Is Contact Tracing Without Personal Data Collection Possible?

[Ryan_m_b]

Thought the members here might like this explanation on how contact tracing apps can be made that collect no personal data. The design is quite clever at preventing situations where centralised databases collect information like who you are, where you’ve been, who you’ve met etc.

 

[russ_watters]

Thought the members here might like this explanation on how contact tracing apps can be made that collect no personal data. The design is quite clever at preventing situations where centralised databases collect information like who you are, where you’ve been, who you’ve met etc.

While it's clever and nice, I guess, that these apps can do contact tracing while maintaining complete privacy, even at their theoretical best they are inferior to more invasive methods. And worse, most discussions, including this video, completely gloss-over the fundamental flaw, which is that they are voluntary at both of the key steps: downloading and reporting positive status*.

The video says you need about 60% participation for these apps to work, but makes no mention of the actual participation rates in countries that have tried them. Singapore, for example, was reporting a 20% download rate 3 weeks ago. India has done even worse:
https://www.reuters.com/article/us-...id-19-show-modest-early-results-idUSKCN2232A0

But there's no word on the certainly much lower rate of reporting positive cases to the app.

So we're headed down a path that we can be reasonably certain won't work. Awesome.

*3rd potential key step: obedience.

 

[Vanadium 50]

I see a few problems:

1. A infects B who infects C. C won't know he's infected until/unless B reports he symptomatic.
2. What Russ calls "obedience" and I might call "participation".

 

[russ_watters]

2. What Russ calls "obedience" and I might call "participation".

I think you may be misunderstanding, so just to clarify: obedience is what is supposed to happen after your phone buzzes an alert or you tell it you are sick. Of course the people coughing in the grocery store are also going to be unlikely to participate, so there's probably a lot of overlap in that part of the group.

 

[256bits]

If one is out and about with the app sending and collecting the contact jibberish contact information while infected but not showing symptoms, AFAIK there is no known length of time to be in contact with another person for the spread of the virus to be 100% certain, or 0%, or in between, the case may be.
Showing signs of the disease, and then alerting the "contacts" in no way ensures anything.
Some contacts may be infected, while others not.

Quarantine of all contacts of the prime infected person only , or also those down the line, who also may be asymptomatic, or not, is the actual question on what to with contact information.

Anyways, is it contact with another person on how the disease is spread, or contact with objects that is how the disease is spread. No contact information is collected from areas ( door knobs for example ) that have been contaminated.

I think it is another one of these sounds good on paper, and jump on the bandwagon of implementation, without much thought, but is pretty much useless in practice in being a step ahead of the disease.

 

[Vanadium 50]

I think you may be misunderstanding, so just to clarify: obedience is what is supposed to happen after your phone buzzes an alert or you tell it you are sick

Yes, I am misunderstanding. I was considering the participation of people who would use the app at all.

There is still the problem that there is no way to track second-nearest neighbors in this scheme without making then first-nearest neighbors.

 

[russ_watters]

Yes, I am misunderstanding. I was considering the participation of people who would use the app at all.

Right, so the user failure points I see are:

1. Failure to download the app.
2. Failure to use/open the app.
3. Failure to report infection status.
4. Failure to obey when the app tells you to isolate because you are sick or have come in contact with someone who is sick.

Limitations of the app's own functionality (largely also due to the attempt to preserve privacy) are separate/additional "problems" the articles/video I've seen also gloss-over.

 

[russ_watters]

If one is out and about with the app sending and collecting the contact jibberish contact information while infected but not showing symptoms, AFAIK there is no known length of time to be in contact with another person for the spread of the virus to be 100% certain, or 0%, or in between, the case may be.
Showing signs of the disease, and then alerting the "contacts" in no way ensures anything...

I think it is another one of these sounds good on paper, and jump on the bandwagon of implementation, without much thought, but is pretty much useless in practice in being a step ahead of the disease.

Contact tracing is a tried-and-true method of inhibiting disease spread. No, it isn't perfect, but it doesn't have to be. It can be really, really good and help a whole lot.

 

[256bits]

Contact tracing is a tried-and-true method of inhibiting disease spread. No, it isn't perfect, but it doesn't have to be. It can be really, really good and help a whole lot.

Isn't location important to know for an investigation of hot spots and the spread of the disease from that area. This app does not have that.

 

[russ_watters]

Isn't location important to know for an investigation of hot spots and the spread of the disease from that area. This app does not have that.

The issues we are discussing in this thread are specific to this disease, yes. The high transmission rate and long incubation time are what make this one unusually hard. Otherwise, you can do a lot of good with named contacts. EG, I went to dinner at my parents' house for Easter and choir practice the week before [I didn't]. That's where the vast majority of new cases come from.

 

[Vanadium 50]

Isn't location important to know for an investigation of hot spots and the spread of the disease from that area.

If the app knows who you have been in contact with, the app doesn't need to know where. The premise, however, is arguable.