On Tuesday, a data collection company took smartphone location data and used it to publish the world's first Social Distancing Scoreboard. This is the first, admittedly rough, attempt to virtually estimate how well states and counties in the US are complying with the official guidelines to stay at home.
The company called Unacast is using anonymous GPS data collected from millions of smartphones to evaluate social distancing scores through this interactive project. While admitting that the current measures are only crude attempts based on the distance traveled by the phone user, the company says it is refining its technology to offer a more accurate picture. This could help enforce social distancing amidst the current coronavirus pandemic. Unacast obeys the EU's General Data Protection Regulations (GDPR) as well as the California Consumer Privacy Act.
After founding the music streaming service TIDAL in Norway, Thomas Walle and Kjartan Slette wanted to understand where users went to concerts to create better playlists and recommendations. As there were no such insights available, or insights of how people move around in the world, in general, the pair built Real World Graph, and Unacast was born. With the arrival of COVID-19 Unacast adapted their product to create the Interactive Scoreboard, to measure and understand the efficacy of social distancing initiatives at the local level.
The Social Distancing Scorecard
The Unacast Scoreboard is simply a set of maps that show how much the average individual movement has decreased after the lockdown. The use of smartphone GPS-mediated location data can reflect such a decrease by measuring the total distance traveled before and after the first case of COVID-19 in each county.
According to Unacast, scores are calculated by combining "tens of millions of anonymous mobile phones and their interactions with each other each day" with further extrapolation of the results. This comparison was intended to tell how well people are following the lockdown norms – the social distancing score.
These scores are compared with the number of cases reported in each county and state, to generate a grade. The resulting analysis is projected as a map of the US with good social distancing success being represented in green and poor compliance as orange.
Currently, the Unacast Scoreboard is based on the overall reduction in the distance traveled per phone user. If people are staying at home except for occasional trips to buy food or medicine, there should be a sharp dip in the distance covered.
The company says it chose to focus on average distance traveled over other scoring metrics because it best matches the number of cases confirmed, and provides reliable measures even with changing data and many unknowns like home locations.
However, at present, this system does not have the capacity to assess how many people the user meets on the way or at the end. The company aims to look into how they can assess the number of people met by the smartphone user within any given area, as well as counting how many locations are visited by the user.
American scores only 'satisfactory'
The Scoreboard displays grades to reflect how well each state or county has been following social distancing measures effectively.
The social distancing scores are based on tracking the current GPS location data to that on any typical weekday, from millions of phones in any locality. Such data comes from the numerous apps on a typical smartphone that tracks one's location.
Overall, the USA gets a 'B' grade. The District of Columbia (D. C.), Alaska, Nevada, New Jersey, and Rhode Island are among the states with proper compliance (shown as an 'A'), as is Illinois, including Cook County, home to bustling Chicago. On the other hand, Wyoming earned the worst score, 'F'. Montana, Oregon, New Mexico, and Idaho are just ahead, but not by much, with a 'D'.
Overall, Northwestern States appear to be the least concerned about social distancing, while the Northeastern States are more compliant, as well as, to some extent, California and Texas.
'A' grades are assigned if the distance traveled within that location has come down by 40% or more. If there is less than a 10% reduction, or if people are traveling more than before, the location gets an 'F'.
The most significant reductions in distance come from heavily hit areas like New York City, where the distance traveled has gone down by 57%.
The benefits and limitations
The advantage is that such comparisons are detailed enough to allow a county-wise view of movements. On the flip side, this is not a scientific assessment, nor has it been reviewed by medical or public health specialists. Neither does it help to tell how far people stay away from each other when they do go out of their homes.
Recognizing the limitations of this early attempt, Unacast says it wishes to help experts judge how far people are complying with shelter in place orders, and whether they are going to work despite advice to the contrary. In the words of the company CEO Thomas Walle, "We can start to see and learn what states are getting this right."
Time trends important says Unacast CEO
Walle emphasizes that the learning from this Scorecard can only be measured as it shows travel trends over a series of weeks, signaling which states and counties are actually taking measures to restrict personal contact between people who are not staying together. This can also be correlated with the change in the number of COVID-19 cases to judge whether such measures are productive.
The US government is already discussing with technology-heavy monoliths like Facebook and Google to try and roll out a way to see if location data from these giants can be used to track and limit the spread of the illness. Such data would be anonymous, enabling the preservation of privacy. Rather than following the movements of individual users, which could be construed as 'spying' or 'infringement of privacy rights', the location data can be processed to yield movement patterns and group gatherings.
In this way, it could supply health and government officials with crucial information, such as whether people in groups are keeping their distance from one another.