Last week I posted on the odd phenomenon of one-way splits, where the two directions of transit service are moved some distance apart. My point was that it's a great example of symbolic transit, in that the lines appear to cover more area, in a way that looks nice to developers, but they actually cover less, if you define "cover" as "being within walking distance of both directions of service." If we assume an acceptable walk to be, say, 400m, then as you can see in the diagram below, the area within walking distance of both directions of service (blue in these diagrams) shrinks as the separation gets larger.
Most commenters agreed, but there was an interesting dissent.
The post is misleading. It makes more sense to calculate the round-trip walking distance, in which case the areas covered in blue would be the same.
As I understand it, this commenter thought I should draw the situation this way:
I have to admit I had never heard this definition of coverage before. It implies that if we have to walk 800m to the bus in the morning but 0m to return in the afternoon, we're within a acceptable 400m walk.
What do you think?
I think you dissenter is wrong. Two 1/4 mile walks separated by 9 hours are not the same thing as a single half mile walk, which is considerably more tiring. Also consider inclement weather. If it’s raining, being exposed to the rain or cold for 5 minutes with plenty of time to dry off or warm up before the next 5 minute exposure is far less bad than a single 10 minute shower or blast of cold.
I think that works, but empircal studies would be needed to say whether the user behaviour of two 400m trips is more equal to a round trip combined of 800m, 1000m, or 600m.
The incentive of the 0 m trip for one leg might more than outweigh the longer trip on the way back, or the longer trip could be a greater disincentive.
I could see similar human logical errors coming into play that leads to the lack of backtracking even it it saves time for the total trip. Also for travel patterns that are primarily in one direction I could see different demand patterns for people whose inbound trip is influenced by the short distance vs those whose outbound trips have the short distance walk.
I think the dissenter has a valid point. That is a 0 & 800m walk is basically the same as 2 x 400m. A nuance may be were you are regularly carrying more, such as groceries, on the 800m leg.
When I read this I immediately thought of my walk to BART. Interestingly enough, I dont mind walking there, but I hate walking back. I always consider riding my bike there, and sometimes do, so I dont have to walk back.
However, as with any change in location of a transit line, an improvement for some people (a reduced walk on one leg of a trip) is a loss for others. Overall I agree with Jane Jacobs (in Death and Life) that splitting bus service is a harm to quality of the experience.
Its a disadvantage to most bus services already that you have to cross the street to get service in the opposite direction (this is sometimes but much more rarely the case with subway service), but making someone walk down to the next block to get service in the opposite direction is certainly worse.
I take different buses to work and home. Walk distance and reliability both play into my decision for each trip. If I only have one route to choose from though, walk distance, reliability and waiting time are a lot more important factors at the origin end than at the destination. There’s no case where I would consider the weighted average of the two.
I think there’s also a secondary issue – how self-explanatory is the service? To a potential rider, if there are stops on a street in one direction but not in the other, it means some research is needed to figure out how to use the service, while two way service on the same street is easy to understand.
I would argue (without any proof) that the dissenter is correct…upto that magic point where the one way walk is too far…for me maybe 1km for statistical transit geeks 400m…which is exactly the point you were making only the dissenter is willing to walk more than your hypothetical transit user (all things being nice and hypothetical, I would walk more on a beautiful interesting walk to comfortable frequent rapid transit and less to poor service).
If this was correct, Vancouver’s north-south buses would be just as useful if they only ran in one direction but with twice the frequency on alternating arterials.
One-way networks are less useful because all trips require an 800-metre walk, which is the worst case for a two-way network. With a two-way network, a shorter walk is required to get to destinations closer to the arterial.
It takes less time to get to places on arterials because only this shorter walk is required. This gives arterials a trip-time advantage, and creates an economic reason for destinations to concentrate there.
I think the confusion that Rob highlights outweighs any concerns about walking distance. Suppose you are only taking the bus one-way? Then it doesn’t really matter that the return trip would be shorter. Plus having the line travel on one-way streets makes it much more difficult to figure out where you need to be, and where you are going to end up.
I think that the round trip argument actually is quite valid. I have what may be a unique station, I have a 350m walk from the bus to my office in the morning. Because of construction of a transit center that is supposed to solve all of this (5 years and we’re told 4 more before it’s done), the bus skips that stop in the other direction, leaving me with a choice of a 750m walk to a relatively frequent bus that drops me closest to my house with a one seat ride, or a 900m walk (the last 300m of which are on a parallel street to the first) to a place where I can catch any of 8 buses that get me a two seat ride home. In my mind, it does average out. I would rather have the long walk after work when I have been desk-bound all day, and I don’t care as much if I get cold, hot, or wet when I can change clothes at home. When deciding whether to take the bus or drive and pay for parking, I take the overall experience into account, not just the long pole.
Well I definitely think more thought has to be given to how people treat walking distance to transit. To put it simply, not all equal distances are treated equally.
For instance, I suspect very strongly that the average person is far more willing to walk a large distance from their place of residence to transit (and back) than they are to walk from a transit stop to their place of work or retail or just about anything else. I willingly walked a kilometre and a quarter to and from the C-Train in Calgary every day, but I would have balked at having to walk anything more than half a kilometre on the other end. Walk 1.25 km to transit and 0.5 km from transit, that I would do, but flip that around so that I have to walk 1.25 km to get to my destination after getting off… no way. It doesn’t seem too rational, but we see it all the time when we transit advocates complain about dealing with suburban shopping malls and office parks surrounded by parking lots that must be crossed by any transit users.
So going back to the topic at hand, if the diagram depicts the residential distance to transit situation, then frankly I think I would buy it. On the other hand, if we’re talking distance to the destination, then probably not. Even there though, I’d more willingly put up with a long evening walk than a long morning walk when time is more constrained.
This assumes that distance is irrelevant except for determining whether you’re in the “blue zone” or not. In reality, riders are going to be drawn more from the closer parts of the zone than the farther parts (and some will probably be drawn from outside of it).
With two-way service on the same road, you’ll have some people with a round trip of nearly zero (hopefully a lot, if the road chosen is one with the most intense development, but even if density is uniform you’ll attract a larger percentage of the people that are living/working/etc. right there).
That won’t be the case for anyone if you separate the directions by 400m. The minimum round trip distance will be 400m (not counting distance added by stop spacing), equivalent to being halfway out into the two-way-on-one-road blue zone. Half the blue zone has the same round trip time as before (the farther half), and the other half is now worse off. You also dilute the ability to concentrate development along the route.
And that’s before you consider that one 800m trip is not going to be perceived as the same as two 400m trips, as others have pointed out.
In my experience, the person at the margin making a mode choice between transit and the auto doesn’t think as deeply as the dissenter did about this problem. That marginal decisionmaking person thinks about whether or not the trip is “too long” to them, and if the long walk is in the morning and they have a car, they choose to drive and never get to benefit from the short afternoon trip because they never really figure out that it exists.
CityBeautiful21 just beat me to the punch. I think that rationally, dissenter is correct, but the actual effect of implementing the service will be reduced ridership because the mode-choice decision is being made at the beginning of the trip and households further than the “acceptable walk distance” will choose not to embark on the transit journey.
I can offer my personal experience; when I was a student in high school I took a bus just in front of my house and then I had to walk about 1 km to get to school. The walk was admittedly pleasant in that I had to cross the medieval center of my town.
In the morning there was no problem; I got to school without even thinking about it. Coming back was a different story though.. I definitely hated the walk back…
After so many years if think back I still do not know the exact reason why; but I have to concur that our perception of distances is not fixed.
It’s an interesting suggestion.
Here’s my real world example: When I worked in DC, I would take metro to the closest stop in the morning. In the evenings, I would sometimes walk further away both to get the exercise (and fresh air) and also to make time so I could pay the off-peak fare. I’m not a morning person, so I would never do that in the morning.
So in the streetcar example, if the streetcar could drop me off at the front door in the morning, I would actually enjoy the longer evening walk. But if it were the opposite? That would cause major problems for my schedule.
I think there are 2 major problems with this reasoning:
1) people do not react linearly to distance. If we had something like “probability of using transit due to walking distance to final destination”, I bet this probability, on average, would be lower if you had uneven walking distances above a certain threshold. Assuming a logarithmic probability curve, the excess distance on either outward or inward (morning/evening) would put off more people from using the system than it would compensate with shorter trip on the other.
2) the “probability of using transit” function is most likely not equal for all people. Some are more easily put off from using transit than others according to how long you have to walk. Others might have sharper threshold cuts (making the function curve “steeper”). Because shorter distance to transit is rarely an issue, using one-way patterns have an overall negative impact.
In practical terms, having one-way patterns mean you risk losing users that are very sensitive to walking distance (like someone leaving stores with relatively large packages, or someone who uses park and ride because his/her final destination is very close to transit but would drive if she/he had to walk 500m).
I live in a similar situation. There is a bus stop 3 meters from my front door that takes me directly to work. However, because of the way the route loops on the return trip, it is faster for me to get off at a different stop several blocks away and walk. I return home before the same bus arrives at the stop in front of my house.
My guess is that a person at my house is MORE likely to use transit under these imbalanced conditions than if the stops were both about two blocks away. An imbalance in the other direction (longer morning walk, shorter evening return) would probably have the opposite effect. Therefor, I think these differences come out in the wash.
The dissenter is correct that round-trip is the better way to measure walking threshold.
I think it would partially depend on whether or not the split route increases overall transit service (either increased speed or frequency for both directions) &/or creates some other public advantage (like less gridlock therefore better throughput for all transportation modes).
I think it would be an interesting theory to test out in Toronto’s downtown with the Queen and King Streetcars, if we could derive mobility improvements. Those streets make a natural ‘loop’ between the Don River in the East, and Ronscessvales in the West, and it is a very short walk between those streets for that entire length. http://www.ttc.ca/images/fixedImages/TTCRideGuide_5.pdf
I live just north of Queen street in the east, and work on King St. in the downtown (at Bay). Walking distance between Queen & King downtown is so short that the difference doesn’t figure into my decisions – only whether the increased speed/reliability of the King line warrants the transfer (it usually does). If, as someone posted above, we could increase frequency by doing this, and at the same time reduce gridlock for vehicles on the other street, perhaps that would be a win/win. In addition, this arrangement could make it more palatable to put the streetcars into a ROW, as there would be loss of only one lane per street. Doing this would greatly increase the speed and reliability, which would, I’d imagine, be a win for everyone.
I don’t know if the above scenario holds water – but I realized that the way I actually use those lines looks very much like Jarett’s diagram: when using the streetcar I usually ride the Queen all the way downtown in the morning and walk further to my destination (I’ll transfer only when I can see the King car right behind me at the Caroll St. transfer point), but frequently take the King car on my return trip and transfer onto the Queen car.
I should have mentioned that I agree with the dissenter; in most cases it is the round-trip that should be measured.
My above post was considering where and when that split route could provide added benefit that would justify the added complication it creates.
I think the answer is somewhere in between. At shorter distances, RT total is less than longer distances. I wouldn’t mind walking 1/8 mi both ways OR 1/2 mile then no walk to reach a loop. But whereas I wouldn’t mind 1/4 mi each way, I think I would mind a half mile walk on cold winter days even if it meant no walk for the remainder of the commute. Also, I think preference would skew towards preferring a longer walk in the afternoon instead of the morning when people want to get to work at a particular time.
Most importantly, even with this alternate understanding, the loop does nothing to improve access to transit. At *best* it’s the same as if both directions were at the same spot. Furthermore I think loops take away the simplicity factor that, in general, has an underrated influence on transit use. So, barring some major engineering reason why a loop is preferred to a service on one corridor, I still don’t see a reason for loops.
The original argument stands if we consider the coverage for the worst case trip chain (going in the same direction on the wrong line). The dissenter’s point stands if we assume the customer will be reversing direction at the destination, and one doesn’t mind a single long walk.
How should we weigh the worst-case trip? It certainly wouldn’t weigh at all if the one-way split is at the end of a dead-end line. It should probably be weighed the very heavily for a commercial area where many trips will involve customers continuing in the same direction.
Seriously? I can’t believe they are even thinking about this. Don’t these people ever ride the bus?
Since the problem of splitting route direction is very similar to poorly spaced stop pairs, there should be a very large amount of anecdotal and experiential experience with this ’round trip’ vs ‘one way’ question. (Which would be interesting to see added to previous stop spacing discussions).
I think it is not appropriate to simply use a round trip distance. From personal experience, I have had are harder time with longer walks at the start of my trip – since I’m racing the clock/bus schedule. I do precieve a long distance to walk *to* the bus stop as more of a barrier than a long distance to walk *from* the bus stop. There is nothing worse than walking over 1/2 a mile to miss your bus.
Well, very simple:
“Out of sight, out of mind” (or “aus den Augen, aus dem Sinn”, as we say in German)…
It is psychology. When stops in two directions are too far apart they do not work.
We do all to change such situations, cause you get less than half the ridership with such situations. Maybe some school kids and people who could not afford a car. So yes, if you want to offer just any transit you can do it. But if you want to get riders, it will not work.
On the whole, the dissenter is right.
There are two other aspects though.
First, most people are more willing to walk at one time of day than another, or in one direction than another.
I daresay most will walk quite a distance FROM the stop TO the destination, and will not walk very far FROM the source TO the stop, for multiple reasons, which others have said. (This matches up with commute behavior too.)
Second, if the “out” stop and the “back” stop are too far apart, most people won’t even consider it, because it’s too unclear and confusing. A one-way pair has to be CLOSE.
I consider the walk from transit to the destination to be more tolerable than the walk from the start to transit. Why? It’s useful to be able to wait inside and do something productive, and use a real-time tracking system to decide when to leave to catch the bus or train. With a longer distance, there’s more room for error (lights don’t go your way) and thus a greater ability to miss the bus, or a longer wait needed to compensate. When disembarking, there is (usually) not the same pressure.
And I would echo what Scott Wood says above–if the lines are separated by X meters, than the minimum round trip distance anyone will experience is X meters. While the total area in the acceptable zone is the same, the average total walk distance within the zone is higher. The boundary case is when the lines are separated by the walk distance–only the area between them is still acceptable, and the average is the same as the maximum.
In practice, I’ve little objection to transit lines in one-way grids with narrow blocks, particularly if counterflow lanes (or opposite-side boarding) is used. But beyond that, the quality of service degrades quickly.
Scott Wood makes a very good point most are overlooking. The one-way pair guarantees that everybody will have to walk some distance, whereas two ways on the same street give some folks the possibility of not walking much at all.
My hunch is that round-trip distance doesn’t really matter. It’s the individual one-way distance. You could say that a 0m walk and an 800m walk average out to 400m each. Likewise, if you stick one hand in scalding hot water and the other hand in liquid oxygen, your temperature probably averages out to a tolerable level. Granted, that’s an extreme comparison, but that’s the danger when you get too into statistics and averages and ignore simple reality.
There is a distance (whatever it is, five feet, five miles) that is simply too far for some people to walk in one go. Whether it is due to actual physical disability, or because they just don’t want to walk that far, if you make people walk beyond a certain distance, then they simply won’t. Even if you add up all the little walks they make in the day, and show that the total is the same (or more) than if they were to walk it all at once, they still won’t do it. It just appears to be too much.
To use another example, yesterday I walked about 10,000 steps during the course of the day (I have a pedometer, so I can keep track). Later in the evening, I got on the treadmill and ran another 10,000 steps in one hour. Technically, the second 10,000 didn’t require any more (total) effort than the first 10,000. But you’d be foolish if you actually believed the two were the same.
@Robert Madson: the thread mill analogy is not quite a good one. The human body acts in two capacities: producing short-term energy for intensive exercise or metabolizing ATP through normal digestion.
Is for that reason that someone walking 4km a day on his office job as a messenger will never get “fitness” benefits someone exercising beyond the “replacement rate” gets by running 4km in a thread mill within 45 min.
I suspect the true answer is somewhere in between: averages are taken into account, but the catchment area shrinks on line splits.
I am in Kitchener-Waterloo, where we are planning some route splitting for LRT but already have some split bus routes due to one-way roads.
I have a high tolerance for walking to meet a bus (if it’s reliable.) On the way to a direct bus, I have a 1km walk, and I tackle that without concern. It helps that the destination is extremely close to my office.
On the return, I have an extra 400m and need to cross 2 major roads. And I frequently balk at this, preferring instead to make a connection that gets me much closer to home.
Does this mean that the average doesn’t matter? Or just that I’m pushing my tolerance limit already at 1km?
(You might also ask, why not use the connection route to get *to* work? Ah, that’s where route frequency raises its ugly head: I can plan to catch a reliable 30 minute bus, and use it to transfer to a high-frequency main line… but taking a high-frequency main line to connect to a 30 minute bus for the last leg forces me to buffer for trip time variation, lest I miss the connection and be forced to wait half an hour.)
one thought of mine: Is ridership equal in both directions? While most people go directly to work, it’s common to go somewhere else on the way home, therefore coming back on a different bus than the one taken to work.
I was the dissenter. I have to say, I’m impressed with the seriousness and professionalism of this blog. Kudos.
Just don’t know. But I want to hear about successful split rapid transit lines (bus or rail). Do they exist, and how far apart are the lines? Does local transit ever go counter-flow in traffic? Does something that confuses people with largely invisible buses (in traffic) work better with dedicated rights of way?
In many situations it would make surface rapid transit easier to implement where roads are narrow. But easier is not necessarily better.
I think we had the discussion before, but distance is only half of the story. It makes the situation more complex, but actual walking time has more of an effect than absolute distance. And there are the rules of thumb, that the transit stop should be within 5 minutes of walking.
Now (I know it gets obscure and illogical), for many people the “5 minutes” are even a variable duration; “felt” 5 minutes can be more if the path is nice, easy to walk and refreshing (and even depend on the time of the day. It can be shorter if the path is rotten, crossing highways, noisy etc.
So, the dissenter has some points, but it may be more on the psychological level than the absolute distance level.
Slightly off topic: about ten years ago, Beijing began replacing a lot of pedestrian overpasses with tunnel underpasses. The local rumor mill “determined” that the reason for the change was that scientists had determined that a pedestrian who would rather dart across a street at-grade than climb the pedestrian bridge would still take an underpass if one was available. Even if the pedestrian knows that he’ll have to climb stairs at the end of the tunnel, the rumor mill reasoned, he’ll discount future costs (even obvious ones). By the time the pedestrian has reached the stairs at the end of the tunnel, Beijingren concluded, the cost of climbing those stairs is effectively sunk.
I don’t think that the Beijing rumor mill actually consulted scientists on this one (I heard it from my barber, who heard it from another customer), but I suspect that it’s possible that some commuters with a 200m walk in the morning may discount their impending 600m walk in the evening. Anyway, that’s why God created bikeshare…
I think that this discussion reveals the pathetic lack of good research into the whole issue of walking distance. Or if there is good research, the failure to disseminate the results.
Until researchers perform a variety of revealed (observed) preference studies in a variety of locales, we are like theologians arguing about the unknowable.
I know my own preferences (similar to Engineer Scotty’s) but without valid research, I would hesitate to impose that view on the system. Of course such research isn’t cheap or easy to do.
One thing no one has mentioned yet is the impact of hills. It’s better to arrive on a street that is uphill, and leave on a street that is downhill. This way, you minimize uphill walking; you might even be able to walk downhill both ways.
This is important in Downtown Seattle, where the area roughly centered around 3rd and James is extremely steep, and most of the avenues are one-way. So a bus that arrives on 5th Avenue (which the furthest uphill) and leaves from 3rd Avenue is better than the other way around. Not coincidentally I am sure, there were more buses arriving than departing from Downtown on 4th and 5th Avenues, and more departing than arriving on 2nd Avenue, when I lived there.
Maybe not round trip, but 200m to the bus and 200m from where you are dropped off might explain why many transit systems place stops so close together.
The real point that the dissenter is missing is this: The 400 meter distance that a “typical” transit user is willing to walk is really the mean of a probability distribution function. Some transit customers are willing to walk 600m, but some are only willing to walk 200m. As a side-point, there are a lot more people who are willing to walk at least 200m than who are willing to walk a least 600m. (since every person who would walk 600m is by definition someone who would walk 200m).
For the sake of argument, let’s assume that someone who would walk 400m one-way would be willing to walk 800m round-trip (I think that’s optimistic, but we can deal with that later). This means that for a system with 400m directionally-segregated service, you lose all the riders who would walk less than 400m round-trip, because that is the absolute minimum distance someone could walk.
In other words, of all the people who contribute to the 400m average walk distance, those who would walk at most 200m won’t ride the bus, no matter how close they live to a stop. Not only that, but people with limited mobility who are looking for a new place to live will find it geometrically impossible to find a location within their preferred walking limit.
Where transit-focused uses like shops and higher densities of residential and commercial development concentrate nearer to two-way transit lines, the average walking distance is much shorter than for one-way transit lines.
In places with uniform residential development, the round-trip average walking distance with a two-way transit line is 1/2 the distance between transit lines, shorter for people living closer to the line and longer for people living further from the line. The round-trip walking distance with a one-way transit line is the distance between transit lines and is the same for everybody.
A system of one-way transit lines might make sense if (1) the one-way lines are at least twice as close as the two-way lines would be or (2) a two-way line is so infrequent that doubling its frequency and converting it to a one-way line reduces the averaging walking time by more than the average waiting time.
A special case: lines which split into a loop at the end of the route. I am thinking of the F-line bus which connects San Francisco to UC Berkeley. At the Berkeley end it does a 4 km loop around the perimeter of the campus, which is about 0.7 km x 1 km.
I think this worked pretty well. The coverage was much better than any single line could have been. I spent my time just outside the loop, so I generally used the same stop to arrive and depart (meaning that my return trip was five or six minutes longer than the trip there). Someone closer to the middle could arrive on one side and depart on the other– or even choose, depending on how they felt on that particular day.
Another (minor-ish) advantage is that none of the stops required you to cross the street from the campus (though of course that worked against me).
I should add that it wasn’t a terribly popular route, since most people making that trip took the train (the station was at the beginning of the loop). But the school provided a free bus pass, but not a train pass, and it worked reasonably well for me.
Any metric breaks down on the microlevel; there will be those who won’t mind the half-mile walk on the front end (young, healthy students in good weather) and those who’ll dread it (an arthritic grandmother in the rain).
The better metric in the aggregate, then, will be the one that measures the prefferred outcome of the population being served, though there will probably have to be access accomodations for those whom the system challenges the most, and conversely, flexibility to mildly inconvenience those whom it challenges the least.
I think the point made by Hat (and others) is key:
“…, of all the people who contribute to the 400m average walk distance, those who would walk at most 200m won’t ride the bus, no matter how close they live to a stop.”
Focusing on average and/or typical riders, obscures the fact that most people aren’t average, and most people aren’t typical. By evaluating the split route based off of how it would impact the ‘average’ rider ignores the fact that there is a significant constituency that will demand a higher level of service than your ‘average’ rider.
Articulating the question as:
“in a split configuration, who is included (medium to long walkers), who is excluded (short walkers or mobility impaired)?”
would be better than:
“how would this affect the average rider who is willing to walk 400m one way to a stop?”.