Privacy-friendly and effective contact-tracing

Privacy-friendly and effective contact-tracing

a proposal in a comic-book format for a privacy friendly contact-tracing that is effective to boot. Go to that link and read the entire post carefully. It has been created with the help of privacy experts.

Apple and Google announced a joint-effort to provide APIs that will enable developers to make privacy-friendly contact-tracing apps.

There are many arguments against such a contact-tracing technology (no technology is perfect):

  • it is not accurate (it is better than nothing – even a certain percentage of false-positives might be acceptable in the absence of anything)

  • if a person self-reports, she may be penalized by the authorities (that is a fault of the authorities, not of the technology). In any case, any behavior or outcome that discourages self-reporting only exacerbates the problem by hiding it and making it into a worse time-bomb

  • the person will lose her privacy because her neighbors will find out and ostracize her (nope, the proposed technology makes sure to protect the privacy of all involved individuals)

And many more arguments, many of them justifiable. But not doing anything is not justifiable at all. These are unprecedented times. The choice is to have a lockdown and kill the economy, and mainly the poor people and service-sector with it. Or to identify the newly-symptomatic and asymptomatic potentials and provide them with immediate care and quarantine so that infection and spread are nipped in the bud.

There were similar arguments against masks (they are not good enough, droplets can get through anyway) but again, something is (almost) always better than nothing.

Public Service Announcement: Practice safe treks, wear a mask when you go out.


What would that comic-book have?
What type of comics?

1 Like

Did you see this:

1 Like

So, is there any voluntary action anywhere? Does the model depend on Alice voluntarily sending all the sent-received-messages to the hospital or does this happen automatically? This seems to be the crucial step. Who asserts that Alice is infected?

1 Like

first, in my view, this is a diagram Nicky should have also included (answers to your questions after)

             Alice                   Bob                      
   β”‚day 0 β”‚  gets infected                                    
   β”‚day 1 β”‚                                                   
   β”‚day 2 β”‚                                                   
   β”‚day 3 β”‚  becomes contagious  - come in contact w Alice    
   β”‚      β”‚                      - gets infected              
   β”‚day 4 β”‚                                                   
   β”‚      β”‚  - shows symptoms   - phone automatically         
   β”‚      β”‚  - gets tested      determines he came in         
   β”‚day 5 β”‚  - confirmed pos.   contact with someone          
   β”‚      β”‚  - uploads data     who was infected              
   β”‚      β”‚                     - self-quarantines            
   β”‚day 6 β”‚                     X chain is broken             
   β”‚day 7 β”‚                                                   
   β”‚      β”‚                     - shows symptoms              
   β”‚day 8 β”‚                     - gets tested                 
   β”‚      β”‚                     - confirmed neg.              
   β”‚      β”‚                     - no need to upload data      

Now, to your questions:

  • So, is there any voluntary action anywhere?
  1. yes, on the part of Alice and Bob, to install that app allow it to do its job, which is to send out the random codes every 5 mins, and to poll the central db for matching codes (this becomes useful in step four below
  2. yes, on the part of Alice to get tested as soon as she shows symptoms
  3. yes, on the part of Alice to upload her data anonymously to the central db
  4. yes, on the part of Alice and Bob to let their app check the central db for matching codes
  • Does the model depend on Alice voluntarily sending all the sent-received-messages to the hospital or does this happen automatically?
  1. Alice has upload her data to the central db once she knows she is infected, which happens once she is tested, which happens once she shows symptoms. More on this below.
  • This seems to be the crucial step. Who asserts that Alice is infected?
  1. Alice gets tested as soon as she shows symptoms. More on this below.

More: All of the above assumes a number of things –

  • cooperation: download the app and let it do its thing
  • education: everyone knows about the app, how to download and install it properly and let it do its thing, what the symptoms, and who to call soon as symptoms are visible
  • availability of testing: naturally, this is a prerequisite

In all of this, there are some very tough assumptions (my comments are in the context of India; other places will be different):

  1. Testing works. However the recommended test is PCR, which seems to have a chancy effectiveness. In the absence of other tests, the choice seems to be some test or none.

  2. Testing is possible. This seems to depend somewhat on chance. Some people seem to be able to access testing within a day or two of either asking or being told to get tested. Others are turned away. The overall test rate is one of the lowest in the world, and there aren’t reliable figures available by town population, although in fact the majority of the authorised test centres are in towns.

  3. Enforced isolation: people identified as needing testing are not allowed home quarantine, and there are uneven reports about the quality of enforced institutionalised isolation. From my immediate home surroundings, a couple of people taken away under suspicion have been returned after testing negative, but their building premises were heavily quarantined during that time. One young child was allowed home quarantine, and is now declared clear. Last I heard, six people from this area, about half a km of road, have been identified (not named officially, but gossip flares) as suspects.

  4. People don’t have phones. Smartphones are estimated to be around a third, feature phones a little more. Shared phones are common among lower income groups. The value of a contact tracing app is questionable, especially since many very low cost feature phone models don’t support Bluetooth, without which proximity sensing using a phone is most probably not possible at all. The usefulness of Bluetooth for determining opportunistic proximity is also in question.

  5. Informed consent and trust: a number of exceedingly questionable promises accompany the β€˜marketing’ of the app. The present downloads seem to be a question of faith, which is, of course, a form of trust. But the second part of having the app is trust in the process that follows an alert about a possible infection (a β€˜contact’). This needs the person to approach the health authorities on her own. And that needs the health authorities to respond sympathetically, or at least empathetically. But the health authorities are under huge pressure, and the overall atmosphere used to attempt to create a lockdown environment has been fear, and these two put together aren’t comforting or encouraging.

It is hard to imagine that something is better than nothing applies to trusting phone app technology in our circumstances.


6 distribution of essentials. Again a herculean task in India. And the primary reason for the wild spread.


There are two voluntary actions, one by the person who has chosen to send the data, and the hospital staff who confirms that the person is corona positive.

well, as I mentioned in my update to the original post (in response to your query), the voluntary actions are required on part of the users. They can choose to cooperate or not, hence voluntary. The hospital staff is actually doing their job, so I wouldn’t call that strictly voluntary.

1 Like