Pete Dyson and Rory Sutherland: Transport for Humans. London Publishing Partnership, 2021
Behavioral science is a growing influence on the planning of cities and transport. Conventional economic thinking views the customer as a rational actor who cares only about travel time and cost, but behavioral science adds two insights: First, time and cost aren’t the only things travelers care about. Second, even when optimizing time and cost, human brains aren’t ideal computers for this purpose, so they take various mental shortcuts that good planning can anticipate and guide.
This new British book is a lively read on how to apply behavioral science to transport planning — a process that’s well under way at many leading agencies such as Transport for London. The authors touch on everything from signage to pricing to the joys of public art. At one point they praise a stairway painted like piano keys that sound a note when you step on them.
Is there anything wrong with putting behavioral science in the lead in transport planning? It depends on how you use it. One danger lies in how planners choose to think about perceived vs actual time:
Cars go fast, so transport engineers put a lot of effort into making alternatives faster. But a 32 minute train ride does not feel all that different from a 36 minute one, and the changes needed to shave off more time can get very expensive.
Can we just make the train feel faster instead?
The only bad word here is instead.
It may be that 36 minutes is the optimal travel time at a reasonable level of investment [1]. And certainly, there’s nothing wrong with doing things that make that time more pleasant and seem to pass more quickly. But if you are making the train feel fast so that it doesn’t have to be fast, you may be making an elite projection mistake with major consequences for equity and social justice. The authors cite the concept of Maslow’s hierarchy but not its most obvious point: Values such as pleasure and “self-actualization” are only motivating when more basic needs are met. Basic needs include the food and shelter that you will only have if you have a job, which in turn may require you to get to work on time. When you get to your low-wage job four minutes late, the timeclock you punch doesn’t care that your trip felt so fast that you feel like you’re on time.
It’s also important to question the polarization that drives the book’s rhetoric. The authors’ opponent is a conventional model-driven transport planning that thinks about passengers the way it thinks about cargo.
When we move things, rather than people, around efficiently, no feelings need to be taken into account. Planning can be mathematically optimized without any consideration for psychology.
There really is a transport planning orthodoxy that is this silly, but there are also a lot of interesting positions in the middle. For example, transport planners like me, who insist that we respect people’s need to get places by actual-time deadlines, aren’t denying the foundational role of great information design. Bad information is a cause of delay as well as stress, both of which are fundamental human needs in Maslow’s hierarchy: stress is bad for health, and delay threatens our access to food and shelter if those require getting to work on time. Obviously public transport authorities are not providing much value if their services aren’t used because they can’t be found, understood, and navigated, or if the experience is so stressful as to be harmful for the customer’s health.
The book offers little evidence to challenge the foundational importance of travel time. People aren’t cargo, but like cargo they have deadlines. A pleasant-but-unreliable public transport service may satisfy both tourists and the relatively fortunate people whose jobs aren’t in danger if they’re late. But such a service is not part of a comprehensive public transport network addressing the full diversity of the society and potential ridership.
I recommend this book for a fun overview of behavioral science insights. It has made me smarter about how to discuss these issues. Fortunately, though, the transport planning world isn’t as black-and-white as their rhetoric might suggest. There are many ways to use these insights while still respecting our need to get where we’re going, at the time we’re expected there.
[1] I’m not saying it should be 32, since a responsible travel time analysis is door-to-door and therefore includes frequency as well as walking or other kinds of access-t0-transit time.
I wonder if there is not an opportunity here in terms of identifying ‘feeling faster’ with ‘frequency’ in terms of the transit not being available when desired as a source of ‘feeling slower’.
This post reminds me of the Swiss concept of the cadence timetable – designed to be as fast as necessary*, not as fast as possible.
* as necessary in order to connect consistently with other lines and modes – they even re-work the infrastructure to allow it
Thinking about elite projection a bit more though – In this model you suggest that the key value for non-elite people might be reliability not speed. If accurate the answer would be to make trains run more reliably not necessarily faster. But It is much easier to sell speed to politicians than it is to sell reliability.
Yes, reliability is essential, and you’re right that that’s a challenge.
Reliability is tricky because you can always improve on-time performance by allowing more time in the schedule to achieve a certain trip – but on days when that time is unnecessary, the vehicle will have to wait at the stop it has arrived early at before it can depart, which is annoying for its passengers. Possibly more annoying than waiting at the stop for a late vehicle.
But frequent service can help us here too by maintaining as far as possible the timespacing between vehicles, not fixed arrival-departure times at particular stops. Incidentally, is there a number of minutes where this becomes plausible? I’m aware 30 minute intervals are often called frequent service, which is true, but I’d still try to arrive at a stop at the correct time for a 30 minute service, although live departure monitoring can help here too. Probably every 12 minutes would be sufficient for me to just turn up.
I agree that there will be day to day variability in running time, even for the same trip. However, schedules based on the average necessary time for each trip and between each timepoint on each trip are ideal IMHO, far better than scheduling for the slowest conditions to ensure the vehicle is always on time, though much if the time it means going 10 MPH in a 30 MPH zone or waiting several minutes at timepoints due to being early. There are situations where differences in average necessary time are predictable, such as school/non-school days, or a place that opens/closes at a certain time. In these cases, schedules should reflect that as well. If you know the 9:15 PM trip from the mall takes longer than the 8:15 PM trip due to a 9:00 closing time, then that trip should be scheduled for the longer necessary time. Yes, the headway will be longer than 60 minutes further down the route, but at least the schedule will be honest and people won’t think they’re going to make connections that in reality aren’t possible. I think you get the idea….
That’s a good argument when optimizing for the situation where you can get all the way origin to destination on just one bus. But, when infrequent routes connect and you need a timed connection to avoid a long wait, there really is often no alternative but to time each trip for the slowest conditions. Trying to adjust the schedule of one route and adjusting the schedules of connecting routes to compensate can quickly cascade out of control, as you have to also adjust the schedules of every route that connects to the connection routes, every route that connects to the routes that connect to the connection routes, etc.
This reminds me of the previous post from a while ago: https://humantransit.org/2022/04/do-you-really-want-to-live-in-perceived-time.html. In both cases these should be considerations, but not the priority. There is no substitute for a fast, frequent, reliable system.
But there are various considerations beyond that, given the fiscal realities of most transit systems. The data is clear that people don’t like walking a long distance to transit. They also don’t like waiting a long time, or making a transfer. Which is worse? What if you increase frequency, but require more transfers? You may get people to their destination faster, but many prefer a one-seat ride.
Personally, I’ve found that perception is important in that regard. It is common for people to accept a transfer — even one that may be slower — if they have the sense that they are moving quickly towards their destination. For example, I have a friend who used to stand on the corner, and wait for one of two buses. The first would take him directly downtown, while the second would take him to the subway station. The latter involved a fairly time consuming transfer, as it meant walking a ways, then into a very deep tunnel. The express bus (directly to downtown) was faster, but since it ran less often, he rarely took it. He never adjusted his schedule, or waited for that bus. Of course if the express was frequent, that would be his choice, but the point is that if you can’t have it all (in this case speed, frequency and a one-seat ride) then perception — specifically the perception that you are always making good progress — makes a big difference.
This suggests that frequency has an even bigger perceived effect than real effect. It is quite possible that if he needed to be at work at specific time, and was willing to adjust his schedule accordingly, then taking the bus was the best bet. It was less frequent, but still frequent enough to make up for the extra walking (and a little extra waiting) involved with taking the subway. But since he could arrive any time, he was effectively a spontaneous rider, and the one thing he hated was waiting. This is another argument for prioritizing frequency over speed.