Quincy Abeyie wrote:
I see now that the Norwegian app uses a combination of Bluetooth as well, but I always keep my Bluetooth off. As do most people, I'd imagine? Now I really can't see this working.
Nothing technical stopping Google and Apple releasing OTA operating system updates that turn it on for this specific purpose at all times. Given their setup they are also capable of achieving much higher rates of use. By contrast, TraceTogether only has 17% adoption in Singapore and it's doubtful whether it has been beneficial at all.
banduan wrote:
These apps will ask you to switch Bluetooth on so that won't necessarily be an issue. The issue is as Clrnc says not enough people install it, and even then it will still be a bit iffy.
The idea for integrating Google/Apple data is not for them to develop their own app, but for govts to check if suspect individuals lied about their travel history or break voluntary curfews. This had happened over here leading to clusters of imported cases (one imported case from Italy caused the death of 5).
This isn't quite correct.
The Google-Apple contact tracing proposal works roughly like this—phones generate anonymous IDs that are regularly changed, and exchange them via Bluetooth when in range. When someone is tested positive for COVID-19, a health practitioner facilitates the upload of their own ID-set, only for the dates they are deemed to have been contagious, and absent of other identifying information, to a central endpoint referred to as the "diagnosis server".
Every day, phones auto-download these new positive ID-sets and check them against their locally stored exchanged IDs. Then when a match is found, an alert and suggestions for appropriate action are displayed to the phone user, without being reported to any central endpoint.
There are a number of grey areas in the Google-Apple proposal—notably the interface with the state—but that's how they framed it. They are very clear that no contact information is stored on the "diagnosis server" and the "diagnosis server" is not alerted in the event of a match on someone's device.
This is by contrast to the TraceTogether model, in which a user's Bluetooth contact data is uploaded and related to an email address in a central repository, then matched server-side, with the administrator determining how people who are assessed to have had contact are alerted or otherwise dealt with. The main issue with this model for me is that there's a central database containing a vast amount of identity-linked contact data, that could be subject to data breaches or government abuse.
I don't trust Google or Apple in the slightest, but I think their control of smartphone operating systems, combined with the design they've put forward probably leads to the best combination of utility and privacy out of the options around. As to the question of whether they might do anything nasty with all that power, at least they're competing companies locked in a kind of prisoner's dilemma that could be extremely damaging for either. Individual states have far fewer incentives to respect anyone's rights.
The status quo where the government does contact tracing manually is also riddled with privacy issues and other difficulties.
