The School of Cities at the University of Toronto and the Institute for Governmental Studies at the University of California, Berkeley have been using mobile phone data to track the recovery of 62 downtowns across North America. This work has been being published at downtownrecovery.com, but it has also been widely cited.

First, to be clear on how this works, the data they are collecting is not dependent on people actually making calls or actively consuming data on their phone; instead it is simply based on people having a phone with them and being physically located in one these 62 downtowns. It also covers the period between January 2019 and November 2022, and includes cities with least 350,000 people.

I'm not exactly sure how long the phones need to be in a particular place or how they treat time in their data, but the unit of measure is something that they call a "Point of Interest." This includes things like restaurants and shops, so presumably this data isn't just saying, " I went downtown and sat in my office for 8 hours." It could also be, "I went downtown and ate good pasta."

I say this because, based on my understanding of the data, having a high Recovery Quotient (RQ) could mean a number of different things. It could mean that more people are back in the office, but it could also mean that the downtown isn't a monoculture and that it has other things going on besides just work.

In any event, here's what they have found:

The headline finding is that San Francisco has the lowest RQ at 31% and Salt Lake City has the highest at 135%. There does appear to be a bias toward higher recoveries with mid-sized cities, and one of the reasons for this is that these recovery quotients appear to be correlated with average commute times:

Some of the other strongly correlated explanations, include the percentage of jobs in professional, scientific, and technical fields:

And the number of days that events were shut down during the pandemic (note the Canadian cities on the right below; welcome, New Orleans):

I suppose one way to grossly oversimplify these findings is to say that some people have been avoiding going downtown if they can't quickly drive there (and have to take transit), if their job more easily allows them to work from home, and if things were shut down for too long during the pandemic. Because if it was, they maybe forgot about all of the fun things that typically happen downtown.

Image: The School of Cities

Sidewalk Labs is currently building out a platform called Replica that will support them in their development plans here in Toronto. Replica is: 

“a user-friendly modeling tool that uses anonymized mobile location data to give planning agencies a comprehensive portrait of how, when, and why people travel in urban areas.”

Here is a preview of the Replica dashboard showing a section of Main Street in Kansas City. I hope the animated GIF shows up for you.

The platform uses a combination of mobile location data (~5% of the population) and on-the-ground checks, typical stuff like manual traffic counts and transit boardings.

The goal is to understand in real-time who is using a street, as well as how (driving? cycling?) and why (going to work?).

Their introductory blog post obviously stresses the importance of personal privacy, but I am curious how they determine where people are going.

I suppose if they pair journeys with destinations (and the durations at those destinations) they can make reasonable assumptions around the why.

I think the benefits to all of this are clear. But does any or all of this worry you from a privacy standpoint?