>Planners can set minimum and maximum amounts of time to be spent at each stop, and buses might even be told to skip certain stops during crowded runs
>Passengers might be encouraged to wait for a following bus, with the inducement that it’s less crowded.
>Northern Arizona University improved its service by abandoning the idea of a schedule altogether and delaying buses at certain stops in order to maintain even spacing.
FYI the real solution is bus lanes so busses don't get stuck in traffic. But for that, you need to make space that isn't for cars. So you won't get it in Arizona.
In Spain most city roads with 2+ lanes per direction have a dedicated bus lane, yet you still get bus bunching and spacing adjustments anyways.
- Imagine bus A is followed by bus B, it's 5PM and people are leaving work simultaneously
- Bus A spends a lot of time on each stop picking passengers up
- When bus B arrives, fewer people are waiting and it tends to take less time
- Eventually, bus B catches up to bus A so A must skip stops to maintain proper spacing
As for why this doesn't annoy people: if both buses are already close together, any ETA changes from spacing adjustments are minimal, and bus A will show no destination on the front display, so no false expectations are created.
If there's bus lanes the bunching happens much more slowly. You fix it by giving slightly more time than needed to the busses for their routes on the schedule, say, 20 seconds per stop. That way if they lost some time on some stops, they fetch it back later. With the inconvenience that all the routes take longer
Could you standardize the dwell time? Hit a max and then maintain regularity
except that bus A must often go to stops that it would skip, if any of the passengers want to get off there.
Bunching also happens on things like subway lines that have no other traffic. Many passengers are irrational about forcing their way onto already overcrowded trains.
>Many passengers are irrational about forcing their way onto already overcrowded trains.
the train that leaves first gets there first
----
although, entering a station once with a friend, we ran for an arriving subway train with a friend. they got on but I did not. however, I was familiar with the station we were in, and the destination station and I realized....
so I ran down the platform, continuing in the same direction, and before I reached the other end of the platform, as I expected, the follow-on train arrived. I boarded it, and now I was ahead of my friend schedulewise: blew their mind when I was waiting for them to exit at the destination station.
> the train that leaves first gets there first
I understand this is what motivates people, and I also understand that, when asked, they would claim that this is in their rational self-interest. As I also wrote elsewhere, I tend to disagree or at least question this assumption in general. Are you really in that much of a hurry? If you have a very hard deadline, wouldn't it have been in your rational self-interest to leave home five minutes earlier? etc.
> the train that leaves first gets there first
Yeah, but in this bunching scenario I prefer taking a train with lots of space than arriving maybe two minutes earlier packed like a sardine. Having screens which also show when the following trains will arrive helps a lot of course...
I don't understand how this is possible. Trains are single track in most subways and metros, and definitely do not stop before an already waiting train.
I think the implication is that the time between two trains was little more than the time it took to walk along the whole platform, which the friend still had to do after getting off, plus someone familiar with the stations not needing time to orient themselves and find the intended exit.
they climbed on the last car of the first train
i ran down the platform and, without any other waiting, climbed on the first car of the next train
they got off the last car of the first train and started to walk in the same direction as the train. while they were walking, the first car of the next train, me, rolled well past them before stopping.
Tell me you don't take public transit without telling me you don't take public transit
Getting on the train that's physically in front of you, no matter how crowded, is the rational decision basically everywhere outside Japan, where you can have metaphysical certitude that the next train will show up as scheduled.
It's not, when you know through announcements that the next train is right behind it and will be much less crowded.
It's not as irrational as all that.
Subway system here has an amazing propensity to send random trains on express tracks, especially during peak traffic. I understand that this is done to alleviate congestion, but the net effect is that when you see a train going somewhere you want, you _seize the opportunity_.
It's a coordination problem - taken in isolation, it's not irrational. If only you refrain from doing it, and 37 other people do, you will still have to wait for the next train, and the train in front of you will be almost as late anyways.
The wait for the next train is one factor, the overcrowding is another one. I'm usually not in such a hurry that a minute's wait would make a difference. But the next train is very likely to be much less crowded and much more comfortable to ride on. From my point of view the decision is clear, in isolation, just from my own selfish point of view. And I think many others are making a choice that makes them unhappy. (Train systems differ, the one where I live has sufficient capacity that you rarely get two overcrowded ones back to back. I know there are places where this does not apply.)
> a minute's wait
Is your subway system running trains every 1-2 minutes (30-60 trains per hour)? Mine has trains at 5-8 minute spacing at peak times and 8-12 minutes off-peak.
No, it's running at 3-5 minutes at peak times. But bunching significantly shortens the interval between the delayed, overcrowded train and the empty one right behind it.
Bus lanes exasperate this problem by fast tracking buses to common choke points. It's one of the very few disadvantages of bus lanes.
What causes the choke points?
Passengers getting on and off, intersections, other buses.
Busses stopping or slowing down at intersections is a bug that should be fixed. Other busses is a symptom of the problem, not the cause.
Passengers getting on and off can be a cause, but that is probably a symptom of boarding accessibility issues.
Yes, as long as buses don't travel in public or pick up passengers, they'll be extremely efficient.
What?
[dead]
Another element is fast fare payment. The faster you can load passengers the lower the penalty for every stop the better
This involves smart cards and fast electronic payment methods. Also removing or discouraging cash payments.
It also might involve designing buses and stops to get people on/off buses faster
I miss tokens. CTA used to use them. Dropping a token in the fare box as you got on the bus was faster than any app or smart card.
> FYI the real solution is bus lanes
The article literally shows that the problem is crowds of people. After work you have 40 people queuing for bus b at every stop so bus c catches up. traffic congestion would mostly affect both buses.
The article shows no such thing. The crowds of people are not the root cause most of the time - if the buses would come at regular intervals, statistically they would all have the same delay because of passengers. But if a bus is already delayed ("if Bus B is delayed by traffic congestion, it incurs a penalty"), there will be more people waiting, which will slow boarding, which leads to more delays, which leads to even more people waiting at the next station etc. But of course, the cause for the initial delay is not always traffic congestion, it can also be a bus breaking down, a large group (e.g. a school class) boarding the bus etc. So dedicated bus lanes (and other measures such as buses/trams being able to influence traffic lights) can't completely eliminate bunching.
Maybe you haven't seen a big crowd. During the day there can be such crowds going to work after 7am that bus must wait for ages then complete cutoff at 9am then some bumps around lunch then another craze at 4-6pm and this even without irregular local events like as you say groups of students
Bus lanes are good but it's driven by crowd sizes and a lot of combatting bunching can be done without adding bus lanes
Where I live the interval at around 7 am is 10 minute, which drops to 15 after 9 am or something and is 20 at the most chill times. Haven't ever seen an empty bus or really overcrowded one either.
Where I lived recently the interval at peak is like 4 min most. One stop boarding can take 3 min when stop is full. My bus skipped my stop until 9am because bus after bus are full. Skipping stops is one way to reduce bunching I guess
EDIT Who said there's only buses? There are cars, taxies and subway
It's always possible to add more buses or people can just go queue at the previous bigger stop it's 2 min walk away and it always stops there.
The point is that crowds cause bunching. I don't know what is your point
That sounds like it's at capacity and buses alone don't cut it anymore.
> FYI the real solution is bus lanes so busses don't get stuck in traffic
The real solution is to stop using transit to move human misery from one place to another. Get everyone an individual car (a self-driving at this point EV, of course) and redesign cities to be human-oriented, not transit-focused.
There is NO mathematical way to make buses robust. They will always be some combination of too slow, too expensive, or too inconvenient for most people.
Buses have low average speed because they need to stop often. And you can't make bus stops more infrequent, because people won't be able to walk from their homes fast enough. And doing complicated systems with local/express buses just wastes time during transit.
Buses can't be frequent, because the average daily load is already just around 15 people per bus. And this is with longer off-peak intervals.
Etc.
> Get everyone an individual car (a self-driving at this point EV, of course) and redesign cities to be human-oriented, not transit-focused.
These two things are in direct, violent contradiction. Mass transit is what allows human-oriented city design. Individual traffic requires car-centric, human-hostile design.
> Mass transit is what allows human-oriented city design.
What exactly is "human" in dense human anthills with 30-story towers?
> Individual traffic requires car-centric, human-hostile design.
Cars have people inside them. And unlike buses, cars allow people to have much more living space, cars are available at any time of day and night, you don't have to wait for a car for 30 minutes, etc.
Cars are human-friendly! Far more so than ANY other transit mode.
And cities should be built for humans, not for transit and bikes.
> What exactly is "human" in dense human anthills with 30-story towers?
Living with and regularly interacting with "people" instead of living in a personal convenience store.
> Cars have people inside them.
Cars kill people. A lot. Like, orders of magnitude more than literally anything other mode of transportation. If you die in a car, it was almost certainly the fault of a driver. If you die walking or cycling, it was almost certainly the fault of a driver. Cars a human-hostile in the most literal sense. Cars kill people. The solution to the problem of cars killing people is not putting more people in cars, because cars are happy to kill people in cars. Because cars kill people.
> And unlike buses, cars allow people to have much more living space
Please elaborate on this. What does "living space" mean to you?
> cars are available at any time of day and night, you don't have to wait for a car for 30 minutes, etc.
Buses only exist because so much money has been spent on car infrastructure, and they are the only mode of public transportation that can directly use that car infrastructure. Once you stop spending insane amounts of money on car infrastructure, more reliable transit can be built (trams, subways, heavy rail, cycling infrastructure, walking infrastructure).
> Cars are human-friendly! Far more so than ANY other transit mode.
Have you - heck, have anyone you know - ever made a friend while driving? Or does human-friendly not include being friendly with humans?
> And cities should be built for humans, not for transit and bikes.
Kachow!
> What exactly is "human" in dense human anthills with 30-story towers?
Certainly an interesting question, but completely irrelevant to the discussion. Unless you're really arguing against big cities rather than against mass transit, but that's an entirely different discussion.
> Cars are human-friendly! Far more so than ANY other transit mode.
No. Cars are human-hostile. Far, FAR more so than ANY other transit mode. Foir one thing, they kill FAR more people than any other transit more. But also because they require So. Much. More Space.
Sure, it's nice to have your own private space while moving around (not worth much when you have to concentrate on driving and are thus stressed, but let's assume that will be alleviated by self-driving cars) but you're not alone. There are 100,000 people who also want to move in the same space. And thus everyone gets a big upside once and a downside that is small but multiplied by 99,999.
The result is this: https://i.dailymail.co.uk/i/pix/2018/02/07/16/48F74A0F000005...
And this: https://lede-admin.usa.streetsblog.org/wp-content/uploads/si...
Now does that look human-friendly to you?
> And cities should be built for humans, not for transit and bikes.
But what you're arguing for is to build them for cars instead. And we have done that and we know the result is horrible. And whe have done the other options too, and we know they are much better.
>What exactly is "human" in dense human anthills with 30-story towers?
You don't really need 30 story towers to have mass transit. Row houses with 3 stories and a few apartment complexes here and there is enough density to not spend forewer to get somewhere and also not necessarily need a car.
None of the claims in this comment are grounded in reality.
On the contrary. I have researched every statement behind them, and ran simulations to prove them. And I was not actually the first person to do that, what I said is well known.
Urbanists just prefer not to talk about it, it complicates the propaganda. After all, urbanism is supposed to have no downsides. It's like violence: if it's not helping, you're not doing enough of it.
Some evidence to support your statements would be very helpful.
As much as I hate buses (trams are better), you seem to hate people more. Society doesn't have to accommodate such tendencies.
No. I want people to be happy and the society to flourish.
Some of the proposed solutions are problematic. A public transport systems absolutely needs to be reliable for the people who use it.
Skipping stops is the worst in that regard and breaks the whole point. No schedule causes issues downstream, since now there won't be a schedule to depend on when needing to switch to trains or other busses.
But in general, the only thing to realistically improve without decreasing reliability is the amount of time spent at a stop (also mentioned in the article).
All in all, I see these suggestions as "what to do in a worst-case scenario", i.e. if the service already has major issues.
No-schedule works fine if (and only if) service is sufficiently frequent, say every 5 minutes. The overwhelming majority of intra-city trips will have 3 transfers or less in a well designed bus network and when planning to catch a less frequent service, it's acceptable to bake in a 15 minute safety margin.
Is someone going to regularly take a trip that requires them to take 4 buses? I largely avoid trips that would require even a single bus transfer (admittedly, the Boston bus system is pretty terrible, meaning bus transfers are daunting when contemplating trips).
Yeah, 4 buses sounds terrible. I already don't like it when my trip to that one fancy office is bus-metro-tram instead of just bus/tram and metro. I would totally hate if buses in question were overcrowded, smelling hobos and not having a schedule.
Is this the usual European supremacy propaganda? Yes, of course it is.
Here in Graz, Austria, if nobody wants to get off at a bus stop and nobody is there waiting for the bus, it continues right on to the next stop. It all seems to work, somehow.
Same in London
> Skipping stops is the worst in that regard ...
If the expectation is that the bus will stop at every bus stop in its path.
Vs. a larger system can easily have "commuter", "direct", "express", "park-n-ride", and other busses, with different expectations.
Plus the trivial case - the bus will roll past a stop which has 0 people waiting for a bus, if no current passenger has pulled the "Getting Off at Next Stop" signal cord.
> Skipping stops is the worst
Depends. If the timetable is packed or the buses are already bunched, skipping a stop is actually preferable - unless you want to hop off at that stop, too bad then! ;)
This is one of the things I find infuriating about the Subway in NYC.
There are local trains and express trains. Often, you'll have to wait on the platform for an extra ~10 minutes for a local (v.s. getting on the first available express).
Then partway through the journey, they'll declare that the local needs to go express to make up time. You'll be kicked off and told to wait for the next train.
To make matters worse, the next train is usually in the same predicament, so you end up waiting indefinitely (or giving up and finding another way home).
Sounds like a dark pattern to push people towards express. If the train changes how fast it moves or where it stops, that's their problem. The passengers should be allowed to get off at the nearest stop to their original destination, and any potential difference in fares should be eaten by the operator
Related. Others?
Why Do Buses Bunch? - https://news.ycombinator.com/item?id=19558482 - April 2019 (150 comments)
Pittsburgh Bus Bunching (2016) - https://news.ycombinator.com/item?id=17589349 - July 2018 (60 comments)
Why do buses bunch? - https://news.ycombinator.com/item?id=9577476 - May 2015 (154 comments)
Digital bus timetables at the stop, so that you can see that your bus is running late, or that the next bus is right behind it, definitely make a difference here, because you can make a sensible choice without the driver having to explain to everyone.
Many subway systems have displays showing that the next train is right behind the current one, and yet many people still insist on getting on the current overcrowded one.
Certainly with buses people have been burned where they are told that another bus will be along in 2 minutes only for that to evaporate and the next bus actually takes 15+ minutes. If that happens to you then you'll squeeze onto the first bus you can physically fit.
It takes quite a long period of good service to undo one bad interaction.
Which makes sense, as the next one may be equally or more overcrowded. In a busy urban area, it may be hours before an uncrowded train arrives.
I'd argue that that's not bunching, it's the entire system being overloaded. A characteristic of bunching is that you have a pair of vehicles which together carry an average manageable load of passengers, but that load is unevenly distributed between them.
Perhaps they should offer a discount for boarding the next train or a price increase for boarding the overcrowded one or even both?
Too difficult to implement - crowds on a busy subway platform are very different from boarding planes at an airport.
The displays are gradually less important because a growing fraction of the population are already carrying a display with wireless networking, on a personal device the UI can show you your exact route and where the bus you need is now etc.
These could still be smarter, I remember a year or two ago stood at a bus stop, watching the position indicator for a bus I wanted to catch and realising as it in real life appeared at the next junction - it was diverted away from my stop, one of the icons I'd been ignoring was a deviations from normal route, it's often set [road construction work] but that week the diversion avoided the stop I was stood at. That bus goes in long loops so I caught it about a quarter mile away after a breathless run, but a smarter app could say
"Hey, you seem to be waiting for the U6H, at the Broadway stop, and it's not going to that stop. Walk this way for a few minutes to reach a temporary stop at which today's U6H will pick you up instead."
> The displays are gradually less important because a growing fraction of the population are already carrying a display with wireless networking, on a personal device the UI can show you your exact route and where the bus you need is now etc.
That is not my experience at all. I use busses every day and I always use the posted times and don't want to look for it on my phone. There are a handful of bus stops on my common routes without displays and I always hate it.
I'm not sure there is a critical mass of people actually looking at the apps for real-time departure information, at least not in my city (Vienna Austria).
Why not both? A QR code with a link to the same content as it would've been shown on a display would work quite well and provide redundancy in case of display hardware malfunctions.
This is what the Berlin system does.
I lived in a city with this once, and beyond the novelty factor of the first few weeks after I found the app, I stopped using it. Looking up routes on your daily commute is just too much friction. The display at the station/stop is by far the best indicator, you can see at a glance how long you have to wait when you get to the stop.
Let's put even more things on a device which is extremely addictive to many.
I love my local region e-ink screens which just show me the info without wasting too much energy.
The displays aren't that important for me. But for visitors to the city, who haven't installed the bus tracker app, they're still very useful.
(Also, signal is terrible throughout lots of central Edinburgh. I can be at a bus stop just off Princes Street and get nothing at all.)
There shouldn't need to be separate apps. I like Edinburgh, only been there twice and both times as a tourist but I don't think "Wow, a separate Edinburgh bus app" would have been a boon, whereas "Oh, my bus app just works here" would make sense.
At one point Edinburgh's bus operator was part of the same legal entity as the company which provided some bus services in my city, though that is no longer true. London has it right, no tourists and almost no locals care about the bus companies. All the buses are painted the same colour, all of them work the same way, who cares which company operates the bus or why?
You can get bus times from Google Maps. But almost nobody seems to know that.
A standard "non-profit" bus app that all bus companies could use would probably be very useful.
Isn't Google just giving you timetables?
That's not useless but it's no substitute for real time information. Seeing "Your bus is six minutes away" is reassuring in a way that "Well, the bus isn't scheduled for another minute, and maybe it's running late" is not.
In that "Oops, it's diverted" case which annoyed me, my bus was, from that point of view, genuinely getting closer, I could see it on the map. And then I realised, with growing horror, that it's on a road which won't pass me. Maybe that's a glitch? Then I saw the bus itself, in the real world, too late it's actually not coming here.
There's both: GTFS is a standard for the regular schedule, and GTFS-RT is a standard for realtime information.
In the Boston area, the bus drivers seem particularly likely to react to this by the second bus passing the first (even by crossing a double yellow: traffic laws are generally optional here).
In the article this is presented as a symptom of how bad the bunching is, but as a rider it feels like this helps the problem: new riders are now getting on the bus that's less full.
Passing would only be a temporary solution since the new front bus will now have the delayed position, and the more crowded bus that was passed will have the expedited position and will shortly overtake the front one. They will essentially travel at the same rate and stay bunched in this scenario.
It might make more sense, when the delay to the front bus gets to within a given time of the rear bus' expected arrival time (maybe 1 minute, maybe more, maybe less), the lead bus simply switches to a state where it only picks up when a stop is requested by a passenger until it gets to within some time differential from its scheduled arrival time.
That would suck for some people to watch it pass but realistically, if it wasn't skipping stops, the wait for leap-frogging platoon would be closer to that off the rear bus anyway.
This solves the issue that one bus is way too full. It doesn't do anything to the issue that you've now doubled the headway in front of the late bus.
It isn't half annoying when you're on the London Underground or a bus, and there's an announcement that you will be waiting for a few minutes at a particular stop to "even out gaps in the service".
It seems so perverse to artificially delay a load of passengers because others are running late. But at a system level it probably makes sense.
If the lane is frequent enough, a 2-3 minutes delay is fine. In many European cities the schedule is not specific times but frequencies per time of the day.
Like "from 7AM to 11AM, one bus every 8 minutes". Then you have the bus app if you want to optimize further (and if you're me, miss it because of being too tight)
I meant when you're already on the vehicle mid-journey, and you just have to sit at a random stop for a while so you don't catch up to the one in front.
I get why, it just feels like your time is being wasted.
Of course, the trade-off is that next time you'll have a shorter wait to catch a bus or train because some other passengers had to sit and twiddle their thumbs for a bit
[dead]
That’s usually how they’re scheduled though, and it’s printed that way to make the timetable shorter and much easier to read.
That's how global optimisation works. Generally speaking you will always miss out on a bunch of local optima when you go for a global optimum. This was well understood by designers in metro systems like the London Underground[0] and road signs.
When each individual strives for the best result for themselves (a local optimum) we often end up in a worse result for everyone (see tragedy of the commons, prisoners' dilemma etc).
Where I live they say it’s “so everyone can make the bus in time” and it’s always made sense to me. Running to arrive at a bus stop just in time, only to find it passed two minutes ahead of schedule, is way worse
I get on many buses and 90% of the time the message is played about "Evening out the service" it's because the drivers shift is about to end and he/she doesn't want to wait at the driver changeover stop too long.
They could probably spread the wait and no one would notice...
1. Frequent service (<5 minutes between buses, ideally) 2. Enough capacity to support rush hours (so each bus can straight up refuse to take extra passengers) 3. Education campaigns and police / fines / other external factor for the short term (to break existing habits) 4. Separate public transport lanes so buses don't get stuck in traffic or behind red lights.
IMO, these are sufficient for a good public transport system. Skipping stops is the worst since it makes the whole network unreliable.
If the above points are too high of an investment and skipping stops is the only viable solution then a proper digital interface is needed. If the schedule is dynamic then the information about it also needs to be dynamic. I need to be able to know that the bus I am on is going to skip my stop and plan my next steps while I am sitting in the bus itself.
bus only lanes should be standard. One or two people in cars shouldn't be able to delay an entire line, nor should they get priority.
Anecdotally, I'm pretty sure this phenomenon occurs even with bus lanes.
It does. Because junctions still slow them down, as do groups of 30 tourists all getting on at once, asking questions about whether this bus goes to the castle.
You can tell exactly where you live based on the castle reference. Happened to me once when a tourist in front asked about the castle and I couldn’t quite believe it was real.
Edinburgh?
> junctions still slow them down
This is also mostly fixable, with signal priority. Except at complex intersections where different roads each have transit lines fighting for priority.
You can't magically create more space out of thin air. The easier solution to the car problem is to get rid of the cars.
And the easiest solution to reduce the use of cars?
WFH. Every company implementing a RTO politic should be declined next time they try to get certified for some green-washing label. Especially when some government grants are tied to those certifications.
> One or two people in cars shouldn't be able to delay an entire line
It happens all the time that one or two people on the bus itself (or, even worse, train) delay the whole line. The fundamental problem is inflexible public transit.
I described this 25 years ago on blush everything2.
The number one thing that has improved my bus riding experience here in Seattle was them getting all their buses onto a real time tracking app.
I usually have a couple different bus route options, and now instead of just waiting blindly at a bus stop I can actually see which bus is going to be getting to me the soonest and go to that stop instead.
I ride the bus 4 to 5 times a week and pretty much never use the actual “schedule”
Discussion on Wikipedia: https://en.wikipedia.org/wiki/Bus_bunching
Discussion of how to solve it in OpenTTD: https://www.openttd.org/news/2024/02/10/unbunching
It's really annoying when this happens on the NYC subway. You can be on an express train that has to wait forever for a local train to go first as they try to restore the schedule.
South Brooklyn B35 needs this bad.
Always upvote Futility Closet!
