The Receding Fantasy of Affordable Urban Transit “To Your Door”

Uber’s UberPool service, which attempts to gather multiple people on a single vehicle going the same way, is undergoing some tinkering that will make it even more like fixed route transit.  Andrew J. Hawkins at Verge has the story:

[UberPool riders] are being prompted to walk to the closest corner or intersection for more convenient pickups, rather than have drivers deviate from their north-south route.

The same goes for drop-offs, where riders are being let out at a proximate corner rather than the exact address of their destination. Uber calls it “dynamic drop-offs,” but the result is pretty plain. If you want those cheaper fares, you’re going to have to be cool with a lot more walking. Uber began testing this feature last year, and has since rolled it out in wider use.

Does the difference between these images remind you of anything?

Walking further for more direct, useful, and affordable service is the basic deal that fixed route transit has offered for more than a century.

What’s more, if you walk to the bus instead of to UberPool, you can get on any bus instead of waiting for your specific UberPool to arrive.

Yet this is exactly what Uber must do to make their UberPool less unprofitable.   As we’ve explored many times here, demand-responsive service is wildly inefficient.

UberPool would be less absurd if we were talking about somewhere other than Manhattan, or any other big city that’s rich in frequent transit.  In places with less transit, this concept could have some use.  But in big cities it’s clearly converging on something for which fixed route transit is already the ideal tool.

Now, New York City bus service has some problems, especially in the delay-ridden way that they handle fares.  I am not defending specific practices of any agency.  But the geometry remains what it is.  If you want affordable transit service, you’re going to have to walk to it.  That’s the math that makes fixed route service inevitable.

Transit people all over the world have understood this since long before Uber’s CEO was born.  They’ve also known about the concept of demand responsive transit, which means “transit that is extremely inefficient because of the degree to which it deviates or circulates based on the needs of a single person.”  Demand responsive service is so inefficient that it arises only in these contexts:

  • extremely low-wage environments, as in parts of the developing world, or:
  • focused on elites who can pay high fares, as in typical Uber or Taxi operations, or
  • for special-needs groups, such as the disabled and seniors, or for other groups that have the power to demand it, always at astronomical subsidies per rider.   (Costs per rider for this kind of service, called paratransit in North America, typically run about 10 times the rate for well-designed fixed route service.)

All the new apps have helped smooth out inefficiencies of communication, but they will never change the math.  Technology never changes geometry.

As UberPool gradually discovers this, the question becomes: Did Uber, and similar companies, really invent anything at all?  They invented apps, and algorithms, but do they have any new answers for the geometry problem that is urban transportation?


(Yes, I said almost exactly this in a post two years ago!  But one needs repetition to break through all the noise)  


51 Responses to The Receding Fantasy of Affordable Urban Transit “To Your Door”

  1. Zagster May 23, 2017 at 11:39 am #

    Multi-modality for first mile/last mile will bring transit to your door, but anything better than the informal 1/4 mile standard should be a premium paid 100% by the rider (with exceptions for rural communities, disabled, and NEMT).

    It would be interesting to see the data on UberPool usage along standard fixed-route lines in urban areas since it is really just offering a slight quality improvement at a massive increase in price. Will Uber passengers realize that they could take the same route on a bus for a fraction of the price? And, knowing that, would they do it?

    • Tom May 24, 2017 at 2:45 pm #

      Why do you say massive increase in price? There are several competing services in NYC that charge $5 for pool rides. The subway/bus costs $2.75. Both can be paid for with pre-tax money.

      • drs June 10, 2017 at 9:31 pm #

        I’d call nearly 2x a massive increase. Yes, it’s just an extra $2.25 absolute, but done consistently it would double your transportation budget. For… what? Coming at about the same time, to be stuck in traffic similarly?

        • Ian Mitchell June 12, 2017 at 4:34 pm #

          I think that people may have some degree to which they value their comfort.

          If the MTA charged $5 for guaranteed seats on buses and $2.75 for no seat guaranteed there’d probably be a market for it.

          Instead, the people who want to pay an extra 2-3 bucks for a seat are in a different vehicle entirely. That is arguably a worse outcome for everyone.

  2. Andrew May 23, 2017 at 2:31 pm #

    Only in the most perfect cases is transit within a block of your start and destinations. Then there is still a limit on what is a one seat ride from you. I am super pro-transit and do just about everything by foot and transit, but I am for having different options at different price points. There are times when we are at a Comedy Club at 2am in Brooklyn, tired, had a few drinks, I’d like having a mid-level option like UberPOOL that can get us home with a one seat ride even of it’s a block walk, than wondering if the bus stop 6 blocks away that is heading home runs at 2am, then transfer to a subway to Manhattan, to another bus to Hoboken, with 20m waits in between, when all you want to do is get home and crawl in your warm bed. Meanwhile, I feel like the author is saying UberPOOL offers nothing new above transit and therefore shouldn’t exist.

    • Alon Levy May 23, 2017 at 3:08 pm #

      Brooklyn-Hoboken is a problem of poor interregional transit connectivity. It can and should be solved with new tunnels leading to an integrated regional rail system; see this example.

      • Ian Mitchell June 12, 2017 at 4:35 pm #

        Those are necessary and good projects- but they’re going to take a while, even if we broke ground today.

    • h st ll May 25, 2017 at 5:53 am #

      exactly! you should start your own transit blog. You just demolished his argument

      • George Lane May 25, 2017 at 3:13 pm #

        This is exactly what Jarrett is already saying though. UberPool is perfect for somewhere (or some time) with low density of trips, and where users are willing to pay a premium.

  3. Adam May 23, 2017 at 5:27 pm #

    Is your assertion that there isn’t a profitable version of Pool that is more convenient than public transit? That seems very unlikely to be true.
    If the assertion is that at the price point of public transit you can’t do Pool, that could be true, but it also doesn’t seem to be what Pool is trying to do.

    It doesn’t seem strange at all that in the trade-off of convenience vs money there could be a profitable middle ground between a taxi and a bus.

  4. smithcorp May 23, 2017 at 8:35 pm #

    I think he’s saying that as the Uber ride-sharing model develops its offer, it starts to look more and more like regular, good transit planning; and this highlights that good transit planning principles are robust and the key to delivering good mobility in cities.

    My worry with the Uber model is not that it will replace transit, (because I think it will inevitably morph into transit in cities) but that it will drive down transit driver rates of pay. Most of the cost of a bus service for instance is the cost of the driver. Using smaller vehicles only results in a small cost saving and quickly runs into capacity barriers. The promise of autonomous buses seems pretty great – the reduced operating costs can be reinvested in more frequent and/or cheaper services with benefits for customers.

    But, in the interim, I worry the private transit model will have to cut driver wages to compete with the ubers; and uber-style services will cherry-pick the most profitable routes and times of day.

    • Ian Mitchell June 12, 2017 at 4:42 pm #

      I don’t see why that’s a bad thing. When someone who is driving for an hour to let the traffic die down after work, to save up to finish his basement into a mancave (my last Lyft driver’s story), that doesn’t demand a 100K salary+benefits package.

      Perhaps it’s better to transition people out of driving as a career due to its crowding out by the gig economy, as that will reduce the degree of the impact when those responsibilities are automated entirely.

      Automation will be putting a lot of Americans to better use- even if that use is on the couch rather than developing repetitive strain injuries, patronizing flying J’s, and occasionally decommissioning pedestrians.

      Truck Driver is the most common profession in 30 out of 50 states- we can do better.

  5. JBM May 24, 2017 at 5:16 am #

    You’re totaly right, M. Walker. However, Uber does something public transport agencies don’t: they brand their operational constaints (“dynamic drop-off”)! I’m a bit ironical…

  6. Joe May 24, 2017 at 6:38 am #

    Funny, Uber’s maps appear to have their cars going the wrong way on one-way streets!

  7. Alan May 24, 2017 at 11:01 am #

    While this is a valid criticism of demand-responsive “Uber transit” in the densest urban areas today, I don’t think that the current Uber/Lyft model has ever really been the main threat to transit systems (and never in Manhattan). Rather, it’s the prospect that autonomous vehicles would re-create the extremely low-wage (indeed, *no*-wage) environment that allows demand-responsive transit to flourish in the developing world.

    At a certain price point, autonomous taxis will destroy the financial viability of fixed-route mass transit systems in all but the dozen or so densest U.S. metros – and would severely curtail the scope and reach of transit systems in the outlying metro area even of those cities, which poses political risks to those systems as well. It is by no means certain that autonomous taxis will actually come in at that price point. But *that’s* the existential threat to transit in most of the U.S. – not the Uber and Lyft driver, but the taxi with no driver.

    • Sluggo May 24, 2017 at 1:23 pm #

      If taxis can be autonomous, so can buses.

      Also, it’s more than a dozen metros that have at least some corridors that require more capacity than ten 5-person cars. And those five cars take up the road space of three buses. Everywhere congestion is a problem, road space is an issue.

      • Alan May 25, 2017 at 11:26 am #

        But if taxis *and* buses are both autonomous, the cost differential between them still closes tremendously. The labor cost of the driver is an important component in both modes, but it’s a larger cost component in private demand-responsive modes (like a taxi) than in a bus. That’s the real problem – if autonomous taxis provide a better functionality than a bus, then all it takes is for the price to get *close* to an unsubsidized bus fare for those taxis to eat away at the transit system. Either directly (by taking riders) or indirectly (by enticing the transit authority to subcontract out service to the taxis rather than running routes).

        In most of the U.S., even where congestion is a problem, there’s just so little transit ridership that the shift from transit to autonomous cars will have a trivial effect on the amount of roadway congestion. So while I certainly agree that this geometry issue is the reason why you’ll always have significant mass transit in Manhattan, I don’t think it applies in most other U.S. metros – and not outside the urban core during peak times.

        And if autonomous taxis do a better job for all the other trips *other than* the urban core during peak times, how long can transit systems persist outside a handful of U.S. cities? And how could they ever *expand* from those core service routes?

        • el_slapper May 26, 2017 at 1:03 am #

          Alan : there is not only the financial cost : there is the cost in terms of space used, which has a tremendous impact upon congestion.

          In other words, if you leave the keays to a financial guy that looks only at the bottom line of the transport side, you’re right. OTOH, if you’re a responsible local leader, and try to get more ROI from your roads – i.e. less congestion for more people tranported, then the automated buses are suddenly very attractive.

          When calculating a ROI, always ask yourself for who is the ROI. The transport company, the city that builds the road, someone else? because the optimal solution will be radically different, depending on for who(and for what) you are optimizing

          Of course, I’m biased, I do live in ol’Europe, in a town built in the medieval ages, and congestion is much quicker to happen here than in a US town with infinite low-density areas. Still. Where density is high, my point remains(and it’s also Sluggo’s and Jarrett’s). You need to get more from your infrastructure if you want to develop your town. More people per key route. For which buses are far better than cars, even opimized ones.

          • Alan May 26, 2017 at 7:44 am #


            There is a cost in terms of space used – but that’s not especially relevant in nearly all U.S. cities. Most of our metros have minimal transit usage (less than 5% of even commuter trips, far less than that modal share for trips over all). You could eliminate transit in most of our cities, and the delta on roadway traffic would fall within the normal day-to-day variation in traffic usage. That describes all but maybe a dozen U.S. metro areas.

            Even in cities where there is somewhat more transit use, it’s concentrated in peak-hour rides to the central business district – the rest of the network has modest ridership that could be moved to autonomous taxis with minimal impact.

            We see this in DC, actually. You’ve had an *enormous*drop in transit ridership – but just like we haven’t seen “carmageddon” when various roadway networks shut down, we *also* haven’t seen carmageddon when you had a reduction in transit usage.

            So if you can have autonomous taxis at a modest cost (by no means assured), it has the potential to directly displace transit in all but a handful of cities. In a few of the cities that have decent peak-hour CBD ridership, autonomous taxis might *still* upend the transit network by eating the rest of their ridership – and it’s not clear that such transit networks can politically get the subsidies they need if all their feeders and suburban routes go away.

        • RossB May 27, 2017 at 8:22 pm #

          >> if autonomous taxis provide a better functionality than a bus,

          But that’s the problem — they don’t. The only way to provide premium service (i. e. a cab that arrives very quickly and takes you directly to your destination) is if you pay more than the average person will pay. There is a limited amount of space — if you allow cabs everywhere, you soon clog up the streets, and no can get anywhere. So cities limit the number of them (and have for years). With a limited resource, you either run it efficiently or run it as a premium service (for those wealthy enough to pay). What is not obvious — but a key part of this discussion — is that running it efficiently means running it like a bus. You want the vehicle to be picking up and dropping off as many people per second as possible. This means a cab doesn’t wait at the corner, ready to pick up a fare. It doesn’t go down empty dead ends. It picks up people who happen to be “on the way”. It runs at a very frequent, regular time, so that riders don’t have to wait very long. You know, like a bus.

          • Alan May 30, 2017 at 7:15 am #

            RossB – they *do* provide better functionality. The question is whether that functionality can be provided at an affordable cost, which is a premise of the debate. Cities generally don’t limit the number of cabs because of space concerns, but for primarily for economic and/or safety reasons.


          • RossB June 1, 2017 at 5:20 pm #

            Check out this bit of history from, which I will quote:

            In just a matter of years, almost every city re-regulated its taxi market because:

            Traffic in high-density areas increased due to an influx of cars on the road and more “cruising” activity.

            End quote. It really isn’t that hard to understand. The *only* reason that taxi cabs are superior to transit is because they are limited. That’s it. Or, to put it another way, point to point service simply doesn’t scale. Take away the driver, and driving on rush hour is still slow. It doesn’t take long for cities to understand this, and adapt accordingly.

            That is why there are bus lanes, and HOV lanes. Those enable a much more efficient vehicle (one holding a lot more people) to move faster. It is also why cities have HOT lanes. Like expensive cab service, they know that only a subset of people can afford it, so it won’t be too crowded. If you lower the price to zero, the special lane is no longer special, and will be as crowded as the rest of the lanes.

            The myth is that point to point service (AKA cabs) can be modified slightly to operate as efficiently as transit, but as fast as a cab. It simply isn’t possible, and even Uber understands this. Because once you go out of your way to pick someone up, you are less efficient. You pick up fewer people per minute. If you wait around so that a fare won’t have to wait, you are being inefficient. If you don’t pick up people along the way, you are being inefficient. That is why the more efficient a cab system becomes, the more it resembles a transit system, with all of the trade-offs found in it.

            Perhaps an example would help. Consider a common one — an airport shuttle. Much cheaper than a regular cab, because a van can pick up several people along the way. Now, imagine you have 100 vans, ready to go, at the airport. When do you leave? With the first passenger? That might work, but then you aren’t very efficient. How about when the van is full. Very efficient, but that could mean the first passenger waits around a long time, waiting for that next ride. So already we are talking about a fixed timing system (e. g. every five minutes) which, as it turns out, in a more complex system means the vehicles are always moving (and thus have the potential to pick up more passengers per minute).

            Now where do you go? It makes sense to drop off the closest passenger first (that is obvious) but what if half the people live north of the airport, and the other half to the south. So now we have separate routes (from a general standpoint). How close to you get the riders? If you deliver them to their door it is great, but that means the last group (those farther away) have to wait a long time. Once you start talking about a general purpose system (not an airport shuttle) it is obvious that you are describing a transit system.

            Again, the key thing is that cars don’t scale. A software engineer would simply say the way to solve the problem is to throw out efficiency. Don’t buy 100 vans, but a million cars. But again, even if driving itself isn’t an issue, the price of the car and the space it uses is an issue.

            This makes it remarkably different than software. Software scales (if designed well) because computing power and memory keeps getting cheaper. But road space is limited. Otherwise, every midsize city in the country would simply add a transit lane, a truck lane, a bike lane, and just like that solve all of their traffic problems. But that is way too expensive, and in most cases, simply impossible.

            It is no different than, say a mansion on Malibu. No matter how advanced we become, we can’t all live like that. All 7 billion people, in various big mansions, living large. Sorry, it can’t happen.

            At best these sorts of system will improve transportation in very low use areas. Access vans could become much more efficient. But as vehicles become automated, the price of fixed route service drops while the quality increases. Run vans every five minutes on ever major street and suddenly you really don’t need a cab. Unless, of course, you want to pay for something special, that won’t make as many stops and will drop you off right at your beautiful house in Malibu.

    • Michael D. Setty May 24, 2017 at 1:55 pm #

      Automated buses on predetermined fixed routes as a technical problem is at least two orders of magnitude less complex that robotaxis. In fact, there a few automated bus routes running in demonstration service today.

      Even a small minibus-worth of patronage in environments that can generate such minimal ridership would render robotaxis too costly, even if they actually work–which I am 95% confident that they never will, for a whole variety of reasons, even some that are technical.

  8. Novacek May 24, 2017 at 12:23 pm #

    Any comments on Austin’s experiments with demand response transit (in areas where they’re cutting routes due to lack of demand/density)?

  9. Jo May 25, 2017 at 2:21 pm #

    Recognizing the gist of your post is that Uber’s “Pool” model is financially unsustainable you veer off on the end to imply Uber Pool didn’t “invent” anything at all but you glossed over two important factors, especially for those people facing bus-lines that are infrequent services.

    1. Communication. “All the new apps have helped smooth out inefficiencies of communication, but they will never change the math. Technology never changes geometry.” — *But* communication is vital, is my bus on time? How soon will it be here? How reliable is it? I can trust that reliability with Uber Pool but can I with regular MTA service? Communication is vital. Transit Agencies constantly miss this.

    2. Ride Comfort & Safety. The app allows immediate feedback. Can you imagine MTA giving you a refund because of the loud or smelly rider on the bus? I may have to walk 3 blocks with both, but the ride once I get there is substantially different.

    Fixed routes may be inevitable but level of service is not comparable. Transit agencies can’t seem to get this ‘back to basics’ idea. No one likes riding on the cattle car that is public buses. And *yes* that changes the geometry for transit users. If buses are unacceptable choices because of the poor level of service then that person lives in a very different city than most. Believing people to be mere consumers of transit and not real people is a mistake that seems common in public transit circles.

    • RossB June 1, 2017 at 5:46 pm #

      Uber is a cab company. The only major innovation was to use unregulated drivers. They took a page out of the pizza delivery market and figured paying ordinary people to drive other people around could work. They took advantage of loopholes in regulations and the general stratification in society (lots of low wage workers) and built a successful company.

      But compared to traditional cab companies, it isn’t exactly a breakthrough, simply more convenient. Instead of calling an operator, you type something on the screen. Instead of paying cash, you pay with your credit card. In some ways it is less convenient (you can’t hail it) but in other ways it is an important progression. But lots of other companies have made the same progression (e. g. I can buy clothing from Lands End over the web!). It isn’t like, say, a spell checker (which lets you know if you have typed a suspect word). So, to answer your questions, the old cab operator would let you know when your cab was expected. You could give feedback about the conditions of the cab. Nothing Uber has done in that regard is very special — if not for their labor model, they wouldn’t be in business.

      The new system allows you to share rides. Again, nothing new. Airport shuttles have done this for years. Advanced software allows the routing to be more efficient, but it still is a premium service at a premium cost (over public transit). That is because no matter how smart the algorithm is, as you get more and more riders, it resembles a traditional fixed route system more and more. The balance between premium service (e. g. a ride right to your door) and efficiency (making you walk a few blocks) remains the same.

  10. Ray May 25, 2017 at 3:28 pm #

    Maybe the above image makes sense in SF or NYC, but in Los Angeles and other road heavy cities, UberPool provides very good service for rides around 5 miles. You may pay $4 more than a bus fare, but you will save on walking a half mile to the bus stop, waiting 30 minutes for your bus, having to make a transfer and another 20 minute wait, and finally walking another half mile to your destination. This is the reality for millions of people, and public transit doesn’t know how to provide a competing service.

    • Dave May 26, 2017 at 5:40 am #

      “…Public transit doesn’t know how to provide a competing service.”

      Maybe that’s because the city itself isn’t designed for public transit? You said it yourself: in other “road heavy cities.” If you’ve already made the decisions to carve up your urban area with freeways and widened all your local arterials as much as possible and abandoned your downtown for many generations and sprawled your way across hundreds of square miles… well, yes, then you’ve already destroyed any natural advantages transit might have and thus transit will struggle to “provide a competing service.”

      • Ian Mitchell June 12, 2017 at 5:00 pm #

        LA’s basin was already platted out and inhabited, from in 1910. With pacific electric and a little mainline rail.
        Since then, its population has increased by over 1000%.

        LA was never not sprawled. It’s always been polycentric. Its downtown is more important to the basin now than it was a century ago, and still the net flow of traffic in the morning is OUT of downtown. The surface streets haven’t been widened much since 1910 (take a look at some old pictures).
        The freeway system was new. But had the PE not been torn out and none of the freeways built, Ray’s point would likely still apply.

        Even in “transit city” alternate L.A., diagonal trips over moderate distances favor smaller vehicles. In some of LA, there isn’t space to do that (hence the speed of subway construction since 1980).
        In a lot of LA, there is, and will be for some time still.

  11. cph1776 May 26, 2017 at 10:10 am #

    Last October, the city of San Clemente, CA (southernmost part of Orange County) contracted with Lyft to provide a transit service in the city. This service replaces two bus routes (OCTA #191 and #193) that were cancelled due to low ridership.

    The city will subsidize fares if the ride begins and ends within 250 feet of a former OCTA #191/193 bus stop….rides in that case will cost no more than $2, same as the local bus fare.

  12. Scott May 27, 2017 at 3:11 pm #

    Thanks Jarrett, and I agree with you. This idea that there must be a “tech fix” to traffic congestion seems to be beloved by the IT industry. But as you say, they need to do some more maths. During the peak the problem is lack of road space. More cars, self directed or not, does not fix that. Outside the peak the problem is lack of trip density if you are trying to service an “anywhere to anywhere” model. It dorsn’t work either. I was involved in some analysis of this for Queensland Transport way back in the mid 1980s in Brisbane. Para transit buses were hopelessly uneconomic in outer suburbs. It was cheaper to give the disabled people taxi vouchers, which they much preferred, and provide park and ride at the train stations for everyone else.

    This blind faith in the unproven ability of demand responsive transit to magically solve capacity problems reminds me of the sales efforts of the IT industry to sell “intelligent Transport Systems” for freeways going right back to the 1980s. They promised to double freeway capacity via platooning vehicles. Presto, our congestion would be gone! Thirty years later and we are still waiting. In practice ITS is useful for incident management, and increases lane capacity around 10%. But hey, they sold a lot of computers!

    Now we hear the same exaggerated claims for autonomous vehicles, as though something that makes it easier for people to drive will reduce traffic congestion. Better safety – sure. But eliminating traffic congestion by getting more people to use cars? Laughable. That would be like trying to fight ISIL by selling arms to its biggest financial backer.

    • Rick R May 30, 2017 at 4:27 pm #

      The idea that modern tech ‘disruption’ is a new and unstopable force is certainly pushed by those leading it. Disruption goes back to when cavemen started making stone tools. There is a line of disruption through walking -> horse power -> steam power -> electric power -> internal combustion power that has seen numerous transport modes come and go, but the new forms have not always obliterated old ones.
      As noted by others here, anything that can be bolted to a car or van to make it ‘competetive’ can also be bolted to a bus. Vast amounts of data are already being generated by transport agencies across the world already, which can be used to increase the effeciency & reliability of existing services, as well as giving passengers certainty about their travel options through the travel apps provided by authorities or third parties, an advantage already claimed by the ‘disruptors’.

  13. Joe May 30, 2017 at 12:14 pm #

    What sucks for drivers is that the only thing that makes UBER Pool profitable is the extra mileage required when making those side trips. Now, we will be picking up people along the way and essentially not getting paid for the extra work.

  14. Andrew Cone May 31, 2017 at 11:23 am #

    Even if you buy this “geometry” argument, why should I believe that municipal governments are better positioned to provide transit service than Uber? Even if you look at the most pro-transit cities in the US, with the exception of Manhattan and parts of Brooklyn, transit is god awful.

    I live in one of the most pro-transit cities in the country, and the busses are unreliable, frequently late, and sometimes not much faster than walking. The drivers are often extremely rude, and there are frequently passengers that shout delirious nonsense or smell like feces. You can say those are social problems that should be solved by this or that economic policy, and that it’s unfair for me to pin them on transit agencies, but the reality is that’s how it is on the bus, and it sure as hell isn’t on Uber.

    Even if Uber wakes up one day and sees the geometric light, their bus-like things will be awesome, and those of most municipal transit agencies will still be terrible. I honestly don’t care if UberPool wants me to walk a few blocks away to get a cheaper ride. That is completely reasonable. Hell, even if Uber wants me to transfer, and the routing starts looking more bus-like, I’m 100% OK with that. I trust Uber to actually get me around far more than I have ever trusted an American transit agency. I don’t care if Travis Kalanick is a tool—he may well be—but then he is a tool that has actually solved a problem.

    My family recently got rid of our second car because Uber is a more economical substitute. It requires fanciful absurdity to believe that our transit agency could or would ever provide service that made us consider getting rid of our second car. Sure, in some alternate fantasy land, in which American cities were twice as dense, and had spent the post WWII era growing up around transit, things would be different. But that’s not how things are, and we should make plans for the cities we actually live in, not some Jane Jacobs neverland.

    Anyway, the “geometry” argument is nonsense that applies at best to high density areas like NYC or SF. It has no bearing on the actual urban (and suburban!) geographies in which 80+% of Americans actually live.

    • RossB June 2, 2017 at 7:52 am #

      You are basically describing the advantages of a public versus private system. The country club is nicer than the public park. No smell of feces, or people shouting delirious nonsense.

      There are two ways to deal with it, of course. One is to do what we (and many European countries) did in the period just before and after the war. Tax the wealthy and improve the public spaces. So taxing Uber (a luxury) and paying to improve the public agency (transit) along with paying for mental health and sanitation services would be one way to go.

      Of course we can also just do what most third world countries do: Ignore the problem. The wealthy hire servants, stay isolated from the poor, never take public transportation and basically live in their own, comfortable bubble.

      As to the geometry argument being nonsense — well, that is nonsense. Already we have seen UberPool move towards a transit style system. Being encouraged to walk a couple blocks — rather than be picked up right by your house — is precisely the sort of geometric trade-off made by transit agencies every day. That is because UberPool is an attempt to make a more affordable, lower quality service aimed at the middle class, rather than those willing to spend money on a more direct trip. Regardless of the particular geographies of the urban or suburban landscape, they face the same trade-offs as a bus network. If the area has very few riders, then it won’t be very efficient (it will operate just like a traditional cab company). If they have lots of riders, then they are more cost effective, but they struggle giving everyone a good, fast ride to their destination. It is what Jarrett Walker and others (including me) have been saying for a while. A smart, point to point network that seeks to be highly efficient will gravitate towards a bus network that does the same. The big difference between it and the bus system is that those who take Uber won’t have to mingle with the commoners or deal with the social problems they have.

  15. David E. Pickeral May 31, 2017 at 11:35 am #

    The KPI is not that on demand services make transit “affordable” because it actually isn’t anywhere near that now as it currently exists. All that needs to happen is that it cost less than low volume fixed route service that for decades has operated under massive public subsidies. That said this Mr. Walker is as usual perfectly correct in his observation that some walking can cut a lot of the OPEX out of the equation for the operator–which is the central context of jitneys or “route taxis” that have operated successfully around the world almost since motor vehicles appeared–only disappearing in North America prior to WWII when local authorities forced them off the road as a concession to the then mostly private transit companies. Jitney service failed in the US recently primarily because it was forced to compete with and not replace inefficient conventional mass transit operations now operated predominantly by governments.

  16. asdf2 May 31, 2017 at 9:59 pm #

    My personal opinion of UberPool/Lyft Line is a lot of skepticism. Of the 10’ish rides I’ve taken on such services to date, all but one of them, I wasn’t matched anybody, so I was effectively riding UberX at a subsidized price (which is nice while it lasts, but obviously not sustainable).

    The one time I was matched, the route we took bore no resemblance to any reasonable definition of “carpool”. We drove two miles south, picked up a passenger, then drove two miles northeast, dropped off the passenger, then drove back south again, passing right by the point where the other passenger got on, then drove another 15 miles to my destination, where, thankfully, we didn’t have to take any more detours for anybody else. All in all, the extra passenger added about 20 minutes to my trip while saving essentially zero vehicle miles traveled compared to if the other passenger were transported in a separate car. And for those 20 minutes, I saved a wonderful $2 off a $29 fare.

    It was this ride that was a perfect example of the big flaw with these shared-ride shuttle services – you have absolutely no idea how long the trip is going to take until you are fully committed, even when traveling during off-peak hours on nearly empty roads. A bus at least has a schedule – you know in advance what route it’s going to take and (assuming the bus doesn’t get stuck in traffic) how long it’s going to take.

    While Uber’s change to require people to walk a block to keep the route straighter is welcome, I am still going to be very reluctant to use such services if I can’t know the travel time in advance and have the ability to opt out if the trip is going to entail an unreasonable detour.

    In fact, I’ve come to the conclusion that if you really want a middle ground between a private ride and a mass transit trip, an old-fashioned transfer between regular old UberX and the regional transit system is the way to go. Even then, the mass transit part of the trip still has to be very fast and very frequent – otherwise, when you get to a station and find a 30-minute wait, it becomes all too tempting to just screw the transit part and pay the driver to take you to your final destination.

    One other thing I’d like to comment about Uber’s change – is that walking a block to make the route straighter for the other passengers is remarkably similar to what happens when I form my own carpool with friends to ride UberX – if two of us live within a block of each other, we get off at the same stop, we don’t waste the 3rd passenger’s time (and run up the bill) by having the driver go door to door for each of us. Sometimes, one person will even ride the light rail for part of the way if doing so avoids a significant amount of detouring in the car for the rest of the group. It’s amazing how much more considerate people are of their fellow carpool passengers when it’s their friends, rather than strangers who just happened to get matched together by a software algorithm.

    • Andrew Cone June 1, 2017 at 8:35 am #

      I’m not sure where you live, but this is not how it is here in the Bay Area. It started like that when they rolled out UberPool, but now the majority of my UberPool rides involve picking up someone else. Rarely does the detour add more than 25%, and it is solidly never as slow as the equivalent public transit ride. UberPool requires volume to work, and that takes time. When self-driving cars happens and the cost of Uber falls, it’s reasonable to expect that the amount of UberPool users, and thus the benefit of UberPool will increase. This is not some “techie” dream; it is something business analysts and urban planners are increasingly planning for.

      UberX to mass transit sounds like a good idea, but in the Bay Area, I have only found it effective for transbay trips, or at times when rush hour traffic is awful. The delay involved in transferring typically eats up any gains.

      Sharing an Uber with friends is great, but I do not think this represents a scalable solution for very many people. I’m friends with my all neighbors, for example, but none of us go anywhere near the same place at the same time, much less close enough to make Uber-sharing a good idea.

      • RossB June 2, 2017 at 8:09 am #

        Right — that’s because bus service scales. UberPool is not very different than an airport shuttle. I’ve ridden plenty, and they take a trip very much like the one that asdf2 described. It is a bit different, in that no one is interested in meeting at the nearest arterial. So going down a dead end street to pick up a passenger is just something they do. But the more passengers they pick up, the more direct it becomes. One van serving the whole city would involve huge detours for those picked up first. But if a van can serve a corridor, everyone gets there faster. That is really the point in this (and similar) articles. The more a service like this is used, the more it starts to resemble a bus network. The advanced technology involved (the ability to know that you have an actual customer at a particular time, not just a potential one) gives them a leg up, but it is only a small advantage. Eventually, as volume increases, the two systems resemble each other. You have the same trade-offs as any transit network. There is no fundamental advantage to Uber. Their entire system is based on low labor costs, and nothing more (adopting the pizza delivery model). Either folks are willing to pay for a more expensive product, or the driver earns less money. Nothing very revolutionary, really.

        When Uber replaces the drivers with machines, their prices go way down. But the same is true for the public transportation agency. The transfer penalty you mentioned goes away. Their is no need to run big buses every half hour, just run vans every five minutes. It is pretty easy to see how the two systems merge together.

      • asdf2 June 3, 2017 at 8:35 am #

        Seattle. Although the bulk of my Uber/Lyft trips are almost never the couple-miles-in-the-middle-of-the-city kind, but longer trips to/from the suburbs, where transit isn’t as good.

        The Seattle area also has a network of regional express buses which are fast, but outside of rush hour, tend to not run very frequently, and the local connections to the express buses are often slow, not super-reliable, and not timed very well (at least on weekends). The result is that there are many trips where UberX->express bus ends up saving a good half hour over busing it all the way, in exchange for about $7 or so.

        I can see San Francisco being different, as there are no freeway express buses, just the glacially slow Muni. Even there, however, I used the combination of CalTrain and Lyft not that long ago to take a trip from San Fransisco to a hotel in Sunnyvale. Uber/Lyft all the way would have been quite expensive, due to the distance, and transit all the way would have added another 30-45 minutes to go those last couple of miles.

      • Crispin Cast-Nine June 26, 2017 at 7:14 pm #

        This isn’t at all my experience with ride-shares in SF. I’ve been on a lot of Lyft Line/UberPOOL trips within SF that ended up either being longer than the equivalent Muni trip, or else maybe 20% shorter but at 2-5x the price. I just checked and a non-surge Line from my home to my work, both within the city limits, costs around $12 with tip — for that price differential, it had better be *way* better than a $2.50 Muni ride. Worse, my travel time has been no more predictable than Muni. I once got a Pool from SoMa to the Castro, naively assuming it’d be better than the bus. Instead, with the extra pickup, it took 30 minutes, making me late for an appointment — a direct car trip would have been 11-12 minutes and the bus would have been around 25.

        As far as I’m concerned it’s nice for getting home from the bars every once in a while, when transit has mostly shut down but the roads are clear and there’s still plenty of transit demand. But the trade-off doesn’t make sense to me for much else.

        I also do find the “regular Lyft plus BART” pattern pretty useful for visiting friends who live in the East Bay burbs in a reasonable amount of time, particularly since East Bay bus service can be especially thin on the ground. It gets expensive very fast, though, especially since BART itself is not exactly cheap.

  17. CX June 6, 2017 at 8:17 am #

    I appreciate the amount of capital companies like Uber and Lyft are putting into their services, creating a massive experiment when it comes to way we move. Jarrett makes a good point that that geometry of efficient transit has not changed. However, technology has improved our relationship to these services and provided dynamic incentives to passengers and drives. One of the key takeaways of this is that these incentives (often related to cost) create different types of services — many that we have ignored in US cities, and that can once again be achieved with greater discussion and understanding about the role of government and the benefits of private corporations. These companies will help us achieve a more efficient transportation system if we also choose not to ignore some of the basics about geometry.

    Additionally, I hope that the private sector will tackle the costs of building subway tunnels in the US.

  18. Ian Mitchell June 12, 2017 at 4:29 pm #

    There’s a (comparatively) new startup called VIA that does something along these lines. So rather than coming to your door the middle of Euclid ave, you’ll be directed to “go to the SW corner of Euclid and 13th st NW”. The fares are very low ($3 in DC, $5 in NYC, serving manhattan below 125th, and the waterfront areas of Brooklyn and Queens, not sure in Chicago), and it’s solely shared-ride, paid per-person, and when you request a ride it’ll give you an option, e.g. “VIA in 5 min, VIA in 15 min”.

    It may be a bit like re-inventing the wheel trying to use vans rather than buses, but I’d love to see that kind of approach produce real, useful data as to designing fixed-route transit.

    • Mike June 13, 2017 at 3:43 am #

      But that’s exactly what I do here in London with my London Transport app. I’ll ask it for all bus departures in the next 30 minutes from my local stop and go and get the most convenient one. If I’m in a part of town I don’t know it will direct me the nearest stop as well as tell the real time arrivals of the next bus. i can board using a pre-paid card or a debit card. I can transfer for free within one hour. For £1.40. Uber are underpaying drivers and massively subsidising in order to destroy city wide go everywhere public transport. When they have their prices will rise. That’s it – they are not trying to improve things. It’s a cab firm.

      • Ian Mitchell June 13, 2017 at 4:30 pm #

        If you think that salary+benefit package of a bus driver belongs in the 100K range, especially when the typical bus in the US carries 11 people, then you and I have differing opinions on how best to expend scarce funds.

        You live in london. You know that uber’s not going to destroy citywide anything in London.

        I live in a college town in North Florida (which does happen to have the highest per-capita transit ridership of anywhere in this 20 million person state- about 4 scotlands, for scale). Here, ride hailing services mean fewer kids die from drunk driving and that people in areas that can’t reliable support a bus route at the hour that they’re departing have an option.

        Before uber came, cabs here would tell folks it was $20 PER PERSON for rides that were at most 3 miles (this town is about 5 miles across in any direction)- that’s when they were waiting outside the bars at closing time. If you called them? You might get a cab in an hour. It might not show. There was no standard for service, no guarantees, no refunds. The last new cab company opened nearly a decade ago.

        If uber isn’t trying to improve things, well, they’re doing a crappy job of not improving things.

        I primarily bike. Because here it’s the most tenable option. I live in a household where I have access to a car- I use it 2-3 times per week. I use the bus for certain trips too. I also walk. None of these things threatens the existence or usefulness of the others.

        When I’m in other city, I travel other ways.
        Here’s how that’s gone:
        I attempt to make transit work in Austin. When (not if) that fails, I resort to either uber or a long, long walk. I bought a folding bike thinking that it would work to bike to the train and take that- then I realized the folly of biking on the roads there.

        When I’m in LA, if I’m within a mile of a metro station, I’ll use metro. Most bus journeys involve multiple transfers, acceptable where buses/light rail have their own lane, but impossibly infuriating where they don’t. I’d be able to use a friend or family member’s car, but to drive and park in LA requires a balance of skill and foolishness upon which I won’t endeavor. It would be financially hazardous to use regular uber, taxis are unreliable and somewhat threatening, but the shared-ride services fill in the gap between how things are and an acceptable experience.

        In new york I’ve used a cab exactly once, everytime I’ve used an MTA bus it’s tuned out to be slower than my (impatient) walking pace, and the vast majority of trips by subway.

        In philly the light rail works well and the buses work reasonably well. The regional rail, albeit a bit too 19th century in operations, is useful too. never tried rideshare or cabs there.

        In DC, I use via to avoid the red line, because 20 minute headways mean that anywhere I travel along the red line would take longer than walking. But otherwise I tend to use the metro. I used bikeshare once- I’d use it a lot more if it were integrated with other transit fares. The buses are typically slower than walking within the district, but are pretty good in connecting the metro to the rest of NoVa.

        In Boston: The buses aren’t great, seem to be on the exact routes as streetcars with no thought endeavored into building an efficient or coherent network to Boston as it is in 2017 rather than 1917. The ferries are fun, and I’m glad they’re fare-integrated, but mostly just fun.
        The green line is an anachronism, but the other subways and the silver line are all fantastic.
        I never needed a cab or rideshare.

        In Baltimore: If you’re not a racist county resident or an impoverished city resident, the transit system works. Unfortunately, that means it only works for a minority of the people it serves. Never needed a ridesharing service, and pretty much only the charm city circulator buses and ferrys (both free!) were ever necessary. Would be a great place to bike, though you’d want a beefy lock and removable lights.

        In south florida: Tri-rail works really well for trips between counties, but is way too far from actual destinations. Buses would work really well if they’d have lanes. Driving is absolutely insane, parking in some places costs more than rent does where I live. Shared-ride rideshare services made it possible for me to see any of the city on my own, rather than only doing anything with friends who have cars. On US-1 and A1A the buses are fine.

        San Fransisco: In the birthplace of ride-hailing, MUNI makes you realize why it exists. Public buses that continue their route even when there’s human feces on them, that stop every 200 feet. Hills and aggressive driving that make bicycling a death wish and driving seem equally insane for a flatlander. A subway that is really more of a commuter rail system. Streetcars that work exactly as well as they did 100 years ago- so yes, they’re better than walking. But this is pretty much where any trip I couldn’t walk would end up making the most sense doing by uberPOOL.

        San Jose: Get in by caltrain, then have a car or a bike waiting for you. Even ridesharing isn’t reliable somehow.

        Seattle: Driving works, but poorly enough that the buses are acceptable rather than demoralizing. The downtown transit tunnel is the only place that really has “show up and go” service levels, but if you’re willing to wait 15 minutes most of town’s fine. The streetcar is absurd and slower than walking. The light rail is useful. The commuter rail is useful. Amtrak is reasonable. It’s perhaps the only place where I’d say having a car and using public transit are exactly equivalent, with neither being actively unpleasant.

        Atlanta: Buses, streetcars, and MARTA exist in parallel universes. The surrounding counties and state want the capital to drop dead, and it shows. Biking is for those with multiple DUIs. If you’re trying to get from the airport to where your friend is willing to pick you up (but they’ll want to know exactly which train you’re on, because it’s not safe to wait near the station), MARTA is useful. ATL’s transit situation doesn’t assuage my total lack of desire to ever return.

        Cincinnati: A walkable streetcar city with an unfortunately never-finished subway that has been horribly scarred by numerous attempts to kill walkability and transit. Mainly due to racism. But it still has “good bones”, so even the horridly neglected buses work well. The new streetcar is a toy. Rideshare is the only good way to/from the airport, or getting around after midnight (some areas, after dark).

        • Jarrett June 20, 2017 at 9:17 am #

          Thank you for sharing your experience. But as a transit planner I get this kind of input all the time. Should public policy be based on your personal experience, and your particular notion of what transit services are worth your notice? Are you sure you’re a sufficiently average person for that to make business sense?

  19. Brian July 16, 2017 at 3:39 am #

    One positive feature of UberPool, as against the New York subway, is that one’s fellow passengers will not be aggressive panhandlers, unmedicated schizophrenics, homeless people drenched in their own urine, etc. So the public-spirited types who rhapsodize about mass transit must also support comprehensive law enforcement on the Giuliani-Bloomberg model, if they want to preserve any chance that travelers will obey their command & control dictates.

  20. Edward Re August 1, 2017 at 8:21 pm #

    The buses that go like a goats track are usually empty, because they take forever to get there. Buses that run straight get the significant ridership.
    If you want to get somewhere off route, you just do a transfer. If you draw the other straight lines they will form a network. Then it becomes useful, and then it starts to look like a plain old bus service.

  21. Jeff Vance January 9, 2018 at 7:53 am #

    I’ve grown interested in transit for a multitude of reasons, a few of which are my disinterest in seeing people waiting in all weather, at all manner of ill-conceived stop, that most or all had to walk to or be dropped at, that then delivered an extraordinarily long ride to another stop, similar or not in form, where another hike would be required to reach the final destination. Who wants that? I suppose that may not be the standard everywhere, but it is here in a fairly populous metro area, 600K, called Louisville, Kentucky. It would seem possible to optimize the mobility of people by bringing them to comfortable depots(for lack of a better term) via a localized Uber-like…or autonomous pickup service, where they might then connect with another local Uber-like service for a close destination, or a medium range city-bus type service, or a more regional express bus which could drop them at a depot 17 miles/kilometers distant, where the final Uber-like service could deliver them quickly. Would seem a private sector role all around, except for maybe the stodgy old city-bus. Pay a monthly fee and not concern yourself with the price per trip, except perhaps a fuel surcharge.

  22. Geetanjali January 25, 2018 at 3:03 pm #

    I was wondering if you could elaborate on why demand-responsive services arise in extremely low-wage environments, or in the developing world.

    • Jarrett January 29, 2018 at 9:11 am #

      Demand-response service is very inefficient in terms of passengers/driver hour. So it’s only tolerable if driver hours are extremely cheap.