It’s a time-consuming and labor-intensive process, given that most Americans carry location-tracking devices with them everywhere they go.
And as it turns out, one doesn’t need to collect all of that data to understand how people are moving. In fact, one doesn’t even need most of it.
About 5 percent of the people in a given area will do.
That’s the thesis behind a new product out from the Google company Sidewalk Labs called Replica. With statistical modeling, one can take travel data from a relatively small slice of people in an area and come up with a pretty reliable view of how the larger population moves.
Nick Bowden, formerly the chief executive officer of the startup MySidewalk, has been working with Sidewalk Labs for about a year on the product. So far, Replica has met with a long list of curious customers: Kansas City and Chicago are on board to start using the product sometime this year, and there are more in the queue behind them. Those are paying customers, not early-stage pilot projects.
“We’re actually already booked out through most of 2019, if not all of 2019 based on some of the demand from regions across the U.S.,” said Bowden.
Essentially, Replica offers models of how people have traveled — by foot, on bicycles, by personal car, by transit, etc. — for what purpose — to work, to school, for entertainment — and how that changes over time. Whereas many local governments might do transportation studies like this every several years, Sidewalk Labs aims to get new information every three months.
“Most places are using models that could be a decade old or even longer,” Bowden said. “So they’re still basing a significant number of capital and planning decisions on [outdated] data from pre-iPhone.”
There are, potentially, a lot of ways government could use Replica. A city, or group of cities, could use it to understand how traffic flows regionally rather than just within the arbitrary bounds of a single municipality. They could use it to look at how travel patterns change as development shifts — how do people respond to big construction projects? How many people in that new subdivision are driving to work, as opposed to taking the train?
It could serve as operational intelligence for planners making big-picture decisions about how the city should grow.
And in the future, Sidewalk Labs wants Replica to be able to plan for scenarios that haven’t happened yet. That way, if a city is considering how a proposed development would impact congestion, parking needs and walkability, they would be able to model it.
It wouldn’t be the first tool to do that — far from it — but it just might be among the most flexible, sophisticated, data-rich scenario planning tools.
That’s largely thanks to Replica’s data sources, which include several vendors that collect location data from smartphones. Bowden and Alexey Pozdnukhov, a scientist-in-residence at Sidewalk Labs’ “Model Lab,” emphatically point out that the data is anonymized before they ever get their hands on it — nobody can use it to track down an ex.
So how can they know, from a phone’s location, without having any idea who the person is, what the purpose of a trip is?
With some advanced educated guessing. If a person takes the same route every day, and that route happens to line up with a set of train tracks, and their schedule matches the train schedule, they’re probably taking the train. If they move along a highway at the same time every morning, and return at the same time every night, they’re probably driving a car from home to work and back.
“It’s all about building models that predict really well how people like me, like you, like him, will behave,” Pozdnukhov said.
That kind of data will likely become more and more important in coming years, as several technologies promise big changes to the way people move. Self-driving vehicles could open up flexible transportation options to people who couldn’t afford them before. Electric cars could change traffic patterns because people won’t be going to the same places for fuel.
And then there are myriad initiatives in urban areas meant to get people to drive less, or at least drive more efficiently. As housing prices rise and people move farther away from where they work, as cities seek to mitigate the nigh-unbearable traffic jams that clog their highways each morning and evening, as local governments try to revitalize downtown areas, they will be looking for better understanding of how their people move.
“We want people to be able to see, is what we’re doing on the ground today impacting the way people move?” Bowden said.
Then there’s the regional planning aspect of it all. Although there are plenty of metropolitan planning organizations (MPO) in the U.S. — which bring together representatives from nearby municipalities to cooperatively plan transportation work — they don’t always succeed at overcoming the piecemeal approach.
“More often than not, if a city wants to understand a regional place, they go to the MPO and the MPO helps them run the model, or they hire a company to do a … modeling exercise,” Bowden said. “It’s not intentional, but what that results in is you have a bunch of cities that do a bunch of micro things that may or may not roll up into a regional model.”
Replica is supposed to help users see the bigger picture because it can capture data from people across a large area. If a city makes a parking change, that could change how people coming from other cities choose to travel to that city.
“The amount of data in those datasets is just enormous and it’s up to data scientists to get this data out into the product,” Pozdnukhov said.
