Skip to content

Google mobility reports, differential privacy and a Berlin-based artist

In the beginning of April 2020, Google started to publish mobility reports for 131 countries and regions in the world. The company uses the same type of data they normally use for their service Google Maps to show mobility in different areas. Many of you may have taken advantage of this type of data in the past, e.g. checking when the grocery store is not too crowded or what is the best time to get a seat in a fancy restaurant.

The data in the new mobility reports is clustered in the categories retail, recreation, groceries and pharmacies, parks, transit stations, workplaces and residential and aims to show movement trends in times of a global pandemic. Although no absolute numbers are displayed, the increase or decrease in percentage points compared to a baseline might help to understand if exit locks work or how movement patterns of citizens changed on a broader level. Eventually, these insights can help governments to support decisions about managing the COVID-19 crisis.

How does the report look like?

You can find reports for 131 countries and regions here.

I checked the report for Germany from 5th of April 2020. Compared to baseline, the movement in retail and recreation areas dropped by 58%. This category includes restaurants and cafes but also libraries, museums and movie theaters, all of which are currently shut down.

At the same time, we can see a 61% increase for mobility in parks compared to the baseline. Here it is important to note that the baseline is the median value for the corresponding weekday between 3rd January 2020 and 6th February 2020 (when it’s freezing in Berlin and most people outside are either hard boiled joggers or dog owners).

Even though my inability to get yeast or flour in the supermarket nowadays paint another picture, the time spent in the category grocery and pharmacy has dropped by 13%.

Who is included in the dataset?

However, one needs to keep in mind that only a subgroup of the population is represented in this data. One might think about people not owning smartphones or similar devices. But the group is even smaller. The mobility report is only based on Google Maps users who have turned on the location history in their settings.

I want to use this opportunity to share how you can stop Google from tracking your location once and for all. Unfortunately, investigations in 2018 revealed that Google is still using your location data even though users opted out for location history. Here you can read up the whole story. And this Wired article perfectly explains how to navigate to the right setting hidden in your Google Account (for iOS and Android) to finally turn off location tracking.

Isn’t this data private and sensitive?

At least, the tech giant affirms to take care of the hoarded data by using differential privacy. But what does that actually mean? In my own words, I’d describe differential privacy as a way of ensuring privacy of published data through different methods such as adding noise to the data. An algorithm can be considered differentially private if it is not possible to obtain from the output or shared results if the information of an individual was used or not. As a consequence, there is no risk for an individual to be part of this database. The concept was mostly shaped by researcher Cynthia Dwork and you can find all the underlying maths here.

“Differential privacy addresses the paradox of learning about an individual while learning useful information about a population”

– The best quote on the Internet about differential privacy by Cryptowiki

A bit more inviting, let me also share a wonderful video that explains the admittedly complicated construct.

Video on differential privacy by Simply Explained

If you want to deep-dive, I highly recommend the following paper produced by a working group on privacy tools for sharing research data at Harvard University. The not only gave their document an appealing name (“A primer for a non-technical audience”) but also succeeded in explaining the consequences and benefits of differential privacy. Next, they offer a broad range of case studies illustrating and the pitfalls of sharing research data and how to overcome them with the working mechanism of differential privacy.

How reliable is Google’s database?

Before you decide for yourself, consider the art installation by Berlin-based Simon Weckert. The artist borrowed 99 phones and put them in a little red handcart. With his handcart he walked up and down a Berlin street to trick Google Maps into displaying a big red line in the app indicating slowed traffic. The beauty lies in the simplicity.

Artist Simon Weckert trolling Google Maps

With that, stay home, stay healthy.