I recently wrote an article stating that “Contact Tracing Apps Are Not Going To Save us“. Some people pointed out that while maybe not a silver bullet, they might still help in fighting a pandemic. I concede there are indeed scenarios where contact tracing apps could be helpful.
Buses and trains are often poorly ventilated and people sit comparably close to each other. Spending a lot of time in public transport is therefore a significant risk. I assume it should be rather easy to detect whether people travel together in a bus, so apps could be quite beneficial here.
Flagging at-risk individuals
Hbkk pointed this out in his comment to my last post: We could just accept that apps produce lots of false negatives and positives, but use them anyway to better target testing. In that sense, the app provides some Bayesian evidence that alters the pre-test probability and allows for a better selection of whom to test. The argument goes like this: People need not worry about false positives if you simply test them right away. If you have limited testing capacity, this would help selecting those most likely at risk of infection. This should also prove helpful for contact tracing, because you can identify where people otherwise would not have known about a possible infection.
People also need not worry too much about false negatives if we don’t treat the app as evidence per se. We simply need to communicate that likely tons of infections will be missed and therefore nobody should put any trust in the fact that their app doesn’t currently show any alerts. Imagine the app only showed 2 out of 10 infections. This means we are still better off in a scenario where 2 infections are flagged vs. one in which we don’t know about any of them. A problem only arises if people believed we could detect 9 out of 10 infections and therefore thought they were safe if their app didn’t show anything.
My personal update since my last article is this: if used correctly, apps can make a positive contribution. I am somewhat unsure about the balance of benefits and harms (think e.g. opportunity costs and privacy issues). Next to the financial costs of such an app, I think identifying the best use cases and communicating them correctly is crucial in making such an app successful.