Jarrett Walker on Redesigning NYC’s Bus Network: Start By Wiping the Slate Clean

Free your mind, and your bus network will follow.

When parallel routes are very close together, people don't have to walk far to reach a bus stop, but they have to wait longer for a bus to arrive than they would if service was concentrated on fewer streets.
When parallel routes are very close together, people don't have to walk far to reach a bus stop, but they have to wait longer for a bus to arrive than they would if service was concentrated on fewer streets.

When New York City Transit says it’s embarking on a “top-to-bottom” overhaul of “the entire city’s bus route network,” that’s music to Jarrett Walker’s ears.

Walker is the consultant who specializes in guiding transit agencies through the process of redesigning their bus networks — most famously in Houston, and most recently in Dublin. At a time when transit ridership is slumping nationally, Houston is one of the notable exceptions. Its redesigned bus network, which brought frequent service to new parts of the city, deserves substantial credit.

With the MTA embarking on a total overhaul of the bus network in all five boroughs by 2021 (the Staten Island express bus network was the warm-up), Walker’s fundamental advice is to start by “wiping the slate clean,” he told a crowd at TransitCenter last week. Only a blank canvas will yield “design choices based on the whole network,” he said, which in turn produce benefits compelling enough to “overcome little problems.”

For transit planners who know their bus network and each justification for all of its quirks inside out, it can be difficult to let go and think freely. At the bus network redesign workshops he leads for transit agencies and city DOTs, Walker insists that participants refrain from discussing historical reasons for routes and steer clear of anticipating political objections (as in, “that parking space over there belongs to a business owned by the city councilor’s cousin”).

In Dublin, that meant completely redoing a system that emphasized one-seat rides at the expense of frequency. With the old network, riders could avoid transfers, but they sacrificed in the form of longer waits. Under the new bus network design, unveiled earlier this week, service will be simplified into core routes that arrive much more frequently, which is projected to reduce total wait time even though riders will have to make more transfers.

The basic concept of the Dublin bus network redesign is to . Image: Jarrett Walker and Associates
The basic concept of the Dublin bus network redesign is to reorient service around a core network that minimizes the wait for the next bus. Image: Jarrett Walker + Associates

The politics of these major redesigns are never easy. (“Beautiful people will come to you with their elderly parents and their babies and say the redesign will ruin their lives.”) But with a sweeping overhaul, the benefits should be substantial enough to win over elected officials and other community leaders.

As Walker put it, “It’s easier to take out half of all bus stops than each stop one by one.” (Strategic advice that’s especially relevant for New York, where stops spaced too close together drag down bus speeds and stop consolidation will have to be embedded in any effective bus network redesign.)

Walker describes his role in the redesign process as a facilitator, and he didn’t want to go into prescriptive detail about how to improve New York’s bus network.

But he did point out the inconsistent distances between parallel routes in Brooklyn. Where parallel routes are close together, riders don’t have to walk far to get to a stop, but they have to wait longer than they would if service was concentrated on fewer streets.

Walker also stressed the importance of a bus network that transcends the boundaries between boroughs. While the waterways dividing the land masses of New York justify separate bus networks to some degree, he also sees artificial distinctions in the current network.

When you wipe the slate clean, he said, “There’s a way to look at Brooklyn and Queens as one thing.”

  • Albert

    If both the slate and bus lanes could be wiped clean of private cars, Walker would find it child’s play.

  • Larry Littlefield

    “It’s easier to take out half of all bus stops than each stop one by one.”

    True that. The same professional objectors will be as likely to object to small things as big things, so better to do the big things.

  • sbauman

    As Walker put it, “It’s easier to take out half of all bus stops than each stop one by one.” (Strategic advice that’s especially relevant for New York, where stops spaced too close together drag down bus speeds and stop consolidation will have to be embedded in any effective bus network redesign.)

    There are several differences between Houston and New York City riding habits. If the Houston “cure” were replicated in NYC, these differences would further add to the decline in ridership.

    A glance at the 2016 NTD, shows that the average bus trip in NYC is 2.1 miles vs. 5.0 miles in Houston. The average bus speed in NYC is 7.1 mph vs. 12.2 mph in Houston. This means that the average time a NYC rider spends in a bus is 17.7 minutes vs. 24.5 minutes in Houston. From the perspective of the amount of time spent in a bus, NYC beats Houston.

    An overall trip consists of: walking to/from the stop; waiting for the bus; and traveling in the bus. Suppose half the bus stops were eliminated. This would increase the average walking distance by 1/4 mile. This translates into 5 minutes more per trip.

    What’s the time savings by eliminating a stop. The dwell time consists of the time getting in/out of the stop plus the passenger boarding time. If the total number of passengers remains the same, there will be no boarding time difference. The sole difference will be the time savings by eliminating the braking/acceleration necessary for stopping.

    The pull in/pull out bus time is 30 seconds, according to the literature.

    There are currently bus stops every 1/4 mile. This works out to 8 stop for the average 2 mile bus trip. This would be cut to 4 stops, after Walker makes his cuts. The time savings would amount to 2 minutes by eliminating half the stops for the average 2.0 mile ride, assuming passenger totals remain the same.

    To the bus passenger taking the average 2 mile trip, the walk to/from the bus stop will take 5 minutes longer but once in the bus, the trip will be 2 minutes shorter. The net effect will be a 3 minute longer commute.

    The story is different for Houston because of the longer bus trips. There will be 20 bus stops eliminated for the average 5 mile trip, for a 10 minute time savings. The longer walk to/from the bus stop will be 5 minutes longer. The net is a 5 minute shorter travel time.

  • ortcutt

    The problem with this analysis is that the stops aren’t currently every 1/4 mile. In my neighborhood, the stops are every two blocks, which works out to every 500 feet (i.e. less than every tenth of a mile). Even if every other stop was eliminated, you’d still be at every 1000 ft, which would still be less than 1/4 mile. The stops are simply too damn close together .

  • Exactly right. Whatever inefficiencies exist in New York’s but bus network are down 100% to an inappropriate allocation of street space that fails to give priority to buses. Imagine a Jamaica Avenue, a Myrtle Avenue, a Metropolitan Avenue that are for buses only (as well as bikes, of course). Travel times would be half of what they are currently; and a lot of those Queens goofballs who think that they “have to drive” would see that in fact they do not have to.

    Another thing that hurts buses is the inexplicable policy decision to tolerate rather than eradicate the dollar vans. As a result, revenue which should rightfully belong to the MTA is diverted into the pockets of an array of sleazy fly-by-night thieves, and rickety deathtraps piloted by unlicenced maniacs endanger their passengers and terrorise the other street users.

  • AnoNYC

    There’s two stops on my usual bus route 330 ft apart. Super annoying on a busy street.

  • Andrew

    There are currently bus stops every 1/4 mile.

    Standard spacing on north-south bus routes in Manhattan is every 2-3 blocks. At 20 blocks per mile, actual stop spacing is half of what you claim. That is the stop spacing that Walker is proposing to double.

    Of course, spacing is wider on limited, SBS, and express routes, and also on non-stop portions of any route that crosses a bridge, has a highway run, etc., but that’s all a distraction from Walker’s point. If you’re coming up with your quarter-mile spacing by averaging everything together, you’re averaging a lot of irrelevant stuff in with what Walker is discussing.

    You’re also ignoring the fact that, if buses weren’t so dreadfully slow, they’d be more attractive as options for longer trips, thereby driving the average trip length up from 2.1 miles. You treat the 2.1-mile average as constant, when in fact it is not.

  • ortcutt

    If you eliminated superfluous stops, put in bus bumpouts at the remaining stops (so buses wouldn’t need to exit and enter traffic constantly), and implemented contactless payment and all-door boarding, buses could be a compelling option in this city again.

  • Joe R.

    I don’t think I’ve ever been on a bus in NYC where the stop spacing was 1/4 mile. Typical is one stop every two or three blocks, or roughly one stop every 1/8 mile. Cutting out half the stops then saves about 2 minutes per mile if you assume 30 seconds per pull in/pull out. On average a person will only have to walk about 1/8 mile more. This takes 2.5 minutes at average walking speed. So basically the break even point occurs for a bus trip of 1.25 miles. That’s far less than the average bus trip, so on balance people will save time.

    It’s also worth mentioning that the pull in/pull out time isn’t the only delay caused by an additional stop. Quite often, an additional stop also means waiting for a light the bus otherwise would have avoided.

    Finally, as Andrew mentions if we got the average speed of buses up doubtless the average trip length would increase as well, giving an even greater net time savings for the average rider despite needing to walk that extra 1/8 mile to a bus stop. Now in many cases people will opt to drive on longer trips simply because the time penalty for taking a bus is too high. We can fix that with greater stop spacing, exclusive bus lanes, signal priority, and a higher speed limit for buses.

  • sbauman

    The problem with this analysis…

    Let’s add a little more rigor.

    First, how close must the bus stops be before the time savings from reduced pull in/pull out time exceeds the extra walking time.

    Let x be the distance between stops in miles.
    Then the number of stops removed in 2 miles is 1/x and the number of minutes saved is 0.5/x.
    Also, the extra walking time in minutes is 20x (assuming a 3 mph walking speed).
    For the extra walking time to be less than the run time savings the following must hold:

    20x < 0.5/x or
    20 x^2 x^2 x < 0.158113883 mi
    x < 834.84 ft.

    The bus stop spacing threshold to receive any time savings benefit from removing half the bus stops is approximately 835 feet. The bus commuter will experience longer trips on average, if the spacing between stops is greater than 835 feet.

    Second, what is the average distance between bus stops in feet?

    The MTA static GTFS bus schedules answer this question. The data shows the route and the number of stops for each trip. One need only divide each trip length by the the number of stops less 1 to find the average distance between stops for that trip. This distance between stops can then be averaged over all the trips for a typical day.

    Yesterday (July 5th) for NYCT buses: 1348 express bus trips were scheduled; 4073 SBS bus trips were scheduled; 2639 LTD bus trips were scheduled and 33498 local bus trips were scheduled. The average distance between stops was: 2990 ft for the express buses; 1977 ft for the SBS buses; 1485 ft for the LTD buses; and 818 ft for the local buses.

    How does the existing 818 foot spacing between local bus stops compare to the 835 ft threshold for any gain from eliminating half the bus stops? The 17 foot difference would fit within a bus length. It's a wash – no real net gain on average.

  • ortcutt

    This is fascinating numerical masturbation, like asking what temperature the average recipe should be baked at. The MTA should look at each line individually and assess what the the proper spacing is for each particular line. My example comes from the Q49 in Jackson Heights, which inexplicably stops every two blocks for no damn reason.

  • sbauman

    This is fascinating numerical…

    Were it not for numerical analysis, doctors would still be blood letting.

    The MTA should look at each line individually and assess what the the proper spacing is for each particular line.

    I agree. I wasn’t the person who suggested dropping half the stops from every route.

    My example comes from the Q49 in Jackson Heights,

    The average spacing between stops on the Q49 is 628 feet.

  • Andrew

    There’s nothing wrong with numerical analysis.

    There’s much wrong with numerical analysis based on erroneous assumptions, such as your assumption that the average stop spacing is in any way typical of every portion of every bus route.

    Nobody is proposing to blindly double the stop spacing everywhere. The concern is with the extensive portions of the vast majority of local bus routes that have eighth-mile spacing, and it is there that eliminating roughly half of the stops – exactly how many and where precisely depends on the details of the route itself – would be generally beneficial.

    Throwing express, limited, and SBS routes, and bridge/highway segments of local routes, into your average renders your entire analysis irrelevant.

  • Andrew

    This is fascinating numerical masturbation, like asking what temperature the average recipe should be baked at.

    I love the comparison. Thanks.

    (This isn’t the first time he’s done this, and I’m sure it won’t be the last. Watch for it.)

  • Andrew

    Let’s add a little more rigor.

    Second, what is the average distance between bus stops in feet?

    Your notion of rigor fails to incorporate any measure of variability. There may – and in this case there is – a vast distinction between average and typical.

  • Andrew

    It’s also worth mentioning that th e pull in/pull out time isn’t the only delay caused by an additional stop. Quite often, an additional stop also means waiting for a light the bus otherwise would have avoided.

    Good point, especially where the traffic signal timing is in a progression. One or two stops will knock the bus out of the progression.

  • ortcutt

    I’m not opposed to quantitative, empirical transit planning. I’m opposed to numerical masturbation. To expand your medical analogy, the average person doesn’t get leukemia in their lifetime. That doesn’t mean that this particular person shouldn’t get treatment for leukemia. And even a person who gets leukemia probably in an average year of their life doesn’t have leukemia. So what? This particular person needs the particular treatment when they need it. The average distance between stops in the entire NYC Bus system is irrelevant. The average distance between stops on each particular line is irrelevant. Quantitative, empirical transit planning is hard. Numerical masturbation is easy. It’s easy because it’s useless.

  • ortcutt

    I also don’t know why you think anyone proposed dropping half of the stops from every route. Walker didn’t propose it. He made an observation about the political process, i.e. that it is easier to make big changes all at once than to change one bus stop one-at-a-time with a separate hearing for every one. Any other interpretation is willful misrepresentation.

  • Wanderer

    One issue that’s particularly bad in New York is the inaccessibility of the subway system. Only about 25% of stations are accessible. This means that disabled/partially disabled/senior passengers wind up on the bus, when they could use rail in a more accessible system. This tends to slow down the buses; it also creates for shorter stop spacing than might otherwise be warranted. It’s a big issue, but it doesn’t help when MTA does subway station rehabs and doesn’t make stations accessible.

  • Jason

    I’m not looking to argue that the MTA does enough on station accessibility, but how many stations are there where it’s not physically possible (due to space constraints) to make the station accessible?

  • Alon Levy

    In your calculation’s parameters, if the stop spacing is x then the average walking distance is x/2: you’re on average x/4 from a stop at either end, assuming uniform ridership distribution. If you double the stop distance then the average walk distance goes up to x. So the time you need to plug in is 10x and not 20x.

    The formula for the optimal stop spacing is sqrt(2*walk speed*stop penalty*average trip length) assuming uniform distribution of ridership. If you instead assume all destinations are concentrated at definite locations like subway stops then the 2 inside the square root is replaced by a 4.

  • Joe R.

    That probably accounts for easily half the stations. While we can all criticize the MTA in this area (and some criticism is certainly well-deserved), the hard fact is much of the subway system was built at a time when nobody expected the disabled to venture out of their homes without assistance, much less travel independently. Retrofitting it would be at best very costly, but in many cases it’s just not possible.

    My personal feelings on this is that perhaps instead of making every place accessible to the disabled (often at great expense which takes money away from other things) we should instead focus on ways to make the disabled better able to negotiate existing infrastructure. Mobility devices similar to those being made by Boston Robotics come to mind. Why not a chair which can climb stairs instead retrofitting ramps or elevators? In the long run I think this approach is superior. Besides being more cost effective, it gives the disabled access to everywhere the able-bodied can go, instead of only those places which are ADA-compliant.

  • Joe R.

    Exactly. Many bus runs go through places where there are no stops precisely because there is nothing to stop for. That might be over a bridge, past a cemetery, or on an expressway. Those distances increase the average stop spacing. In reality the local bus which has an average of 818 feet between stops might have a lot of stops spaced closer than 500 feet because portions of the run are devoid of stops.

  • Joe R.

    If we designed our streets right, ideally buses should operate just like trains. The only time they’re stopped is when loading passengers. When they’re done loading passenger, they proceed unfettered (i.e. at the maximum speed/acceleration capabilities of the vehicle) to the next stop. No reason we can’t have most bus runs averaging something like 25 mph. That would make them a viable option over driving for lots of trips. A street optimized for just buses would also be great for cycling. Without a constant stream of cars pedestrians wouldn’t need traffic signals to cross the street. They would just wait for a gap in between buses. Most of the time they wouldn’t have to wait at all. Bikes would fly along without traffic lights, besting average travel times by car now.

    But of course the motor heads still dictate transportation policy. If only it wasn’t so.

  • Andrew

    While I absolutely wish we had more ADA accessibility on the subway, New York is not “particularly bad” when compared to its international peers (by which I mean very large systems dating primarily to the early 20th century). London is also around 25%. And Paris is closer to 0% (with perhaps a very few exceptions, only the very newest stations have elevators).

    Yet bus stop spacing in London and Paris is still greater than in New York.

    We certainly need to make sure that people who can’t climb stairs and can’t walk a few blocks still have a means of getting around, but designing a bus system to cater solely to that market will make it highly unattractive to everybody else.

  • Wanderer

    I disagree strongly with the idea that New York can’t make stations accessible. Boston and Chicago have systems that date from the same era as New York. Roughly 70% of the stations in each city are accessible, and they continue to work on others. Berlin has a huge, old transit rail transit system which it’s in the middle of making accessible. MTA is now being sued for renovating a station and not making it accessible, which shows their lack of seriousness about it.

    Making stations accessible also benefits more than just the disabled. It benefits people with strollers and pushing any kind of luggage or cart. New York is a leader in many aspects of transit, but not this one.

  • Joe R.

    One thing you’re neglecting is the bus travel time from a removed intermediate stop to the next stop. Yes, there is extra walking time, and worst case it’s 1/2 the distance between stops divided by 3 mph. However, the person will walk to the next stop further down the line in the direction they’re traveling. Therefore, you can subtract the time the bus took from the removed stop to the next stop from the total walking time. The makes the break even distance somewhat longer than 835 feet. It all depends on the average speed of the bus between stops. And with stop spacing set further apart, the average speed in motion is going to increase also. With stops 1/8 of a mile apart it might only be 10 to 15 mph. When the stops are 1/4 mile apart it could be in excess of 25 mph. So you might be saving another minute or two per mile just in travel time.

    Also, as others have already said, nobody is proposing to just blindly remove every other stop. It needs to be done on a case by case basis. The stops which merit removal might be those 2 or 3 blocks from the adjacent stops in a spot where the buses would otherwise be able to pick up a good deal of speed. You might not gain much removing a stop on portions of a street which is always congested.

  • Jason

    “That probably accounts for easily half the stations.”

    Thanks, I was pretty sure that was the case but it’s been long enough since I looked into it myself that I wanted to double check that. I also seem to recall that a lot of stations don’t see any major improvements because it’d create the impossible situation of triggering ADA-mandated accessibility overhauls that, as we just said, aren’t always physically possible. It’s a great example of how the ADA is well-intentioned but often has the practical effect of achieving equity by making things suck for everyone.

  • Joe R.

    You can do anything. The question is whether or not the cost justifies it. One big part of the problem is construction costs in NYC are sky-high compared to just about everywhere else. This gives a good example of the costs:


    Nearly $5 billion has been invested to make subway stations ADA-accessible, including the nearly $1 billion already approved for the 2015-2019 MTA capital plan. The approved 2015-19 capital program also includes more than $400 million to replace 69 existing elevators and escalators for better reliability. Future capital plans will include funding for accessibility improvements to additional stations.

    From where I stand that’s a lot of money. If it’s “extra” money the feds are giving us solely to make stations accessible then at least it’s not taking away from any other improvements. If not, then think what $5 billion can buy in terms of new signals or other things the system sorely needs.

    Another problem is the MTA’s failure to maintain things. In my opinion that’s the biggest reason not to move forward on accessibility. Why bother putting in an elevator if it’ll be broken most of the time a few years from now? The station still won’t be accessible, except now we’ll be in an even bigger financial hole. We should do accessibility improvements only to the extent of implementing ones which don’t require ongoing maintenance. That mostly means just sticking to building ramps. That can mostly be done for the underground portions of the system. Not sure in how many places you can build a ramp up to an el.

  • Joe R.

    The els represent one of the biggest problems. A ramp to go up 50 or 60 feet would be huge and unwieldy. If you use the ADA standard of 1″ rise per 2 feet then you’re looking at a 1200 foot ramp to go up 50 feet, not including the landings. Where do you put it, especially when it gets near street level? An elevator has a smaller footprint, but it’s so tight around lots of stations that there might not even be room for that. On top of that, where do you put the machinery for the elevator? I’m sure it can be done, but only at great cost.

    It basically comes down to a cost-benefit analysis. Do you spend $500 million to make an elevated station ADA compliant when perhaps only a few hundred people annually might benefit? You can make a fairly good case for installing ramps to most underground stations as those benefit others besides the handicapped. They also don’t require ongoing maintenance, which is another issue with installing elevators. Most of the time they’ll be out of service if the MTA’s current track record is anything to go by.

    It’s a great example of how the ADA is well-intentioned but often has the practical effect of achieving equity by making things suck for everyone.

    That’s really the problem. I certainly agree in principal with the intentions of the ADA. I just think we should focus now more on improving mobility devices for the disabled rather than retrofitting existing infrastructure for the reasons previously mentioned.

  • Jason

    From what I’ve seen in Boston their smallest stations are generally quite a bit bigger than New York’s smallest stations, and crucially, the MBTA seems to control more of the space above the stations than the MTA does, the latter having entrances that are frequently crammed in between other buildings or simply holes cut open in the sidewalk. It’s not just about the age of the systems–most MBTA stations had an inherent advantage over most MTA stations in terms of the feasibility of adding elevators and what-not in after the fact.

  • Wanderer

    The ratio of benefit to inconvenience is much better with taking out a lot of stops. If, to take the extreme case, only my bus stop is taken out then I have to walk further to the bus. But I don’t get a faster ride. If my stop and a bunch of other stops are removed, it’s different. I’m still inconvenienced, but I also get a faster ride from the other stops that are skipped.

  • ortcutt

    “Do you spend $500 million to make an elevated station ADA compliant when perhaps only a few hundred people annually might benefit?”

    I think you greatly underestimate the number of people who would use elevators if they were in place. Disabled people, elderly people, parents with strollers, etc…

  • Joe R.

    The problem is the elevators don’t scale well. The number of people using them is going to be limited by their capacity and how long a round trip takes. A round trip up and down an el would take probably at least 30 seconds (assuming a 50 foot climb and ~4 feet/sec travel speed). Add in time for a full load of passengers to get on and off. I think you would be hard pressed to make more than 50 round trips per hour. Most of the subway elevators I’ve seen hold about ten people maximum. So that’s 500 people per hour either up or down. I’m sure there might be demand for more than that, but the elevators couldn’t handle it. Multiple elevators might, but now you’re increasing the cost by some multiple also.

    On top of that, you’re haven’t addressed the maintenance issue. The realist in me knows elevators will break and the MTA will take weeks or months to fix them, like they do with everything else. So in the end you’ll spend money on something which (optimistically) will move a few hundred people per hour maybe 50% of the time. The rest of the time you’ll be taking up platform and street space while the elevator is waiting to be fixed.

    All of the above is why I prefer ramps. They can move far more people per hour and they don’t break down. If it’s feasible to put a ramp up to an el station then I’m certainly all for it. I’m just hard pressed to see where we might have the space. Remember the ramp needs to rise at least 7 feet over a sidewalk before there’s clearance for pedestrians, and at least 15 feet over a street to clear motor vehicles. At 1″ rise per 2 feet that’s 168 feet and 360 feet, respectively. The former is particularly relevant. Even with switchbacks you need to clear at least 7 feet before the ramp can go above the one under it, more likely 8 feet when you add the support structure. So the ramp needs to rise half that, or 4 feet, before a switchback. Starting at sidewalk level then, you’ll need a run of 96 feet before the first switchback. Add in the landing and you’re around 100 feet. After that the next part of the ramp is adjacent to the first part before hitting the next landing at around 8 feet high. Basically you need a space of 100 feet times twice the ramp width at ground level. And to go up 50 or so feet the ramp will end up with about 16 or 17 landings.

  • Lawrence Gulotta

    My wife used to walk blocks in a few minutes. Now as we are older 500 feet is a painful excess that takes 4 times as long. Plus you chance missing a bus that might run every 20 minutes if you have a longer walk. That means more demand for Access A Ride and cabs which will clog streets with more traffic.

  • ortcutt

    There are costs and benefits to any changes and they differ for different riders. The MTA needs to take all of that into account and balance it as best as they can.

  • Cain McDougal

    More frequent service would likely balance this out.



How Long Will Dallas Take to Make a Better Bus Network?

A lot of cities and transit agencies have been inspired by Houston’s recent effort to overhaul its bus network with a grid of high-frequency routes. One of those cities is its neighbor to the north, Dallas. According to Brandon Formby at the Dallas Morning News’ Transportation Blog, DART is planning to redesign its bus network, with many of the same goals: transforming from […]

Columbus Prepares for a Dramatic Transit Makeover

Columbus, Ohio, population 820,000, is technically America’s 15th largest city. But its transit system has never lived up to its big-city aspirations. The Central Ohio Transit Authority (COTA) carries just 64,000 riders daily. By comparison, Cleveland’s Red Line, a single rail route, carries 26,000 people a day. Now, like Houston, Columbus is considering a reorganization of […]