Can We Ensure Fairness With Digital Contact Tracing?
This article originally appeared in the Monday Magazine of 3Quarks Daily (see here) and on my personal blog (see here).
COVID-19 has forced populations into lockdown, seen the restriction of rights, and caused widespread economic, social, and psychological harm. With only 11 countries having no confirmed cases of COVID-19 (as of this writing), we are globally beyond strategies that aim solely at containment. Most resources are now being directed at mitigation strategies. That is, strategies that aim to curtail how quickly the virus spreads. These strategies (such as physical and social distancing, increased hand-washing, mask-wearing, and proper respiratory etiquette) have been effective in delaying infection rates, and therefore reducing strain on healthcare workers and facilities. There has also been a wave of techno-solutionism (not unusual in times of crisis), which often comes with the unjustified belief that technological solutions provide the best (and sometimes only) ways to deal with the crisis in question.
Such perspectives, in the words of Michael Klenk, ask “what technology”, instead of asking “why technology”, and therefore run the risk of creating more problems than they solve. Klenk argues that such a focus is too narrow: it starts with the presumption that there should be technological solutions to our problems, and then stamps some ethics on afterwards to try and constrain problematic developments that may occur with the technology. This gets things exactly backwards. What we should instead be doing is asking whether we need a given technology, and then proceed from there. It is with this critical perspective in mind that I will investigate a new technological kid on the block: digital contact tracing. Basically, its implementation involves installing a smartphone app that, via Bluetooth, registers and stores the individual’s contacts. Should a user become infected, they can update their app with this information, which will then automatically ping all of their registered contacts. While much attention has been focused on privacy and trust concerns, this will not be my focus (see here for a good example of an analysis that looks specifically at these factors, drawn up by the team at Ethical Intelligence). I will instead focus on the question of whether digital contact tracing is fair.
Michael Klenk and Hein Duijf argue in a recent publication that digital contact tracing, often put forward as an effective mitigation strategy with respect to COVID-19, may pose a fairness risk when the possible consequences of the technology are situated in a broader social context. When discussing digital contact tracing, much focus is often placed on the efficacy of such a strategy in terms of its ability to save lives, that is, its ability to contain the virus as to avoid overwhelming the healthcare system. In conjunction with this, there is a second priority: saving livelihoods, which involves maximising the ratio of free (unquarantined) persons in a society.
When it comes to digital contact tracing, it seems plausible that it may save lives, and there is evidence that such strategies are indeed effective when there are few infections. The system works by identifying infected individuals and notifying their contacts, which reduces the number of transmissions. While effective during the SARS outbreak of 2003, it is doubtful that a similar strategy will work with COVID-19. With SARS contact tracing was much easier, as transmission occurred only after the onset of symptoms. With COVID-19, however, current studies indicate that up to 47% of infections occur before the onset of symptoms. For this reason, simply isolating infected individuals (who are often only identified after the onset of symptoms) is an ineffective way to mitigate the spread of the virus. People may be unaware that they are infected, and by the time they realise this and tracing kicks in, it will be too late. A further potential issue with digital contact tracing is the tipping point at which it becomes effective: up to 60% of the population would need to use the app in order for it to be efficacious. However, these levels of uptake are unlikely to occur, as the meagre 16% that was achieved in Singapore attests to. Even Iceland, with a population of around 356 000 only saw 40% of its residents make use of its tracking app Rakning C-19. In addition to these uptake worries, for tracing techniques to work, a reliable testing infrastructure must also be in place. While such infrastructure might be possible in more developed states (although it is by no means guaranteed), developing nations will struggle to put the same amount of resources into this, potentially intensifying existing inequalities.
What the aforementioned suggests is that it is by no means certain that digital contact tracing will be effective in saving lives nor saving livelihoods. As noted earlier, a further question that emerges in this regard is whether digital contact tracing is fair. That is, are the costs of the pandemic allocated fairly across society? Who are the potential beneficiaries of this technological deployment, and who foots the bill? As Klenk and Duijf astutely note:
“The costs of the SARS- CoV-2 pandemic are distributed unequally across countries and within societies. Within some societies, people of colour and migrants are disproportionately affected by the effects of COVID-19, for example by being represented disproportionately amongst COVID-19 deaths.”
Based on this, the authors go on to claim that there is a risk that the composition of those who are required to quarantine will be unfair. This is due to social and economic features (and the relation between these two) that are already prevalent in society that could be exacerbated (as is often the case with new technology). Once again, this draws our attention to the question why technology as opposed to what technology. It is not the case that all jobs can be done with equal ease from home. Specifically, those jobs that can be done from home tend to be done by workers who are in middle to high income industries (basically white-collar jobs). Service staff (for example in restaurants) cannot work from home, and do not earn very high salaries. Subsequently, they may rely more on public transport, which would significantly increase the amount of registered contacts they would have, subsequently increasing their chances of coming across an infected person and being forced to quarantine. Without a large social safety net that only developed nations would be able to implement (whether they in fact have such safety nets in place is a separate question), it would result in a loss of income. Developing nations simply do not have this luxury, as
“Many poor and middle-income countries face a balance-of-payments crisis and a collapse in government revenues as they need to raise health-related spending and imports (to reduce the death toll) and welfare (so that workers can isolate themselves without running out of money). Whereas governments in rich countries can borrow cheaply in a crisis as investors flock to safety, poor countries see their borrowing costs soar.”
While this is only one example, it is easy to see how such a sequence of events could play out in many industries where workers already find themselves in economically precarious situations. Moreover, there is evidence which suggests that social networks, from a behavioural perspective, will come to significantly influence the effectiveness of any given technological intervention. This is due to the fact that, to put it rather simply, birds of a feather flock together: individuals are more likely to form bonds with those who are similar to them (the technical term for this is homophily). Specifically, people are more likely to interact with those that they share working spaces and similar socio-economic challenges with. Significantly for our purposes, such network clustering means that those who finds themselves in economically precarious circumstances will form social ties with others in similar situations far more easily. An implication of this, as it relates to digital contact tracing, might be that already disadvantaged workers will be forced into quarantine, as they usually engage in work that is difficult to do from home and puts them at higher risk of infection, and, significantly, of coming into contact with infected persons.
While initially so promising, it seems that digital contact tracing is not the panacea that many thought it would be. Due to the specific nature of COVID-19, it is doubtful that such a technological intervention would be as effective as it was in the case of the SARS outbreak of 2002. Moreover, even if digital contact tracing were to work (where “working” means saving lives and saving livelihoods), it would still fall short of being fair. Specifically, it would result in an unfair composition of individuals who would be forced to quarantine. Evidence suggests that those who already find themselves in disadvantaged economic circumstances would be the hardest hit by such measures, further exacerbating existing social and economic inequality. Such concerns, therefore, do not necessarily turn on questions regarding the “technology itself” (whatever this might mean), but rather bring into sharper focus how existing societal conditions can be exacerbated by a myopic form of techno-solutionism.