We know #ContactTracing and #testing testing will be a core part of our #COVID19 pandemic response, which could involve contact tracing apps. Implementation everywhere has been a mess and unlikely to get better soon, and I’ll explain why here.
1/
1/
By luck I am uniquely qualified to write about this; I have both a medical and software background and am part of a team trying to solve the issue. I’ve also been toiling away at my own startup @thenewsraven, a news and social feed aggregator which will be out in beta soon
2/
2/
Contact tracing itself in its modern form, dates back to Dr. John Snow (not the Targaryen) and tracking down sources of cholera in 19th century London. Using similar means, a team was able to track spread among infected guests at a restaurant in China
3/
3/
It seems natural to use smartphones to track potential contacts . It can keep tabs on who you were in contact with, how long, and how close using GPS and Bluetooth low energy signals. Encryption can keep identities private.
4/
4/
Singapore was the first out of the gate with their TraceTogether program, used Bluetooth signals shared between phones to track contacts. Their own team took great pains to emphasize was not a replacement for standard ‘shoe leather’ contact tracing
bluetrace.io/policy/
bluetrace.io/policy/
Closer to home the @covidsafepaths team, led by a team at MIT, formed and took on 1200 volunteers, @PeppPt formed in Europe. Both were privacy-oriented, keeping user data private and avoiding centralization of data. So here are the problems:
6/
6/
So when Apple and Google decided to fix this by creating their joint ‘Exposure Notification’ protocol and building bluetooth tracing into the phone, this problem would be solved and all would be well, right? It was just released to billions of iOS and Android users this week.
Unfortunately, no. Apple and Google threw a wrench into this process. Apps are
- forbidden from using GPS
- Limited to government entities, no way to test
- prohibited from storing location data.
- Onus on users to explicitly opt-in and then identify themselves when infected
- forbidden from using GPS
- Limited to government entities, no way to test
- prohibited from storing location data.
- Onus on users to explicitly opt-in and then identify themselves when infected
What does this mean? As Britain and France have discovered to their frustration, valuable insights into infection hotspots and spread are impossible… they are entirely at the mercy of Apple and Google, national sovereignty be damned.
10/
theverge.com/2020/5/5/21248…
10/
theverge.com/2020/5/5/21248…
Even MIT’s Prof. Raskar goes over some problems;
- GPS data can rule out false positives (i.e. you were wearing a mask when you were exposed)
- GPS can help track and predict hotspots
- You don’t need as much uptake with a GPS based app
11/
- GPS data can rule out false positives (i.e. you were wearing a mask when you were exposed)
- GPS can help track and predict hotspots
- You don’t need as much uptake with a GPS based app
11/
Basically American hang ups with privacy are being exported worldwide. Citizens in many other advanced democracies (like mine) that have stronger oversight and greater trust in government are stuck in a US-driven paradigm until the grandees at big tech change their minds.
12/
12/
That’s not to say there are real concerns about centralization and privacy; any government using this kind of data should have robust data-security policies. But a typical Android phone user leaks all kinds of data that’s picked up by ad companies
pindrop.com/blog/the-gory-…
13/
pindrop.com/blog/the-gory-…
13/
Privacy commissioners in Canada wrote a set of guidelines that encourage:
- Limits on use of data after the crisis
- Oversight and accountability
- Openness and transparency
- De-identification and privacy
priv.gc.ca/en/privacy-top…
14/
- Limits on use of data after the crisis
- Oversight and accountability
- Openness and transparency
- De-identification and privacy
priv.gc.ca/en/privacy-top…
14/
Issue 2: Uptake of the app. Oxford researchers estimate 60% of the population would need to use it. Semi-authoritarian Singapore is at 25%, homogenous Iceland is at 40% Everyone else is somewhere in between
research.ox.ac.uk/Article/2020-0…
15/
research.ox.ac.uk/Article/2020-0…
15/
Issue 3: governments are terrible at IT projects. Senior officials lack expertise, contracts are awarded to expensive consultancies who outsource the work to bottom dollar contractors overseas. This is an example: a failed payroll system:
en.wikipedia.org/wiki/Phoenix_p…
16/
en.wikipedia.org/wiki/Phoenix_p…
16/
It’s worse in Canada as decision making is split between the feds, provinces and local health units. The country is massive, one region or province can’t pick one solution in isolation to others, as they may not share data with each other. Leadership and guidance is needed
17/
17/
Issue 4: phones need to be updated, and capable of broadcasting bluetooth LES, which is a new feature. Singapore discovered their outbreak among marginalized migrant workers working in crowded dorms, unlikely to have the latest smartphones.
theconversation.com/this-is-why-si…
18/
theconversation.com/this-is-why-si…
18/
Hong Kong gets around this by using tracking wrist bracelets on suspect cases and returning travelers, which is coupled with public health surveillance. Densely populated Hong Kong has had only 4 deaths so far. This is something our group is testing
nytimes.com/2020/04/08/wor…
19/
nytimes.com/2020/04/08/wor…
19/
The Israelis used the domestic spy agency Shin Bet to track, using cell tower monitoring with software made by NSO and others, who famously hacked Whatsapp. This approach is unlikely to gain acceptance here but could be used as a last resort.
privacyinternational.org/long-read/3747…
21/
privacyinternational.org/long-read/3747…
21/
Issue 5: Contact tracing software needs to be made as part of a larger group of technologies
- Workplace monitoring (keep workers safe, trace infected colleagues)
- Entry to public areas
- Dashboards for public health units (some are using pen and paper and fax machines still)
22
- Workplace monitoring (keep workers safe, trace infected colleagues)
- Entry to public areas
- Dashboards for public health units (some are using pen and paper and fax machines still)
22
Issue 6: There’s so much asymptomatic spread, population-based antibody tests are showing we are missing perhaps 90-95% of cases. What is the point of contact tracing if we are only tracing 5-10% of possible cases? I'll leave this to the experts @dfisman
Solutions?
- Governments should be more open. A lot of expertise is sitting on the sidelines, but they need to ask for help. A group of volunteers at @Shopify just put something together in a week. Contact them, and be open about it.
@tobi
ottawamatters.com/local-news/tea…
24/
- Governments should be more open. A lot of expertise is sitting on the sidelines, but they need to ask for help. A group of volunteers at @Shopify just put something together in a week. Contact them, and be open about it.
@tobi
ottawamatters.com/local-news/tea…
24/
- Start a debate on privacy - a lot of intrusive technology might be needed to get this under control. Better to have that discussion now.
- Encourage the adoption of open-source projects like MIT’s @covidsafepaths, so we have an insight into the tech
25/
- Encourage the adoption of open-source projects like MIT’s @covidsafepaths, so we have an insight into the tech
25/
I’m going to be organizing a debate into the privacy aspects of electronic contact tracing, reach out if you want to be involved. It will be similar to a debate I moderated on universal masking which was largely civilized and well received
end/
end/
