(Director’s note: Jordan Frith, Pearce Professor of Professional Communication in English, researches mobile technologies, social media, and infrastructure, particularly where those topics intersect through questions of space and place. He’s the author of 30 peer-reviewed articles on these topics and three books, the most recent of which was published by MIT Press in 2019. This is Clemson Humanities Now.)
Back in March when much of the world was starting to realize that the COVID-19 pandemic was not going to just disappear, various governments and corporations began exploring plans for digital contact tracing applications. Digital contact tracing involves using the data from mobile phones to track whom people come in contact with on a daily basis, so if someone is later diagnosed with COVID-19, anyone they shared a space with can quickly be notified. The apps mostly work through Bluetooth, meaning the phones of anyone who signed up for the contact tracing app would constantly broadcast a Bluetooth signal to nearby phones and record the unique ID of each phone within Bluetooth range. Those unique IDs would link to a database of users, which is how people would be notified if a positive COVID-19 case did occur.
With the COVID-19 outbreaks across the globe, digital contact tracing began being talked about as maybe the number one tool for containing outbreaks without going into full lockdown. A successful system would enable health officials to quickly identify case clusters and quarantine people who were at risk. And in a few cases, digital contact tracing has been useful, particularly in places like South Korea that put these plans into place early in the pandemic and got significant public buy-in. But many of the much-hyped digital contact tracing plans have mostly failed. For example, the UK digital contact tracing app has not made much of an impact, and the United States still does not have a robust Federal digital contact tracing system. So why have these systems not succeeded despite getting so much public attention? I argue that it’s partly because the plans relied too much upon data as a good in itself without the necessary critical reflection necessary for the success of most big data projects. That reflection and interrogation is where the humanities come in. As I explain below, the failure of digital contact tracing projects at the time we need them most shows why we need more data-literate humanists.
Digital contact tracing is obviously somewhat unique because it’s not every day we have a global pandemic that has killed hundreds of thousands of people. But the hype surrounding digital contact tracing is less unique when considered in the greater context of what has been called the “big data revolution.” The supposed big data revolution represents two major shifts in society in the 21st century. The first is that we are producing far more data than ever before about everything from energy usage to mobility patterns. For some context, according to Viktor Mayer-Schonberger and Kenneth Cukier, ‘‘Google processes more than 24 petabytes of data per day, a volume that is thousands of times the quantity of all printed material in the U.S. Library of Congress.’’ And that’s just Google. The second development is closely related. We now have vastly improved computing power necessary to process and act upon huge datasets. These two trends have combined to usher in an increased reliance on data in many walks of life, and according to some proponents, if people just collect enough data, the “data can speak for itself.” This blind belief in data without diving deeper into the unique contexts of data sources is part and parcel with the hype we saw about digital contact tracing.
There’s nothing wrong with the use of large-scale datasets to solve problems. Data informed decision making is often far superior to just winging it and guessing. However, the hype surrounded big data often went too far and did not involve humanists to ask questions about data projects that could have helped them succeed. After all, despite beliefs that data could supposedly speak for itself, that can simply never be the case. Data always needs to be interpreted to be acted upon. In addition, datasets are often a partial picture of a phenomenon, and data collection also often has privacy and personal liberty implications that are not closely enough considered. As Rob Kitchin argues, data projects are often best served by combinations of data scientists and humanists/social scientists who can interrogate the data and how it is used to inform decision making.
The COVID digital contact tracing is an important contemporary example of what critical, data-literate humanists can add to data projects. After all, these digital contact tracing apps, despite all their hype, faced some major issues from the very beginning that were not fully addressed. For one, as anthropologist Genevieve Bell argued, the design of these apps matters significantly for user privacy, and governments did not communicate design choices clearly to the public. The lack of explanation of where data was coming from and how it was stored likely hampered public adoption, and contact data is only useful when enough people buy into the system to provide a comprehensive dataset. Secondly, and maybe most importantly, the design choices in these apps were inherently exclusionary. They required mobile phone data (unlike focusing on a human-driven contact tracing system), so children, some elderly people, the homeless, and other non-adopters simply could not participate. In addition, many cheaper mobile phones do not have low-energy Bluetooth capabilities necessary for the applications. For people without the right devices, they are simply left out of the dataset, which can have major consequences for epidemiological tracking. And finally, digital contact tracing itself cannot work successfully without a rapid testing infrastructure to quickly identify positive cases, and—in my opinion—a social safety net that can financially support people staying home from work if they receive a notice they could be infected. Without these broader systems in place, the data provided through these systems and how actionable the alerts are was always going to be limited.
The issues in the previous paragraph have all limited the efficacy of digital contact tracing, and that’s not even mentioning the major surveillance concerns surrounding location data and whether systems are dismantled once (well, maybe if) the pandemic is under control. After all, Congress just renewed parts of the Patriot Act almost 20 years after it was first passed in the shadow of 9/11. But regardless, many of these issues could have been addressed through more critical engagement and less of a blind embrace of data. In other words, I argue part of the reasons these projects have not succeeded is that they often did not involve the necessary types of critical literacies that are the expertise of the humanities. By interrogating the plans for these datasets, issues of inequality, concerns about design and data storage, fears about surveillance, all could have been addressed from the start and communicated to the public. But instead, governments too often pushed forward with these plans without the types of critical analysis required to make them work.
In conclusion, I’m using digital contact tracing here as an example, but I could point to so many other data projects to make the same case. As humanists, we need to play a role in how data is collected and how data is interpreted. We also have to not be afraid of data. Humanists don’t have to be mathematicians or data scientists, but some of us need to know how to question datasets, how to explore where data comes from, and how to interrogate what it may be missing. And we need to move past antiquated quantitative/qualitative divides that act like numbers are outside the purview of the humanities. Most data projects aren’t going to have major public health consequences like a digital contact tracing app. But regardless, we need to fight for a place at the table for future data projects in both industry and government. We have a role to play to shape the collection and interpretation of data of various types.