The blog post by the tech lead of the TraceTogether app starts out very reasonable, but ends up with a vague non-sequitur ("additional API refinements will be needed").

https://blog.gds-gov.tech/automated-contact-tracing-is-not-a-coronavirus-panacea-57fb3ce61d98

A few thoughts on digital contact tracing, hoping to clear up some confusions. 1/n

Yes, digital contact tracing (or proximity tracing, which is what it really is) is no panacea. It should be explored to complement existing solutions, of course. Yes, existing public health infrastructures and expertise must be used, of course. 2/n

But the post ends by saying “consequently, additional API refinements will be needed”. It’s not clear which ones, and by whom? Presumably by Apple and Google? And why? Not arguments are given. This is really confusing, at a time when we should be very clear on the details. 3/n

The stakes are high, as contact tracing has two interesting properties. 1) It could make a real contribution to keeping COVID in check. 2) It has the potential to be an epochal privacy disaster. 4/n

This is why many groups have sprung up, arguing privacy-preserving proximity tracing is possible. One of them is #D3PT, and the arguments have been out in the open from day 1. It favors a decentralized model, the one now adopted by Google and Apple. https://github.com/DP-3T/documents  5/n

It’s great that such proposals are now debated, specifically on the grounds of privacy. This is how it should work. But I see two lines of arguments that are different. 1) That Apple and Google tell governments what to do, and 2) That much more data is needed to fight COVID. 6/n

I won’t go into any discussion about private tech companies. In liberal societies, there is an established independence of the private sector, and the government. I personally think Apple and Google have made the right decision in the choice of their API. 7/n

The argument that more data is needed to fight COVID needs to be unpacked. Yes, from an epidemiological perspective, more data generally helps you understand the situation better, which in turn helps containment. But it’s a slippery slope. More data also means less privacy. 8/n

The droplet route (through close proximity) seems to be the main route of transmission. This, and only this, is what proximity tracing is trying to capture. If you have been in close contact to someone contagious for long enough, there’s a chance you may have been infected. 9/n

A quarantine protocol would then help to minimize ongoing transmission. Now, airborne transmission is probably possible. Data from the flu increasingly tell us that this may be an important route as well. But this is *not* what proximity tracing is designed to capture. 10/n

Proximity tracing is designed to capture close contact transmission. No proximity protocol in this world will capture airborne transmission. For that, you’d need location data, not contact data. Airborne transmission does not require close contact. 11/n

The contact tracing API proposed by Apple and Google allow any app to do contact tracing in a privacy preserving way. It will capture close proximity contacts, allowing to stop transmission chains started by a potential droplet infection. Not more, not less. 12/n

I will be suspicious of any argument along the lines of “A & G stop telling us how we can respond to COVID”. If you have solid epidemiological arguments on how this particular API stops you from doing privacy-preserving proximity tracing, please put them on the table. 13/n