Tuesday, January 4, 2022

Exposure Notification Update, January 2022

Gerrit De Vynck and Cat Zakrzewski:

The tech giants managed to build and launch the “exposure notification” framework in months, a previously unheard-of level of collaboration for the rivals.

[…]

But nearly two years later, as the omicron variant sweeps across the United States, adoption of the system is still far behind what its creators and proponents envisioned. More than 20 states don’t use it at all, including large states like Florida and Texas that have reported millions of cases and tens of thousands of deaths. Even in states where millions have activated the notifications, only a fraction of people who test positive for the virus report it to the Apple and Google system. California’s system, for example, has been activated on more than 15 million devices, but only about 3 percent of the nearly 3.9 million cases reported since launch were logged in the system.

[…]

The apps are “not really talked about as part of the mitigation strategy,” said Kameka Dempsey, a co-chair of the Covid-19 Technology Task Force, a group of prominent technologists and venture capitalists working on efforts to combat the virus.

Anecdotally, there still seems to be little awareness of these apps outside of tech circles. And without lots of people using them they’re not very helpful.

Via Nick Heer:

Similarly, this Post report paints a bleak picture of the framework’s poor adoption in the U.S., which Karen L. Howard of the GAO blamed partly on a lack of privacy protections in U.S. law. But I have had a hard time finding similar information about other countries’ responses.

The Apple–Google exposure notification framework is the system adopted by national COVID apps in Germany, Ireland, and Switzerland, where there are much stricter data privacy laws than in the U.S., which has perhaps played a role in driving higher adoption rates. Even still, finding evidence that this framework has played a meaningful difference in this pandemic is hard to come by. Irish authorities were understandably proud of their country’s rapid adoption rate, but a report earlier this year found that only a quarter of cases in Ireland were registered in the app.

Previously:

Update (2022-02-11): Anthea Katelaris:

Now, almost two years after its launch in April 2020, we publish in The Lancet Public Health our evaluation of the app’s effectiveness and usefulness in New South Wales.

[…]

A total of 22% of cases were using the app. Most (61%) contacts the app registered as “close contacts” turned out not to be epidemiologically linked to a case. The app detected only 15% of true close contacts identified by conventional contact tracing.

In total, COVIDSafe detected only 17 additional true close contacts in NSW during the six-month evaluation period.

It caused substantial additional work for contact tracers and overall, did not make a meaningful contribution to the COVID response in NSW.

5 Comments RSS · Twitter

Anecdotally, my daughter’s friend got a contact notification, which led him to be tested this week, which gave him another contact notification (presumably some of the people waiting to be tested were positive and told the app). But yeah, he’s in tech.

It’s worth noting that in England and Wales, at least, there’s been at least one peer-reviewed paper that estimates that exposure notifications via the NHS COVID app (which uses the Apple/Google framework) prevented between ~300k and 600k cases, and between 4200 and 8700 deaths, in its first three months of being available. That’s not nothing! https://www.bbc.co.uk/news/technology-57102664

This is a conundrum. Tech companies invested a lot of effort into ensuring that their frameworks preserve their users' privacy, potentially at the cost of effectiveness. But people still didn't install the contact tracing apps, often due to not understanding how they worked, and assuming that they would violate their privacy. So in the end, countries that just ignored their citizens' privacy ended up shipping these apps sooner, and also had both better working apps, and higher install rates.

Contact tracing apps are not the only area where technology suddenly allows authoritarian countries to outcompete free societies.

The design is also based on the initial assumption that the virus spreads through droplets. Aerosol transmission happens indoors within a room or venue, not just face to face. The framework is useless for super spreading events that turned out to be so common.

Anecdotally, most people who were successfully warned by the app in Germany already knew to get tested, because if you spend 10 minutes face to face, you are probably not strangers and we’re warned personally.

For these reasons, Germany switched focus to a (often) mandatory check-in system where you just scan a QR code when you enter a venue or event. Much simpler to understand and to trust.

Andrew Abernathy

Regarding the Feb 11 update on Australia's COVIDSafe app: it looks to me like it doesn't use the GAEN framework. At least, I can't readily find anything which claims that it does, and apparently it requires that you keep it "active" (they warn against closing the app, and say that your phone might close the app on its own), which suggests it's not using the framework.

That's not to make any claims about how useful the GAEN framework has been — it doesn't seem to be any kind of rousing success. Certainly the limited (and slow) uptake has hampered it, and it may well have the same kind of sensitivity issues (falsely identifying people in adjacent apartments or restaurants as contacts, for instance). Still, some of the issues with the COVIDSafe app seem like they would be reduced or bypassed had they gone with the GAEN framework. No issues with closing the app or the app being ejected due to memory pressures, for instance; and the startup costs would have been reduced had they used the framework instead of creating their own (though perhaps the ongoing costs would be similar).

Leave a Comment