Maybe Apps Are Not Transforming the Urban Transport Business

Revised February 19, 2018, based on excellent comments.

We’ve all heard that the most important transportation innovation of the century is the smartphone.  Who can doubt that apps for ride-hailing, navigation, and payment are making it easier to use shared transportation services, whether buses or Uber/Lyft or anything in between?   How can anyone who remembers waving helplessly at rushing taxis, or wondering when the bus would come, possibly doubt that this transformation has fundamentally changed all the products it touches?

From a customer’s point of view, I don’t doubt any of these things.  Apps have transformed the customer experience totally.  But that says nothing about whether they’ve transformed the bottom line of the provider.

Len Sherman has a nice short piece in Forbes explaining why Uber can’t make money.  Key quote:

The taxi industry that Uber is seeking to disrupt was never profitable when allowed to expand in unregulated markets, reflecting the industry’s low barriers to entry, high variable costs, low economies of scale and intense price competition — and Uber’s current business model doesn’t fundamentally change these structural industry characteristics.

Standard Uber/Lyft ride-hailing service is made of two main ingredients:

  • Taxi service, minus the protectionist regulations that kept some taxi fares artificially high.
  • An app that expedites hailing a taxi and paying the fare.

The relationship with drivers is also a difference, but not as much as it may seem. Uber and Lyft let drivers use their own cars, but many taxis are driver-owned as well.  Both Uber/Lyft and taxis pay the driver based on fares, not based on hours worked.

So really, the big difference is the app.

The app has transformed customer experience — by taking the friction out of the hailing, routing, and paying — but it doesn’t seem to be transforming the fundamental nature of the task, or its potential to be profitable.

That’s because transportation happens in physical space.  The dominant element of cost is the time it takes to drive someone to their destination, and to travel empty between jobs. The app does nothing to change this.  At most, Uber and Lyft have turned their efficiencies into fares slightly lower than taxis, due to intense competition between them.

If ride-hailing companies had the potential to be profitable — short of creating the same monopoly for them that taxis used to have — someone surely would have done it by now.  But Sherman notes:

Every major ridesharing company in the world is still experiencing steep losses after five or more years of operation, including Lyft (U.S.), Ola (India), 99 (Brazil), and Didi Chuxing (China).

We are seeing the same thing on the microtransit side.  So far, microtransit is doing no better than demand-responsive transit has always done, generally worse than 3 passenger trips per driver hour, compared to 10 for the typical outer suburban fixed and 20-100 for fixed routes in dense and walkable places.  In fact, the most widely promoted recent experiment, the Bridj pilot in Kansas City, did not reach 1 passenger / hour and managed to spend about $1000 per passenger trip,[2] compared to less than $5 for a decent fixed route.

This gap is too vast to be a marketing problem or something that can be solved by tinkering.  It’s a fact about the intrinsic spatial inefficiency of demand-responsive service, which has little to do with the communications tools used.

It’s time to notice a pattern:  Tech boosters treat solutions to a communication problem as though they were solutions to a spatial problem.

Certainly, communicating via telephone calls was part of the inefficiency of taxis, but if the smartphone app were enough to make taxis profitable, we’d be seeing the results by now.  Likewise, it’s great that apps are improving the communications side of demand-responsive transit, but so far, there’s no sign that this is making a difference on the bottom line.

Remember: Urban transportation is a spatial problem, and (until automation) a problem of the efficient use of labor.  If you’re going to transform it, you have to transform those things. Nothing about the standard Uber/Lyft product, or “microtransit,” is touching those fundamentals.

So have apps transformed the customer experience of urban transport?  Yes!  Have they transformed the urban transport business?  Maybe not so much.




[1]  There is a vast range of hybrids between a fixed route and a fully “to your door” demand-responsive service, all of which are very old ideas.  I was designing and revising these 25 years ago.  Everything that’s known about the math of that problem was well understood back then by the people doing it.

[2] This appalling number is from Eno Center’s report “UpRouted: Exploring Microtransit in the United States,” p.7, which is generally upbeat about microtransit prospects.   More commentary on this report soon.

100 Responses to Maybe Apps Are Not Transforming the Urban Transport Business

  1. Stephen Rees February 2, 2018 at 4:17 pm #

    There are some taxi regulations that are related to passenger safety – getting into a car with a stranger is not something a lot of people are comfortable with. Some relate to driver safety, others to the safety of other road users. There are also other considerations such as the ability of a driver to help people with disabilities, or the accessibility of the vehicle. Many of the regulations came into existence to deal with problems experienced with taxi operations. Regulations are not simply
    invented to be vexatious or contrary to profit making even if that is sometimes the result.

    The biggest threat presented by Uber is their desire to create a monopoly and thus the ability to impose surge pricing whenever it suits them. They will achieve that if they are allowed to compete with public transit.

    • Georgist Economist February 3, 2018 at 3:17 pm #

      While I mostly agree, I doubt that abuse of monopoly powers would be a significant long-time concern. If e.g. Uber achieved a monopoly, and started to regularly abuse this, it wouldn’t take long to regulate away that abuse. And because the leaders of the company are smart enough to realize this, they will restrain themselves to only mildly abusing their monopoly, so that they are only a mild irritation, not warranting a large effort.

      • Novacek February 5, 2018 at 6:33 am #

        I also doubt Uber’s ability to sustain a monopoly (if they were flagrantly abusing it).

        Austin is an example in point. When they put in regulations that Uber/Lyft refused to operate under, plenty of startup “rideshare” apps popped up to replace them. Turns out writing such an app really isn’t that hard to do.

        Most (not all) of those have since gone under after U/L returned (presumably because competing with Uber’s investor driven spending is really hard). But if Uber ever tried to double/triple its base prices, I think you’d see a new round of competitors.

  2. Andy Chow February 2, 2018 at 4:32 pm #

    While fundamental hasn’t changed, apps themselves have transformed customer service, which is not the main core business of transportation companies but a necessary component of it. Part of the problem with taxis in the old days is that there’s no single customer service interface. Every cab company has its own phone number and dispatch. Each city has its own phone numbers for the dispatch.

    The hotels make it easier with bellhop acting as a broker between the customer and cab. That’s why in the old days cabbies paid the bellhops when they wait at the hotels.

    I certainly have issues in regards to how TNC rides are delivered, but there’s no doubt that the apps are a powerful marketing tool. When that marketing tool has reached a certain penetration, then there’s a network effects where the system could be operated in a more efficient level.

    • Jarrett Walker February 2, 2018 at 4:51 pm #

      Andy. My point is precisely that I think we’re at the limit of any “marketing” effect. The apps haven’t fundamentally changed in five years, and the business is still unprofitable.

    • Eric February 5, 2018 at 2:30 am #

      Taxis can also use apps, and in some places they do.

    • Mike February 6, 2018 at 11:16 am #

      Only if the app is provider-neutral. If you have to install a separate app for Uber, :Lyft, and each taxi company and “for hire” company and check all of them, then you haven’t solved anything over carrying a list of taxi companies’ phone numbers. And you’re giving the company your account information and whereabouts information even before/if you book a ride. With a list of phone numbers, the companies don’t know who you are or when you’re consulting the list.

      • Tim February 6, 2018 at 12:32 pm #

        At least in my area, it seems pretty universal that TNC drivers have both the Uber and Lyft apps on as needed when looking to pick up a fare. From the user side, there doesn’t seem to be a meaningful difference between the two in terms of service availability. This would probably not work in a more fragmented market.

  3. baselle February 2, 2018 at 4:55 pm #

    “The app has transformed customer experience…”

    Carole Lombard developed an even more efficient ‘app’ of hailing a car over 80 years ago.

  4. Andy Chow February 2, 2018 at 5:07 pm #

    It is unprofitable because it is deliberately underpriced to try to kill off competitors. I have seen people standing in the street corner, starring at their phone waiting for a TNC car to show up, next to an empty cab waiting for customers.

    From a policy prospective, you want that guy to just hop on that cab, because there’s environmental cost for all the inefficiencies due to competition (car deadheading to a ride, and driver not compensated waiting for customers who is right there but for pricing reason chose a competitor).

    When you think about it, that cab waiting at that spot and the TNC car offer essentially the same thing: a private ride on a motor vehicle, so there’s no public reason why one should be chosen over. Therefore they belong to the same system as much as nurses and doctors (regardless who they are employed with) belong to the healthcare system.

    • Mike February 6, 2018 at 11:23 am #

      The cab pays extra fees to park in an approved area waiting for somebody to see it and engage it. So that’s another imbalance, although there are arguments on both sides on what the significance of this is and whether the regulators should do anything about it.

    • Kenny Easwaran February 22, 2018 at 9:16 am #

      I think that might be an issue in New York and San Francisco. But most places where Uber and Lyft operate, there really aren’t visible taxis anywhere other than hotels, airports, and maybe a few taxi stands. The app makes the service viable for passengers and cars to match up than was possible when both were at a density of only one or two every few blocks.

  5. Andy Chow February 2, 2018 at 5:11 pm #

    A reason for the taxi regulation is to ensure the drivers work as a system rather than companies competing with each other. When you remove things like price competition, that guy would’ve hop on that cab rather than waiting for a TNC ride. You can have a more efficient pick up areas like taxi stands rather than people waiting at random street corner and cars stopping at random bus stops or on bike lanes.

    • asdf2 February 3, 2018 at 8:45 am #

      There are plenty of taxi regulations out there that result in more, not less, driving around empty.

      Typically, they arise as a result of different jurisdictions (city/county/etc.) awarding exclusive pick-up licenses to different taxi companies. Drivers are typically allowed to drop passengers off in neighboring cities/counties (otherwise, it would be impossible to take a taxi ride across a county boundary), but they are then forced to drive empty back to their “home” area before they are allowed to pick another passenger up.

      Uber and Lyft are largely free from having to abide by silly rules, like this.

  6. asdf2 February 2, 2018 at 10:01 pm #

    The claims that companies like Uber are unprofitable does not mean every trip is unprofitable. A typical trip (for simplicity, I’m referring to the basic, non-carpooling service), the driver’s pay is simply a percentage of the fare, so the company *has* to make a non-zero profit on every trip (assuming its fare cut exceeds the cost of insurance).

    What makes companies like Uber unprofitable (for now) is a combination of special promotions and below-cost pricing for the carpooling services. Basically, if you’re paying full price, the company is making money from your trip, if you’re using a coupon to get a deal that seems too good to be true (e.g. ride 10+ miles for $5), the company is probably losing money off your trip.

    The promotions are mostly temporary efforts to gain market share after expanding to a new area. It is totally expected that periods in which the company is rapidly expanding are going to involve lots of free and discounted rides to get new customers used to the service. When the promotions go away, the profits will come.

    One other source of subsidy may be the carpooling services. They often offer significant discounts, even though the “carpools” frequently go unmatched (at least from my personal experience). The carpool services also decouple the fares from the driver pay, so Uber/Lyft still pay the drivers full fare for unmatched “carpool”, and take the discount out of their bottom line.

    The article also overlooks a big reason why taxis have such high fares, and that’s because cities traditionally use a “cap and trade” scheme to limit the number of taxis, the same way the EU tries to limit carbon emissions. When the supply is kept down, the price goes up, only the higher prices go not actually to the driver’s pocket, but to the loan payments to cover the cost of the medallion. Simply getting rid of the “cap” brings prices down significantly, in and of itself.

    • Jarrett Walker February 3, 2018 at 10:41 pm #

      Yes, competition in the TNC space pushes prices down below the level of profitability. This is a feature of healthy capitalism. If it weren’t for this competition, prices would be higher and one monopoly player would do better at breaking even, which would make it pretty much like the old taxi monopolies, wouldn’t it?

      So losing money by being cheaper is intrinsic to how TNCs differ from taxis.

      • asdf2 February 6, 2018 at 10:11 pm #

        You’re assuming that because Uber is losing money, so far, in the *aggregate*, that they are subsidizing each and every trip. That is not the case.

        For each trip, Uber takes a published 25% of the fare. Since Uber does not actually own or operate the vehicles, the only marginal cost to Uber from each fare is the insurance. So, unless a $20 ride costs Uber $5, just on insurance (which seems really excessive), the company *is* making money on each trip, or at least trips where the customer is paying full price.

        Yes, there are some trips that do get subsidized. Like, when you are redeeming a special promotion, or when you get a big discount for UberPool, but don’t actually get matched with any other passengers. But those rides are not common, and promotions are easy for Uber to shut off the moment they decide that the PR benefit isn’t worth the money their spending.

        So, the profitability of Uber ultimately does become a availability problem, trying to generate as many trips as possible to generate enough revenue for the company to offset the various *fixed* costs, such as developing the app, recruiting the initial set of the drivers, lobbying the various cities to make the service legal, etc.

        Furthermore, Uber is rapidly expanding into new cities, so whatever profits they do make in the cities in which they are well established is (for now) getting spent setting up operations in other cities. Occasionally, they have to write off big losses spending a bunch of money trying to get set up in a city, only for the local regulators to say “no”. They are also spending money experimenting with new types of services (e.g. UberPool, UberEats, etc.). Some will be successful, some won’t. The losses from the unsuccessful services is the price to be paid for finding out what works and what doesn’t.

        If Uber *really* wanted to start posting profits immediately, I’m sure they could do so, simply by cutting all future expansions and experiments, and just doubling down on the services and cities that are proven money-makers. But, what’s best for a company’s balance sheet *right now* isn’t always what’s best for a company’s long-term profitability. Amazon lost money for many years before it finally started posting profits, once it got big enough. Without access to detailed information on Uber’s balance sheet, breaking down all the profits and losses by city and service type, I have to assume that this is basically the same – their goal is to be profitable in the long term, not the short term.

    • Sailor Boy February 6, 2018 at 2:27 pm #

      “(assuming its fare cut exceeds the cost of insurance)”

      For most trips a driver is losing money (gas, insurance, wear, depreciation)

  7. Grandview Citizen February 3, 2018 at 8:15 am #

    The fares charged by Uber, Lyft etc. will remain low until the legacy taxi companies are thoroughly bankrupted. At that point, fares will rise to the level needed to meet profitability goals.

    Uber, Lyft etc. don’t need to have goals to solve “spatial problems”, unless that is a side-effect of their financial goals.

    In the meantime, it’s still useful to compare the non-financial part of the business. In that regard, your article lacks data.

    Here’s something to try, and report the actual results:

    Pick a random city that has both Uber/Lyft and legacy taxi service. Let’s pick Portland OR, just for giggles. Over the course of a month, take a ride a day, alternating between the two service types. Note the differences.

    In my experience, legacy taxi companies with Apps still suck. Software doesn’t seem to change their entitlement mindset. What’s different about Uber/Lyft (and their twins) is their attitude that customers need to be treated with respect.

    • Iain Perkin February 3, 2018 at 8:54 am #

      I wouldn’t wait for the legacy companies to go bankrupt. If in 2 years Uber runs out of money, then the legacy taxi services won (over Uber but Lyft might still be around).

      The biggest issue is the tokens that are treated as investments. Here is a ideal:
      1) Legislate low cost of license tokens while ‘leasing’ them to those seeking them. Distribute using lotteries and the token is then tied to a vehicle and ‘token holder’.
      2) App time: 2a) Drivers must use Legislated Apps to retain their tokens. 2b) Users must use the App to ‘call the taxi’ but can see driver rating. 2c) Users can rate rides that the App has recorded that the App indicates they have taken (thus avoiding competitors being able to downrate any other drivers). 2d) Drivers grading will be their for all to see in the App.

      Now a driver who has won the lottery and got a token will for the sake of positive ratings maintain good or better service. Failure to do so would be expensive as renewal would require ‘out of pocket’ money if you haven’t got the riders. Also the idea of multiple drivers on a token would have to be dealt with to allow compensation to the vehicle owner. The rating of each driver would encourage careful selection of co-drivers. The devil is going to be in the details.

  8. yq February 3, 2018 at 11:14 am #

    Maybe someday TNCs will sell data as a potentially profitable “side-business”.

  9. Bjorn February 3, 2018 at 2:38 pm #

    For regular people in the rest of the country, the IRS requires that businesses actually make an effort to be profitable in order to not be treated like hobbies, which don’t have access to normal business deductions. As Uber keeps losing money year after year, perhaps it’s time for the IRS to take a closer look at if they are actually a business.

    • Mark Bahner February 3, 2018 at 7:43 pm #

      Amazon went public at $18 per share, and didn’t make a profit for the first 6 years of its operation.

      • Jarrett Walker February 3, 2018 at 10:44 pm #

        The TNC business is not like Amazon’s business. TNCs need one employee per customer, so as the customer base grows the operating cost grows with it. There’s no economy of scale such as Amazon enjoys.

        • Mark Bahner February 4, 2018 at 12:34 pm #

          “TNCs need one employee per customer, so as the customer base grows the operating cost grows with it.”

          I agree in general, which is probably one reason why Travis Kalanick was and is so obsessed with autonomous vehicles. (But I think it was a yuge mistake to try to start in Pittsburgh. He should have started with a flat, warm and sunny location, with a young population, where vehicles could drive very standard routes.)

          Where I don’t agree 100 percent is that it is of course possible to boost the number of customers per employee to at least a few.

          One place where Uber and Lyft should be focusing is on high school football and basketball games. In that situation, many, many people have a common destination or origin point and time. (Plus, the parents would probably prefer their kids don’t drive anyway.)

          Another place Uber and Lyft should be focusing on is convention centers, particularly at universities. Again, there is a common destination or origin point and time.

          And don’t get me started on Research Triangle Park, NC. There are literally thousands of educated and affluent professionals coming from the nearby cities of Raleigh, Durham, Chapel Hill, Cary (etc.) to major employers (hundreds or even thousands of people) in a central location at more or less fixed times every work day. But virtually everyone comes alone in a car. Amazing! (And the inefficiency is very frustrating to an engineer like me.)

  10. Xianhang Zhang February 3, 2018 at 2:47 pm #

    Bridj provided 597 rides during its 6 months of operation which is the main reason it cost $1000 a ride. That very much is a marketing problem. If they had achieved their planned capacity of 200 rides per day, they would have delivered 36,000 rides in the same time frame for the same $600,000 cost which is $16 per ride, not the best but also within striking distance of $5 fixed routes, especially for a first phase pilot project.

    Reading through the whitepaper, the persistent lesson is that government agencies have no innate expertise in marketing which is the core downfall of all of the government run microtransit experiments so far. Because microtransit is so dependent on scale for the economics to work, for-profit ride sharing services are much better equipped to provide it because they can initially aggregate scale via their traditional taxi-style services before adding on shared rides and microtransit.

    I think there’s one huge advantage to microtransit over fixed routes that people have yet to appreciate: guaranteed seating. A significant proportion of my trips not taken on public transit are because I have no idea if I’ll be sitting or standing for the duration of the trip. In the same way that consumers will prefer the fixed, predictable ETA of ride sharing vs the stochastic process of trying to hail a cab, even in cases where the average case for cab hailing is significantly faster, I predict that consumers will also be drawn to knowing ahead of time whether their trip will be a sitting or standing one vs the stochastic process of fixed lines.

    • Georgist Economist February 3, 2018 at 3:30 pm #

      If there are enough passengers that you don’t know whether you will have a seat, then quite simply, there is not enough room for everyone to sit. Especially if they aren’t on large vehicles (buses), but a multitude of small vehicles, because the latter takes more space. Basically, in such a case microtransit is not viable for spatial reasons.

    • Peter L February 3, 2018 at 6:39 pm #

      Yes, and if they’d reached a crush-load of 400 pax/day they would have only lost $8/ride.

      But I think that’s the point: They didn’t.

    • Mark Bahner February 3, 2018 at 8:03 pm #

      “Reading through the whitepaper, the persistent lesson is that government agencies have no innate expertise in marketing which is the core downfall of all of the government run microtransit experiments so far.”

      They don’t have any innate expertise in data handling or GPS technology.

      Combined, Uber and Lyft have operations in hundreds of cities around the world. A city government has expertise only in their city. Plus, companies like Uber and Lyft will likely form natural partnerships with automobile companies. A local government isn’t going to partner with automobile companies.

      A revolution in transportation is coming, and will likely be completed within the next 2-3 decades. Private vehicle ownership will be replaced by transportation-as-a-service provided by fleet(s) of autonomous vehicles. I don’t see how local governments will compete against those fleet owners. Local governments simply don’t have the necessary expertise.

    • Hugh February 4, 2018 at 6:50 am #

      The question is, how much would you be willing to pay for a gauranteed seat?

      • Jo Walton February 6, 2018 at 5:21 am #

        Interesting question of why buses and subways don’t have “business class” with premium pricing and guaranteed seats. People are fine with it on planes and Amtrak.

        • Mike February 6, 2018 at 11:35 am #

          Because they’re publicly-subsidized services that are supposed to focus on the Everyman, and especially those who are poor and have no other choice. Business-class service is a distraction and a subsidy to those who least need it. It’s a perfect niche for private companies to pursue if they want to and can make it profitable. It also exist in some cities in the form of express buses that overlap with a regularly-stopping subway, as in Jersey City and some others. These can be premium-priced and run by the same agency or usually a different agency, or theoretically a private company.

          Now before you say, “But Amtrak has business class and it’s publicly subsidized”, people have different expectations of long-distance transit or large-region transit than they do about city or metropolitan transit.

        • Kenny Easwaran February 22, 2018 at 9:22 am #

          Does finding a designated seat add a few seconds of dwell time at each station? If so, then it won’t work for a service that relies on many stops for many people along their way. It’s fine to add a few seconds (or even minutes) of dwell time at an airport for an airplane, because the takeoff and landing is so slow that you generally only make at most one or two stops between origin and destination. Even intercity rail doesn’t have too many stops per minute of travel time. But once you’re making a stop every minute, adding a few seconds per stop isn’t worth the comfort and convenience for the few that would pay for it.

  11. Georgist Economist February 3, 2018 at 4:09 pm #

    Not a problem of space, but one of time here. If/when automation happens, it will be possible to run fixed routes with van-sized vehicles, significantly cutting headways, while providing the same capacity and spatial efficiency. I wonder whether this will lead to a stratification of service?

    I mean that a route with e.g. one bus every 15 minutes would become a route with a van every 3 minutes. Now, this is so short that it could be stratified, with two overlapping routes, each with a van every 6 minutes. One for passengers willing to pay more for the right to only travel with others who are willing to pay for the same; and one for those who only want to pay the lower fare. This would provide much the same advantage as the private car—or taxis, for that matter—of only having select fellow passengers, but could be cheaper due to spatial efficiency.

    • Hugh February 4, 2018 at 6:45 am #

      I don’t get the two classes of vans. Why would you pay more just to travel with other passengers who also want to pay more? That doesn’t make any sense. In the absence of any extra advantage for the customer that comes with the higher-priced van service, everybody would just choose the cheaper van.

      • Qantas 94 Heavy February 5, 2018 at 10:42 pm #

        Say there are two buses:

        * Bus A: arriving in 5 minutes, costs $10
        * Bus B: arriving in 15 minutes, costs $2

        If you were in a rush or a rich person who doesn’t want to wait, you’d probably pay extra for Bus A.

        But if you weren’t as rich and wouldn’t mind waiting longer to save money, then you’d wait for Bus B.

        If they just charged $5 for both buses, they’d lose money from the rich people who would have otherwise paid more. Also poor people mightn’t be able to afford the extra fare for the bus.

        There are also cases where people pay more for practically the same service. For example, some first class train carriages are pretty much the same as the economy class carriages. The difference? Less “lower class” people around them, or more space for themselves as it’s less likely to be full.

        • Jo Walton February 6, 2018 at 5:31 am #

          Right — but say Rich Person gets to the same stop ten minutes later. Now there’s a bus in 5 mins for $2 and a bus in fifteen minutes for $10. They’re going to take the inexpensive bus, right? If time is the significant thing? But if it’s a case of comfort — guaranteed seat, air conditioning, while the rest of us stand and swelter? Then they might time themselves for the business class bus.

          When Britain privatized buses and you got to a stop and had to watch a “blue” bus go by while waiting ages for a “red” bus so you could use your return ticket, the result was rage and a drop in bus use, as well as the abandoning of unprofitable coverage routes.

      • Mike February 6, 2018 at 11:44 am #

        “Why would you pay more just to travel with other passengers who also want to pay more?”

        To get away from smelly people who talk loudly to themselves and assault others or are a different color. It depends on the city and how widespread the stereotypes are, but some people will take anything except the lowest-priced service. That has been the problem with attracting people to local bus service throughout the US. It may not matter as much in the UK, or on long-distance buses (although some people avoid Greyhound for the same reason). And there is also some truth to it because the US doesn’t provide as much housing or mental health services to people who need it, so they’re more present and stressed everywhere you go.

        • Ben G February 7, 2018 at 11:37 pm #

          Sadly I find that these stereotypes on long distance busses in the US are true. I’ve taken Greyhound many times and there is *always* at least one jackass that talks loudly and can’t shut up for the entire trip (or making other obnoxious noises, like constantly clearing their throat or coughing). Where in Northern Europe the intercity bus passengers are generally so quiet you can hear a pin drop.

          • Mike February 21, 2018 at 12:19 pm #

            I rode Greyhound quite a bit between 1998 and 2006. I found that the quality went up and down: both passenger behavior, ticket booth experience, station cleanliness, on-time performance, the flexibility of unlimited Discovery Passes (at first it was walk up to a bus anytime during the validity period, then you had to declare your itinerary up front and it was more cumbersome to make changes), and the policy if a transfer was overbooked (because people who were already on the bus had priority, so sometimes transferees would be bumped and they’d either bring a second bus or make you wait four hours for the next scheduled run; sometimes it required 12 people to activate a second bus, other times it seemed to be at the local manager’s discretion even if there were 30 people. On a few runs like the Los Angeles – Portland overnight express, they just always had two buses).

            The best-behaved, lowest-stress routes tend to be the secondary ones: e.g., not the packed long-distance expresses on I-5 and I-80, but the lesser-used ones like Seattle-Missoula and Seattle-Denver, and short-distance locals.

  12. Peter L February 3, 2018 at 6:34 pm #

    Heretic! Lol!

    “I was designing and revising these 25 years ago. ”

    The rest of this sentence “and they weren’t new then”.

  13. Brent February 3, 2018 at 8:44 pm #

    “I was designing and revising these 25 years ago.”

    40 years ago, “dial-a-bus” was going to be the magic bullet to solve the last mile problem feeding subway and commuter rail stations. The hype about demand-responsive transit reminds me of this.

    • Peter L February 7, 2018 at 8:28 am #

      Exactly what I had in mind when I posted above – I remember visiting Toronto as a kid and being impressed at seeing the signs for Dial-a-bus (or whatever the agency called it – forget if it was TTC, GO, or one of the other providers. This was 1977 so 40 years is spot on.

  14. asdf2 February 3, 2018 at 8:53 pm #

    I read the leaked Uber financials and the data really doesn’t say much:
    – We have no idea how much of the expenses are “fixed” stuff, like app development, driver recruitment, and how much is promotions, insurance, unmatched UberPool subsidy, which represent a real marginal cost per ride.
    – The data does not break down profits and losses by type of service. It is entirely possible that UberX is profitable, while UberPool is not – without more detailed numbers, we just don’t know.
    – The data does not break down profits and losses by city, so there is no way to distinguish between cities that Uber has just entered, or trying to enter vs. cities where they have an established market. It is entirely possible that the company is making money, hand over fist, in cities where they are well-established – that they’re just spending it faster than it comes in by expanding to new cities to grow the business. If the latter is, indeed what it is (and again, with the limited data we have, we really don’t know), Uber may very well become extremely profitable, once they’re done expanding everywhere worth expanding to.

    The arguments that you need taxi-like pricing for a car service to be self-sustainable don’t hold water either. Taxis are ridiculously expensive, with each mile of travel costing on the order of $2-4 – on top of a fixed charge for the pickup and per-minute fee for wait time. Even after taking into account the cost of labor ($0.25/minute at $15/hour), it does not cost anywhere near $2-4/mile to operate a passenger car. It just doesn’t. If a 10-mile trip has to charge $35-40 in order for the service to be self-sustaining, that should be an inherent red flag that something about the way that the service is operated is extremely bloated.

    • Eric February 5, 2018 at 3:07 am #

      Taxis are not in use all the time. You also have to pay the salary for the times they are not carrying customers (waiting for a customer, or else returning from a customer’s trip).

      • Dorian February 5, 2018 at 6:58 am #

        “Both Uber and taxis pay the driver based on fares, not based on hours worked.”

      • Mike February 6, 2018 at 11:47 am #

        But they have to pay a high enough salary to recruit drivers who are bearing the cost of deadheading, and worse, the underacknowledged wear and tear on their car.

    • Sailor Boy February 6, 2018 at 3:24 pm #

      “it does not cost anywhere near $2-4/mile to operate a passenger car”

      My company charges ~$0.80/km at 0% profit margin. That’s $1.30 per mile. If we assume that the car is empty approximately 50% of the time, then that is $2.60 per mile, before anyone makes a cent of profit or before Uber pays their fixed costs.

      Motorists almost universally underestimate the cost of operating a car.

      • asdf2 February 6, 2018 at 9:56 pm #

        According to the IRS, the cost to operate a motor vehicle in the United States is 53.5 cents per mile, when the vehicle is being used for business purposes. For services that intend to put a lot of miles on their vehicles, the overall per-mile rate should be less, as 1) The fixed costs of car ownership becomes amortized over more miles and 2) Spending more on a vehicle purchase price to get a car that burns less fuel and requires less maintenance becomes more cost-effective.

        Maybe in Europe, it actually does cost $0.80/km to operate a motor vehicle, but in the United States, with our cheap gas, it costs nowhere near that much.

        To illustrate the problem from another perspective, consider pizza delivery drivers. They typically use their own car, and get paid minimum wage, plus mileage reimbursement, plus whatever tips the customers are willing to give them. The mileage reimbursement that a pizza delivery driver gets is nowhere near the per-mile fare to ride in a taxi. There is no reason why a taxi should cost 3-4 times as much to move for each mile as the same car, with a different paint job, delivering pizzas.

        • Sailor Boy February 7, 2018 at 3:17 pm #

          I had actually forgotten how little of the cost of driving is actually borne by the motorist in the US. Sorry.

      • Adam Tauno Williams February 20, 2018 at 2:48 am #

        This! I have this conversation sooo often.

  15. Hugh February 4, 2018 at 2:34 am #

    Isn’t this obvious? Why would uber be anything more than app that A) make it easier for a customer to get a taxi B) bypasses regulations in many places on the licenses required for taxis ?

    Why not move on from the same old Uber and Elon Musk story, and start blogging more about public transportation, which is what this blog is supposed to be about.

    • Mike February 6, 2018 at 11:57 am #

      It makes it easier for people to get an Uber car, where Uber sets the rates and takes a middleman’s share. On the other hand, a vendor-neutral ride-hailing app that’s run by a nonprofit and funded by a subscription to just the ride-matching services, with drivers from several companies and independents, would make a better rideshare that I could more fully support. It’s similar to the concept of municipalities running fiber or coaxil/copper cable to people’s houses and allowing open access to several ISP companies and services.

  16. Chuck Mose February 4, 2018 at 5:49 am #

    Driverless cars would totally change thus equation.

    • Mark Bahner February 5, 2018 at 11:46 am #

      “Driverless cars would totally change this equation.”

      I agree, but mostly because:

      1) Driverless cars will encourage transportation-as-a-service, rather than private vehicle ownership.

      2) Transportation-as-a-service will encourage vehicles that are more closely tailored to the particular use. For example, there will be many single-seat, single-occupancy vehicles, rather than people driving alone in 4-door vehicles. The single seat-single occupancy vehicles might even fit two-abreast in a single lane, particularly in low-speed traffic. They will also likely be far less expensive in terms of initial purchase cost and operating costs.

      3) Transportation-as-a-service will also provide knowledge of the origin, destination, and time-of-desired-arrival for thousands or millions of people in a town or city. This will allow adjustments in routes and vehicles to dramatically increase efficiency with respect to cost, travel time, energy usage and emissions, etc. For example, imagine what could be done in New York City if the origin, destination, and desired arrival time was known for every person entering, leaving, and traveling within the city at every minute of every day.

      • Ben G February 7, 2018 at 11:43 pm #

        You mean… a bicycle?

      • Georgist Economist February 8, 2018 at 12:54 am #

        I agree about driverless tech changing transportation, but not with the particular points you bring up.
        1) Is possible but I wouldn’t count on it too much.
        2) Today the average suburbanite could own a small car (a really small car, like the Tata Nano) for daily commuting and grocery shopping, and rent a larger vehicle for a few days per year when they actually need the speed or range or cargo capacity. This would be cheaper than commuting in a large car or SUV every day. Yet they own and drive such large vehicles. Personally my opinion is that many people have some sort of “irrational” desire for driving a “mine’s bigger” car.
        3) 99% of that data is useless. If you knew a representative 1% of that data, you would have a very good idea about when and where people want to go, and you would move on to solving the logistical problem of actually moving 100,000 people from residential areas to the CBD. Just because you know a bit more exactly where people are going, doesn’t make it any easier to actually move the 100,000 people. Indeed, you might not get anything whatsoever out of that data. If you are moving 50 people per bus, it really doesn’t matter which building passenger #42 is going to. There are 49 more people on the bus, whose time you are not going to waste by doing a detour. You just stick to the arterial and passenger #42 will walk to the building he wants to go to. And at that point you don’t even need to know what the exact answer is, because it doesn’t matter one way or another. Look up the “Explainer: The Transit Ridership Recipe” on this site, in particular “Diversity, Not Specialization”.

        • Mark Bahner February 8, 2018 at 8:01 pm #

          1) “Is possible but I wouldn’t count on it too much.”–>Of course autonomous vehicles will encourage transportation-as-a-service. Why would you own a vehicle that you only drive 2 hours a day, and has to compromise between a large number of duties (i.e., hauling kids, groceries, going to movies, traveling alone to work) when you could rent an autonomous vehicle that is perfectly suited to your current task, and might even travel 10+ hours per day, thus greatly reducing per-mile cost?

          2) “Personally my opinion is that many people have some sort of “irrational” desire for driving a ‘mine’s bigger’ car.”–>We’ll see. Autonomous vehicles providing transportation-as-a-service will provide some very clear choices between vehicles that cost very much per mile, and vehicles that cost very little per mile.

          3) “99% of that data is useless. If you knew a representative 1% of that data, you would have a very good idea about when and where people want to go,…”—> No, if you know everyone’s location, destination, and desired time of arrival every instant, you can do all sorts of things that are not possible if you only know where 1% of the population’s location, destination, and desired arrival time. You could have 4 single-occupancy vehicles that arrive at some location along a street in Ft. Lee NJ at precisely the same time. They could get into a minivan or 4-door car and drive across the GW Bridge to some location approximately at the centerpoint of all their destinations in Manhattan, then single-occupancy vehicles could drive them to their destinations. (Or one or more of them could be delivered to a subway station with *exactly* the right amount of time to take them to their destination.)

          Just yesterday, I watched a full-size diesel city bus go past where I was eating lunch. It had *zero* passengers in it. It went to a bus stop on another street that I could also see. It waited there for easily 5 minutes, picking up *one* passenger. That would absolutely never happen in a situation where a fleet of vehicles provided transportation-as-a-service.

          Now you might say, “That’s a very unusual situation!” But the average number of passenger on a city bus in the U.S. is nine passengers, per this website:

          If the transportation-as-a-service provider knew there was only going to be one person at the stop, they certainly wouldn’t provide a bus that can fit more than 40 passengers. Or sit at the stop for more than 5 minutes. That’s the beauty of knowing where everyone is, where they want to get to, and when they want to get there.

          • Georgist Economist February 10, 2018 at 3:59 pm #

            1) I think you overestimate the savings. Even in non-American cities without strict single-use zoning, there is a significant difference between the CBD and the residential belt; in American cities, this is strictly enforced by zoning. This leads to a strongly tidal traffic pattern. Let’s say that the morning peak lasts two hours or so, during which the cars can shuttle back and forth usefully. If a single-direction commute is 20 minutes (quite fast), the car could carry THREE people in the peak direction. In the afternoon peak, the same people travel in the opposite direction, and there are some additional errands that people run. Let’s be generous, and not press the fact that people run errands with the same car they commute with. Even then, with all these generous assumptions, each autonomous car replaced five or six human-driven ones. So, you split the fixed costs of the car in five, but add in the profit the fleet owner must make. And this does absolutely nothing to the marginal cost of the car. Indeed, now the car runs around empty a lot of time. Sure, the end result is cheaper than owning your own car, but not by all that much.
            3) “If the transportation-as-a-service provider knew there was only going to be one person at the stop, they certainly wouldn’t provide a bus that can fit more than 40 passengers.”
            The transit operator cannot sell the bus when they don’t need it and re-buy it when they do. They only have one fleet, and that has to be sized to the highest regularly occurring demand. (A purely for-profit company can ignore that and tolerate regular overcrowding.) Outside the peak, they run the same vehicles with a lower occupancy; what else could they do?
            Sure, automation will lead to smaller vehicles on most routes. A van every 10 minutes is a lot friendlier than a bus every 30 minutes, offers similar capacity and has a similar cost. But even with vans, you are unable to use 99% of the data. When thousands of people live along the route, you can’t adjust the schedule to fit any single passenger much better. Even if you actually had all the data (which would be quite the police state; “resident #123 has left his/her house and wants to go to [address]”), the maximally efficient way to run the vehicles would still come out fairly even. Unless people are somehow magically linked to each other, so that everyone spontaneously leaves their homes within a short interval, there will be a fairly steady “trickle” of passengers. It’s just easier to skip all the invasion of privacy, all the number crunching, and just run the vehicles at a steady interval. And that can be done just fine with only 1% of the data.

            To address your specific example: LOTS of people want to go from Ft. Lee to Manhattan. If every van took four, there would be quite a column of vans over the bridge. Eventually, it doesn’t really matter which person is sitting in which van, because they all go to Manhattan and all the passengers connect onwards. So why bother invading their privacy, when you could just run a van every so often along a fixed route? That way, people wouldn’t have to tie themselves to any particular van. If they are on the street corner when they realize “they left the stove on” or something else, they can go back and turn it off without holding up the other passengers in the van you would have designated them. They are not forced to do exactly the same thing every day. One week they can go to work half an hour sooner to finish their report or whatever. This is a perfectly normal thing, and causes absolutely no problem with a fixed schedule service. With the demand-responsive approach, it either causes an upset, cannot be fit, or requires so much surveillance that is downright Orwellian.

          • Mark Bahner February 10, 2018 at 7:59 pm #


            “To address your specific example: LOTS of people want to go from Ft. Lee to Manhattan. If every van took four, there would be quite a column of vans over the bridge.”

            If every van took four people, traffic on the GW Bridge at peak hours would be reduced to about the level it is at approximately 5 AM and 10 PM.

            Here’s a graph of hourly traffic on the GW Bridge. At the peak time of 5 PM, there are about 20,000 vehicles per hour on the bridge. (Let’s just say they’re all cars for convenience of calculations.)


            The cars over the GW Bridge carry, on average 1.75 people (I previously provided a reference for this number), so the 20,000 vehicles carry approximately 35,000 people. So if the cars instead carried on average 4 people, the number of vehicles per hour at the 5 PM peak would be 8750 vehicles per hour, instead of 20,000 That’s about the number of vehicles that cross the GW Bridge at 5 AM and 10 PM. So congestion on the GW Bridge will be completely eliminated.

            There would be no lines waiting to get on the bridge because all vans/4-door-cars would be meeting at places specifically chosen so that no entrances to the bridge would be particularly clogged.

            In the immortal words of Judge Doom, “My God, it’ll be beautiful!” 🙂


            20,000 x (1/

          • Mark Bahner February 12, 2018 at 11:25 am #

            “To address your specific example: LOTS of people want to go from Ft. Lee to Manhattan. If every van took four, there would be quite a column of vans over the bridge.”

            If every van took four people, traffic on the GW Bridge at peak hours would be reduced to about the level it is at approximately 5 AM and 10 PM.

            Here’s a graph of hourly traffic on the GW Bridge. At the peak time of 5 PM, there are about 20,000 vehicles per hour on the bridge. (Let’s just say they’re all cars for convenience of calculations.)


            The cars over the GW Bridge carry, on average 1.75 people (I previously provided a reference for this number), so the 20,000 vehicles carry approximately 35,000 people. So if the cars instead carried on average 4 people, the number of vehicles per hour at the 5 PM peak would be 8750 vehicles per hour, instead of 20,000 That’s about the number of vehicles that cross the GW Bridge at 5 AM and 10 PM. So congestion on the GW Bridge will be completely eliminated.
            There would be no lines waiting to get on the bridge because all vans/4-door-cars would be meeting at places specifically chosen so that no entrances to the bridge would be particularly clogged.

          • Georgist Economist February 13, 2018 at 12:41 pm #

            Apparently, I was ambiguous. When I said that “there would be quite a column of vans”, I meant that there would be thousands of vans per hour (8750, as you said). An average of 129.+ per minute, more than two per second. At that frequency, there’s no reason to insist that Joe Schmoe travel in van #1234, because another one will be around in seconds. Which is why my opinion is to just run frequent service (where the land use pattern justifies it) and let passengers board whichever vehicle they wish, without the transit agency having to know them better than their own spouse.

          • Mark Bahner February 13, 2018 at 9:07 pm #

            “When I said that “there would be quite a column of vans”, I meant that there would be thousands of vans per hour (8750, as you said). An average of 129.+ per minute, more than two per second. At that frequency, there’s no reason to insist that Joe Schmoe travel in van #1234, because another one will be around in seconds.”

            Yes, there are vans running at more than 2 per second over the GW Bridge itself, but there may be only a handful of people each minute in a particular section of Ft. Lee NJ that are wanting to go over the bridge.

            For instance, I’m too lazy to do detailed calculations, but Google Palisades Park, NJ. Let’s say that in rush hour there are 4 people from Palisades Park that are trying to get into Manhattan. I’m saying that 4 incredibly small, single-seat, single occupancy vehicles might take them from their doorstep to a minivan that’s close to the Sixth Street entrance to Route 46. That minivan would then get onto route 46, and then from route 46 onto whatever road then crosses the GW Bridge into Manhattan…is that U.S. Highway 1, I-95 lower level? (IMPORTANT NOTE: I’ve never lived in NYC, or even visited more than a handful of times. So I don’t know the roads.)

            Now, as I think you’re saying, if the person missed the van that left Palisades Park at 8:59 AM, he or she could get another van that left Palisades Park at 9:00 AM, but in my mind that van might start from some place other than the Sixth Street entrance to Route 46.

            When people fly on airplanes, they have no “privacy” in the sense that the airlines know exactly who they are (unless they’re flying under a clever alias) and when they’re departing and arriving. But I’m not aware of people who avoid flying to protect their “privacy.” I see this transportation-as-a-service the same way. I don’t see any significant “privacy” concern, just like I don’t see a privacy concern with flying on airlines.

            P.S. Here’s a similar comment I just made on the IndyWeek (Raleigh NC) news site about a proposed extension to I-540, the “outer beltline” around Raleigh. In my comment of Feb. 13 at 9:25 PM, I go through some alternate means to get from a particular neighborhood in Raleigh to Research Triangle Park, in the middle of the Raleigh-Durham-Chapel Hill triangle of cities. I propose that people will want to get into “flash” vanpools to travel on I-40, because it will be much cheaper to do so, and won’t take significantly more time than driving alone. (I tried to provide what I thought would be some realistic costs and times involved.)


          • Georgist Economist February 17, 2018 at 8:05 am #

            “I’m saying that 4 incredibly small, single-seat, single occupancy vehicles might take them from their doorstep to a minivan that’s close to the Sixth Street entrance to Route 46.”
            You mean, they ride their own bike to the bus stop or rail station (or vanpool pickup point) and board a vehicle coming from further down the line? Or walk to the station? P+R, K+R, whatever+R?

            I don’t have any problem with vanpools. My problem is that there is absolutely no need for transit agencies (or taxi drivers, or vanpool operators) to know who goes where. Let me quote Jarrett Walker directly:
            »But in the end, it’s less important to high-ridership planning than a simpler question: “Where can we find lots of people, and places where lots of people are going, located in ways that are cheap for us to serve?”«
   “Diversity, not specialization”
            The same is true for long-distance trains and airlines. The latter used to work just like this, too, and as transportation-that-always-had-been-a-service, that was fine. It’s only due to hijackers and other security issues (khm) that plane tickets can no longer be bought with just “good morning, I’d like a ticket to [destination]”. But in principle, they could; and for decades, had been.

          • Mark Bahner February 19, 2018 at 10:06 pm #

            “You mean, they ride their own bike to the bus stop or rail station (or vanpool pickup point) and board a vehicle coming from further down the line?”

            1) No, the transportation service provider owns more small, occupant vehicles than any other type of vehicle in the fleet. But they’re bicycles. They can go 30-45 mph, and they have protection from the elements, and protection from low-speed collisions with other vehicles. Bicycles don’t move quickly enough, they have no protection from the elements or low-speed collisions…and they even need the riders to be coordinated and reasonably fit.

            The four people that are meeting for the van might each be traveling 3-5 miles to get to the van meeting place. Virtually no one wants to ride a bicycle for 20-30 minutes in New Jersey in the winter, when the streets are wet and slushy like the opening scene of Hill Street Blues. But they don’t mind at all getting in a car that, in 5-10 minutes, gets them to the van for their van pool. (And is already heated or cooled, and has the radio tuned to their favorite station.)

            2) …and board a vehicle coming from further down the line?–>It’s less clear to me how the van got there. My crystal ball is a little fuzzy about that. 😉 But one thing I think is very bad (except for the people who choose the super-low-budget option) is for a vehicle to be stopping at multiple places, such as stopping at every exit on Route 46.

            3) “My problem is that there is absolutely no need for transit agencies (or taxi drivers, or vanpool operators) to know who goes where.” –> It’s necessary to know where everyone is coming from and going to, and when they want to get there, in order to optimize the system. For example, I go to breakfast meetings on the third Thursday of every month, in Research Triangle Park. There are typically maybe 25 people there. If a transportation provider knew that 25 people were trying to get to the same place at the same time (very early in the morning), but it was only happening this one particular day of the month, the transportation provider could arrange some better situation than if vehicles were simply running fixed routes at fixed times.

            4) Let me quote Jarrett Walker directly: »But in the end, it’s less important to high-ridership planning than a simpler question: “Where can we find lots of people, and places where lots of people are going, located in ways that are cheap for us to serve?”« –> I think Jarrett’s philosophy is fine for a situation where drivers are a significant cost to the system, and the system is not providing door-to-door service. I think that philosophy will be rendered obsolete by driverless vehicles operating in transportation-as-a-service mode. The reason it will be rendered obsolete is because it will be out-competed in cost/door-to-door-transit-time by a system that knows where everyone is starting from, going to, and when they want to get there. Such a system will take tremendous computer power, but computer power is already cheap and will only get more so. Perhaps some fixed system will be preferred, such as on the grounds of privacy concerns. But I think people will prefer a system that optimizes their travel according to their desires, even if there’s a record somewhere of where they came from and where they went. (Bank and house robbers, murderers, etc. will probably choose some system with greater anonymity. ;-))

          • Mark Bahner February 19, 2018 at 10:21 pm #

            Oops. It’s very late:

            “1) No, the transportation service provider owns more small, occupant vehicles than any other type of vehicle in the fleet. But they’re bicycles….”

            Should have been (with emphasis on the correct wording added):

            “1) No, the transportation service provider owns more small, *single*-occupant vehicles than any other type of vehicle in the fleet. But they’re *not* bicycles….”

        • asdf2 February 8, 2018 at 8:51 pm #

          “Today the average suburbanite could own a small car (a really small car, like the Tata Nano) for daily commuting and grocery shopping, and rent a larger vehicle for a few days per year when they actually need the speed or range or cargo capacity.”

          Part of the reason why not is that the price difference between a smaller car and a bigger car just isn’t enough for most people to make a compelling case for the smaller car. Much of this is due to the fact that the car manufacturers *want* people to buy bigger cars, which have bigger profit margins, hence they have no financial incentive to spend money figuring out how to get the price of a smaller car down.

          Smaller cars would also be cheaper if we could get away with a less powerful engine, that would maybe work on surface streets, but not the freeway. Unfortunately, too many communities are structured so that many trips *require taking the free (or a huge detour). For instance, Seattle is adjacent to a giant lake, and any vehicle that is too fast for the bike trail, yet too slow for the freeway effectively cannot cross the lake’s bridges, by any means, and must detour all the way around.

          • Mark Bahner February 9, 2018 at 8:50 am #

            “Smaller cars would also be cheaper if we could get away with a less powerful engine, that would maybe work on surface streets, but not the freeway.”

            Yes, I think laws can and will be changed such that single-seat, single-occupancy vehicles will be built and licensed to operate only in low-speed situations (such as below 45 mph). This will allow much less expensive vehicles, because they won’t need the power to accelerate to highway speeds, nor the energy (battery capacity) for long-distance travel, nor the safety equipment to operate safely at highway speeds.

            “For instance, Seattle is adjacent to a giant lake, and any vehicle that is too fast for the bike trail, yet too slow for the freeway effectively cannot cross the lake’s bridges, by any means, and must detour all the way around.”

            Yes, this is somewhat similar to the GW Bridge in NYC I gave. (Except the situation I was describing was to avoid congestion on the GW Bridge.)

            I don’t know Seattle at all, but looking at a map, let’s say a person wanted to go from Redmond to Ravenna. I’m picturing that person would take a single-seat, single-occupancy vehicle inside Redmond, to get to route 520. At route 520, the person would catch a multiple-occupant vehicle such as a minivan to cross Lake Washington on route 520 and go north on I-5 to the Ravenna exit. At the Ravenna exit, the person might catch another single-person single-occupancy vehicle to get around Ravenna. The two transfers would be close to instantaneous, with the receiving vehicle maybe saying, “Hi Jill, I’ll be taking you to Ravenna.”

            With transportation as as service, I’m expecting many more situations where people change vehicles once or even twice in travel from one place to another. This will be acceptable mainly because the transfer will not involve long wait times (e.g. waiting 10-20 minutes for a bus in the present situation).

  17. Malcolm February 4, 2018 at 9:59 am #

    How does PRT fit into the equation, something like Ultra at Heathrow Airport, able to pick up at frequent stops but bypass those the passenger doesn’t want to go to. Possibly work out something like this with autonomous vehicles?

    • Eric February 5, 2018 at 3:11 am #

      We already have PRT, it’s called cars/taxis.

      The futuristic versions of PRT will never be worth building when the road network already exists. (And if the road network is overloaded, the solution is to build mass transit, not another low-capacity network.)

      Autonomous technology will make taxis much cheaper, but they won’t create road space. (Except to the extent that they can eliminate parking spaces)

  18. Lisa February 5, 2018 at 5:55 am #

    From a European perspective: The only thing that could possibly make ridehailing profitable (at least in transit-rich European cities) is if private car ownership goes down and a majority of the population becomes “multimodal” (using transit and bike for most trips and combining this with ridehailing). In this scenario, which would be the ideal scenario in terms of climate protection and liveable cities, the demand for ridehailing would rise – but not by reducing the number of transit trips but by reducing the number of private car trips. Cities could initiate a development in this direction (as proposed by Bruce Shaller) by implementing road pricing, strengthening public transit and bike infrastructure and reduce parking requirements, thus making private car trips less attractive.

    • Georgist Economist February 8, 2018 at 1:03 am #

      I think the above is mostly correct, but leaves out a crucial element. Namely, that the marginal cost (the cost of a single additional trip) is much higher with ridehailing than with anything else, including private cars. (Private cars have a high fixed cost, i.e. having a car and only using it very rarely is enormously expensive.) Thus many trips that are today worth making in a car *if you have already paid for the car*, that is, the trip is more valuable than the marginal cost, wouldn’t be worth making with ridehailing. The result would be a decrease in the number of trips. Which is somewhat counterintuitive.

  19. Dave February 5, 2018 at 8:07 am #

    “The app has transformed customer experience, but it does not seem to make much difference to the bottom line. It takes some friction out of the hailing and routing, but it doesn’t transform the fundamental nature of the task, or its potential to be profitable.”

    It also takes a lot of friction out of the payment side of the ride, by billing all of the fare to your credit or debit card instead of requiring passengers to carry around cash. Yes, I know that many cities have mandated their regulated taxicabs to accept cards… but, until Uber/Lyft came around, cab drivers would balk all the time about accepting cards. I can’t tell you how often I got into a cab and tried to pay with a card, only to have the driver tell me “the machine is broken”… and then “I don’t have change for a $20” after I had him take me to an ATM.

    There are definite benefits to not ever having to carry cash around with you, namely (1) you’re less at risk for being successfully mugged, (2) you’re less at risk for being solicited by homeless people who can hear coins rattling around in your pocket, (3) you have a record of every transaction in case there’s a problem later and (4) you can take rides more impulsively (important in those situations where running out of cash and not being able to find an ATM meant you’d be walking farther than you would have liked). This ease of payment was the entire reason why I switched from cabs to Uber/Lyft.

    • Jarrett Walker February 5, 2018 at 5:33 pm #


      Good point. I will add a word about payment.

    • Carl Stanton February 20, 2018 at 9:13 am #

      The flip side to this is Uber/Lyft are unusable for people who do not have credit/debit cards.

  20. JJJJ February 5, 2018 at 8:35 am #

    asdf2 is correct that you are missing one small detail: Uber/Lyft drivers, by virtue of ignoring regulation, are not geographically limited. A, official Boston taxi picking up at the airport and going to a suburb MUST return empty. They are not allowed to pick up in the suburb. That is hugely inefficient, especially in areas where there are a bunch of very small cities all close together (ie, Northern NJ, Boston Area, Miami). Meanwhile an Uber can pick up near every drop-off spot, if demand exists. Additionally, the software does have a “go-home” feature where they will try to match a driver with a rider heading in the direction that is desirable to both.

    So that does alter a small part of the profitability argument.

  21. John D February 5, 2018 at 10:45 am #

    The app changed the whole market industry. It’s dramticly increase the supply which help to lowe the cost. So I belive the app changed the industry by providing more and easy access to both drivers & riders.

    • Mike February 6, 2018 at 12:01 pm #

      Only because the companies’ investors are subsidizing it heavily. The question is whether this service model can be sustained indefinitely. The estimates are looking negative.

  22. John Niles February 6, 2018 at 7:01 pm #

    Key is Jarrett’s sentence “Urban transportation is a spatial problem, and (until automation) a problem of the efficient use of labor.”

    There is no way that Uber and Lyft and the other TNCs become profitable until paid drivers are removed. Both Uber and Lyft are working on automated driving. Uber is working with Volvo’s automated vehicle and Lyft is working with a variety of car automation companies, last I heard.

    Think of future robo-cabs as Uber without a driver.

    The company farthest ahead in developing robo-cabs is Waymo, a branch of Google. From the evidence I can find, the company is running a growing fleet operational trial of Chrysler Pacifica vans providing free rides on demand to registered customers in the Phoenix region with a Waymo employee so far always in the back seat just behind the driver’s position connected by wireless communication to some sort of control center. That Waymo employee can order a hard stop if necessary, and it will happen. But that is probably pretty rare because of the low speeds and calm suburban streets that is the apparent service territory so far. The service territory is very well mapped and “known” by the vehicle’s computer. A current requirement of Arizona state law is that a licensed driver must be in the control loop of a motorized road vehicle. That licensed individual could be remote. Removing that requirement is going to be a big, big step for both State of Arizona and Waymo.

    Look at the video posted prominently at for visual evidence of what I am describing.

    Robo-cabs are a part of the human transit of the future, but are not the entire future because of the geometry thing that Jarrett keeps us all aware of. The public transit agency I am most familiar with, King County Metro in Washington State serving Seattle and vicinity, is well aware of transit’s robo-cab future component at the senior management level.

    • Jarrett Walker February 6, 2018 at 9:25 pm #

      If robo-cabs are the future, robo-buses will need to be the future as well. It’s the only way everyone will fit down the street in a city as dense as Seattle.

      • asdf2 February 6, 2018 at 9:44 pm #

        I would not assume that. Human-driven buses are protected by powerful unions…which donate large sums of money to the campaigns…of the elected officials…which pass laws…requiring the transit agency to use union labor. Breaking that cycle to replace human bus drivers with robots is just not going to happen anytime soon. The NYC subway still has two employees on board every train for absolutely no reason except that the union contract requires it.

        If autonomous buses exist in the next 10-20 years, they’re going going to be privately operated jitney services, not government run public transit, since the private companies are not subject to the same union constraint.

        • Jarrett Walker February 9, 2018 at 12:29 pm #

          Unions know about the history of labor. They weren’t able to prevent the automation of factories. If transit agencies are being destroyed by driverless car competition — with ruinous impacts on society and cities — it will be an emergency, and the right thing will be done.

          • asdf2 February 10, 2018 at 6:08 pm #

            I think this a different case. The factories were run by for-profit companies, while transit systems are run by government agencies, funded largely by taxpayers. If a factory is getting undercut by competition, it has to either adapt or die, and both the owners and the unions know it.

            A transit agency, on the other hand, is different. As long as taxpayers are willing to keep paying for the human drivers, the human drivers will continue to get paid. They’ll justify on grounds related to passenger safety (need an official presence on the bus to discourage bad behavior) and access for people with disabilities (with no driver, who straps in a wheelchair user if there are no other passengers on the bus).

            The United States in particular, has an annoying habit of prioritizing, nearly universally, accessibility to disabled riders over maximizing the number of total riders.

      • John D February 12, 2018 at 10:03 am #

        If robo-cabs are the future, robo-buses will need to be the future as well.

        If this is will be true, then the transit agencies will be in danger. Why? for many reasons;

        1/ Autonomous buses will bring transit industry to profitability. But, this will attract everyone to jump in, and compete with transit.
        2/ Autonomous buses will not run by itself, it requires a very sophisticated technical infrastructure and technical users. Most of the transit agencies are really behind when it comes to technology.
        Again, what prevents Uber, Tesla, or Apple to run their Roberts all time. They have the technical infrastructure and the resources.

        • Georgist Economist February 17, 2018 at 8:16 am #

          Then the companies will BECOME transit. Who cares what logo is on the side of the bus? It still works like a bus, i.e. runs on a fixed route with a regular-ish headway and seats a multitude of passengers. If the current transit agencies disappear, that’s OK. Horse-drawn streetcar companies have been replaced by electric streetcar companies, with no change in geometry.

        • Mike February 20, 2018 at 4:11 pm #

          It’s not clear that autonomous buses will be profitable. But even if they are and private transit buses come back from extinction, that’s not necessarily a bad thng. But tax-funded agencies and private transit companies have different mandates. The agency’s responsibility is to serve the entire service area, or at least the parts that meet its ridership/coverage goals, and to serve lower-income neighborhoods that don’t have a lot of money but do have a lot of mobility needs. A private company’s mandate is maximum profit, and that means serving only the highest-volume affluent areas. Georgist asks, “Who cares what logo is on the side of the bus?” What matters is whether the bus exists at all: is it deleted from lower-income areas? Does it charge a higher fare than the previous public bus, a fare that is harder for lower-income people to afford?

  23. Stephen Richards February 7, 2018 at 7:52 am #

    Uber has transformed the taxi business in London where the traditional black cabs were highly regulated, expensive, and only available inside the wealthy central core of the city, and then only if the destination was somewhere the cab driver wanted to go. The arrival of uber has meant cheap easily hailed taxis in all parts of the city willing to go wherever you want. If that’s not transormation, I don’t know what is.

  24. Dorian February 7, 2018 at 9:59 am #

    One thing that could be investigated is the degree to which TNCs are replacing cabs, versus worsening congestion in a net-sense.

  25. Aaron Antrim February 12, 2018 at 8:16 am #

    My guess is the preposterous valuation of Uber represents that investors anticipate it will provide a fairly complete transportation *marketplace*, rather than being a single transportation provider, or, rather, broker within a larger and highly competitive marketplace. It’s a scenario and assumption about which we should be concerned. Default API terms for Uber and Lyft prevent 3rd party apps from representing a full marketplace — their services are not allowed to be presented alongside direct competitors.

    • asdf2 February 13, 2018 at 11:45 pm #

      I read the terms of use. The fact that it’s illegal to create one app that shows all the ride-hailing companies side by side is sad.

      Although, in the modern world where Uber and Lyft are now essentially operated by the same drivers at nearly identical prices, the side-by-side would not be all that useful.

      What would be more useful is an app that would make suggestions on how to avoid surge pricing. Of course the official apps don’t do this because they *want* you to pay for surge pricing so they can make more money. In fact, the Uber app, about a year ago, stopped bothering to even tell when surge pricing is in effect; they will quote you a price that takes surge pricing in account; they just won’t tell you whether the quoted price is the normal rate or the surge rate. (It’s amazing how many more people will pay $12 for a $6 ride when the fact that the $12 ride is usually only $6 is disguised).

      Armed with sufficient information, there are lots of surge-pricing-avoiding tricks that an app might suggest. For example, you might be able to avoid the surge by walking a couple blocks, or by hopping on an express bus that happens to be coming at the right time, and is traveling vaguely in the direction you want to go. Of course, the necessary data to train such an app cannot be harvested without violating the companies’ terms of use.

  26. Wanderer February 12, 2018 at 4:20 pm #

    A couple of thoughts about driverless buses:

    1. There’s likely to be a substantial period where passengers are resistant to a bus without a staffperson. One way to get around that might be substituting a conductor/security guard, to help and monitor passengers. That might reduce, but not eliminate labor costs.

    2. My cousin was a printer for a major American newspaper with a strong union. The newspaper wanted to automate the process as much as possible, the union of course didn’t. Ultimately they reached an agreement where everyone who wanted to keep his job could do so, for as long as they wanted. People were transferred to new jobs. The paper was not required to replace people when they left. So gradually the workforce declined. You might see something like this at transit agencies with strong unions.

    3. If companies really wanted to run transit privately, the function of the transit agency would shift, not disappear. The big problem would be making sure that service was equitably distributed and coordinated. So the agency would pivot to being primarily a planning and coordination entity, though it might run some lines that weren’t attractive to businesses.

  27. Melanie Curry February 21, 2018 at 10:02 am #

    I lived in Singapore a number of years ago, and at that time taxi companies already had what would have been a proto-app. No maps, no smart phones, but you could call (phone-hail) a cab from your landline and you’d get a call back to tell you a cabbie had accepted your request, and another call when the taxi arrived. It was no less convenient than Uber and Lyft, and this was before smart phones and even before most people had cell phones.

    The difference there was a tightly regulated environment for owning vehicles, in a small country with impermeable borders–being an island–and a maxed-out transit system. Singapore had already begun reaching maximum vehicle saturation at that point, and had just started a congestion pricing program in its CBD. Owning a vehicle was out of reach for many people, who relied on transit and taxis to get around. Driving taxis was also a form of semi-reliable employment–with its own problems–for many people who didn’t or couldn’t find a fit in the country’s rigid employment expectations.

    That tight market is likely what made taxis at least somewhat profitable, and phone-in options just helped make the customer experience better. But they were still taxis. Uber and Lyft haven’t really innovated much beyond that—however, they are looking forward to the day they get rid of drivers altogether.

  28. Todd Medema February 23, 2018 at 11:09 pm #

    An important fact not mentioned here is that the Uber / Lyft app experience has been dramatically increasing “ridership” for this type of service; as ridership for any service scales, so too will its efficiency. For example, an Uber / Lyft driver doing a pooling route in a highly walkable area like SF might easily service 10 people in an hour, for a net cost of about $10 (about 10,000x better than the Kansas City trial).

    But, I agree that that still does not solve the geometry problem. Even at a perfect efficiency of 3-4 passengers in a pool at all times, it’s still 10x as many VMTs as a single bus carrying them.

    The real question in my mind is: If Uber and Lyft are capable of dramatically increasing usage of taxi services through an app, and it’s a given that transit operations become more efficient as more people use them, how is it not then given that apps hold the potential to transform urban transit by increasing their ridership scale?

    For the reasons you point out, Uber / Lyft aren’t profitable yet, and it’s possible they will never be due to geometry. But urban transportation provides a TREMENDOUS amount of value – arguably the entire value of a city is in its more efficient transit. So a sufficiently efficient transit provider (capturing a large portion of the transit market in a geometrically efficient way) /should/ be able to be profitable. Otherwise, if urban transit as a whole is a net loss instead of a net positive… how would cities even exist? That defies our previous assumption that one of the primary values of cities is their improved transit vs the alternative.

    To tie it all together, I would postulate that transit agencies have a business model problem. It’s clear that they provide a critical service worth a significant portion of the entire city’s economy, and yet they are unprofitable and shrinking! Rather than thinking about TNCs as terrible evils, what if we consider them a demonstration of one of many different user types whose needs are not being met by transit agencies – the user who values their time at more than $10/hr, and is willing to pay for that Uber / Lyft to get to a destination that’s not perfectly aligned with a rapid fixed route.

    Any business that wants to continue to thrive needs to offer multiple products to fit different and changing market needs. We see this with Uber and Lyft themselves with their more affordable but slower options (as well as more expensive premium options)…. but we don’t see this with transit agencies! Although transit agencies have different frequencies of routes, I am never given a choice as a consumer: am I willing to pay the agency an extra $5 to guarantee I arrive to work on time? Not at all. Quite frankly, if nothing changes, TNCs offerings may expand to more and more transit-agency-like services and eventually consume transit agencies if they continue to insist that they only provide one product that doesn’t quite fit most people’s transit needs.

    So it’s not just “an app” or “increased IT spend” – it’s giving the freedom to choose your transit and financial priorities. The very foundation of your valuable approach to transit planning! I would love to see a similarly in-depth analysis from you on past attempts transit agencies have made on providing choice to users, since I think that may lead us all down a more beneficial path than the TNC point-to-point model alone 🙂

  29. Paulo Ferreira dos Santos February 28, 2018 at 2:21 am #

    Regarding urban mobility apps, I see that most of the analysis tend to put very different things (apps, products, services, and issues addressed) in the same basket. I tend to agree that some apps minimize communication problems and others improve riders’ experience. I also agree that most of the apps don’t solve the spatial problem (reducing the number of vehicles in the same space).

    I believe that smartphones and apps will have quite soon an important role in promoting the transition from cars to transit, addressing on that way also the spatial problem. This can be achieved through an agnostic approach regarding all mobility offer available on each place (transit, car, car parking, bike, ferry, etc.) which solves, at the same time, the fragmentation issues (intermodality, multimodality); the visibility issues (where is the best place to leave my car and ride? what is the cheapest and faster way to arrive to my destination?); the information issue, for riders, agencies; the stakeholders’ involvement issue (e.g. employers, as they represent often a lot of riders and pay often for ridership of their employees); and the mediation of incentives and costs to make attractive to use transit instead of cars ….. and a profitable business model.

    Stay tuned.

  30. Charu April 9, 2018 at 2:22 am #

    The companies like uber are making a great profit by providing delivery services as well. But as far as other services are concerned I believe app based tracking system allows users to get better visibility and real time order tracking.

  31. Felix April 19, 2018 at 8:21 am #

    On the empty travel time topic, a few years ago I was looking into minimising that, by, yes, building an app, that uses statistic to guess what the best place to go is after you drop someone off.

    Curiously, this was also included in Uber’s very early stage pitch deck, and given that not much have came of it, I assume that is a bad idea.

    Does anyone know why though ?

    My hypothesis is to the extent there’s a consistent, or obviously predictable demand (end of a concert or a plane landing, for example), taxi drivers “price in” that in deciding where they go.