The company building Ireland's contact-tracing app says the original centralised model, which was abandoned in favour of a design created by Apple and Google, would have been doomed primarily for data privacy reasons.
In a blogpost today, Nearform is also warning that Bluetooth is not “perfect” for the purpose of contact-tracing and that it “came up against a wall” several times when designing a technical implementation. However, despite technical “limitations”, the Waterford firm says that it is “available and effective”.
It also says that it has been working with “other governments” on similar apps.
The contact-tracing app, which is expected to be released within days, detects proximity between smartphones that use it. A person who tests positive for Covid-19 can register that finding within the app, which then starts a chain reaction to send alerts to other individuals, whose phones also use the app and who were in proximity to the person who tested positive within the last 14 days.
Apple and Google have built the technology on which the Irish app, as well as many other European contact-tracing apps, will work. It replaced a “centralised” app design, originally favoured because it gave governments and health agencies more control and visibility of how the data from the phones could be used and shared.
“The centralised model was favoured because it more closely aligns with how manual contact tracing works,” Nearform says in its blog. “Every country based their apps on the use of centralised data — as we did with the early iterations of our apps. However, while various countries pushed to release those early apps, we quickly determined that the centralised model simply wouldn’t work for governments and health authorities. Regardless of how the apps themselves are built or how restricted their use of the data is, privacy concerns remain that could stop people from using the app.”
Researchers from Trinity College Dublin have cast doubt on the effectiveness of Bluetooth within contact-tracing apps to accurately pick up other phones’ proximity. Research from TCD Professor Doug Leith and Stephen Farrell shows mixed results from app-enabled phones on buses and indoors.
“The reality is that all these smartphones were never designed to facilitate contact tracing. Bluetooth technology was not created to enable this kind of communication between devices,” says the Nearform blog. “Yet, it’s what we have. In the course of designing, building and testing contact tracing apps with our collaborative partners, we have come to understand these technological limitations very well. And we’ve worked tirelessly to devise solutions and workarounds.”
“For example, early on we came up against a wall when trying to ensure the Bluetooth signals would work even if the phone was inactive in someone’s pocket, so we reached out to Apple to see how we could work together to find a way around the restrictions.
“Other problems surfaced with letting Apple and Android devices communicate seamlessly with each other — which became part of the motivation for Apple and Google to come together to devise a protocol that allows the efficient use of Bluetooth for proximity detection. So the technology may not be perfect, but it is effective and available.”
It also explained why it abandoned its first design attempt at the app, a “centralised” model.
“Some aspects of the centralised model could be seen as favourable,” it says. “For example, such apps could provide public health officials and epidemiologists with a view of everybody an infected person had been in contact with, where and when. They would then have a far better idea of who was infecting whom, how many people were being exposed and who any super-spreaders were. That data could be graphed and mapped to see plumes and spreads, all thanks to the app. However, that would require identifying individuals (even if not using their real names) and collecting their data in a central source. And that would mean having specific information about an individual that could, with work, be used to profile and identify them.
“Luckily, when we highlighted these issues early on, the governments we were working with agreed that they could not be resolved and that another solution had to be found. We were already in touch with Apple when they and Google announced their joint API effort, so we were able to be one of the first to use that technology.”