14 April 2020

Electronic Frontier Foundation: “The Challenge of Proximity Apps for COVID-19 Contact Tracing”

We cannot solve a pandemic by coding the perfect app. Hard societal problems are not solved by magical technology, among other reasons because not everyone will have access to the necessary smartphones and infrastructure to make this work.

Finally, we should not excessively rely on the promise of an unproven app to make critical decisions, like deciding who should stop sheltering in place and when. Reliable applications of this sort typically go through many rounds of development and layers of testing and quality assurance, all of which takes time. And even then, new apps often have bugs. A faulty proximity tracing app could lead to false positives, false negatives, or maybe both.

Andrew Crocker, Kurt Opsahl, & Bennett Cyphers

Another hot topic for discussion lately, besides the benefits of wearing masks in public, has been the various proposals and initiatives to track people infected with coronavirus using mobile technologies – in this case individual tracking, not aggregated tracking like we saw from Google. Some of them are already in place, for example in Hong Kong and a growing number of countries around the world, others are being discussed, as in the European Union and San Francisco. And, as everything moves at an accelerated pace in 2020, late last week, Apple and Google announced a partnership to create a decentralized contact tracing system that will work across iOS and Android. There are many critical takes online, starting with the article above, because this is a situation where purely technological solution can be of limited use.

TraceTogether is a community-driven contact-tracing tool to help trace COVID-19 cases in Singapore

Manual contact tracing – interviewing infected individuals about their recent contacts, in order to track them down and isolate them as well – is an effective tool when dealing with a rapidly spreading epidemic, but it’s limited in both scale (in an outbreak this large there are not enough health workers to do the tracing) and effectiveness (even healthy people have trouble remembering everyone they came into contact with for the past two weeks, let alone sick people lying in a crowded hospital). The selling point of proximity apps is that they could potentially solve both these issues: first the app silently tracks people we meet on our smartphones, then, if someone gets infected, automatically sends an alert to her or his contacts from the past 14 days, so that they may seek medical help and isolate from the community.

All that said, I suspect the tracing apps are really just do-something-it is. Most countries now seem past the point where contact tracing is a high priority; even Singapore has had to go into lockdown. If it becomes a priority during the second wave, we will need a lot more contact tracers: last week, 999 calls in Cambridge had a 40-minute wait and it took ambulances six hours to arrive. We cannot field an app that will cause more worried well people to phone 999.

Ross Anderson

This ideal scenario can fail on many levels, even when leaving aside privacy concerns. For it to be effective, each country first needs a quick and accurate test for the infection, an area where many are still lacking.

Secondly, just like vaccines and herd immunity, a sizeable proportion of the population would have to participate in the program for it to be effective. This is impractical for various reasons, both technological and social. The Apple and Google-developed APIs will not be available soon enough to help in this phase of the epidemic; after launching, it remains to be seen how fast the update will be delivered, to how many devices, and how many can be convinced to install an additional app on top of that. Despite wide-spread adoption, many people do not own a personal smartphone, often those from the most vulnerable social groups, the poor and the elderly. Obviously, some people will refuse to install it for fear of government surveillance, or, if pushed automatically by Apple or Google, will turn off their Bluetooth signal, or simply leave their phone at home, thereby rendering the collected data useless. The experience so far in Singapore makes me suspect that any voluntary tracking measure will not gain sufficient mass to provide accurate information.

Apple Google COVID-19 smartphone tracing Apple Google COVID-19 smartphone tracing

There are also several practical details the app needs to account for, starting with: what kind of human interaction leads to an infection, how long does it last, how close the people involved? This is unfortunately a topic without a clear medical answer, so attempting to codify it universally will certainly lead to faulty data. There are many factors involved that the tracing system has no way of evaluating: are people wearing masks or other protective gear (as in supermarkets or at a doctor’s office); are they in closed, poorly ventilated quarters (like in a church or at the office) or meeting outside; are they hugging and shaking hands or keeping a safe distance while chatting (distance measurement using Bluetooth between smartphone is not particularly accurate, from what I understand).

Another potential issue highlighted in a comparison with other proposed tracing protocols is that a OS-level solution is more inflexible, harder to update as new methods are put in place, or new information becomes available. This could make the gathered data more fragmented over time, as some people fail to update the OS to take advantage of API improvements. Even with a common underlying API, some countries may choose different approaches to these questions, or to data retention, which would make worldwide data analysis tricky. It could complicate foreign travel as well: if the tracing data from your origin country is incompatible with your destination, you may be asked to quarantine, either on arrival or upon return, or denied entry…

Already there are reports that limitations of the tracing system jointly developed by Apple and Google are interfering with app development by governments in UK and France.

But my main issue remains the fact that such a tracing system would not prevent me – or anyone using it for that matter – from being infected, unlike social distancing, or wearing a mask. Getting notified that you’ve been in contact with a confirmed patient does not provide an immediate benefit, hence the incentive to use proximity apps is low. This downside may be improved with proper presentation and messaging to emphasize the social benefits – possibly less strict distancing measures. Depending of the state of available testing and the overall health system in each country, the notifications may even exacerbate fear and panic. A more proactive approach would be to notify people of congestion in the areas where they plan on going, like supermarkets and parks, thereby encouraging them to pick less crowded times to shop and go out – a Waze for foot and indoor traffic. This could be directly integrated into the Maps apps on mobile via a simple update, no need to download new apps. Tracing systems may improve in the medium to long term, as some of these issues get clarified, but I don’t think these can become more than an auxiliary measure in a complex set of solutions for managing future pandemics.

Post a Comment