Re: Privacy-protecting contact tracer for COVID-19?

I hope my comments won’t be taken out of context here, but a concern that has been growing for me is that we are looking at this problem primarily through the lens of technology and missing the extremely important social and emotional dimensions.  I believe this is a very dangerous mistake.

I understand the desire to solve a truly difficult technical challenge (and it is clear that all have put real thought into it), but I believe these other dimensions must be of equal, or perhaps even greater, weight of that than the technical problem.

What happens when someone(s) use a system as outline here, to go out to find and remove the ‘problem’ (a la the individual in Missouri who was going to blow up a hospital)?  Will the people who are  now spitting on police officers use this information to go and beat up the infected? Do people who have have/had the virus (or any other attribute that the system can track) become social pariahs?  Would parents move their children away from all the ‘dots’ on the map?  

If we can’t solve the social and behavioral problems that this kind of capability exposes, may be this technical problem should be left alone.  Are we inadvertently creating a problem bigger than the problem we are trying to solve?

I can’t get Christopher Allens recent email about the Dutch Archive out of my mind, what was started as a great good was taken and twisted to great evil.  In our current global situation where it appears democracy is under threat, and populism and nationalism is on the rise, not factoring in societies baser characteristics whether into any solution is foolhardy at best, and gross negligence at worst.

With greatest respect to you all,

Michael Shea.



> On Mar 28, 2020, at 12:15 PM, Ouri Poupko <ouri.poupko@weizmann.ac.il> wrote:
> 
> Here is a third approach:
>  
> 1-      Everyone's path is recorded locally on their smartphone
> 2-      A public bulletin board (public ledger?) publishes the tracks of infected people in the following manner:
> a.       Each track is divided into segments
> b.       Each segment is represented as a 4d ball – just center and radius
> c.       Each segment is signed with a different temporal DID, derived from the master DID of the patient
> d.       Each ball is enlarged and offseted by a random displacement (as they do in differential privacy)
> 3-      When my smartphone finds that my path intersects with one of the segments, it requests a peer-to-peer anonymous communication with the owner of the segment.
> 4-      In the peer-to-peer communication both sides break their segment into smaller segments as in step 2 (sub step c is redundant) and communicate the scrambled sub-segments with each other. They do this iteratively for any overlapping sub-segments, until they get an intersecting point (2m radius) between their true paths.
>  
> Ouri.