Designing and prototyping a digital system for contact tracing leads us to see the value of a solution based on a mobile application through BLE.
Contact tracing can work, but it presents with challenges and limitations:
Current effectiveness is low
The effectiveness of currently deployed solutions, seems to be as low as 10.4%: it means only 10 infectious notified on 100 really infectious citizens, in our rough calculation. This is due to technical limitations in use of BLE and another key factor that follows.We deepened that here.
2. High effectiveness requires high adoption
It will be key to ensure the highest adoption rate possible among the population. Mass adoption of digital systems is never a matter of imperatives, but a matter of long, faceted, unpredictable and multi-disciplinary negotiations. This is not, at the moment, a topic in the news feeds and we believe a discussion should be opened. Leveraging on design techniques out of abundant literature, we advance a possible solution in our paper. More here.
3. Introduce a probabilistic approach
We strongly believe a useful tool shouldn’t only be able to track back a chain of contact, but give authorities the information to sort responses by priority. We introduced the concept of “risk index” as a probabilistic tool to do so. You can read more here.
Two main topic arise as next steps in this process. On one hand, to improve the effectiveness of the system, all developers should openly discuss technical challenges and adoption strategies. We have some ideas to share about this.
On the other hand, Apple and Google are working on some of the most impacting technical issues: they are the only one that can change the limitations imposed by their operating systems. We trust and admire their good will and ability, but we spotted a few potential weaknesses that could be mitigated by:
a) shard keys in broadcasted lists, while changing theIDs encryption flow to allow it;
b) scrambling the intensity of the radio signals (not toomuch, to preserve reliability)