The Flexibility Industry Discovers the Virtues of Rigidity

An important belated update from the world of ridesharing – Uber is now testing a feature they are calling "Suggested Pickup Points", which directs customers to walk to nearby locations that are easier for their drivers to reach, saving time for both the driver and (in the case of UberPool) for other passengers on board.  Lyft takes this even further, offering discounted rides on its Lyft Line service for people who come to meet it.   

You may be familiar with an identical concept in the public transit industry, called a "stop" or "station" — a location near to destinations, but maybe not at their front door, that is cost-effective for a transit vehicle to reach.  This saves the driver the time it takes to drive to the precise preferred location of each passenger, which is especially crucial if there are other passengers on board whose travel time is also valuable.  (It also encourages a bit of walking where that's easy to do, which is good for you!)

This new feature illustrates how a demand-responsive service like UberPool can evolve to resemble the very fixed route bus that it often pretends to be supplanting, particularly when serving high-volume markets.   Discounts for walking to a pickup point make perfect sense, for the same reason that fixed route transit should be even cheaper; the customer is taking on inconvenience in return for a more efficient transit service.  

These moves show these companies recognizing the geometric logic of rigid, fixed transit: that when you connect places where many people want to go together, along a fast, direct path, the resulting service is both efficient to provide and useful to vast numbers of people.  

Flexible transit sounds like it's more responsive to our needs as customers, but if you want it to be affordable it has to be efficient.  The vehicle that comes to your door is intrinsically a low-efficiency concept when efficiency means "passengers/driver hour", as it will so long as the cost of service is mostly labor cost.  

That's why, for decades, transit agencies have  sometimes deployed flexible services in low-demand but growing markets but then replaced them with fixed routes as demand grew beyond what the low capacity of flexible services could handle.   Transit agencies also know about "fixed stop Dial-a-Ride", which is the specific phase of this inevitable evolution that Uber and Lyft are exploring now.  

The cool kids at Uber and Lyft are showing that for all their pretense of having invented something new, they live in the same geometric and economic space as their ancestors, and will evolve the same solutions that worked in the past.  Data is cool, and technology is cool, and enraptured high-paying customers are very cool, but none of that changes the facts of space, biology and economics, ever. 

Will driverless cars change all this?  Not if we also have driverless buses.  In that case, the math and geometry, and the nature of efficiency, will be largely the same.


26 Responses to The Flexibility Industry Discovers the Virtues of Rigidity

  1. Rob August 11, 2015 at 8:11 pm #

    I wonder about this. I think it’s hard to understand all the dynamics that would have to come to balance in a driverless future. Would large buses still make sense if labor isn’t the cost driver, and wouldn’t large buses still require staffing even if a driver isn’t needed? I’d guess that automated jitneys would take over for group transit (except rail and BRT) and that buses would be less frequently seen. If true, whether there are clear pickup points or a navigable route system would depend either on having a monopoly provider or aggressive regulation – and the whole business model intends to avoid the latter.
    I’m not sure we’re ready for this, or will be before it’s unleashed. That’s the disruptive business model – make facts on the ground before society or competitors can respond.

  2. Willem August 12, 2015 at 6:54 am #

    The beauty of transit comes from it’s ability to aggregate the travel needs of everyone. By transferring a small amount of travel discomfort onto each passenger, the total benefit for everyone can increase. It’s a simple fact that a full bus is more efficiently moving people (while it’s moving) than a car with a few passengers.
    There will always be enough demand between two certain points to warrant having a fixed route between them. I don’t think eliminating the driver has any consequence other than making operations cheaper, and therefore allowing transit to compete even better with other modes of transportation.

  3. Willem August 12, 2015 at 6:56 am #

    I had meant to add – for a more eloquent version of what I’ve said see my post:

  4. Jim August 12, 2015 at 9:47 am #

    I think you’re giving transit way too much credit. Uber and Lyft are not, will never try to be transit or anything like it. It will exceed conventional transit in every single way, feature, and dimension. So they try to be more efficient by getting customers to walk out of an inaccessible area to one that is more auto accessible? Big deal? Obviously, if you’re at Disneyland, the Uber driver will suggest you leave the park to meet up and if you want to save money and time, meet somewhere the driver doesn’t have to cue up for half-an-hour. You can’t take credit for this obvious innovation. If transit will not innovate and create flexible computerized stop-to-stop routing, and there is very little evidence all but say several transit agencies nationwide are considering this, then someone will create intelligent bus service that will replace transit, after transit contractors sue them in all manner and spend millions to lobby against them. Transit agencies are monopolies loathe to innovate. It doesn’t surprise me that they are so defensive and hostile to an innovation that may one day replace them. Instead of innovating themselves, they waste their time pointing out how much Uber and Lyft are borrowing their ideas, reaching so far to the brink of disbelief that it’s almost hilarious.

  5. Paris August 12, 2015 at 10:44 am #
    This is an interesting post that highlights the innovation Uber is thinking through to make their service more reliable

  6. uno August 12, 2015 at 11:05 am #

    Starting to think Jarrett might need to do a post explaining what the word “innovation” means…

  7. Jarrett August 12, 2015 at 12:15 pm #

    Your comment is a reaction to the current culture of some transit agencies.
    I’m talking about the geometric nature of efficient transit.
    Geometry lasts longer than culture.

  8. EngineerScotty August 12, 2015 at 1:45 pm #

    It will exceed conventional transit in every single way, feature, and dimension.
    Uber will have vehicles capable of carrying 80 passengers, or more?
    In some contexts, large capacity may well be a bug–if the capacity goes unused, it’s just wasted space. But in dense urban environments, it is a feature, given that a single bus or train takes up far less square footage than the equivalent number of cars, or even vans.
    And given that dwell times are generally not insignificant, large vehicles operate most efficiently with consolidated stops. To continue the Disneyland analogy, the parking lot tram only makes two stops: the entrance to the parks (Disneyland, California Adventure, and the overpriced shopping mall), and the entrance to the parking garage. It does NOT drive through the massive sprawling garage, getting customers close to their automobiles. You can probably walk to and from your car faster than the time it would take the tram to traverse the garage.
    If Uber were to operate buses (that were full), even if with flexible routing (for an extant example of this, see the “public light buses” of Hong Kong), it would likely still insist that passengers board and disembark at demarcated stops, not wherever passengers may stand or ask to be let off.

  9. Rob August 12, 2015 at 6:49 pm #

    I think labor costs drive transit to use large vehicles. If labor was a much smaller cost driver, many more smaller vehicles could be deployed and operate with some flex to their routing, eliminating many of the customer compromises and inconveniences inherent in bus transit. Also, while vehicle fleet fuel use and emission standards have improved over the years, there’s been little improvement for buses, which still get 3-5 mpg – barely better than the average car on a person-mile/gallon basis in some places. There’s just a lot stronger incentive to innovate for a mass consumer market then for a limited government market for transit buses.
    The reason I think the jitney model is most relevant is because I agree that its just a lot more convenient for customers to got to a set location and expect service with a predictable interval. I’d expect more frequency, but also more route choice, and less willingness to transfer (due to low cost of robot taxis); and for work trips I’d expect very customized routes replacing peak overlay express routes. But – I’m guessing the unexpected outcomes will be more pronounced than any we can imagine in this sort of thought experiment.

  10. uno August 12, 2015 at 8:46 pm #

    Rob, I think you’re conflating what would be efficient for you personally with what would be efficient for everyone in the city/region/whatever.

  11. P August 13, 2015 at 3:31 am #

    This is interesting. You might see the reprivatisation of transit in the future, because the operating costs will become so low, even low density areas would have some service, and still be profitable.
    A city might simply decide to award a private monopoly to a firm to do the city PT network plan and operations, and not pay any subsidy. You might have Google running the PT operations for your city.
    It is clear that Uber and Lyft are on the path to obsolescence. Think about it, all those ‘contractors’ won’t have a job because the car will drive itself. People might not even have their own car, but lease one long-term from a club managed local pool, in a similar way to how people rent houses.

  12. Dave August 13, 2015 at 7:29 am #

    Rob – “I think labor costs drive transit to use large vehicles.”
    What about the actual cost of the vehicle itself?
    A large articulated bus today that holds 100 people usually costs in the ballpark of $1,000,000, which means each space on that bus needs to earn $10,000 in fares before the bus is fully paid for.
    A small pod that holds only 4 people still will need a lot of the same technology that makes the artic bus cost so much (plus I’d argue it will need more if you’re making it driverless). Let’s say it costs about twice as much as a regular car, or $50,000. That means each seat on it will need to earn $12,500 in fares before that pod is fully paid for.
    But both calculations above depend on EVERY SPACE BEING USED ALL THE TIME. That’s unrealistic… and I’d argue that it’s far more likely the pod will have 1-2 riders on it most of the time than the artic will have 25-50 riders on it (because the company who bought it wouldn’t have bought an artic if they didn’t have a fixed route that would fill it up).
    So the pod experience will need to be more expensive than the artic bus experience, to make up for people using it more like a car and less like transit.
    Transit uses large vehicles both because labor costs AND (assuming they are put onto high demand routes) capital costs are lower per person transported with artics vs regular buses. Driverless pods are going to have to make up in price what they lose in volume, if the costs of the pod is to be fully amortized over the lifetime of the pod.
    Therefore, pods will likely be more expensive on a per-person-per-trip basis than transit’s large vehicles, even if both go driverless. This is why I think there will still be a need for large vehicles on fixed routes in the future (i.e. what we call “transit”), even with the advent of driverless technology… the only way I see this not being true is if the future no longer contains ANY price-sensitive people (e.g. everybody can afford the pods).

  13. Jarrett August 13, 2015 at 8:32 am #

    Rob. The logic of larger vehicles is not just labor efficiency. It is also the efficient use of scarce urban space. Eventually, we will have some form of free market for Road space, and in that context the larger vehicle moving more people in less space will prevail. Political demand for bus lanes would achieve more or less the same outcome.

  14. Fraser August 13, 2015 at 9:28 am #

    @P – Your ideas around PT being turned over to the private sector are interesting. On the surface it appears that it would be more efficient for transit agencies to turn over “coverage” routes (all low-density intra-suburban transit) to ride-sharing companies. I would love to see an experiment with some kind of voucher system (especially for needy individuals).
    Your comments about Uber & Lyft’s coming obsolescence only apply to their current business model of having drivers. They are driving the obsolescence themselves, evolving to be that future service provider of driverless transport.
    @Jarrett – It’s pretty optimistic to believe that there will be a change in road pricing. People are very reluctant to give up something that is currently free. If anything, the rise of driverless cars will reduce the need for street parking & free up scarce urban space that way.
    Fixed bus routes are an optimum configuration given an inability to dynamically route. Large capacity vehicles with lower frequency are an optimum configuration given high marginal operating costs.
    One of the biggest trade-offs you aren’t mentioning yet is that the more individuals you have on a vehicle, the more time it spends stopped for loading/unloading (increasing trip time). Alteratively, you can reduce the number of stops which increases the average walking distance to/from the final destination (increasing trip time). With driverless vehicles and dynamic routing, there must be a new optimum somewhere between a car and existing buses. It will be interesting seeing what Uber & Lyft end up settling on.*
    *From a conversation with a friend that works for Lyft, they haven’t tried larger vehicles yet because of questions around vehicle sourcing. Who provides the vehicle, the driver or Lyft?
    Will this impact the driver being classified as a contractor?

  15. Miles Bader August 13, 2015 at 11:24 pm #

    We don’t have to guess about how well “dynamic routing” will offset anything, because we already have an overwhelmingly studied example in the form of private cars.
    The answer of course is that in dense urban environment, they’re an unworkable mess.
    There will almost certainly be specific niche cases of not-so-dense cities where some uber-like service can serve as a replacement for conventional public transit, but in cities where density makes public transit the only viable solution today, it will likely remain the only viable solution tomorrow.
    Driver or no, geometry is pretty hard to escape from.

  16. JJJJ August 14, 2015 at 7:47 am #

    @Jim, no, Uber is not talking about disney, theyre talking about Manhattan. They want people to walk to the Avenue and wait, rather than in the middle of a small one way side street. You know, like buses.
    “Uber and Lyft are not, will never try to be transit or anything like it”?
    Really? Right now when the driver owns the car they wont, but in their automated future, they absolutely will be pushing their car-pooling pricing by buying vans. They will absolutely resemble jitney service, such as in the Caribbean, where the buses and taxis are one and the same. You can offer the driver $20 (aka, uberprivate)and get a private ride for yourself and your partner, or pay $5 and sit around until all 12 seats are occupied (uberpool).
    @Rob, Dave is absolutely right.
    Transit agencies get asked ALL the time why they stick to one type of bus (40 feet) rather than buying 20 and 30 foot models and deploying as needed.
    1) The smaller models are more expensive in the long term. Aside from Robs point, they have much shorter lifespans. When you spend $500,000 on a bus youre also buying 15 years of service life.
    2) Maintenance. Its cheaper to maintain one fleet then a whole bunch of types.
    3) Flexibility. The 20 foot model is great for off peak times and low ridership routes, but if anything changes and you need more capacity…youre out of luck. Meanwhile, the 40 foot model might be at 30% capacity today, but hey, in 3 years a new development is opening up with 500 potential riders and you can carry them!

  17. Fraser August 14, 2015 at 8:49 am #

    @Miles – Apart from Manhattan, the majority of urban trips in NA are still done by car. People PREFER point-to-point on-demand service. In SF right now, every single Lyft Line I take is matched with other people and costs $5 (alone) or $6 (two people). On a few occasions i’ve been matched with 2 other parties. As @JJJJ points out, it’s jitney-esque service that with vans could hit a lower price point ($2.50 – 4).
    There is absolutely a role for traditional fixed-line buses on major arteries that have originations & destinations along their length. There is absolutely a role for subways and commuter rail. What is going on is a disruptive innovation that transit planners should acknowledge will create BETTER service for the end user. I can live without a car because Lyft/Uber give me a) the freedom to reach destinations I can’t easily with transit and b) the option to pay more and get to places faster. I’ve frequently gone to a meeting in a rideshare & taking transit back at a more leisurely pace.
    @JJJJ – there is no logical reason why a shorter bus should necessarily have a (much?) shorter lifespan. If that is the case, it’s a sourcing problem.

  18. EngineerScotty August 14, 2015 at 8:49 am #

    Transit agencies get asked ALL the time why they stick to one type of bus (40 feet) rather than buying 20 and 30 foot models and deploying as needed.
    This ought to be a FAQ somewhere… it came up AGAIN over at PT, someone accusing TriMet of wasting money because the (mostly empty) coverage route to Outer Suburbia is a 40′ bus rather than a van.

  19. JJJJ August 14, 2015 at 11:36 am #

    @Fraser, Im not a bus manufacturer so I dont know the mechanical/logical reason, but I know when a transit agency goes out and buys those minibuses they only last 3-5 years while the full size buses can last up to 20 years. From what Ive heard they literally fall apart.
    As it is not cheaper to buy 4 minibuses to serve the life of one big bus, they dont do so. Thats why as Scotty pointed out, those suburban feeder lines with 3 passengers an hour use the big buses. The exceptions are when the transit agency has their hand forced by politics (penny wise, pound foolish) or actual physical constraints (bridge with low weight limit, narrow streets, hills)
    Could they build a better minibus? Maybe? I dont know. But assuming they can, they probably dont because its a chicken egg demand issue. Too few agencies buy them to warrant the R&D needed to make them better so more agencies buy them. Im just speculating.

  20. Miles Bader August 14, 2015 at 9:17 pm #

    … and the result is an unworkable mess. >< In car-oriented American cities road congestion is a serious issue, and attempts to address it by endless rounds of road-widening have not proved to be very effective (and have had serious deleterious effects). Obviously this is related to density, with denser cities being easier for mass-transit to cover efficiently and at the same time exacerbating the problems of cars. You can draw the line wherever you like, but my point is that one doesn't need to guess about the effects of using custom-routed individual transit, the answer is already present. Of course another factor is trends in how cities are developing, with some factors encouraging sprawl and lower density, and others encouraging density. Mass-transit can play an important role here as well, as the natural effect of aggregated routes is to encourage aggregation of commerce and residences as well. Preferences for density (or lack of density) encompass a lot more than just ease of transportation, so it's quite reasonable for somebody to prefer mass transit even if it’s “less convenient.”

  21. Dexter Wong August 17, 2015 at 9:55 pm #

    @EngineerScotty, You say that the Disneyland tram only stops at the entrance of the parking garage? That’s a change from my experience. Back in 1968, my family visited Disneyland and stayed at a motel across Katella Ave. from the park. We were advised to ride the tram to the edge of the parking lot and then cross the street. (In those days there was no garage, just an immense parking lot.) Well, after a long ride, we exited the parking lot and reached the sidewalk, where we saw the motel directly across the street. But, the traffic was too busy to jaywalk, so we had to walk a long block to the corner, wait for the traffic light and then walk the long distance up to the motel.

  22. EngineerScotty August 18, 2015 at 11:16 am #

    Last time I was at Disneyland (about 3 years ago), the tram ran between the entrance plaza and the pedestrian entrance to the garage, where all the escalators and elevators converged. It did not enter the garage.

  23. Fraser August 18, 2015 at 2:08 pm #

    @Miles you are ignoring (true!) ride-sharing. @JJJJ and I have been talking about multi-party rides, extrapolating to jitney-esque services. In combination with dynamic routing, they offer faster point-to-point service at a price approaching that of public transit.
    You are conjuring your own straw-man in car congestion. Ride-sharing, like Lyft Line and UberPool, takes vehicles off of the street (2+ parties in 1 car). In addition, it reduces the need for parking as people give up car ownership (like I have). Increased density only make these services even more efficient (higher multi-party match rates with more optimal routing).

  24. Colin August 26, 2015 at 2:45 pm #

    My home city of Ottawa Canada experimented with dial-a-bus service in the 1970s when service was being extended out into the low-density suburbs. All of it was later converted to fixed routes.
    In most of the communities, a typical 30-minute cycle might look like this: 10 minutes to travel from the nearest major station to the community, 10 minutes to circulate through the community, and 10 minutes to return to the station (or continue to a different terminal at the other end of the community).
    In most communities there was enough of a collector road network within the community that most residents could be served reasonably well by travelling only on those streets. Thus it is easy to see that the service would be faster and more efficient by travelling only on those streets instead of circulating on all the other back streets. The only way that the dial-a-bus would be more efficient is if you could service two communities with one bus (or 3-for-2) rather than each community having its own bus.
    I wrote “most residents” above. Transit agencies have to accept the reality that there will be some people who are beyond the walking distance target, and a handful who will be a very long distance away from service. In many cases it would literally be less expensive to provide those people with a conventional taxi ride than to run routes that would provide total coverage.
    The fixed routing also means a fixed timetable, which is more convenient for passengers who don’t have to book a trip an hour ahead and then have a half-hour pickup window, wondering when the bus will arrive.
    A decade or two ago, ideas were floating around again to replaced conventional buses by dial-a-bus during low-demand times (evenings and Sundays, for example). The analysis showed the same calculations as above, but another aspect emerged as well. If the ridership was low enough that conventional service would be cancelled, then it was also low enough that it would not meet minimum standards for r/c ratios even if using taxis. A taxi only accommodates a small number of passengers, so if your ridership is above that level (even if only occasionally), then costs rise.
    Many cities do offer taxibus services. It all depends on the context of your urban design and your service standards.
    The fixed-route multi-passenger trips on Uber/Lyft are starting to sound an awful lot like a jitney service. The history of jitneys in North America is that they tend to concentrate on higher-demand corridors, siphoning off passengers who would be waiting at a bus stop anyway. On a societal basis this is not a good use of resources.

  25. Colin August 26, 2015 at 3:15 pm #

    Regarding the use of smaller buses:
    Lifespan is one factor, as buses <25 passengers tend to be conversions of trucks or vans that don't have the durability of a standard transit bus (12-18 years at 12-18 hours/day). Some bus manufacturers do produce 30' and 35' versions of their standard 40-foot buses, with performance and durability similar to the 40 footers.
    But there is more to it than that.
    The operating costs of a 30' bus are not much different from a 40' bus, since the driver, tire usage, brake wear, etc. are all very similar. There may be small savings in fuel consumption. Capital costs may be similar since the 40 footers have the benefit of mass production even though they require a bit more material to build.
    Here are some areas where the small buses really work against you:
    (1) You have to be able to use the small bus all day. If there is one trip at 4:00 pm that gets overloaded due to a school (or factory) letting out, you either have to substitute a larger bus or add a second bus for that one trip. Your savings just became net costs.
    (2) If you can use smaller buses only during evenings and weekends, then you have to buy, store, and maintain two fleets of buses: large buses for the daytime usage, and small buses for the other times. Not efficient at all. Plus, you have to pay to deadhead them to/from the garage twice as often (for evening service).
    (3)If you only have a few routes that can use the smaller buses, you may lose interlining efficiencies. For example, you may have a route that is 25 minutes long and one that is 35 minutes long (I'll ignore layover for the sake of simplicity). You can interline those two routes and run half-hourly service with two buses. If one route requires a larger bus and one doesn't, then assigning a smaller bus means that you will need three buses to operate the service (one on the 25-minute route, and two on the 35-minute route, with long layovers=inefficiency).

  26. LucaGuala September 22, 2015 at 2:59 am #

    Very few trips really require a door-to-door service: disabled persons, hauling heavy luggage etc. In most cases, we can walk a little distance both at the beginning and at the end of the trip, if this is convenient for any reason. This will increase the efficiency of the motorized travel service that we use: public transport, Uber, parking location, even a friend giving us a lift (“I’ll pick you up at the corner of the main thoroughfare, wait for me there”). It will also contribute to keeping local streets free from traffic.
    So, most residential streets, and also many streets in mixed use areas, could be completely pedestrianized, safe for allowing access to few authorized vehicles. In a community where most people can walk and do not normally carry very heavy loads, pedestrianization of local streets should be the norm, and not the exception. So why is it not so?
    Because, given the opportunity, no one will refuse to be picked up at their front door, or to park their car right in front of their house or office. The advantages of doing this are quite evident. But also the disadvantages of allowing motor-vehicles to use each and every steet of a town for travelling and parking are clear and well known.
    What is often misunderstood, is that while the advantages (or benefits – let’s start talk economics) mostly accrue to to who does the action (park their car in front of their office, for example), the disadvantages (or costs) are distributed among anyone passing by (and being disturbed by the presence of the parked car).
    We all bear the same disadvantages (pay the same costs) because each one of us bears the sum of little parts the cost of the action of every other person who does this selfish action (unless I am the only one to park my car in front of my office, and everyone else leaves it at the parking lot 1 km away).
    this is a typical situation of “market failure” best known as “the tragedy of the commons”: the person who performs an action which guarantees a personal gain and distributes the costs of this action among a community, still has an advantage in doing it even if the sum of the costs is much higher than the gain. The bigger the community over which the costs are distributed, the greater can the difference between individual gains and social costs.
    In short, I have convenience in doing an action that makes me gain 1$, even if it costs the community 100$, provided that the costs are distributed among at least 101 persons. Of course if all other 100 individuals think and act as I do, they will do the same “convenient” action that I did, dumping 99% of their costs on me and the other 99 persons. The result will be that the overall benefit for all the individuals will be of 101$ while the overall cost will be of 10100$. Each one of us is losing 99.99$! The bad thing is that we all think that by behaving this way, we are doing our good, and will continue acting this way, unless we agree all together to stop, or some superior authority, acting on the behalf of the community, prevents us from doing it. After all, what do we have to gain by being the only ones not doing it? If I don’t park my car under my office, someone else will do it in my place. I will pay a share of their cost, but not get any benefit. I will get out of home a little earlier instead, so that I will park in my favourite spot before anyone else does it.
    If all this looks familiar, it is because it is. It happens continuously,everywhere and in every context: traffic congestion; occupation of public space for parking; forest cutting; overgrazing; depletion of natural resources; committing crime; mafia feuds; buying lottery tickets; standing up at a football stadium; raising your voice at a cocktail party, telling lies; occupying the toilet for longer than reasonable…