Governments across the world are hunting for an exit strategy. Denmark and France plan to test every citizen with coronavirus symptoms. Australia has hired an army of contact tracers. Germany has begun antibody testing across the population.
Pandemic resilience is a better description of what is needed. Exit implies that we won’t be in this situation again, but exiting without the right infrastructure could mean sliding back into lockdown. This argument is put forward in a paper, released on 27 April, by Harvard University’s Edmond J. Safra Centre for Ethics. (I am co-author of this paper, along with with other ethicists, public health experts and economists.) Returning to lockdown would be bad for business, bad for morale and bad for the health service.
A widespread regime of testing, tracing and supported isolation (TTSI) is the only way to keep an economy open and live with the virus until we get a vaccine. We need 100,000 tests a day, a target that health secretary Matt Hancock said the UK government would reach by May, although this is increasingly unlikely. We need everyone to wear masks. We need an effective system for contact tracing across the UK. But how can we achieve all of this?
This is where the much-vaunted NHS app comes in. The government has described the app as a central pillar of its plan to test and trace. It would enable people experiencing coronavirus symptoms to report their symptoms, alerting other users with whom they have been in contact. Those who receive alerts will be offered information and instructions on how to self-isolate.
A contact tracing app has obvious superficial appeal: an automated, instant, anonymous system to keep us safe. A one-way ticket out of lockdown. However, there are critical problems with the underlying technology and the legal regimes that would oversee it. No exit strategy should depend on digital contact tracing. What we need is an army of people.
The first problem is one of equality: technology tends to reinforce existing inequalities. The NHS app would be a benefit, a tool that helps people stay safe. People who do not have or cannot use smartphones would be denied that benefit, and often, those are exactly the people who most need protection: those in low-income jobs, often in densely populated or rural areas, many of whom are essential workers.
This is a familiar pattern. In the US, the city of Boston developed an app called Street Bump, which uses accelerometers in smartphones to detect potholes and automatically report them to local authorities. It quickly became clear that potholes were most effectively reported in areas which already had fewer potholes. In poorer neighbourhoods where roads were most in need of repair, residents often did not have smartphones or did not regularly drive with them. Places where access to technology is limited are precisely those that need effective contact tracing regimes.
The second issue is what role public health authorities should play. Apple and Google have very different answers to this question than the NHS. The two technology companies are developing a system that is almost completely decentralised. Each phone will have a rotating, anonymous ID that changes every 15 minutes. When two phones come within Bluetooth range, the will swap IDs and keep a digital record of the contact.
If someone tests positive, their IDs over the last 14 days are sent to every phone in the system to ask: have you seen this phone? Phones answer by checking their stored list of recent contacts. This all happens in the background, without users being notified or having to do anything. The IDs are also anonymous and not linked to a particular user. The system would effectively prevent governments from storing or analysing information on a central database.
The NHS system will be “anonymous” but not completely decentralised. Users will be sent alerts about close contacts who report symptoms or test positive. Whereas Apple and Google’s system would store data on each individual phone, the NHS plan would send anonymous data to government servers to anlayse geographic hotspots and near-misses. If this violates Apple and Google’s restrictions, the NHS app will likely be prevented from accessing a phone’s Bluetooth. This would make it close to useless, as the app would only work if a phone’s screen is turned on and no other app is being used at the same time. Apple and Google want to protect privacy by tying the government’s hands and throwing away the key; the government wants to keep the key if they promise not to use it.
The most important problem is that the app probably won’t work very well. Firstly, not enough people are likely to use it. At least 60 percent of the population need to use the app. They all need smart phones, to download and register on it, and ensure Bluetooth is enabled. About 95 percent of people in the UK have a smartphone. About half of those have Bluetooth enabled. If a third either can’t or don’t want to download the app, that is a lot less than 60 percent. For comparison, 12 percent of the population use Singapore’s contact tracing app.
The app will also generate a huge number of false positives. This is partly because the NHS plans to use an algorithm to estimate what combination of symptoms accurately predicts that someone has coronavirus. Given how little we know about the virus, and how often the most sophisticated machine learning systems get things wrong, people are bound to get alerts about contact with people who do not have the virus. Bluetooth is also a blunt tool. The app will send alerts about people on the other side of plasterboard or the other side of the street.
An app is still worth trying. It needs pre-emptive legislation that strictly limits the use of data, grants powers of oversight to the Information Commissioners Office, and prevents an over-eager project manager at Google from using the technology for effective, targeted advertising in a few years’ time. It would be the first step in a long, slow process of learning how to effectively use technology in a pandemic.
But the app must aim to complement human contact tracers, not replace them. We’ve done human contact tracing for decades. It works. People trust a phone call from an NHS official who explains why they are calling and explains what to do. People will not trust an app that often gets it wrong and which they cannot engage with or question. And trust is key to an effective regime of testing, tracing and supported isolation.
Vague promises about catch-all apps should not distract us from the hard, basic work that needs to be done. Much of this is about scaling up human contact tracing, probably hiring and training 20,000 more contact tracers.
Here’s a thought. Why not train some of the 750,000 NHS volunteers to do this? That would do more than any app to enable us to leave lockdown and keep our economy open.
Josh Simons is a fellow at the Harvard Centre for Ethics and the Berkman Klein Centre for Internet and Society