grasp. Do the math and go 10 years plus into the future. When full autonomy is available and you can send your care (small passenger, van or bus) away to earn money for you, while ideally you let it run on solarenergy from your roof or a supercharger like system where the car can refill itself, without pollution (because completely solar). Where is your geometry problem gone?
    Also the Boring company tries to solve just that problem by taking autonomous traffic 3d. He reduced the diameter so that to optimum sized people-to-space ratio is achieved with AI-backed route calculations, constantly getting input from the traffic situation and routing the best available route. NOT necessarily door to door. But maybe at the BEST place to get a connection. Like you do with Google maps today, which suggests you the best route from door to door. Taking into account all available modes of transport. Hence Musks comment: you didn’t do the math my friend. I don’t think you are an sanctimonious idiot but it sure doesn’t to write an article based on comments while this is about simple math. But you are an sanctimonious idiot for telling him he didn’t take those things you are writing about into account. He’s just decades in the future with his thinking, and thats just what people getting money to do constantly fail at. Just look at the current public system and reason back to how it could made less sucking. If you are able to do first principle thinking it becomes painful obvious. His comments may be hard to grasp for normal-thinkers and yes sometimes a bit clumsy. But he sure does his math before he elaborates on those things. And he follows through by pioneering the things needed for this vision. So get into AI, robotics and physics. Then you’ll understand this. Oh and you might have to forget most of the things you were taught. Reasoning from the past doesn’t work on that scale. Because this are things that have never existed before in human history. You might be better ar your job for the next 20 years to come. It indeed is called future with a reason. As a transit planner we could need more unbiased people, that are able to project into the future instead on some elitism bias. That only harms. And hurts. Like Musk pointed out. Thats a weapon one should never use: putting someone else into a box without asking about their *individual* daily life circumstances. You don’t know, hence it’s your own projection. I’m sorry to point that out, because we all do to a certain extent . And I think its harmful and it’s creating separation. There is nothing unifying in that. Plus: for a poor kid in Africa you sure ARE part of this very elite. Think about. It’s not only first principle thinking it’s also about human empathy. Especially when you’re getting paid for planning on behalf of others. Maybe the elites and those who can not even afford public transport are also part of the society that you are planning for. Do your job and imply them in your thinking too. Cause maybe you are reasoning from your own personal view of society, missing some variables that matter. Thanks for your contribution. Have a good day and please let’s all improve on something.

    • Mark December 17, 2017 at 3:46 am #

      Sorry I made some mistakes in my comment. Was sitting in a train while writing this. Also I don’t want to offend you or anyone. The phrase normal-thinkers doesn’t help too. I wanted to elaborate on the different background of thinking and thats something we are not being thought usually when we go through regular education. Also don’t blame things on Futurists. Cause that doesn’t exist. There are people who talk and people who do. Judge everyone by their contribution to the matter. Musk sure does with all his ventures. Not saying its the best available solution always but the spirit is right. As opposed those holding on to the objectivly bad things of the past. Thanks. Now it feels better

  37. Dan December 17, 2017 at 9:36 pm #

    All this talk of Elon not understanding geometry – that there’s not enough space for his vision…… Isn’t the dude working on economical production of underground tunnels? And doesn’t that expand the available “geometry”? Sounds to me like he’s already well along the path of solving these “intractable” problems the article is criticising him for ignoring…

  38. Hansen Lu December 18, 2017 at 4:43 am #

    To Mark, the above commenter, that’s some fairly abrasive stuff.

    But I broadly agree on the intent. The key thrust of Jarret’s arguments in various articles on this topic seem to be “people who have much more experience in this than Musk have already proved this doesn’t work.” And honestly, that approach just doesn’t cut it. It has not been proved that Musk’s ideas won’t work. At most, there are some studies showing that personalised PRT won’t work under certain sets of assumptions. But we don’t really know what Musk’s suggested solution will actually look like.

    I appreciate the value of past work in informing our view about the future, but I think you are using the evidence base incorrectly. Yes, studies have shown that the geometry problem for PRT is difficult. But does Elon seriously think that he can get everyone travelling from door to door using a PRT system based on the assumptions in that study? Highly unlikely. You give him far less credit than I think he deserves.

    Instead, I think a more accurate assumption to make is that he is highlighting an ideal, from an engineering perspective. Then, the goal is to design something that achieves as close to that ideal as possible, that works. What that will be, we do not know. But the process for designing the system (minus any social ramifications) seems to be the right one – improving user experience compared to existing bus transport, and creating a system that is highly efficient in transporting people across cities.

  39. Thomas Earle Moore December 19, 2017 at 6:26 am #

    At BWI airport, and I’m sure many others, there is a choice between large buses operated by the airport and small independent buses, both of which ferry passengers between parking areas and the terminal. The small buses move relatively quickly, pick up directly at one’s car, and drop off at any specific terminal door. The larger buses pick up at a few widely separated “bus stops”, and stop at designated terminal spots only. There is no contest for which I prefer to use, especially when I’m toting luggage for a longer trip. The airport seeks to hobble the inherent advantage of the smaller independent operators by restricting them to one or two pickup points for arrivals seeking transport to their cars. I don’t feel elitist in my preference and would challenge you to show me how it is elitist.

    I seriously doubt that a smaller number of larger vehicles enjoys an efficiency advantage over a larger number of smaller vehicles, especially considering that the larger vehicles are often forced to circulate with far less than a full load of passengers. They may be more efficient when fully loaded, but inherently have less flexibility than the smaller vehicles do to adapt to loading conditions, unless they wait until they are full and therefore become widely spaced in time.

    Perhaps you’ve done some math on this and I will look for it, but my sense is that the question of optimal granularity in transit is completely separate from the question of energy efficiency and emissions of noise and pollutants. Solutions to the former should not be allowed to obstruct solutions to the latter, in my view. Moreover, solutions seeking the most humanly agreeable transit conditions should not be dismissed as “elitist.”

  40. Kenneth Palmestål January 4, 2018 at 3:23 pm #

    For some people the current public fixed route transportation modes with large buses, trams and subways works, fine, let them continue to use them. (But in many (all?) cities it costs the tax payers a lot)

    But the main traffic and pollution problem is that most people choose to use their private fossil fuel cars, right?
    If these people were offered a door-to-door transportation mode that were more efficient, economic and convenient than their private car, most of them would probably use that alternative for their daily commuting.

    Autonomous electric minibuses with separate cabins for each passenger (I call it CabiBUS on my website) can offer the privacy and safety required to attract all types of passengers.
    With six passenger in each minibus it reduces the number of cars on the road with 80% (assuming an average of 1,2 passengers in a private car)

    If cities arranges this as a public transportation service they could charge more than for the current public transportations, it would still be less costly than taking a private car. And with the low costs for running self-driving electric vehicles it could even be profitable for the cities, no cost for the tax payers.

    But the politicians must start now and request this type of solution. It will substantially reduce the sales of cars so the vehicle industry will probably not be so willing to move in this direction without a clear demand.