Posted / 26th May 2020
Oliver Smith, Strategy Director and Head of Ethics, Alpha Health
At Alpha Health, we have been watching with close interest the unfolding debate about Covid-19 contact tracing apps. It is clear that they can form an important part of strategies to help manage the next stages of the pandemic. In addition there is some great technological innovation being brought to bear. However, what has really caught my attention has been the ethical aspect of contact tracing, since these apps need to be trusted enough to be downloaded by at least 60-70% of a population if they are to effectively contribute to tackling Covid-19.
Ethical technology has been a pretty hot topic over the past couple of years. One common theme in ethical technology discussions has been the importance of moving beyond theory and principle, to actually apply ethics in practice; well we are having a crash-course in that right now.
I had thought about assessing various contact tracing app efforts against Alpha Health’s own ethical framework but then realised that that would be an unfair comparison. This is not to say that governments aren’t interested in some of the principles that we have committed to, such as user control over data, transparency, and accountability. Rather, if form should follow function, then it’s not fair to apply a framework designed for behavioural health to contact tracing by governments. Instead, I have been reflecting on what we have learnt so far in developing and deploying our own ethics strategy, and whether any are applicable to efforts to create contact tracing apps.
I talked a bit about our lessons learnt when I introduced our ethical framework, and also spoke about this at Intelligent Health UK 2020 – back in the day when speaking live at a conference was possible. Overall there are three lessons that I think are relevant to the contact tracing app discussion: the need to make trade-offs when principles or objectives are in conflict; the all or nothing nature of being ethical and trustworthy; and balancing perfect with good enough.
Let’s start with objectives. From the debates on contact tracing apps it seems that there are at least six objectives that governments are trying to achieve: advising those who have been in contact with an infected individual so that they can self-quarantine; accuracy of the proximity measurement that underpins the contact tracing; high user uptake; protecting privacy; speed of deployment; and identification of ‘hotspots’ for virus transmission. From what I have read it is the addition of the last objective that is causing the biggest headache for governments as it requires that some level of individual level data is shared at a central point in order to identify the ‘hotspots’. Not only is this centralised approach raising privacy issues, but because it conflicts with how Apple iOS and Android work, it potentially undermines the accuracy of proximity measurement amongst other issues (The Register has a good explanation in this rather strongly worded article). I know from our own work on balancing privacy and bias, that such trade-offs have to be identified and managed from the beginning, otherwise they come back to haunt you.
You can’t be 50% ethical
I’m not going to try to propose the best approach to meeting all of the above objectives, there are many others better qualified to do so. Rather, I’d like to see governments be explicit about the various objectives for their contact tracing apps and how they are making trade-offs between those objectives. This links to my second lesson, the all or nothing nature of being ethical and trustworthy; you can’t really be 50% ethical. We’ve been challenged by this in our work, where we initially put more emphasis on achieving some of our principles than on others. For instance, we put a lot of effort into privacy right from the start, but we’re playing catch-up a bit on publishing as much as we’d like about our work. In the context of contact tracing apps, I think that transparency is also the area where more effort is required. To take the NHS in England as an example, they have quite a lot of information on their contact tracing app available now, including the Data Protection Impact Assessment. It would be great to see the NHS go even further by publishing all of its objectives for the app, the trade-offs it made in reaching the final scope and design, and its justifications.
Perfect vs good enough
The final lesson – balancing the perfect with the good enough – probably most relates to speed of decision making and deployment in the case of contact tracing apps. The need is immediate and it’s easy to see why governments may feel that they don’t have time to wait for untested systems to be created, nor to consult in the normal way. As befits this lesson, there isn’t a perfect answer to this challenge. Countries like the UK and France are attempting to forge ahead with their apps; others, such as Spain, are taking a more circumspect approach to see if they can learn from the early movers. For instance, many early mover countries have seen pretty low uptake: 40% in Iceland, 25% in Singapore; and 20% in Norway (see this Financial Times article for more – it’s behind a paywall). In all cases, it’s important that contact tracing apps are seen within the wider context of a government’s efforts to tackle Covid-19. Going back to the first and second lessons, even a fast-moving government should be seeking to explain what it is doing, why, and how it fits into wider efforts. I would also highlight that moving quickly is not an excuse for forgetting good design principles: in the Indian State of Kerala, use of their contract tracing app is voluntary and the app includes an SOS button that people can press if they are in an emergency or need essential goods, which surely must help drive uptake.
Returning to my comment at the beginning of this blog, I don’t want to fall into a wide-eyed techno-solutionism and suggest that contact tracing apps will solve all of our problems. As with any digital technology, it’s important to marry the digital and the real world to get the best results. In this crisis we must deploy all of the tools that we have available, and contact tracing apps, when combined with more testing and other measures, have the potential to play an important role, but only if people trust them enough. There are many routes to building that trust but some themes common to all of them: identify your trade-offs from the beginning; be explicit about what you’re doing; and don’t let the perfect be the enemy of the good.