As we navigate the COVID-19 crisis and its broad impacts, we are understandably focused on what appear to be our most acute concerns – flatten the curve, save the economy, and develop a vaccine. Daily, we are reminded of the massive toll in lives and livelihoods that the pandemic has exacted worldwide. At the time of writing, more than 10 million cases of COVID-19 have been confirmed, and more than 500,000 lives have been lost. Economically, we are facing the largest global recession in decades and a historic contraction of per capita income.
But as the days and weeks push forward we are also beginning to understand more subtle impacts across communities. Many of these are consequent to the physical distancing and quarantine measures which have been put in place to slow the spread of disease. Crisis, coupled with isolation, has raised important concerns about mental health. An April 2020 poll of US adults reported that 56% of respondents had experienced a negative impact on their mental health associated with worry and stress over the coronavirus. Text messages to the US government’s Substance Abuse and Mental Health Services Administration’s (SAMHSA) Disaster Distress Helpline increased nearly 1000 percent . Data suggests a significant increase in cases of domestic violence, particularly among marginalized populations.
Our group, amongst others, has suggested that social distancing has also magnified individual privacy risks. Specifically, we have observed that social media users appear to be sharing more personal information online. The expanded breadth and depth of online activity during the pandemic has brought increased opportunity for privacy violations, but also we suggest that people may be looking for ways to stay close during isolation. This is a concern because we know that personal information shared voluntarily through social media is vulnerable to collection and misuse.
We are living out this crisis online
Many have drawn comparisons between COVID-19 and the influenza pandemic of 1918. In many ways the comparison is apt, including the use of government-mandated social distancing as a primary intervention. A recent piece in The Atlantic describes the loneliness of that time. Elements of the author’s depiction resonate today. People alternating between fear and frustration, lamenting over lost experiences, and mistrust of leadership was pervasive.
But for all of the similarities, the influenza pandemic of 1918 and today’s coronavirus pandemic bear as many differences – modern hospitals, medical equipment, antibiotics and antivirals, and highly specialized doctors. And beyond significant advances in the field of medicine, a fundamentally different social fabric. The Internet and its associated technologies have furnished a global village. The number of internet users worldwide has now reached 4.57 billion, or 59% of the global population. Over 3.8 billion people use social media. In a moment when we are asked to stay at home, when we are asked to distance from our friends and neighbors and colleagues and families, this connectivity has become a lifeline. The last few months have seen a highly publicized rapid growth of videoconferencing companies and surging social media usage. Facebook has reported increases in messaging more than 50% in countries hardest hit by the virus. Twitter also reported an increase in daily active users by 23% from this quarter last year. That is to say, we are living out this crisis online. We gather and share information, we work and we shop, we connect and socialize through Internet media.
The benefits of modern connectedness are countless. Internet technologies have allowed many of us to continue to work through this period, and our kids to complete their school year. Telehealth has expanded access to essential health services. Contact tracing apps and digital surveillance measures offer hope for virus tracking and containment.
The toll on privacy
Pervasive, persistent connectivity brings risks as well. Amongst them, risks to privacy are perhaps most apparent. Privacy concerns related to contact tracing protocols have now (rightfully) gained widespread attention. Inherently, the contact tracing process involves a fine-tuned accounting of one’s location history, relationships and health information – each sensitive information in its own right. When considered together, in the context of a global crisis, the listing of potential harms lengthens. A number of nonprofits and human rights organizations have raised flags about the potential for these tools to exacerbate domestic abuse, intensify stigmatization, and expose vulnerable populations.
With a few notable exceptions, most countries are following protocols that rely on voluntary reporting. In the US, it is unlikely that a unified nationwide contact tracing protocol will emerge, but rather that exposure notification efforts will be state-directed. Privacy advocates stress a few key requirements. The first is transparency. Users must understand exactly what data will be collected, how it will be used, and who will have access it. The second is decentralization. That is, an individual’s data should be stored only on their personal device and not on a central server. This mitigates risk of data breach or misuse at the central authority. Third, data should be destroyed after a preestablished period of time. This requirement acts in service to the first, helping to ensure data is not later used in unforeseen ways. Ultimately, the success or failure of any opt-in contact tracing technology will depend on sufficient participation (by some estimates more than 50% of the population), so that user trust will be paramount and privacy will take center stage.
Recently, our team has suggested that, in addition to risks posed by contact tracing protocols, a parallel set of privacy risks are emerging associated with information sharing through social media. Our work focuses on the phenomenon of so-called self-disclosure, or the voluntary revealing of personal information, e.g. motives, desires, feelings, thoughts, experiences, in exchange for social reward, i.e., engagement, support, legitimacy. Through analysis of nearly ten million tweets related to the pandemic between late January and early May 2020, we have observed that the practice of self-disclosure has been closely linked to the crisis lifecycle. Specifically, as the crisis intensified throughout February and March, rate of self-disclosure increased sharply – from around 13% to around 23% of per-day tweet volume. Through April, rate of self-disclosure in COVID-related tweets plateaued at a significantly higher rate than baseline.
When we studied the topics represented by instances of self-disclosure, we discovered an emerging focus on stay at home measures and personal, emotional experiences of the crisis. A large cluster of instances of self-disclosure in our pilot study of Twitter data in March and April are support-seeking. Words like “support” and “help” are amongst the most common in that set of tweets. This finding is generally consistent with the thesis that social media users are using digital platforms to elicit support and stay connected during isolation. Ongoing studies further exploring these initial observations are underway with my collaborator Dr. Anna Squicciarini, thanks to new support from the NSF RAPID program.
Self-disclosure through social media carries routine risks even on the most ordinary days, leaving users exposed to discrimination, harassment and bullying, cyberfraud and other crimes. High-profile examples of collection and misuse of self-disclosed personal information (e.g., Cambridge Analytica scandal) forewarn the dangers of user targeting and manipulation. We argue that heightened rates of self-disclosure, coupled with increased anxiety and stress, may leave users particularly vulnerable to these harms. For one, we know that state-backed actors are already leveraging social media to spread disinformation related to the pandemic. It is difficult to measure the impact of these campaigns. We can measure engagement and spread, but the extent to which malicious posts meaningfully impact hearts and minds remains an open question. What is relatively certain, however, is that the efficacy of these efforts depends on accurate targeting, an art in turn facilitated by user sharing.
Mitigating risk
When we weigh the risks to privacy implicit in contact tracing protocols, we wade into complex ethical discussions about privacy-security tradeoffs. Sometimes, sacrifices in rights and liberties need to be made in service to security, in this case public health. Certainly, facing the existential threat of a global pandemic might be one such case. Although, these sacrifices should be met with scrutiny, and that is particularly the case during crisis when fear may exert disproportionate influence. Ultimately, our goal should be to engage measures that are maximally effective, but which we can ensure are adequately regulated and overseen. While the questions are really hard, they can at least be sketched out for concrete choices we can make – whether to engage in contact tracing and selecting amongst viable approaches to doing so.
The risks associated with self-disclosure on social media are substantively different. There are no clear set of options in front of us. There is no clear decision to make. Existing, voluntary participation in these platforms is already widespread. Government regulation of Facebook and Twitter is a topic of continual discussion, but even aside from the social media giants, there are no shortage of opportunities for individuals to share personal information online through ad hoc engagement with websites, from comments on digital news content to product reviews.
Efforts to mitigate risk will likely need to take the form of user education and awareness. For some, learning will be experiential. Privacy risks are difficult to measure and highly contextual. Individual disclosures may seem innocuous, but increasingly sophisticated inference algorithms running over aggregations of innocuous bits of personal information can translate to deep personal insights. For many, intangible and unspecific future harm is outweighed by the immediate gains of sharing. In the extraordinary circumstances we currently share, this may be especially true.