Information measures and cognitive limits in multilayer navigation

Riccardo Gallotti points me to this interesting working paper in arXiv:

Information measures and cognitive limits in multilayer navigation

Cities and their transportation systems become increasingly complex and multimodal as they grow, and it is natural to wonder if it is possible to quantitatively characterize our difficulty to navigate in them and whether such navigation exceeds our cognitive limits. A transition between different searching strategies for navigating in metropolitan maps has been observed for large, complex metropolitan networks. This evidence suggests the existence of another limit associated to the cognitive overload and caused by large amounts of information to process. In this light, we analyzed the world’s 15 largest metropolitan networks and estimated the information limit for determining a trip in a transportation system to be on the order of 8 bits. Similar to the “Dunbar number,” which represents a limit to the size of an individual’s friendship circle, our cognitive limit suggests that maps should not consist of more than about 250 connections points to be easily readable. We also show that including connections with other transportation modes dramatically increases the information needed to navigate in multilayer transportation networks: in large cities such as New York, Paris, and Tokyo, more than 80% of trips are above the 8-bit limit. Multimodal transportation systems in large cities have thus already exceeded human cognitive limits and consequently the traditional view of navigation in cities has to be revised substantially.


My take is this greatly supports things like Grid networks and network simplification (see the work of Jarrett Walker). This looked at rail. Think about buses. In a few years, people will just let their apps navigate them, and human cognition limits may fall off the chart.

Transit shelters and other amenities affect perceived wait times

CTS Catalyst reports  Transit shelters and other amenities affect perceived wait times on our soon-to-be-published report.


The overall results: “Perceived and actual wait times are clearly related, but the relationship is variable,” Guthrie says. “The waiting environment can change perceptions.”

Nearly 85 percent of those surveyed waited 10 minutes or less. Even with waits under a minute, however, people tended to perceive at least a minute or two, and they tended to estimate in round numbers (5, 10, 15 minutes). “This creates an initial ‘penalty’ of overestimates,” he says.

Researchers also found several variables to have statistically significant impacts. The presence of a shelter—even a simple one—made waits seem shorter, especially for waits less than 10 minutes. “The biggest difference in perception was between any shelter and none at all,” he says. The presence of a NexTrip real-time information sign also shortened perceived waits.

Posted schedules produced a “really interesting pattern,” Guthrie says. For shorter waits, schedules caused people to overestimate wait time, but after about 10 minutes, people began to underestimate it. “It’s possible that for short waits, people compare the clock and the schedule and get impatient, but for longer waits, they are reassured to know the bus or train is coming,” Guthrie says. “This implies that posting schedules is more important for routes with less frequent service.”

Gender alone was not significant, but there was a stark difference for women in less safe environments. “Most sites in the study were rated as safe, but at those that were not, there’s potential to improve the experience for riders and potential riders,” he says.5 Minute Wait

“With several major initiatives currently under way to expand the number of shelters at bus stops and to improve the quality of transit schedule information across our entire network, the timing of this project could not be better,” says Marilyn Porter, director of engineering and facilities for Metro Transit. “This study provides important insight that is directly applicable to the work that we are doing to ensure that our customers have the best possible experience using transit service in the Twin Cities.”

The model developed in the project includes many other variables such as household income, trip purpose, and the presence of benches and route maps. “Users of the model will be able to choose criteria and predict the impacts of hypothetical feature mixes,” Guthrie says.

A final report is planned for publication in March. Humphrey School associate professor Yingling Fan was the study’s principal investigator; David Levinson, RP Braun/CTS Chair in the Department of Civil, Environmental, and Geo- Engineering, was co-investigator.

The Many Flavors of Transit

Public policy for mass transit in the United States is largely focused on a few modes of travel: commuter rail, urban rail, urban bus, and paratransit requirements. These few modes certainly carry most of the transit riders in the country, but do not represent a full understanding of the breadth of options that are required to make a truly transit-oriented city. New York is the most transit-oriented city in North America, and it is likely that when most people not from New York think about transit in New York they think about the subway system, or perhaps they include iconic yellow cabs or remember that there are a lot of buses. If you ask most New Yorkers, they will probably add many other modes, but even then there will likely be many modes left out.

An underappreciated reason why New York functions so well as a transit-oriented city—and can grow transit ridership without new expansion of core services (yet)—is that there are oodles of transit options available. Mode choices for travelers is not a binary choice between driving and transit, even though this has been the general attitude toward transit policy over the past few decades. Observing travel in New York suggests just how complex the required systems are to actually provide meaningful alternatives to automobility.

Below are 33 different categories of mass transit offering regular service in New York City (I have reviewed this list with native New Yorkers but I am sure others will have constructive comments about my categories). This is what it takes to create transit-orientation for a city, and I suspect many of these exist in cities everywhere but planners and scholars are not aware of them. In New York lots and lots of operators offer many different services to many different types of people. Not all technologies work for all places, so transit technologies should reflect the problems to be solved.

  1. MTA subways
  2. PATH subways
  3. MTA buses
  4. New Jersey Transit buses
  5. Metro-North Rail Road
  6. Long Island Rail Road
  7. New Jersey Transit trains
  8. Staten Island Ferry
  9. Staten Island Rail Road
  10. Water taxis
  11. Commuter ferries (Five licensed operators)
  12. Access-a-Ride (MTA and other transit provider contracts)
  13. Yellow taxicabs (Medallion cabs)
  14. Green taxicabs (Boro cabs)
  15. Liveries for Hire (Uber, Lyft, Carmel, etc.)
  16. Executive Limousines
  17. Liveries (informal)
  18. Commuter vans (licensed and pre-arranged fares; e.g. Mario’s Transportation)
  19. Dollar vans and local jitneys (informal immigrant services)
  20. Chinatown buses (intercity)
  21. Low cost intercity buses (Bolt Bus, Mega Bus)
  22. Conventional intercity buses (Greyhound, Peter Pan)
  23. Apartment shuttles (CoSo, etc.)
  24. Company/corporate shuttles
  25. University shuttles (Columbia University, New York University)
  26. New Jersey commuter jitneys
  27. Long Island commuter jitneys
  28. Roosevelt Island Tram (Gondola)
  29. Roosevelt Island Red Bus (Publicly owned development corporation)
  30. CitiBike bike share (public access for a fee)
  31. University bike share programs (free access for a designated group)
  32. Amtrak
  33. Executive helicopters

When planning local transportation systems we now commonly say that multiple modes are required. We underestimate how many modes this is and how challenging it is to accommodate everything. Each of these 33 categories represents different customers, fare policies, public/private ownership, terminal capacity, vehicles, road access, curbside access, infrastructure needs, etc. Most of these different types of transit are regulated under municipal or state laws, too, and require the allocation of public space (roads and waterways) more than large-scale capital investment. I outlined some of these challenges in a recent CityLab piece.

The main takeaway from this is that for transit to be useful it must reflect the many ways people need to get around the city. Multi-modal transport doesn’t mean cars-transit-bikes-pedestrians. There are multiple modes of transit, too.

Green Line / Green Lights

Westgate Station

Suppose you have a train moving along (parallel to) an East-West (EW) signalized arterial.

Westgate Station
Westgate Station

Case 1: If the signals are pre-timed, and the timings are known in advance, the train should never have to stop for the signals (aside from emergency signal pre-emptions and other edge cases). Instead, the train should be able to adjust its speed so that it doesn’t have to stop. It might go at an average speed of say 10, 20, 30, or 40 MPH in order to ensure it hits a green light or better a green wave from whenever it departs a station. The train driver can be apprised of the optimal time to leave the previous (upstream) station, and the speed to travel to hit “green” lights.

Typical Signal Schedule and Traffic Flow Diagram, North-South across Market (1929)

Green waves have been around since the 1920s (See Henry Barnes’s autobiography: The Man with the Red and Green Eyes. Dutton. 1965. OCLC 522406). Static signs to tell travelers the speed of the green wave has been in standard use in some places (e.g. Connecticut Avenue in Washington, DC) for almost as long. Dynamic real-time signs which tell travelers what speed to adjust to to make the green wave has been recently patented and tested in simulation for automobiles: Always Green Traffic Control. The time is ripe for some carefully controlled field experimentation.

Still, pre-timing with information certainly doesn’t guarantee the fastest speed possible for the train, but it does guarantee no stops except at stations, which is good for a variety of reasons, including both travel time (avoid acceleration/deceleration loss), traveler comfort, energy use, and train wear and tear.

Case 2: If the signals are actuated, that is, their phase and perhaps cycle timings depend on traffic levels, and traffic “actuates” the signal, usually through an in-ground loop detector, transit signal priority from a fixed upstream distance should be sufficient to ensure the train doesn’t stop at a “red” light. The traffic light controller would know that a train was coming, and either keep the lights in the direction of the train green (if they are green), or change them to green and hold them, if it is currently red and the green is coming up. The train, knowing when the green will be on, should be able to adjust its speed (faster or slower) to make the green without stopping.

The distance that trains can currently notify a downstream signal controller is when they depart the upstream station, which is up to 1/2 mile or so (the spacing between stations). 1/2 mile at 30 mph takes 1 minute. With a cycle time of 2 minutes, and at least half the green time (1 minute) for the signalized arterial, a green can be guaranteed. If the light is currently red, it will be green within a minute. If it is currently green, it can be kept green for up to a minute. The worst case is it was just about to turn red and instead the green is extended for an additional minute. Alternatively, if it is currently green, a shorter than usual red phase can be inserted to clear the crossing traffic, before the light is turned back to green.

For traffic signals less than 1/2 mile downstream (say 1/4 mile) the warning time is only 30 seconds at 30 MPH. The same logic applies, but it is potentially more problematic as there is less lead time to adjust the timings, so the phase shortenings might be more severe. On the other hand, if more than 50% of the green time goes to the EW movement (say 75%) you aren’t really any worse off.

At 1/10 of a mile the warning time is less, but train departure from the station should be able to be coordinated with the light directly.

Case 3: But let’s say your traffic engineers are incapable of making this work. Should the train and its passengers suffer? This is where traffic signal pre-emption comes in. Most widely used for emergency vehicles, this potentially changes the sequence of phases, so maybe a phase is dropped (it doesn’t occur within the cycle, or within the usual place in the cycle).

This system does ensure that the vehicle requesting the pre-emption gets a green light as quickly as possible (safely turning the conflicting movements to a red phase) and thus can drive at as high a speed as possible. While trains should not need to stop at traffic lights with priority and speed adjustments, with pre-emption, they neither need to stop nor adjust their speed.

What could go wrong?

Pedestrians. Thus far we have been talking about a system with cars and trains. Pedestrians too can actuate signals, though “beg buttons“. These may function similar to vehicle actuators, in telling the traffic signal there is someone who wants to cross. The difficulty for priority or pre-emption is that a pedestrian phase may need to be longer since pedestrians take longer to cross the street than a vehicle does, especially if the street is very wide. So a pedestrian actuator may also extend the green time, in addition to calling for green time. This makes it more difficult to quickly change lights from red to green, since for safety reasons you can’t strand a pedestrian. This makes the ability to adjust train speeds in concert with the traffic signals more important.


Firetruck on University Avenue blocked by LRT train
Firetruck on University Avenue blocked by LRT train

Emergency vehicles. Emergency vehicle on emergency vehicle crashes are a known problem, and pre-emption may make it worse as firetrucks approaching a scene from two directions may both demand a green light, but only one gets it. The driver of one vehicle, not realizing he didn’t get the green (especially if he had the green as he was approaching), fails to yield. There are solutions to these problems.

Any of this will likely lead to additional delays for conflicting vehicle movements (cars making left turns or North-South traffic crossing our East-West arterial). With priority, this may even lead to extra delay for some vehicles on the parallel arterial who have been given a short green so the conflicting traffic can also get a short green before the EW arterial returns to green.

However the train usually has more people on it than are queued up at the other directions, so total *person* delay will generally be reduced.

For a variety of reasons, delay is bad (unless your goal is punishing drivers and air-breathers), we want to minimize total person time (or weighted total person time – recognizing long weights are more onerous than short weights) in the system (because time is money), and minimize pollution outcomes as well.

In short, the Green Line not getting green lights on University Avenue is a solvable problem. It should have been solved already. It eventually will be solved.

Further reading, with math: See Fundamentals of Transportation/Traffic Signals

Study says Blue Line’s development impact is minimal | Star Tribune

Adam Belz in the Star Tribune writes: Study says Blue Line’s development impact is minimal, discussing a recent study by Sarah West and Needham Hurst of Macalester. I get quoted:

“My first sense is the Green Line is a better line,” said David Levinson, a civil engineering professor at the University of Minnesota. “It’s going to a denser area.”

Levinson said that the Green Line — and any subsequent additions to the light rail system — will also improve the value of land along the Blue Line, thus making it more ripe for development as the rail system grows and more people use it.

“That positive feedback system sort of kicks in, and it reinforces the growth,” Levinson said.

The Land of Make-Believe –

Cross-posted from  “The Land of Make-Believe


The Land of Make-Believe

Minneapolis and St. Paul are considering a number of streetcar lines. Three have risen to the forefront. The Central/Nicollet Line in Minneapolis, the Midtown Greenway in Minneapolis, and the Seventh Street Streetcar in St. Paul. In part this is viewed as an economic development tool. In part this nostalgia for an earlier simpler time, when we were all children, growing up in Pittsburgh.

Mr. Rogers' Neighborhood. Life in pre-war Pittsburgh.

We at have gone back and forth and back and forth on whether streetcars are a good idea. Obviously it is context dependent, and some ideas are better (less bad) than others.

As with anything, resources are not infinite (I wish they were, but if wishes were horses we would all be riding around on horses). Thus there are trade-offs.

The general trade-off is that streetcars are more expensive than buses, particularly in initial capital expenditures. Thus with a finite amount of resources you can do fewer streetcars or more buses. Those rail-based rides are slightly nicer (particularly smoother) (since with streetcars you are building some of your own infrastructure, while with buses you are literally free-riding on infrastructure built by someone else, in various stages of disrepair).

Streetcars do have a number of other properties in the current world that makes them less advantageous in many situations. They are starting from a base of zero lines. Thus until they are as ubiquitous as the bus, you are more likely to be able to have single-seat ride on a bus than a streetcar if you want to go anywhere past where the initial streetcar lines go. In a world with a near ubiquitous streetcar network, building an extension is potentially a better decision than in a world with a near ubiquitous bus network. Thus while it may or may not have made sense to dismantle the streetcar system, if it didn’t doesn’t imply it makes sense to build one now. The contexts are different.

Bus lines can also branch. Sometimes a city is a tree (in contrast to the famous Christopher Alexander article), and branching systems coming together on an exclusive bus-way provide high frequency in high demand areas and lower frequency (but a single seat) in lower demand areas farther from the center, as illustrated by this schematic map of the Ottawa BRT.

Ottawa Busway - Sometimes a City is a Tree

However if you are going to put a line in a place somewhere, you probably want to find a corridor where there is a lot of local short-distance demand so that transfers are not the main issue.


Now I am dubious of the specific forecasting results (like why does Enhanced Bus have fewer corridor riders than No Build – Table 2, or why is the thumb on the scale for rail up to 25 minutes of In-Vehicle Time (p.2) when average journey-to-work time in the Twin Cities is less than 25 minutes, among others) that have been presented, and am in general skeptical of forecasts (having spent the first five years of my career building such models, hopefully less internally inconsistent), but nevertheless, even if the modeling doesn’t show it, this is obviously one of the best corridors in the region to make a transit investment of some kind. It is plausibly the main North-South spine of Minneapolis, certainly downtown and Northeast. So if you were going to build a streetcar, this would not be the worst place you could do it. In addition to the conventional home-to-work and shopping and restaurant markets, there is a large entertainment market (sports, theater, museums) and tourism market (conventioneers) which this serves.

But we run into the general problem of streetcars (as opposed to LRT), shared right-of-way.

Imagine instead we extended the car-free Nicollet Mall south of 12th Street down to Lake Street (yes, through the K-Mart). Local merchants and restauranteurs would undoubtedly scream for a while, since clearly many of their customers do arrive by automobile (hence the crowded parking lots). But there are many cross streets, and a number of alleys, which could be used to access parking (structured if necessary), while keeping Nicollet (Eat Street) clear of cars and better for transit (which would operate at somewhat higher effective speeds with less auto-interference, pedestrians, and bicyclists. 1st and Blaisdell would of course get more traffic. So this solution is not without consequences.

Extending the car-free Nicollet Transit Mall deserves its own post.

Even then, it runs into many at-grade crossings, which can be assisted with transit signal priority or preemption, but still creates conflicts. Given the grand visions we do have in this city, it is surprising that the (arguably) fifth densest downtown in the United States doesn’t have a subterranean transit line either existing or in the plans.

In other cities, there are of course subways, which are not cheap, as well as bus tunnels, as in Seattle, which are also not cheap. But then you can go fast. Studies in the 1970s considered this option, and it was also considered for Washington Avenue at the University, and perhaps for one of the rejected alternatives of the SWLRT, but nothing I have seen recently even raises this option. Now I am not the type to say “Go Big or Go Home”, I am surprised no one else is saying it aside from a few forum posts at UrbanMSP.

So while I am not advocating a subway, or a streetcar, this is among the least bad routes if you win the lottery and are somehow flush with cash.

Midtown Greenway

If there did not exist an almost fully grade separated corridor between the proposed Southwest LRT and the existing Hiawatha Corridors, planners would be dreaming of one. So given there is one, this isn’t really a question of if, but when. Something will be done here. For all the usual reasons, this should be a busway (with e.g. electric buses), but it will (for all the usual political reasons) of course be an LRT of some kind (perhaps using streetcar sized vehicles). There can be some interesting multi-modalconnections to the Hiawatha Line. There might also be opportunities to extend this to St. Paul.

Seventh Street

Schmidt's Brewery Castle, via Google Streetview

If Minneapolis gets it, St. Paul wants it. Seventh is St. Paul’s Main Street, so this is a plausible corridor to do something.

The problem here is implied reading between the lines of Bill’s recent post, there isn’t enough demand in St. Paul. Somehow streetcars are supposed to solve that. For all the usual reasons this should be an arterial BRT (as the Metropolitan Council has proposed), but even more so, since the demand here is so much lower than the demand on the Central/Nicollet corridor. That is, Minneapolis has a line that connects to St. Paul (dubbed Green) and Bloomington (dubbed Bloo). While there is some dumbbell-like shape to the demand patterns (higher at the ends than the middle), they are asymmetric dumbbells, the Minneapolis weight is far greater than St. Paul or Bloomington. So though there is a certain triangular symmetry to completing a “Yellow” (Yellow+Blue=Green) rail connection between Bloomington and St. Paul, it’s demand will inevitably be lower than the other two so long as Minneapolis is larger than the other two cities in the Triplex. There is no guarantee that a Seventh Street line would connect to the Bloo Line. (To be fair, the Midtown Greenway might also be called Yellow, since it connects the Bloo and Green lines as well, maybe one can be the Turquoise route). [There is I recall a fantasy transit map to interline the 7th Street Streetcar with the Midtown Greenway, via Ayd Mill Road, thus enabling there to be a single Yellow Line, but it won’t do well for the Bloomington to St. Paul market].

The Land of Make Believe, the other end of the Streetcar in Mr. Rogers' Neighborhood

Just like Mr. Rogers’ neighborhood, the streetcar would go to a castle, in this case the Kingdom of Schmidt’s. There are differences: Schmidt’s produced beer, the Castle in Land of Make-Believe intoxicated children with something else — visions of streetcars.


Commuter Rail Ridership Declining Despite Increase in Lines

Caroline Cournoyer at Governing Magazine writes: Commuter Rail Ridership Declining Despite Increase in Lines:

“While the public may love the notion of commuter rail lines, they are perhaps the least popular form of transit for politicians. The subsidies for commuter rail are tremendous, says Michael Smart, a researcher with the Institute of Transportation Studies at the University of California, Los Angeles. A study of the Minneapolis Northstar line concluded that taxpayers were paying a subsidy (which included capital costs) of $89 per passenger. Other studies showed subsidies of much lower rates, but still significantly higher than those for bus or subway riders.”

The study referred to is this 2011 blog post.  Northstar Ridership is of course up since the fares were cut by 25%. (In 2013 it was  787,239, up 17% … so ~700,000 riders pay less so that ~87,000 pay at all)  I don’t think revenue is up, though the cost per passenger is of course lower.

The influence of light rail transit on transit use: An exploration of station area residents along the Hiawatha line in Minneapolis

My student Jessica Schoner and colleague Jason Cao recently published:

Cao and Schoner (2014) The influence of light rail transit on transit use: An exploration of station area residents along the Hiawatha line in Minneapolis. Transportation Research Part A: Policy and Practice
Volume 59, January 2014, Pages 134–143


  • We compare transit use of residents in LRT corridor and control corridors well served by bus transit.
  • People moving into LRT corridor before its opening use transit more than those in control corridors.
  • Transit use of people moving into LRT corridor after its opening is similar to that of urban controls.
  • LRT-related land use and transportation policies are necessary for ridership growth.

Rail transit is often implemented in the corridors already with high transit demand. When evaluating their ridership benefits, previous studies often choose the city/county/region as control groups, rather than comparable corridors without rail, and hence overstate their impacts. In this study, we employ propensity score matching to explore the impact of Hiawatha light rail transit (LRT) on transit use. We find that compared to residents in similar urban corridors, the Hiawatha LRT promotes transit use of residents who have lived in the corridor before its opening, and that residents who moved to the corridor after its opening use transit as often as new residents in the comparable urban corridors without LRT. We conclude that besides LRT, land use and transportation policies are necessary for ridership growth.

Propensity score matching; Self-selection; Transit-oriented development; Travel behavior; Urban form