A team in Switzerland today became the world’s first to introduce a contact-tracing app built on a protocol jointly designed by Apple and Google.
The new SwissCovid app is now being tested by Swiss soldiers, hospital workers, and government employees before being rolled out to the wider population.
Contact-tracing apps work by automatically recording other nearby devices that have the same app installed. Users who later test positive for COVID-19 can use the app to automatically send alerts to those other people they were in contact with, advising them to self-isolate for 14 days or to get tested themselves.
Technologists see the new app as a breakthrough that may help Switzerland prevent a second wave of infections as it reopens its economy—while at the same time preserving citizens’ privacy. A Latvian team also says it will have an app out based on the Apple-Google framework within days. And these are just the first of some 22 nations and a number of U.S. states that have requested access to the Apple-Google framework.
But, unfortunately, there’s a hidden flaw in these contact-tracing apps—in fact, in all contact-tracing apps—that means they are unlikely to live up to their promise: To work effectively, they require adoption rates of well over half the population, which few mobile apps, even the most viral and fun, ever achieve.
The Apple-Google protocol is popular with civil liberties groups that fret about the Big Brother implications of governments collecting in a central database citizens’ location information and records of everyone they’ve met. The Apple-Google solution works without holding this information centrally.
The Apple-Google protocol also has some built-in technical advantages: The two technology giants, which make the operating systems that underpin most of the world’s smartphones, restrict mobile handsets from sharing certain kinds of data, like location information, with third-party developers unless the user keeps the app open. As a result, governments building their own contact-tracing apps may find their software won’t work unless users keep it open on their phones, which can more rapidly deplete battery power.
A number of European governments have bristled at those restrictions, arguing that they ought to at least have the option to track users’ locations and store data centrally. They also want more insight into exactly how the Apple-Google app works.
“The use of digital technologies must be designed in such a way that we, as democratically elected governments, evaluate it and judge it acceptable to our citizens and in accordance with our European values,” the digital affairs ministers from Germany, France, Italy, Spain, and Portugal wrote in a joint letter published in European newspapers Tuesday.
Europe is counting on contact-tracing apps from different countries being interoperable as a way to help revive the region’s devastated tourism sector. That way, a vacationer from Germany who travels to Spain but later tests positive will be able to automatically alert people she came into contact with while on holiday.
Yet all these debates may be moot because they largely ignore what the epidemiological models indicate: The apps make a big difference only if a large percentage of the population uses them.
A widely cited study by a team at Oxford University found that apps must be adopted by 56% of people to have a dramatic impact on coronavirus transmission rates. Once people under the age of 10 and over the age of 80 are discounted—since it is assumed they won’t have access to smartphones—that figure translates into about 80% of all smartphone users in most places.
Another model—from a joint British-U.S. team—came to a similar conclusion, with some reduction in transmission occurring with 53% coverage, but really dramatic declines in transmission rates occurring only with adoption higher than 80%.
The number of apps that have ever achieved these kinds of penetration rates can be counted on one hand: Messaging app WhatsApp is installed in about 80% of smartphones in its top markets. WeChat is used by about 79% of Chinese smartphone users. And rival Baidu’s Tieba and Tencent’s QQ messaging apps are within spitting distance. But it has taken these apps years—more than a decade in the case of WhatsApp—to hit those kinds of numbers.
The adoption rate “needs to be almost improbably high to really capture all of the contacts that might be relevant,” Allison Black, a genetic epidemiologist at the University of Washington at Seattle, told Science magazine.
Technologist Ashkan Soltani recently coauthored an essay with Ryan Calo and Carl Bergstrom, professors of law and biology, respectively, at the University of Washington, in which they concluded that “even if we had 100% installation of these apps (which is extremely unlikely without mandatory policies in place), we would still only see a fraction of the total exposure events,” or cases where people are exposed to the virus.
A survey of U.K. citizens by professors at London’s Cass Business School did find that if configured in a way in which only the health service had access to the data for a period of 14 days, no location data was transmitted, and use of the app gave users priority in getting tested for the coronavirus, as much as 73% of the population may adopt the app. But that is not the app configuration the U.K. is currently testing on the Isle of Wight, a small island off England’s south coast. That configuration, the professors found, was likely to achieve at most 51% adoption.
Meanwhile, the highest penetration achieved by a contact-tracing app so far is Iceland’s official Rakning C-19 app, which has been downloaded by about 38% of the tiny nation’s 364,000 people. And, sure enough, the local official charged with coordinating contact tracing told M.I.T. Technology Review that the app “wasn’t a game changer for us.”
Singapore’s TraceTogether app, which was one of the first contact-tracing apps to premiere in mid-March, has been downloaded by only about 25% of the population there—even though the country’s Prime Minister has asked all residents to use it and the country is known for normally high levels of compliance with government recommendations and rules. “Adoption has always been one of TraceTogether’s weaknesses,” the app development team, part of Singapore’s technology agency GovTech, said in an online post.
Part of the issue may be that Singapore and most other governments have not made the use of the app mandatory. One of the few exceptions is India, where the government has required citizens to download the app and has made its use a condition for travel on trains and for admission to offices, with criminal penalties for noncompliance. But, despite the requirement, less than 10% of India’s 1.3 billion people have downloaded the app so far.
In many European countries, requiring citizens to download contact-tracing apps may be illegal. “To make use mandatory on current evidence would not only undermine public trust and confidence, but likely fall foul of human rights standards,” the Ada Lovelace Institute, a British think tank that researches issues related to artificial intelligence, data, and society, wrote in an April report about the use of technology to help fight the pandemic.
Assuming the app’s use can’t be mandated, privacy does become an important issue. If people are worried about the government having access to their location data and insights into their contacts, they will be less likely to download and use the app.
The British government’s tests of its contact-tracing app, which does collect data centrally, have been stymied by the fact that while it wanted to restrict its trial to just the 148,000 people who live on the Isle of Wight, many mainlanders have also downloaded the app.
Although it is facing a self-imposed June deadline to roll out the app to the wider population, the government is reportedly considering abandoning the whole effort in favor of a backup app being built on the Apple-Google protocol.