Historic US GIS transportation SHP files

Jeremy Atack writes to EH.net:



I have posted my historical GIS transportation SHP files for the Lower 48 states from this nation’s founding through (approximately) 1911 on my www site https://my.vanderbilt.edu/jeremyatack/data-downloads/.  Each transportation mode–canals, steamboat-navigated (as opposed to navigable) rivers, and railroads–has its own archive ZIP file which contains the complete series of files (projection, database and polyline files, etc.) required by ESRI’s ArcGIS and ArcGIS Pro.  These are collectively referred to as “a SHP file” though there are actually multiple files for each mode of transportation.  Once unpacked, these files for each SHP must be kept together and should only be edited using a GIS program.  If corrupted, the entire SHP file will become unusable.

The metadata file (.XML) briefly describes the contents of each SHP, the manner in which it was created, and summarizes any edits since these files were originally posted.  Issues relating to the creation of these SHP files are discussed in much greater detail in the documentation file also appearing at https://my.vanderbilt.edu/jeremyatack/data-downloads/


This looks like a fantastic resource for anyone doing historic analysis of intercity network evolution. Now if only someone would digitize and standardize historic urban transit networks with modern GTFS coding …

Our use of impact factors is backwards

Academics incorrectly use impact factor (IF) to judge the quality and impact of an article.

The only factor to judge the quality of an article is the article itself, and that is subjective.

The only way to measure its impact is its citations (in peer-reviewed journals or elsewhere, like popular media).

If an article published in a low impact factor (IF)  journal has 100 citations, and one in a high IF journal has 100 citations, the first should be more appreciated, since it does not benefit from the spillovers of the journal reputation. It is overcoming a disadvantage.

Thus a high citation article in a low IF factor journal is likely better than one in a high IF journal (if citations are an indicator of quality, and citations are increased with spillover effects — which people must believe otherwise why try to publish in high IF journals). Impact Factor should only be used to discount article citations, not as a positive metric in tenure and promotion.

I have heard all the reasons. IF is an indicator if the author publishing in the “right places”. It’s harder to get published in a high IF journal (likely true, but not as true as you think). There aren’t enough citations by the time a tenure decision has to be made, so this is a surrogate.


Read the actual papers. If you won’t read the papers (and/or don’t know enough to ascertain originality), trust their colleagues. At any rate, if it’s not in your area, why is the system structured to give you a vote on their academic merit? Judge their teaching, or their collegiality, but not their output.

Where have all the masons gone?

The quality of masonry in the built environment has dropped significant in the past century.

I would like to blame this on the rise of the Anti-Masonic Party and William Wirt, unfortunately for my desire for a tidy history, that was in 1832, and preceded the decline of masonry by about a century. Furthermore, freemasonry and stonemasonry in practice are not terribly related by this time (though freemasons were once stonemasons back in the 14th century). Freemasons like George Washington did little actual brickwork.

So instead, let’s turn to the rising price of labor, as men who once would have become stonemasons, as their fathers were, were instead attracted to other businesses, and the real estate sector found that high quality detailing was no longer worth the premium it cost. Today, masonry is often a non-structural skin which is pre-manufactured, what my wife calls “brickaneer“. Yet even pre-manufactured brick veneer seems to lack style, and is just a boring layer. Better perhaps than some alternative skins, but nothing like it once was.

The more interesting question is perhaps why the market doesn’t reward aesthetics on the exterior of buildings now, when it once did.

Consider the four apartment buildings shown below, they are all in the same Powderhorn Park neighborhood, of similar size, but were built  in different decades. The level of detail on two of them is far greater than the other two. At some point interest or willingness to pay for Masonry detail failed. This is unfortunate.

New buildings don’t do much better. Compare some 21st century structures with Thresher Square. Whatever you think of aesthetics, detail is clearly lost.  Perhaps there were many older simple buildings that were just lost to history because of their unimpressiveness, and only the best bits were saved. I think it is more significant though than just survivor bias. No new construction seems to have the same level of exterior architectural detail we once saw.

For all the attention to detail paid to computer design, where has the real architecture gone? I am not a huge fan of Victorian frills. Bauhaus aesthetics were a response, simplifying the ornate form without function, but seemed far more skilled than what we get now. Why did detail (not frills, but details) never recover. Notably, the cornice disappeared with masonry.  Whatever we call late 20th century and early 21st century architectural styles,  future decades will not appreciate the way we appreciate the surviving buildings of the late 19th and early 20th centuries.

Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park
Apartment building near Powderhorn Park


The Edge on Oak Street
The Edge on Oak Street
Mill Quarter Municipal Parking Ramp
Mill Quarter Municipal Parking Ramp
Thresher Square and Old Spaghetti Factory
Thresher Square and Old Spaghetti Factory

Temporal Sampling Intervals and Service Frequency Harmonics in Transit Accessibility Evaluation

Recent working paper by my Accessibility Observatory colleagues:

Box plots for sampling strategy performance over all blocks at each sampling frquency. Boxes show inter-quartile range (25th – 50th percentile) with horizontal medial line; whiskers extend 1.5×IQR above and below. Outliers are plotted individually. Mean is indicated by a dot.
Box plots for sampling strategy performance over all blocks at each sampling frquency. Boxes show inter-quartile range (25th – 50th percentile) with horizontal medial line; whiskers extend 1.5×IQR above and below. Outliers are plotted individually. Mean is indicated by a dot.


In the context of public transit networks, repeated calculation of accessibility at multiple departure times provides a more robust representation of local accessibility. However, these calculations can require significant amounts of time and/or computing power. One way to reduce these requirements is to calculate accessibility only for a sample of time points over a time window of interest, rather than every one. To date, many accessibility evaluation project have employed temporal sampling strategies, but the effects of different strategies have not been investigated and their performance has not been compared. Using detailed block-level accessibility calculated at 1-minute intervals as a reference dataset, four different temporal sampling strategies are evaluated. Systematic sampling at a regular interval performs well on average but is susceptible to spatially-clustered harmonic error effects which may bias aggregate accessibility results. A constrained random walk sampling strategy provides slightly worse average sample error, but eliminates the risk of harmonic error effects.

How I spent 200 minutes last night, or Tim Cook and Jonny Ive should buy their iPhones in the Apple Store

Apple is famous for its user experience. I was at the Apple Store (Rosedale, Minnesota) for 3 and 1/3 hours yesterday upgrading my phone. (5s to 6s, the 5s worked fine, but the battery is clearly nearing the end of its useful life, and getting to 0% before I go to bed and recharge, plus my contract was up, and I wanted to change carriers).  Somehow, 3.5 hours seems too long. This was not Apple Staff’s fault. It’s the system. The system for getting an upgrade should be re-engineered. This will not happen until senior Apple staff actually experience what it is like to go through the bureaucracy required to get a new phone. (Of course Tim Cook and Jonny Ive would likely be recognized, so they will need costumes and have to go incognito).

I got an appointment for 6:00 – 6:30, signing up online last week. I would have done it through the mail like last time, but, (1) There is some complexity with switching carriers (Sorry AT&T, it’s not you, it’s international coverage and the Great Firewall of China. Here’s hoping T-Mobile works better.), and (2) I wanted to trade-in the old phone which seemed more complex via mail. So I signed up to get it in person. This was a mistake.

Like a visit to the Genius Bar, I sort of thought they would actually serve me at the appointed time, so I got there at 5:50. (10 minutes early) Ok, there was a queue. It was the first day of the 6s sales, which I hadn’t quite realized (I knew it was coming out, I didn’t realize it was the first day or I would have avoided and done this over the weekend, though evening should be better than when the doors open in the morning, no?). I made it into the store at 6:40. (40 minutes). Apple should have a better estimate of how long the set-up process takes so it schedules the right number of customers for the right number of staff.

By 7:40 I was signed up with T-Mobile. This should not have taken an hour. (This alone must have cost Apple $25 per customer to sign up just in labor costs (I’m guessing), leaving aside my time.) (60 minutes)

  1. Why cannot the signup software scan the old phone (a photo from the setting screen e.g.) rather than the Apple staff typing everything from one phone into an iPad? (Or isn’t this in an Apple Database already?)
  2. Why must my information be entered more than once? How many times does the system need my name and phone number and iCloud account information, etc.
  3. Why does the T-Mobile signup app ask for my PIN as if I have an account with them. I don’t. If it wants me to create a new PIN, it should say something like ‘create a new PIN’, not ‘enter your PIN.’
  4. Staff said the T-Mobile app was better than the others. This is absurd. Apple should have a straight-forward App that populates the carrier databases later, rather than waiting for their laggy experience.
  5. If I am buying my phone from Apple and service from carriers, somewhere on the Apple website should be a comparison of the different carrier services and prices. I can’t find it. I don’t really know what its going to cost without going to their sites, which are hardly paragons of clarity.

Ok, so now I have a new phone in a box, and am ready to turn in my old phone for a trade-in. First I want to restore my most recent iCloud backup to the new iPhone. This was slow, and failed the first time. Apple staff reinstalled the OS and 9.0.1 update and we did it again. It worked the second time, but this took a while. In the mean time I surfed the web on my old phone with 6% battery. At least they give you free charging at the Apple Store, and free WiFi. Note, the progress bar telling you how many minutes this update will take steadily lies. Surely someone can come up with a more accurate predictive algorithm.  This took from 7:40 until 9:10 (90 minutes).

I understand downloads and installs take time, I did it at home overnight last time. But if you want us to do it in the store (and I don’t want to lose my old phone until my new one is set up), this should be faster. I don’t know if the bottleneck was the local WiFi (which was certainly being slammed), or iCloud servers (which are also likely being slammed) [I suspect the latter], but this is not unanticipatable. And my being in the store is using up some Apple staff time that could be better used.

In short, this would be a much simpler process if Tim Cook and Jonny Ive and Eddie Cue and Phil Schiller and Angela Ahrendts and the other relevant executives at Apple eat their own dogfood and buy an iPhone like the rest of us, rather than being issued phones in-house. They would redesign the process, and save tens or hundreds of millions of human hours now wasted globally at the in store iPhone sign-up/set-up process. If Apple sells one-hundred million phones through stores, and each takes an hour, and this can be cut to 1/2 hour, they will save 50 million human hours for the customer and 50 million staff hours. A human life is less than 1 million hours.

The main thing is just they should take old phone, scan it, take in all your information (enter once your icloud account & password) in one fell swoop, and your new choice of carrier and plan, and set you on your way, and then however long it takes to install properly, and then issue you a set up new phone. You can get a text from them once they are done and return to the store. The old phone scanning can even be pre-done online, as can your personal information intake, they just need to validate in person you are who you say you are, and your phone is what you say it is. I am idealizing a bit perhaps (maybe you must enter your iCloud id and password multiple times, though I am not really sure why), but something closer to this would be standard practice were senior executives to experience this.

So far the new phone is fine, mostly like the old one. The wallet for Apple Pay (with an interesting if trivial bug) does not use my primary credit card, so still is more proof of concept than something that allows me to leave my real wallet at home. There is functioning health app that I did not have before on the 5s, so I can track more than Pedometer++. There is  a slightly bigger screen, but not enough to move me off the iPad when reading at night. And hopefully a longer battery life, which just comes with it being new. A few apps allow force, er, 3D touch.  I will not be profoundly affected by this the way the iPhone 1 or 3g made a difference.

The Apple OSes have their annoyances, mostly related to security and two-factor authorization. How often do I need to do this. I know I opted in to it, but this is a huge waste of time, and I am likely to opt out. Also separate passwords for each app is especially annoying. When I first start up the OS, it asks me for my iCloud password, which is long and complicated (for good security) and I dutifully enter. Then it asks again (apparently for the other apps like Messages and FaceTime), but this is a different password, and it doesn’t say that, so I enter the wrong password several times before it reminds me to get a 1-app use password from appleid.apple.com, which requires two-factors (a text to your phone with a code) to log in to, and cannot remember its the same browser from day-to-day though you tell it to. And since I do OS maintenance and installs for the family, I revisit this whole process multiple times. And this isn’t even including the PIN to enter the device, or the passwords for other apps that want them when you set up a new OS release. Somehow I don’t think Apple executives experience this either. If I were John Siracusa, I would document this exhaustively with screenshots. But this is a blog about transportation.

Symposium in Urban Systems (October 1, 2015 in Tempe, Arizona)

I am participating in a Symposium in Urban Systems Thursday October 1st, 2015 1-4PM in Tempe, Arizona (Arizona Time).

Speakers include

  • Michael Batty, University College London
  • Charlie Catlett, Argonne National Laboratory (Urban Computation)
  • Kevin Gurney, Arizona State University (Emissions)
  • Nancy Grimm, Arizona State University (Climate)
  • David Levinson, University of Minnesota (Urban Accessibility)
  • Jose Lobo, Arizona State University (Urban Economies)

Urban system and networks

Cities throughout the world are growing vertiginously in number and in size. These demographic and socioeconomic trends present important challenges at various levels: they pose a potential threat to the natural environment and generate a long range of urban issues, from congestion to poverty and from increased health risks to crime.

However, it is precisely at the core of these fascinating complex systems, that we also find new opportunities to tackle such long-standing challenges. In a city, diverse crowds are permanently interacting and at a pace that accelerates with urban scale. This makes cites the natural attractors and producers of technological and social innovations. The latest trends in technology are primarily about objects and people in their daily life equipped with sensors; that can be inventoried and analyzed by computers.

Our goal is to analyze cities and develop a common language that allows us to separate the different scale of analysis and to map and discover intrinsic principles of urban organization. We think this is an effective way to analyze cities in quantitative ways opening new unique opportunities to help manage them via new on-line applications and real time communications.

Current challenges to construct reliable theory and models of cities result from the lack of integrated data sources and comparative studies at different scales that would allow us to integrate the building principles of a science of cities. In this workshop we bring together the latest contributions on analysis of urban systems at different scales: from individual mobility to urban infrastructures, including social networks and spreading dynamics. We will arrange presentations that expose the latest methods of analysis and the most important ubiquitous findings related to urban infrastructures, social interactions and human dynamics. Speakers will be selected to represent the most important findings in a particular scale and layer of interest. Panel discussions will be promoted in order to foster interactions and summarize the current state of the art, future challenges and opportunities. Being a multidisciplinary topic, we will also include participants with the latest advances in understanding cities from computer science and urban planning.

Why is the walking man white?

Waiting to cross the street westbound in the early morning, my walk sign was impossible to read because of the sun. That’s a shame, but of course, I could just look at the traffic light for guidance, or just make the assumption that if it doesn’t say “Don’t Walk”, then walk. That’s a dangerous assumption, the light could be out. It is much safer just to look both ways, and ignore the lights altogether, then just rely on the lights.

Pedestrian crossing signal in Portland, Oregon. Newer signals use symbols instead of word legends [from wikipedia https://en.wikipedia.org/wiki/Pedestrian_crossing ]
Pedestrian crossing signal in Portland, Oregon. Newer signals use symbols instead of word legends [from wikipedia]
But think about it:

Stop signs are red. Red lights are red. Don’t walk is red “portland orange”.

Go is green. Green lights are green. But “Walk” or the “Walking Man” are yellow or white, even at new intersections. Why?

Is it simply that green lights are more expensive. Or is it that green would somehow be confusing? Other countries sometimes have a green walking man.

The MUTCD Chapter 4E describes Pedestrian Control Features. It says the color should be white. Why?

03 The WALKING PERSON (symbolizing WALK) signal indication shall be white, conforming to the publication entitled “Pedestrian Traffic Control Signal Indications” (see Section 1A.11), with all except the symbol obscured by an opaque material.

If I were paranoid, I would say it is a micro-aggression against American pedestrians (leaving aside racial implications), but somehow I think the explanation is far more pedestrian (er, mundane).

The East German Amplemannchen https://en.wikipedia.org/wiki/Ampelmännchen
The East German Amplemännchen

(Why the walking man is a man is another day).


Update 9/24/2015. Eric Fischer sends this along, indicating the walk signal has been white for a long time in the US, to avoid driver confusion.

1939 MUTCD revisions, https://ceprofs.civil.tamu.edu/ghawkins/MUTCD-
1939 MUTCD revisions, https://ceprofs.civil.tamu.edu/ghawkins/MUTCD-

Does telecommuting alter travel behavior and residential choice?

CTS Catalyst summarizes Jason Cao’s Travel Behavior Over Time study in “Does telecommuting alter travel behavior and residential choice?

The potential of telecommuting to alter travel patterns—and even mitigate congestion during peak hours—has sparked interest among transportation planners. Despite this potential, however, the actual impact of telecommuting on traffic has remained an open question.

“In practice, the relationships between telecommuting and travel behavior vary,” says Jason Cao, an associate professor with the Humphrey School of Public Affairs. “For example, an individual might replace a commute trip by working at home but make another non-work trip because of the time savings from not making the commute trip.”

In a recent study, U of M researchers examined the impacts of telecommuting on travel behavior in the Twin Cities metropolitan area. One key finding: regular, non-daily telecommuting is on the rise. While those who telecommuted every day dropped, the number of people who telecommuted once a week  or more increased, Cao says.

The research team’s first task was investigating how telecommuters and non-telecommuters differ from each other in terms of demographic and land-use characteristics. To explore the impact of telecommuting on other household members, the team divided the data into one-worker households and multiple-worker households. Some sample findings:

  • In one-worker households, telecommuters tended to be more affluent, more highly educated, older, and more likely to have multiple jobs than non-telecommuters.
  • Multiple-worker telecommuting households tended to be more affluent, have more members, and live in more job-rich areas than non-telecommuting households.

Next, researchers set out to determine how telecommuting replaces or complements auto use. Their data analysis shows that the effects of telecommuting on travel are limited for multiple-worker households, but that it tends to increase travel of one-worker households, particularly for non-work travel. The team also found that the share of one-worker telecommuting households increased from 13 percent in 2000 to 15 percent in 2010, and other telecommuting frequency categories also increased, Cao says.

Finally, researchers examined how telecommuting affects where people choose to live. Research results indicate that while telecommuting has no influence on commute distance for one-worker households, it tends to decrease average commute distances for multiple-worker households. “This suggests that the ability of one worker to telecommute may motivate the household to seek a location closer to the workplace of other household members and reduce the average commute distance,” Cao says.

“This research has revealed the complexity of telecommuting patterns and the real differences in travel between occasional, regular, and daily telecommuters,” says Jonathan Ehrlich, planning analyst with the Metropolitan Council. “We need to be aware in planning how Internet and mobile technology both replace some and necessitate other travel.”

This research is part of a five-part report sponsored by the Metropolitan Council and the Minnesota Department of Transportation based on data produced by the council’s Travel Behavior Inventory household travel survey.

Additional components of the report examine how changing accessibility of destinations has affected travel behavior, changes in walking and biking, the effect of transit quality of service on people’s activity choices and time allocation, and transportation system changes.

Related Links