Turn it off and on again: lessons learned from the NHS contact tracing app 

The decision to delay the app’s launch is the right one, both to address the problems and to preserve future public trust and confidence in public health approaches that involve technology.

The writing has been on the wall for some time now. For the past month, with each successive public reference to the NHSX contact tracing app, the Government’s positive rhetoric has been dialled down. The app’s status has declined from central to the test and trace regime to an ancillary feature of it – a cherry on top of a ‘world-class’ cake.

Wednesday’s concession that the app’s development is no longer a priority, and that its launch isn’t expected until ‘the winter’, was the natural progression in a shift in political emphasis that once posited the app as a tool to get us out of lockdown, and now views it as part of the arsenal needed to respond to a second peak. Today it was announced that the bespoke centralised protocol for the app will be abandoned in favour of the decentralised API developed by Google and Apple.

The decision to delay the app’s launch is the right one. Its development has unearthed a range of problems: some of them technical (the limitations of Bluetooth technology assessing actual proximity between devices), some behavioural (the Isle of Wight trial finding that people prefer to receive notifications of potential exposure to the virus and instructions to self-isolate from a human, rather than an app), some political (the push-back against the decision to build a centralised system).

Many of the issues were foreseeable (see Exit through the App Store?, which we published in April). Pausing now to reconsider the best path is not only necessary to allow for readjustment to address problems, but critical to preserving future public trust and confidence in public health approaches that involve technology.

As Government, policymakers and NHSX take stock and decide how to move forward with the contact tracing app, it is prudent to ask what lessons can be learned from the experience to date.

1. A tool for public health should be led by public health
   Technology development should be led by an informed analysis of the problem, rather than by the availability of a solution. If digital contact tracing apps are to be a tool for protecting public health, the starting point must be articulation of the public health challenge. The nature of the challenge should inform what data public health experts need to respond to and control the disease. It is only once the public health purpose has been identified that any technology can be appropriately designed, and success metrics developed. A robust public health tool presents a stronger argument for the collection of sensitive information.
2. Open, honest, transparent = a better tool and more trust
   Exploring how technology could aid the fight against the pandemic is the right thing to do. But given the little political and public tolerance for failure of new government policies, it is critical that policymakers are upfront about the risks, challenges and pitfalls of experimental innovation. In the case of the contact tracing app, testing and evaluation should have been done in public, with transparency around the options under consideration (such as the Google/Apple protocol), and the mechanisms (such as the Ethics Oversight Board) through which challenge was being plumbed into the app development process. Public communication should have reflected the Government’s measured stance on other novel approaches to managing the virus, including the development of a vaccine (cautious optimism, lots of resources, patience and low expectations).
3. Politics shouldn’t set the pace
   The NHS contact tracing app was announced in early April in the midst of the PPE crisis, and confirmed for national rollout at the end of April. Announcements at daily press briefings triggered a range of questions and concerns – many of which the app development team were working on, but were not yet ready to address in public. Tying political capital to a single technological approach at an early stage and announcing a rapid timeline made it harder to explore other models or acknowledge the issues that emerged. Experimental approaches need time to test, refine, and – where necessary – reconsider.
4. Apps to shape social behaviour cannot be separated from policy decisions
   To support adoption and adherence, an app designed to inform users of risk of infection of a deadly virus, and to instruct users to stay home to avoid spreading that virus, must be underpinned by policy commitments. Separating the task of creating an effective app from the broader public health strategy, including financial support and employment rights, divorces the technical considerations from the questions of wider societal impact, including the effects on specific groups of people. With forms of ‘public health identity’ now under discussion it will be vital that the social, policy, legal and ethical questions are bound intrinsically into development and testing of any tech tool.
5. Stronger governance could enable more options
   By declining to adopt legal safeguards, scope restrictions and sunset clauses at an early stage, the Government boxed itself into a technically restrictive app. The choice of centralised technical architecture meant that data collection had to be minimal to balance out genuine concerns about data security and access. Stronger legal safeguards could have enabled more options to expand the collection of data to meet the needs identified by public health experts.
6. Oversight mechanisms need power to be effective
   The establishment of an Ethics Advisory Board to oversee the development of the app was a novel and laudable approach, but depriving that Board of any power to adjudicate the most fundamental questions surrounding the app – including its efficacy and its value – prevented meaningful input.
7. Be led by the social science, too
   Too often, the development of the app has felt like a conversation between epidemiologists (understanding the disease), technologists (creating the tool) and security experts (protecting the data). Excluded from that conversation have been the public health experts, social and behavioural scientists, psychologists, economists, lawyers, community organisers and civil society organisations. Interventions that aim to deliver public health outcomes, behaviour change and broad public participation need to be guided by people who understand people and society, as well as people who understand science and technology.

Developing experimental, advanced technology to respond to a global pandemic, at pace, is a novel historical endeavour – one that NHSX has approached with faster speed, greater investment and higher stakes than ever before. It must now be given the space to pause and reconsider its approach. If the problems encountered in the past can inform the development of technology in the future, they won’t have been incurred in vain.

Find out more:

-> ‘Towards a public health identity?’ – our work to explore whether the introduction of immunity certification could lead to a ‘public health identity’.