The UK is poised to announce it has abandoned its attempt to build a centralised coronavirus contact-tracing app and will instead switch to the model preferred by the technology firms Apple and Google.
The embarrassing U-turn comes after British officials concluded it was technically impossible to create an effective app that did not conform to the Google and Apple model, but that a straight switch to their model would not solve all the problems.
They said there were also significant problems with the Googleâ€“Apple approach and they hoped to find a â€œthird wayâ€ by working with the two firms to create a tracing app that would work.
The idea behind the app was to track anybody that a person with coronavirus symptoms came into close contact with, using the Bluetooth connectivity on their smartphones.
But the ditched app designed by the government only recognised 4% of Apple phones and 75% of Google Android phones. While the alternative Google Apple model identified 99% of phones, it could not accurately measure the distance between two phones, making it unclear when somebody would fall in or out of the 2-metre physical distancing range.
Officials would not immediately disclose how much money had been spent on developing the failed app. No launch date for the new app has been announced, but officials are hoping that the new approach will allow a contact-tracing app based on the Google Apple model to be launched in the autumn or winter.
The UK had hoped to build an app that kept an anonymised central database of anybody an infected person had come into contact with. But Apple and Google refused to endorse that approach and said they would only redesign their operating systems for governments who used a decentralised approach, where no data was held in a single official database.
British developers spent weeks trying to find a way of making their approach work, but it failed because once Apple phones had â€œgone to sleepâ€ because they were inactive, they stopped communicating via Bluetooth.
Matt Hancock, the health secretary, had promised the app would be ready by mid-May and reports in the Guardian that the government was considering switching that month were vehemently denied. But on Wednesday the digital minister Lord Bethell admitted it would not be ready until winter, in a sign of how little practical progress had been made.
The decision to switch to the Google-Apple model will have wide-ranging ramifications for the governmentâ€™s contact-tracing programme. The Californian tech companiesâ€™ unbending rules greatly limit the amount of data the NHS has access to, preventing the health service from using the app to increase its understanding of the spread of coronavirus around the UK.
It will also force a shift in focus: the current contact-tracing app operates on a two-tier system, where users are notified if they have been in contact with someone with Covid-19 symptoms, and then warned again if that person has a positive test. That choice was made in part because the government wanted to minimise the delay between onset of symptoms and sending out a notification.
Without a centralised gatekeeper to prevent malicious users from falsely claiming they have symptoms, the NHS will be forced to only allow users who have a positive test result to send out exposure notifications â€“ a problem cited by GCHQ as one of the main reasons to use a centralised model in the first place.
The switch will also do little to solve one of the most pressing problems, which has plagued not only the Isle of Wight tests, but also other contact-tracing apps around the globe: the fact that the Bluetooth signal the app relies on to work is a very unreliable way of estimating distance.
This problem, which is unrelated to the question of whether the app uses a decentralised or centralised approach to data, means two phones kept in pockets on a crowded train can â€œthinkâ€ they are very far away from each other despite being within 2 metres, while two phones in active use outdoors can â€œthinkâ€ they are very close, even if they are in fact well out of the danger zone.