The COVID-19 Task Force is considering powerful new data that, if used here, could help measure stay-at-home compliance deep into Memphis and Shelby County’s many diverse neighborhoods.
A sampling of the data obtained by the Institute for Public Service Reporting demonstrates its broad potential in battling the spread of the novel coronavirus:
Residents responded enthusiastically all across the county to last month’s series of safer-at-home orders, but that compliance appeared to ease in early April in numbers of pockets including the Evergreen Historic District, an affluent neighborhood in Midtown Memphis near Overton Park.
“To an extent it doesn’t surprise me,’’ said Sam Goff, recent president of the Evergreen Historic District Association, who was out mowing his lawn on a bright Saturday morning. He offered an innocent explanation for the data’s findings: Many of his neighbors are simply outdoors walking and riding bicycles, not flagrantly violating Mayor Jim Strickland’s safer-at-home order.
Yet the task force’s medical advisers are excited by the data offered free of charge by San Francisco-based SafeGraph. Mobility trackers like Unacast have become popular during the COVID crisis for grading stay-at-home performance at the state and county level. Yet SafeGraph hopes to distinguish itself by offering granular detail down to the neighborhood level.
“The city of Memphis is interested in this data to just understand how social distancing is going: What neighborhoods might be not socially distancing as much,’’ said SafeGraph spokesman Nick Singh. “And in those neighborhoods, you could do more enforcement. You could do more outreach. You could run more ads. You could tailor your messaging.’’
But if Memphis hopes to use the data it may have to first come to grips with its long history of political oppression, including intrusive surveillance of activists and ethnic minorities.
Memphis City Councilman Dr. Jeff Warren, who initiated discussions with SafeGraph, said some officials fear use of the data might run afoul of a longstanding federal consent decree that corrected years of abuse by Memphis leaders who spied on political activists during the civil rights era and again on Black Lives Matter protestors in the modern age.
“They feel you can’t be monitoring anything,’’ said Warren, who was attempting on Sunday, April 19, to consult with the city’s legal advisers.
Singh said SafeGraph uses “anonymized’’ data that measures “aggregated behavior at the neighborhood level’’ not individuals. He emphasized the data can’t be used for technology assisted contact tracing, the controversial use of smartphone data and other technology to investigate the spread of disease among individuals.
Nonetheless, longtime Memphis civil rights attorney Bruce Kramer said the city should move cautiously before using or taking possession of any mobility data.
“There’s a trust issue,’’ said Kramer, who filed the lawsuit that led to the landmark 1978 Kendrick consent decree that shut down a Cold War-era domestic spy operation run by the Memphis Police Department and forbids the city from collecting or storing information “relating to any person’s beliefs, opinions, associations or other exercise of First Amendment rights.”
Warren, a physician who lives in Evergreen, said he’s hopeful officials can work around the decree because he believes the data can help save lives.
“Anything that we can do to allow our citizens to know whether they’re reaching the bar that we want to achieve in lowering this COVID transmission rate is very important,’’ he said.
According to data maps SafeGraph released to the Institute, broad areas of Shelby County are experiencing wide compliance with stay-at-home orders.
Sixty to eighty percent of families are staying home in sections generally corresponding with Germantown, Collierville, Lakeland and Arlington.
The maps show a sudden darkening – meaning more people were staying home – in the days after March 24 and 25 when a series of municipal and county safer-at-home orders took effect. The orders require residents to stay home unless they are performing essential services such as police, emergency and food supply work or they’re out buying groceries, exercising or engaging in other permitted activities.
But between April 5 and April 11 much of the county began to lighten up again before rebounding on April 12, the maps show.
The data appears to confirm official suspicions.
“The city of Memphis for some reason anecdotally thought that people are taking it pretty seriously, but started kind of slacking off after a little while,’’ said SafeGraph’s Singh, who’s noticed similar trends in other parts of the country. “So, that’s kind of what they’re worried about (in Memphis). And you can kind of anecdotally tell that, but you won’t know until you look at the data.’’
SafeGraph did not share actual data tables with the Institute.
But when the firm offered to share its data to the city of Memphis, Warren, a member of the COVID task force, jumped on it. He even signed a non-binding contract to get the discussions rolling.
Memphis was the latest in a series of cities SafeGraph contacted since it began working with academic researchers to develop a “social distancing metrics dataset’’ that attempts to measure movements from defined residential neighborhoods – Census tract block groups, each containing an average 1,500 households.
SafeGraph specializes in “points of interest’’ data that includes “really accurate building footprints,’’ Singh said, “polygons that really define the outline of a McDonalds or a Wal-Mart.’’ SafeGraph uses “anonymized’’ location data from 45 million phones, tablets and watches nationwide to measure foot traffic into the operations of clients like Choice Hotels as well as the businesses of competitors.
Public health researchers used SafeGraph data for a paper published last week by the Centers for Disease Control that found stay-at-home measures were helping Seattle, New York, New Orleans and San Francisco – four cities ravaged by coronavirus – to rebound from the pandemic. Officials in Chicago and Denver have also tapped SafeGraph’s free data.
Warren was impressed by what he saw.
“We have to get this epidemic under control,’’ said Warren, who believes neighborhood-level details will help.
Dr. Manoj Jain agrees the data could help save lives.
“The more granular we can get the data, the more helpful it will be,’’ said Jain, an infectious disease expert hired to help guide the task force.
Those granular details can help authorities focus on neighborhoods like Evergreen, where data indicates some residents may be ignoring stay-at-home provisions.
SafeGraph data indicates one in five or fewer families stayed home on a typical day between April 5 and April 12 in a portion of Evergreen immediately west of Overton Park and south of the Memphis Zoo.
Stay-at-home compliance was a little better in other parts of the neighborhood. More than 20 percent but less than 40 percent of families stayed home on average over that same period across several blocks of Evergreen immediately to the west.
SafeGraph maps display similar findings – 20 to 40 percent compliance – across broad areas in Whitehaven and north and south Memphis.
At the same time, the data may be skewed in places.
One particularly puzzling quirk involves vast open areas like Memphis International Airport, Shelby Farms and Overton Park. Though few if any people actually live in those areas, the data indicates stay-at-home compliance is low there.
“We are likely picking up a few devices as living there (incorrectly),’’ Singh said in a follow-up email. “When you have low numbers, you see weird things with the data. I can ask about editing it out.’’
It’s unclear, but the low compliance rating in part of Evergreen might be skewed because it’s contained in the same Census block group encompassing Overton Park. Neighborhood leader Goff said he believes another metric may be affecting Evergreen – the data counts a device as not at home when it strays 500 feet or more away.
“People here seem to me to walk more. Ride bikes more. They walk their dogs every day. I see dozens of people up and down this street walking their dogs. And people going to the park to take advantage of the park,’’ Goff said. “The park is certainly large enough that people can get exercise and still social distance. Right? So, I think that that may be part of the issue with the numbers being skewed.’’
However, Singh said SafeGraph’s data could also help authorities measure away-from-home social distancing.
“We have that whole foot traffic component, which is another measure for social distancing,’’ he said. “Because as much as you could leave your house, another interesting metric would be how much are people visiting parks or restaurants or grocery stores? And different departments are using that data as another way to proxy how social distancing is going? That is only revealed through our data.’’
It’s just that sort of detail that the ACLU’s Kramer finds troubling.
“It’s so scary,’’ he said, questioning how such information would be controlled. “Where is this data kept? Is it in a central repository? How long is it kept? Who has access to it? For what purposes is it being collected? Is it really collected just for the health issue, the COVID19 issue, or is it going to be used for some other purpose?’’
Kramer’s concerns mirror those expressed nationally by civil libertarians regarding offers by large tech corporations like Google and Apple to assist in contract tracing. A white paper released last week by Daniel Kahn Gillmor, ACLU senior staff technologist for the organization’s Speech, Privacy and Technology Project, warns that the COVID-19 pandemic is “an opportunity for would-be authoritarians and powerful corporations to expand their power,’’ recommending that any technology assisted contact tracing include safeguards such as voluntary participation and “minimal reliance on central authorities.’’
SafeGraph’s Singh said the firm’s data is not used for contact tracing.
“…Because it’s aggregated, this data is not able to help with contact tracing. We do not work with anyone around contact tracing,’’ he said in an email.
He said SafeGraph uses data from a variety of apps “in the safety, health or navigation space’’ that collect “anonymized mobile GPS” location data. He declined to name those apps or SafeGraph’s data suppliers. Though Kramer questioned how voluntary the data had been given by phone users, Singh said “it’s 100% opt-in data and compliant with the California Privacy Act….
“Our data is all about aggregated behavior at the neighborhood level; are people staying home in a Census Block Group, or not? It’s not about tracing individual user journeys – that’s not possible with any of the data we sell,’’ he said in the email.
Determining a device’s “home’’ is determined through scientific inference not literally through records, he said.
“If we detect a device that spends nighttime hours in one location for about six weeks, we have a pretty good inference that that’s the device’s home location. So that gives us a really good idea of where is the device’s home,’’ Singh said.
Still, trust is thin for some in Memphis.
The 1978 consent decree roared back into the news two years ago when U.S. District Court Judge Jon McCalla found MPD had violated the decree while investigating a range of activists, including several involved in the Black Lives Matter movement. Evidence showed an undercover MPD officer created a fake Facebook account to “friend” activists to help monitor their activities. Several of the activists landed on a “blacklist’’ requiring an escort for them to enter City Hall; information on their activities was routinely circulated to the military and major employers like FedEx and AutoZone.
“Their ability to surveil people is vastly beyond what it was back in the day,’’ said historian Michael Honey, a former Memphis activist who was monitored by police and who was a plaintiff in the original lawsuit that led to the 1978 decree. Special precautions are needed for any technical COVID assistance, he said.
“There should be some strong safeguards built into it if it is used, because we know from the past there is government abuse of information, government abuse of power. This is trying to save people. But you never know down the road how that information may get used.’’
Warren said late Sunday Memphis chief legal officer Jennifer A. Sink told him “worry about the consent decree’’ was “slowing down’’ approval to use the data.
Sink released this statement Monday afternoon: “No decision has been made at this time, but SafeGraph has approached the City of Memphis with a data tool, at no cost, that could help us measure how well our Safer at Home order is working during this public health emergency. However, use of the information could potentially violate the 1978 Kendrick Consent Decree. Before moving forward, we would first need to present this proposal to the court appointed monitor, and possibly bring it before the Court.”
Big changes coming
Warren said such fears may be overblown because smartphone companies and many others already have vast amounts of personal data on many Americans.
“In this particular instance, you have people who are elected that are making sure this remains anonymous. And you have people you can hold accountable if it isn’t,’’ he said. “So, in my mind this is a much safer way to share data than it would be by how we’re already doing it.’’
But pressures are building to contain coronavirus, which by Sunday had killed 37 people and infected another 1,729.
“We’re focusing on educating the community about the safer at home orders,’’ Memphis Police Director Michael Rallings said last week. “…But we will make arrests if necessary. We’re doing this by conducting high visibility patrols at hospitals, testing sites, grocery stores, drugstores, essential businesses and our parks.’’
Police have made three arrests so far for violations of safer at home orders, he said.
Another high-tech option authorities hope will take off is a free smartphone app developed by the University of Memphis’ MD2K Center of Excellence called mContain that helps promote social distancing. Launched April 7, it currently has 400 active Android users; the iOS version is under review by Apple.
“We are waiting for 1,000 active daily users to provide crowding hotspot reporting for the Memphis region,’’ chief software architect Timothy Hnat said in an email.
Among other things, the program alerts smartphone users who download the app under an anonymous identifier when they have close encounters that could jeopardize health. The system only works among the pool of people who download the app. It could be particularly useful in workplaces.
“In a workplace setting, employees could all download and use mContain and it would let them measure their own encounters with each other,’’ Hnat said. “The results of these measurements could allow employers to adjust their workplace policies to better maintain social distancing and they would be able to see the results of these decisions the following day.’’
Whatever technology officials may incorporate into the coronavirus response, residents can expect major changes, said Jain, the infectious disease specialist.
“This goes to the heart of our healthcare system and to who we are and how we do things to one another,’’ Jain said. “I mean, you can’t you can’t get any deeper than within our own bodies and within our own social interactions. … The only lucky thing is our food supply and our fuel supply hasn’t been damaged. But our healthcare and healthcare systems have clearly been disturbed.’’
This story first appeared at dailymemphian.com under an exclusive use agreement with The Institute. Photos reprinted with permission of The Daily Memphian.