What led the UK Government to switch to a decentralised COVID-19 contact-tracing app?

Question

It was reported in late March that the NHS had been developing a contact-tracing app with an Oxford-based team since mid-January, encouraged by the success of a similar app in China.

The app, developed by NHSX – the health service’s digital transformation arm – with academic and industry partners is in advanced stages of evaluation and is weeks away from being ready to be deployed, the Guardian understands.

Around 60% of the adult population would need to sign up and engage with the app by registering their symptoms or positive test results for it to be effective. Their proximity to other users would be logged, and they would follow advice given in alerts to self-isolate – even in cases where they were not aware of having been in contact with someone infected.

It seems that, although initially the NHS said that it was "working with Apple and Google on their welcome support for tracing apps around the world", it decided ultimately to pursue a centralised approach, holding user data on a government/NHS database, rather than use the API developed by Google and Apple, which necessitates a decentralised approach.

The UK is once again bucking the trend in its quest to conquer the coronavirus as it opts not to use the framework created by Apple and Google for its NHS COVID-19 contact-tracing app.

Instead, NHSX — the digital arm of the nation's health organisation — is creating a centralised app that strays from the Apple-Google model.

NHS officials hope their app will provide better insight into the spread of COVID-19 and help flatten the curve of coronavirus infections.

According to Professor Chris Fraser, one of the epidemiologists advising NHSX, speaking to the BBC:

"One of the advantages is that it's easier to audit the system and adapt it more quickly as scientific evidence accumulates,"

"The principal aim is to give notifications to people who are most at risk of having got infected, and not to people who are much lower risk.

"It's probably easier to do that with a centralised system."

However, in the last week, the UK has now decided to instead pursue the Google/Apple decentralised approach after all:

Britain on Thursday said it would switch to Apple and Google technology for its test-and-trace app, ditching its current system in a U-turn for the troubled programme.

The government has been forced to abandon a centralised coronavirus contact-tracing app after spending three months and millions of pounds on technology that experts had repeatedly warned would not work.

In an embarrassing U-turn, Matt Hancock said the NHS would switch to an alternative designed by the US tech companies Apple and Google, which is months away from being ready.

What reasons have the Government/NHS given for scrapping the existing centralised application, and pursuing this new approach?

Answer 1

The Government says:

Following rigorous field testing and a trial on the Isle of Wight, we have identified challenges with both our app and the Google/Apple framework.

This is a problem that many countries around the world, like Singapore, are facing and in many cases only discovering them after whole population roll-out.

...

Through the systematic testing, a number of technical challenges were identified – including the reliability of detecting contacts on specific operating systems – which cannot be resolved in isolation with the app in its current form.

What they mean is they made two apps, one based on their preferred centralised approach and the other the plan B based on the Apple-Google API, due to fears the former wouldn't work.

A BBC article about the change in plan says:

The NHS has been testing both systems against each other, over the course of the past month.

The centralised version trialled on the Isle of Wight worked well at assessing the distance between two users, but was poor at recognising Apple's iPhones.

Specifically, the software registered about 75% of nearby Android handsets but only 4% of iPhones.

By contrast, the Apple-Google model logged 99% of both Android mobiles and iPhones. But its distance calculations were weaker.

In some instances, it could not differentiate between a phone in a user's pocket 1m (3.3ft) away and a phone in a user's hand 3m (9.8ft) away.

Experiments in Ireland have flagged a similar issue.

At the Downing Street briefing, Health Secretary Matt Hancock suggested the original plan might have worked had it not been for Apple's restrictions on third-party apps' use of Bluetooth.

I understand the UK Gov app required always-on Bluetooth for the precision. On Apple this only works if the app is in the foreground and the screen is kept on too. UK Gov wanted Apple to make an exception - Apple refused. I will try to find a source.

Also note some confusion (I alluded to above):

Baroness Harding added: "What we've done in really rigorously testing both our own Covid-19 app and the Google-Apple version is demonstrate that none of them are working sufficiently well enough to be actually reliable to determine whether any of us should self-isolate for two weeks [and] that's true across the world."

In response, Google noted that it and Apple had developed an application programming interface - a set of functions and procedures for others to build on - rather than a fully-fledged app.

The BBC's technology correspondent Rory Cellan-Jones wrote a good article about what went wrong.

The FT has a wider, European perspective.