How Can We Study Things in Isolation? They’re Connected!

Whenever I present a bus network redesign plan, I’m always accused of ignoring important things.  How can I design a bus network, people say, without also planning for bus lanes, or bicycle parking, or road pricing, or parking policy, or urban structure? These things are all connected, they say!

Yes, they are all connected. But despite being connected, many planning tasks are separable:

  • Two projects are connected if they affect each other’s outcomes. For example, a network redesign and a bus lane project will certainly improve each other’s benefits over what either could do alone.  A rail line and a bus line parallel to it are competitors that will undermine each other’s outcomes, so they are connected too.  (Deep ecologists would say that almost everything is connected in this sense.)
  • Two projects are separable if one can be done before the others, and will achieve some benefits  by itself, even while waiting for the other connected parts to happen.

I know why people get anxious about this, because we all see situations where things were separated that really were inseparable. A rail line and a freeway are built side by side, without noting how each will reduce the demand for the other.  Maybe bus routes are designed without thought to connections between them, or worse, great infrastructure for bus connections gets built in a place where it’s not actually useful to the bus service.  A public transit service ends at a political boundary even though the demand doesn’t end there.  These are all examples of projects being separated when they were not really separable.

On the other hand, no human brain can focus on everything at once.  If we tried to do bus network redesign, fleet modernization, bus lanes, bike parking, road pricing, and parking policy as part of one project, it would never get off the ground.  Just co-ordinating the hundreds of experts needed to deal with all dimensions of such a project would consume most of our effort.

More important, in any project, everything moves at the speed of the slowest element, which is why it so often takes forever to get things done.

So separating projects is the only way for anything to happen soon. We are not denying that everything is connected. We are saying we have to start somewhere, and make some progress, even as other pieces of the puzzle are in the works.

Like any plan, a good network redesign effort requires clear thinking about separability.  A redesign is mainly a revision of the patterns in which buses run, but this process always identifies infrastructure and policy changes that are also needed. Sometimes these are truly inseperable:  The specified number of buses can’t meet at point A unless the facility there is enlarged to have room for them.  If the plan requires people to change buses at an intersection, we need to make sure there’s shelter and safe street crossings, and so on.  If the fare structure is penalizing changing buses, that needs to be fixed if our plan wants to encourage that.

But we fight to make the list of inseparable things as short as possible, because every time we decide that something is inseperable from the plan, that becomes one more thing that could stop the whole plan if it hits some kind of snag.  We ask:  Would the redesign still be possible, and worth doing, if some infrastructure or policy element doesn’t get done?  Sometimes this leads to good tactical thinking:  Can we do this necessary interchange quickly on-street, even while waiting for the funding and consensus to do the permanent facility that’s really needed?  Can we make some patches to the fare system while the ultimate system is still being worked out?

Another test is:  Does doing Project A without Project B actually make things worse?  If not, this is another signal that the projects are probably separable. The answer, for bus network redesigns, is almost always no.  By itself, redesign will achieve significant improvement even as it leaves a lot of other frustrating problems in place.  But getting it done may make other improvements politically easier if the result is that public transit is more visible, more used, and thus more widely valued.

So when people respond to a network redesign proposal by being angry that it doesn’t talk about bike lanes, electric buses, or road pricing, they’re confusing connectedness with inseparability.  Our network redesign study isn’t ignorant of those things just because we’re not talking about them.  We’re just talking about something different, something that’s also important and needs some attention.  A good network redesign, if allowed to succeed, will make all those other things easier.  And in any case, the redesign itself is important enough, and hard enough to explain, that it deserves the public’s full attention for a few weeks.

Everything is connected, but many things are still separable.  That’s a good thing, because if they weren’t, nothing would get done.

10 Responses to How Can We Study Things in Isolation? They’re Connected!

  1. Kenz July 12, 2018 at 6:26 am #

    I’ve always been a bit challenged by studies of first/last mile solutions in isolation rather then in the context of the entire network. Seems the two are inseparable.

    • Henry July 16, 2018 at 9:06 am #

      Depends. If your *only* thought is to connect the last mile of a very frequent service, you don’t really have to worry about the network other than how to time transfers with the very frequent service.

      In practice, a lot of routes do double duty as neighborhood or crosstown connectors as well as feeders, or they feed multiple lines, or serve dense corridors not served by the frequent service, and then that becomes more involved out of necessity.

  2. Sick Transit July 12, 2018 at 7:55 am #

    Thank you for a well-written article. As someone who spent a lot of his career designing large-scale IT systems I was frequently baffled by people who could only focus on local behaviors; it was a major (and not always successful …) challenge to encourage them to step back from a proposed fix or enhancement and view it in a larger context.

    It’s interesting that you mention the issue of transfer charges. Our regional system (Philadelphia) is currently embroiled in active plans to actually _increase_ the cost of certain transfers even though the system’s layout is such that changing vehicles is almost unavoidable for many passengers. Various advocacy and consulting groups have pointed out the counterproductive nature of transfer charges yet the transit authority is charging ahead regardless of the consequences.

    Another issue where separability is ignored occurs in the regular calls from ideologically-driven politicians to scrap Amtrak’s long-distance routes. The reasons most often cited are that they require higher subsidies than corridor trains and that “hardly anyone rides them end to end”. Both points are true so far as they go, but again are not being made in the context of the entire rail network. Severing those secondary routes would eliminate passengers who use them as connectors to or from the corridors, thus reducing ridership and revenue. In addition Amtrak’s figures indicate a seat-turnover ratio of about 2.5 on long-distance trains, meaning a considerable part of their business comes from passengers who don’t ride end to end but DO use long-distance routes for intermediate travel. Focusing solely on overall revenue and end-to-end usage is as poorly conceived as deciding to shut down I-95 outside of the Northeast because hardly anyone drives from Maine to Florida.

    • Mike July 12, 2018 at 10:14 am #

      At least 50% of Empire Builder riders get on and off in North Dakota and Montana, which has no east-west bus transit unless you go 90 miles south to the Interstate. I’m skeptical of the claim that many regional riders are taking them to/from long-distance trains because it seems like the majority of riders are on regional routes only. But the market for long-distance trains is certainly not end-to-end. Many people take them from Spokane to Montana, northern Oregon to southern Oregon, Seattle to Oakland, etc. Amtrak’s reservation system is limited in that it can’t sell a ticket for a partial trip if the seat is booked anywhere in the segment, at least not until the train is moving and the seat is empty. But that’s a failure of the ticketing system, not an indication of how people ride. When I’ve taken the Empire Builder from Seattle to Chicago, it’s only a quarter full in Washington but gets 100% full in North Dakota, to the point that they apparently have to turn people away, or at least every seat is filled.

      • asdf2 July 15, 2018 at 8:27 pm #

        One could argue that if the Empire Builder ceased to exist, one of the long-distance bus companies would fill in the gap, at least in getting people between North Dakota and the nearest large city (I guess, Minneapolis). If enough people are buying tickets to fill up the train, which has the capacity of several buses and passes through the North Dakota stations at awful hours in the middle of the night, the market would most certainly exist for a bus service that would pick people up at hours when people are awake, and likely, have better speed and reliability than the train, and cheaper tickets.

        Of course, the flip is is that people going between North Dakota and Seattle would be largely out of luck. At best, they’d have to take an 8-hour bus ride to Minneapolis, then hop on an airplane.

  3. Alon Levy July 12, 2018 at 2:22 pm #

    Yes and no. I do think certain aspects need to be discussed at the same time as a bus network redesign – namely, various treatments to increase bus speed. These are germane to the network, in any of the following ways:

    1. If the buses are faster, this gives you more revenue-km, which you can choose how to distribute across the network: running more routes, raising base frequency on the entire system, or raising frequency even more on the strongest routes.

    2. Stop consolidation affects route spacing. It’s invisible if the buses run on a relatively widely-spaced grid of two-way streets, as in Vancouver or Chicago or Los Angeles, but if there’s an imperfect grid, or if buses run on couplets, you might end up seeing a route intersect several other routes in rapid succession.

    3. Dedicated lanes and signal priority may be more feasible on some streets than on other streets, and then you’d want to design the network to use these streets as much as possible.

    • Jarrett Walker July 13, 2018 at 1:24 am #

      Alon. These are great examples, but all are separable in practice if not always in theory.

      1. Yes, when bus lanes comes come later they will release more resources, and at that point we can decide how to spend them. In fact, it’s good to make this nexus explicit as part of the bus lane project, promising more service with the resources freed. But this isn’t a reason not to proceed with a network design before we have the bus lane, and get what benefits we can. In most cases we can achieve a lot just with the network design step, as we are certainly doing in Dublin. We get a 20% increase in where people can get to in 45 minutes, independent of bus lanes, because traffic delay is the same before as after. That’s worth doing. The bus lanes can be the next step, improving this even more.

      2. In theory stop spacing affects route spacing, but in practice the scales of the problems are different.
      Close route spacing is massively more expensive than close stop spacing, so the low-hanging fruit is the route spacing, and that’s what we address in the redesign. Sometimes we do them at the same time, as recently in Richmond, but it’s always fine to get route spacing right first, and come back to stop spacing later.

      3, Yes, there is a nexus here and sometimes we’ll move a bus route a block over in the context of better infrastructure, but again these issues are at different altitudes and are therefore mostly separable. At the altitude of a network design, it doesn’t matter much if a bus route is moved over by a block. Usually we can note that this route might later move the route a block, but this has little impact at the higher altitude where solve the problems of network structure that drive the big design. If you might later move a bus route over by 1/2 mile to get to better infrastructure, but in practice we don’t do that because it means it’s no longer the same route and no longer serving the same places.

      Thanks for asking these questions, because they are great examples of things that look inseparable but that we can in fact separate in practice, with a good interim outcome and little harm to longer term goals.

  4. Georgist Economist July 15, 2018 at 8:51 am #

    “Can we do this necessary interchange quickly on-street, even while waiting for the funding and consensus to do the permanent facility that’s really needed?”
    “Nothing is as permanent as a temporary solution.” I’m certain you have encountered political bodies that work reactively, i.e. always focus on whatever is burning with the most smoke at the moment, rather than having any foresight and focusing on whatever has the best return to investment on the long term. Chronic deferred maintenance is one tell-tale sign of these processes. Wide disparities between a few “display” lines and the bulk of the network is another. (My guess about the underlying process: the occasional influx of money is spent on dazzling top-of-the-line projects, so that the politicians involved can …essentially, brag about it, even while the rest of the network is crumbling (mostly figuratively, sometimes literally). Mere maintenance is not sexy, thus the minimal possible budget is allocated to it. This creates a cycle of expensive, small-scale projects that degrade very quickly for lack of maintenance. The bulk of the network is sometimes run with an obsolete fleet that would be considered to be past its end of life, simply because it’s already there and cannibalizing for spare parts can keep a gradually shrinking fleet running despite zero capital (re)investment.)

    This is not necessarily an argument against separating the separable. If you separate the projects, and the “temporary” solution is merely bad, then the second part probably won’t be built for decades (“never”). If you don’t separate the projects, or otherwise the second part gets built, then the permanent facility will be some silly gold-plated thing that is initially very flashy and follows all the architectural fads, but will look run-down within a few years because it’s difficult to clean and get replacement parts for (i.e. uniquely curved panes of glass). And to top it off, the actual function of the facility could sometimes be downright impaired by these flashy (and expensive) choices for the sake of visual design.

  5. MB July 16, 2018 at 7:31 pm #

    I think part of the reason for the poor performance of US transit systems, is because much of the planning is done in isolation of other factors.

    Take, for example, the ridership success in Kingston, Ontario. They did not just redesign the bus network in isolation. They also implemented a parking plan in the downtown and university areas, where monthly parking must be priced higher than a monthly transit pass. They also reduced parking supply.

    Such parking TDM programs are done throughout many Canadian cities, including Calgary, which has some of the highest parking rates in North America, yet one of the highest transit usage rates.

    American cities do not do this nearly as well, and it shows with poor ridership levels.

    All this can be done in isolation, and yes you need to get projects off the ground. But when done all in isolation, your results may be worse.

    Kingston has actually saw the share of commutes by transit increase. Has any US city that has done a bus network redesign seem the share of commuters using transit increase? I think maybe Seattle? But Houston is probably stagnant or in decline. Portland has been in decline for decades.

  6. Vladimir July 21, 2018 at 9:03 pm #

    Excellent points! I would add that if network redesign comes first it creates a roadmap for future infrastructure changes. If there’s a bunch of routes on the street converging and separating from each other in every which way on intersections it’s hard to come up with a transit priority plan especially if it is not clear which route is more important than the other. If there’s one route and we know it”s important it is simplier to prioritize it.

    Also after the network redesign it is possible to separate some route improvements as priority goals. For example if we have a north-south subway line running through the city, then improving surface transit that compliments it might not be the top priority. Instead the city could concentrate on improving frequent east-west lines that are only served by surface transit.