Apple and Google, which built APIs for apps to use to do contact tracing have switched plans – rather than rely on third party apps to implement the feature, both companies will roll out contact tracing apps to all phones over the next few weeks (except possibly slightly older phones which seem to rarely get any updates).

Supposedly you will be able to opt out.

Source: Apple, Google: Contact tracing to become standard smartphone feature

As I have described extensively on this blog, these apps will have a high false positive and false negative rate due to fundamental limitations of the technology. Additionally, these apps are unable to detect contacts across time. Example – someone with (unknown) Covid-19 sits outside at Starbucks, coughs on the table once, then leaves. You then come along and sit in that seat and your hands touch the table. The app cannot detect this simple scenario. (Or replace the table with a bus or train seat…) Similarly, someone sits on the opposite side of a window from you, tests positive, so you receive an alert – even though you are outside and separated by a window.

The systems require that you have a smart phone and the phone be turned on, with Bluetooth enabled. Many public schools, where students sit in close quarters for extended period of time, require that student phones be turned off during class hours, thus missing many potential Covid-19 contacts. Similarly, I have been told (by someone in the know but not independently verified) that up to 20% of confirmed cases in my state are of residents of institutions (such as prisons) where smart phones are prohibited.

Researchers have also noted that while the tracing app itself does not retain your location information, Android does, and forwards that information to Google for other uses.

There have been no controlled or randomized controlled trials of the technology – yet it will be used for a life safety critical function with large ramifications including putting you in quarantine for two weeks – or missing potential infectious contacts that could give a false sense of security.

Some researchers claim it can find some cases if as few as 15% of people use the app (.15 x .15 = 2.25% of cases are then detectable). That low figure must be compared to the problems of false positives and negatives that I note, above. Several countries that have tried using Bluetooth based contact tracing apps have found so few cases that they concluded it was not worth the trouble.

The system is less likely to produce useful results without human contact tracers and tracing. In my state, up 75% of contacts, as of last month, were untraceable. The states said this is because people had been in large group settings (like protests) where human-based contact tracing is impossible or because people are no longer cooperating with public health. A phone based app might detect potential contacts but will those using the app change their behavior?

The latter could be because public health has damaged their own credibility with inconsistent, contradictory, and frequently incoherent messaging.