The Fantasy of “Service to Your Door” in Dense Cities

Customers love our new invention!  You have to start listening to the customer!  How often have you heard this line as though it ended any argument?  I certainly hear it all the time as an explanation of why “service to your door” will sweep away large parts of the fixed route transit industry.

The answer is:  People want all kinds of things that they can’t all have, because those things just too expensive per customer to provide.  Wealthier people can have them, but the tastes that wealthy people can afford are a terrible guide to what will work for everyone.

A great example is “service to your door,” when applied to dense cities.  There is a different issue when applied to suburbs, to which I’ll return in another post.

As far as we can tell, neither Uber nor its competitors can make a profit, even though they focus heavily on dense cities where the geography is most favorable to them.  Startups have lots of good arguments for why we should wait a while for them to be profitable, but Uber is running out of them.  We are not waiting for Uber to scale up; it is already huge.  We are not waiting for it to become more labor-efficient; it has already squeezed labor so hard that it can’t retain drivers.  We are not waiting for more efficiency in communications; the app already works fine.  What are we waiting for?

As Len Sherman argued in Fortune recently, the real answer is simpler.  Urban transportation is just not a profitable business.  Transit isn’t, and taxis and taxi-like services usually aren’t either.

But transit is supremely efficient at one essential thing: it uses scarce urban space efficiently.  By contrast, “service to your door” is becoming a new way to strangle our cities with congestion. Congestion is a spatial problem; it will still be there in a coming age of automation.

So yes, everybody would like to have service to their door.  But the true price of that, in dense cities, is likely to be something that only relatively wealthy people can afford.  Pre-automation, labor is an irreducible cost.  Post-automation, in dense cities, there will still be the problem of space. Uber and Lyft are already increasing traffic in dense cities that don’t have room for it. If they suddenly become cheaper, the resulting induced demand would be the death-knell for the functioning of cities.

To its credit, Uber understands that only road pricing will solve this problem even in the post-automation world.  This, of course, would push the price of their service back up, and thus out of range for many people.  But that would indeed be the true price.  Which is why the “service to your door” fad must not be allowed to undermine fixed route transit systems that can work for everyone because they use space so efficiently.  (Post-automation, too, we should also think of autonomous taxis competing with autonomous buses, which would be vastly more frequent than buses today.)

Advertising glorifies the tastes of the wealthy, not just to sell to them but to help less wealthy people form unrealistic tastes. “Service to your door” is yet another example of that kind of marketing.  And whenever we are told to design things around technologies that only the fortunate can afford, we’re being asked to make a mistake called elite projection.  Cities do not work for anyone unless they makes room for transportation that works for everyone. So they must be designed around what works for everyone.  They must also be designed around solutions that are financially sustainable, which “service to your door” — when properly priced to account for its inefficient use of street space — is probably not.

But is “service to your door” relevant to suburban needs, or to the distinctly suburban “first mile last mile” problem? I’ll cover that in an imminent post.

22 Responses to The Fantasy of “Service to Your Door” in Dense Cities

  1. Clicky Freewheel December 29, 2017 at 11:03 am #

    I have cheap service to my door — it’s called a bicycle. ?

  2. Asher December 29, 2017 at 11:52 am #

    UC Davis has a research program devoted to shared, automated electric car service. It’s pretty clear that ride-hailing firms will make this the mainstay of their service in cities – but will a shared car be high capacity enough to avoid both congestion and high congestion pricing? Under what conditions would that be more or less efficient than automated transit?

    It will probably hinge on factors like population density, typical travel distances and road space. Of course, the more we value urban space, the more we should rely on high capacity mobility to free up space for other uses.

    • Jarrett Walker December 29, 2017 at 12:09 pm #

      Again, assume we’ll also have electric buses.

  3. John Wall December 29, 2017 at 12:49 pm #

    For Uber question, What are we waiting for?
    We are waiting for autonomous cars. that will change the game, and it will make it a profitable industry. Then let’s see how public transit respond when Uber own autonomous buses with autonomous cars plus the algorithm and software that operate them.

    • Sailor Boy January 7, 2018 at 10:51 am #

      “Then let’s see how public transit respond when Uber own autonomous buses with autonomous cars plus the algorithm and software that operate them.”

      Uber could buy buses tomorrow and compete against human driven buses and cars using human driven buses and cars. How would making them driverless change anything?

  4. Ben G December 29, 2017 at 1:36 pm #

    John above: “We are waiting for autonomous cars. that will change the game, and it will make it a profitable industry.” —— based on what? Because somebody said so? Or imagines so? And if Uber becomes profitable, so what? If the choice is paying $15 for an Uber ride, or $2.50 for a bus ride to the same destination, a person willing to pay $15 probably isn’t a core transit rider anyway. If anything, Uber is primarily competition for taxis, not public transit.

    I’m so sick of this autonomous vehicle baloney. We have the tools to solve every public transit issue RIGHT NOW. Look at cities that get it right, like Tokyo. Amazing service, clean, reliable, affordable, better than driving a car. Public transit doesn’t suck inherently. It sucks (especially in the US) because we underinvest in it, we don’t set aside proper space for it (like exclusive lanes for BRT and trams), and we prioritize private car travel to an insane degree. Yet the solution is to wait for some unknown future date where private autonomous vehicles will magically save transit? Just because cars will be driving themselves does not change the spatial and mathematical constraints involved.

    And to tackle the “last mile” problem, as another commenter above said: BICYCLES. Why this simple solution is still such a mystery in the vast majority of the US, and done so well as a complement to public transit elsewhere in the world… is truly a mystery. For some bizarre reason I will never understand, Americans love to stare at boring roadway for hours on end instead of kicking back and reading a book on a train.

    • John Wall December 29, 2017 at 5:07 pm #

      Well, if you cut the human uber driver, then the autonomous car will drive more efficiently, and effectively. The profit margins can be huge since the operation cost will go down (they only need a machine, and a software only). Also, the cost for consumer would be less with algorithm and membership subscribers.

      Once these companies figure out the secret sauce “algorithm for how people move”, then the same thing can apply to autonomous bus. But, unfortunately, these companies will compete against public transit and not helping them . They will start to accuired big autonomous busses, and do both. With big and small cars, they can transform the public transportation to private one. These companies now own a huge amount of data which make it easy for them to predict the future. Meanwhile, public transit own non accurate data or bad data. It’s sad when you see some transit don’t even know exactly how many riders they have or where they coming from. That why many transit agencies can not tell you why there is a ridership decline or where the ridership gone.

      • MaxUtil December 29, 2017 at 6:03 pm #

        There isn’t really some “secret sauce” for how people move waiting to be discovered. We know that already. It’s important to remember that Uber, et. al. are just taxis. There’s no fundamental difference. The reason Uber loses money is because they charge much less than a regular taxi. Uber is using investor money to subsidize low per ride prices in order to build up business. They’re not doing anything fundamentally different.

        If/when autonomous vehicles become available, it will change the cost structure for everyone to some degree. Taxis/Uber may become cheaper and this may affect transit. But it will also help reduce costs for transit as well so agencies may be able to provide better service at the same budgets. But if taxis become so cheap that large numbers of current transit riders switch, there will be huge impacts on traffic, congestion, etc. There isn’t really that much evidence that autonomous vehicles will move through dense cities in any more “efficient” a way that would allow for higher throughput of vehicles. Under certain, simple controlled conditions they may be able to achieve higher throughput. But in dense cities again, they will mostly just add to existing congestion.

        Autonomous vehicles may have a long term impact on the cost of certain types of rides and may change how we own/operate vehicles. But they won’t change the geometry of cities. At least not until they have been around long enough to fundamentally change development patterns.

        • John Wall December 30, 2017 at 8:14 am #

          How people move = customer profiles data = secret sauce

          If you know that I go from work everyday from 10 to 2 not 8 to 5 and I stop between to get my coffee. Also, I attend all clippers games except the one on Monday because I have to take my kids from school. Then you know how I move. If you could cluster other profiles with other patterns, then you understand how they move and you can predict their future trips. So basically more data you get, the algorithm will know more about how people move in details. The public transit failed to do this because they don’t have any profile data. They don’t know what their customers like or dislike, or who pay or ride for free.

          • Crispin Cast-Nine December 30, 2017 at 11:23 am #

            Public transit agencies definitely collect information about boardings across the system, and I’m not convinced what private transit companies collect is much more valuable. If you look at the routes for private transit like Chariot in SF, it’s largely paralleling corridors like Geary/Van Ness that the city is planning to upgrade or is currently upgrading.

          • Justin December 30, 2017 at 12:37 pm #

            That’s what a smart card (or just a driver with a “fare paid” button) is for – all you need to know in order to optimize the network is where passengers get on and off, when they start and finish, how long their trip takes. What they do when they get off is not only irrelevant, but invasive.

          • Georgist Economist December 31, 2017 at 8:57 am #

            There are several problems with that. You are basically advocating the benefits of overfitting.
            1) The data about people who are already your passengers doesn’t necessarily tell you anything non-obvious about the people who aren’t already your passengers. I mean, the first group will tell you your service is very useful; try to generalize from that, and you’re in for a surprise.
            2) You basically want to make a really accurate model of when each person will want to move where. The problem is, the prediction of summed movements you get that way is not _that_ much more accurate than saying that “lots of people want to move between dense residential areas and dense office&commercial areas”. The extra effort does get you something, but isn’t worth the hassle.
            3) The more detailed the models, the more prone they are to break when you generalize onto the people who you haven’t gathered any data from yet. If your early service is quite limited, both of the following models will be good fits for the data: I. “people want to go where many jobs are”, II. “people want to go to [this, that, yon] office building”. Which one will break when service is extended?
            4) This very detailed per-person model can break down quite easily over time. Joe Bighead got a promotion, so now he commutes to work one hour later. “So what,” you could say, “he’s just one rider out of 100,000 who travel that way in the morning peak.” Which is exactly the point of 2).
            5) In slightly more formal terms, there are so many passengers that their aggregate trips can be modeled pretty well with a smooth distribution plus some noise. In which case, you only need a handful of points to figure out a good enough approximation.

          • Michael Robinson January 1, 2018 at 1:14 am #

            In addition to trip information from smart travel cards, transport agencies can also track passenger movements through WiFi. Your smart phone doesn’t need to be registered to a WiFi base station, the base station can detect the phone as long as the WiFi is switched on. London Underground have started doing this – while they know the entry and exit station from the smart card, they don’t know the route the passenger took between these points and WiFi data can provide this.

            Additionally it provide more data about passenger movement through the system which can be used to target and price advertising displays.

    • Darius January 9, 2018 at 8:47 pm #

      It’s cultural. In the US the unspoken paradigm is that transit is supposed to suck. It’s so ingrained to think of a car as a vital organ that most people don’t even realize it. They just make decisions based on that thought process. That’s why the US does transit poorly with some exceptions such as where Jarrett works. Transit is just for poorer darker people. Isn’t it?

      • Georgist Economist January 15, 2018 at 6:49 am #

        No, it is simply a severe case of stupidity. Once cars became widely available, in the US freeways were built through cities. This opened up an enormous amount of land around the city, from which the city was reachable by car. The population spread out—this might have been partly intentional, to make it harder to nuke the population, but talk about bad reasons to screw your cities—, and now America is stuck with urban sprawl that is unservable by transit. It is only natural that now, when the vast majority of the population lives in car-centric areas, more attention is paid to car-based mobility than to transit. But it is to some degree accidental that mostly the poor (and disproportionately dark) people were left in the area where transit is still viable. I don’t mean that they could have been the ones moving into the suburbs, but the suburbs could have been, and there HAVE been, streetcar suburbs, where the core ridership of transit lines were middle-class whites.

  5. Alex December 29, 2017 at 7:05 pm #

    These companies only have interest in maximizing profits for their shareholders. There is no reason to believe they will keep costs lower than taking the bus.

  6. Steven December 30, 2017 at 10:27 am #

    I wouldn’t mind a drone to deliver my goods to a window in my apartment on the 20th floor.

    • Georgist Economist December 31, 2017 at 4:09 pm #

      Airships would be fine, but heavier-than-air vehicles get severe congestion. Because they fly by putting their weight onto the surrounding air, they create a wake of turbulent and downward-moving air; any aircraft that flies into this wake will tumble and/or lose altitude, possibly crashing in doing so. It’s a bit similar to how, at highway speeds, cars need far more space than their physical size.

      And that is beside security problems. Some guy flying a crosswind kite, or throwing a few spools’ worth of fishing line off some skyscraper, could probably down a freight drone (or a passenger drone, as some people like to imagine). Which would fall onto whoever happened to be standing at the wrong spot underneath. Luckily, congestion in the sky would mean that there wouldn’t be several layers of vehicles that would turn into a sort of avalanche; still.

      And according to this article (I couldn’t post the link because the comment was refused as spam), even if congestion wasn’t an issue, economically, drones wouldn’t happen in cities but in rural areas.

      • Eric January 11, 2018 at 3:24 am #

        “Some guy flying a crosswind kite, or throwing a few spools’ worth of fishing line off some skyscraper, could probably down a freight drone (or a passenger drone, as some people like to imagine). Which would fall onto whoever happened to be standing at the wrong spot underneath. ”

        Or he could drop a rock out his window, and also kill people.

        (Personally, I wonder how the new condos have balconies on the 40th floor. They aren’t worried that a kid someday will drop his metal toy train over the edge and kill people on the sidewalk below?)

  7. asdf2 January 6, 2018 at 8:59 pm #

    If “transit to your door” is intended to be a shared ride, rather than like a taxi, it quickly becomes like one of those airport shuttles, where you painfully realize just how slow detouring to everybody else’s front door on the way to your front door actually is. In many cases, an airport shuttle can actually be slower than riding a plain-old city bus that stops every other block – even when riding the bus means transferring to another bus – if you have the bad luck of everyone else in the shuttle van getting dropped off before you do.

  8. calwatch January 10, 2018 at 7:20 pm #

    I would like to see you talk about service to your door in a suburban context. For example in your work at VTA, the board ultimately adopted a plan that withdrew service to many outlying neighborhoods. Couldn’t you withdraw more service – say go to the higher end of ridership vs. coverage splits – and let ride sourcing handle areas and time periods that can’t support even a bus every 30 minutes? You would have a core network of buses that would run, say, no worse than every 20 or 30 minutes, and a skeletal network to key outlying centers, but outside of those areas, go to ride sourcing, but only to the backbone or skeleton.

    • Eric January 11, 2018 at 3:29 am #

      About that he wrote:

      “But is “service to your door” relevant to suburban needs, or to the distinctly suburban “first mile last mile” problem? I’ll cover that in an imminent post.”