There are two far-too-common traits in our community – domain experts thinking that their tool is a solution to every problem, and politicians who want to be seen to be doing something – anything – to combat the current crisis. This short article builds on a recent exchange on Twitter that I had about why apps are not likely to be useful for Coronavirus contact tracing – and might be positively harmful.
Contact tracing can be an important way to help combat the spread of a virus. If you know who has the disease (through testing) you can identify the people that they might have transmitted the disease to. If you identify those people, you can then isolate them (until they have been testing) to reduce the spread of the disease in the community.
Given that we want to do good contact tracing, the Government needs to prioritise resources to best achieve that. The Isle of Man Government should avoid using apps to do this, and it would be better off prioritising resources to more fruitful areas to protect our community.
There should be an app for that!
The first problem of domain experts thinking that their tool is a solution to every problem is captured in the saying, “When you are a hammer, every problem looks like a nail.” When you are a tech-geek who lives in the world of apps it is tempting to think that every problem can be solved by an app. I have been living in that professional world, so I see this far too much: Need a taxi? There should be an app for that. Need home delivered food? There should be an app for that! Need contact tracing of disease? There should be – wait.
In addition to being aware of the strengths of phone apps, we also need to consider their weaknesses, and in 2020, there are many that limit the ability of automated phone apps to perform effective contact tracing. Phone apps trying to do contact tracing will hit against a fundamental problem that I encountered in my days leading teams dedicated to security in online gaming – you will have a large number of false positives (people who get alarms wrongly) and a large number of false negatives (people who fail to get alarms). The combination of these two things gives you an app that does not work, and which will quickly be distrusted by the public. Thus, when we need high trust in efforts to combat the virus, destroying that trust will be worse than having no app at all – these will be very harmful results for our community’s efforts to combat the virus.
When the programmers write the app, the app will need to be given an exact definition of a contact. For example, you could imagine a contact alarm being triggered if the app thinks you were within two metres of someone who subsequently tests positive* for five minutes.
We will get many false positives because GPS and Bluetooth aren’t accurate enough for such delineations: Bluetooth devices in phones can have effective ranges of up to ten metres, and GPS is entirely useless indoors (which is where most such contact would take place: in a restaurant or supermarket or similar).
In addition, we will get false positives because the app won’t consider walls and other physical barriers, and we will also get false positives because some contacts cause no transmission. One very likely way to get no transmission is if someone has already unknowingly had the disease, and are therefore happen to be essentially immune to new transmissions (if that is possible).
This means that if a person triggers an alarm contact on 1 May, 15 May, 1 June and 15 June, are they seriously expected to isolate for 14 days each time? Despite potentially being immune? That’s clearly absurd. Consequently, anyone who triggers a contact alarm will need to wait for symptoms and/or be tested. Which is the same as the current (no app) situation, but with the additional expense of more false positives, more costs, and more distrust in an app.
It would be possible to reduce the harm of false positives by having widely available tests – as I write this in early May 2020, the Manx Government continues to average less than 70 tests per day, and has been overtaken by the UK in per capita testing. If the Manx Government acted to increase testing, a ‘false positive’ alarm would require you to get a test – but given how many false positives would be generated, it is difficult to see there being sufficiently high testing numbers for some time yet.
*It might also be possible to alarm people further down the chain: You could imagine also triggering an alarm if you were within two metres for five minutes of someone connected to someone who tested positive. This exacerbates the false positive problem in a very obvious way.
We will also get false negatives of not recording when all infections occur. It is possible for you to contract the disease without your phone app alarming you of the contact. This again comes from several areas for error.
Firstly, your phone might not correctly identify that you were in proximity to another phone. This happens in obvious ways with using GPS receivers and Bluetooth devices all the time – we have all seen the variability of location recording in different locations (especially indoors, where most disease transmissions take place) and we have all suffered from being unable to link two bluetooth devices together. These are not new problems.
Further, some people will get infected by disease transmissions from people who do not have the app with them at the moment of transmission. Not everyone has a smartphone (most obviously children, but many adults in our community do not have smartphones too) and not everyone who has a smartphone will install the app. Compounding this, some people do not carry their phone with them literally 100% of the time – for example, to meetings, to the bathroom, to the gym, or any one of a million other places.
Finally, false negatives will arise when the disease is transmitted outside of the app-programmed contacts. This might be through a brief sneeze or cough interaction, or it might be transmitted using fomites.
The End Game
The result of all these false positives and false negatives is that when you come home, and have one of these apps, you will have no idea of whether you have been infected or not. If you come home and receive no alarm, you will not be able to be confident that you are “clean”. Consequently, you will need to wash your hands and maintain all the precautions that you current need to abide by. If you come home and do receive an app alarm, you will not be able to be confident that you are sick – certainly not confident enough to self-isolate in your home for two weeks. At best, you might want to get a test – but given that Governments currently refuse to test asymptomatic people anyway, why would they test asymptomatic people after this app is available?
The end game, then, is that people will need to rely on symptoms. Like we do now.
Some more reading on the limitations of apps here:
We are concerned by this rising enthusiasm for automated technology as a centerpiece of infection control. Between us, we hold extensive expertise in technology, law and policy, and epidemiology. We have serious doubts that voluntary, anonymous contact tracing through smartphone apps—as Apple, Google, and faculty at a number of academic institutions all propose—can free Americans of the terrible choice between staying home or risking exposure. We worry that contact-tracing apps will serve as vehicles for abuse and disinformation, while providing a false sense of security to justify reopening local and national economies well before it is safe to do so. Our recommendations are aimed at reducing the harm of a technological intervention that seems increasingly inevitable.Contact-tracing apps are not a solution to the COVID-19 crisis
April 27, 2020 Ashkan Soltani, Ryan Calo, and Carl Bergstrom
“My problem with contact tracing apps is that they have absolutely no value,” Bruce Schneier, a privacy expert and fellow at the Berkman Klein Center for Internet & Society at Harvard University, told BuzzFeed News. “I’m not even talking about the privacy concerns, I mean the efficacy. Does anybody think this will do something useful? … This is just something governments want to do for the hell of it. To me, it’s just techies doing techie things because they don’t know what else to do.”We Need An “Army” Of Contact Tracers To Safely Reopen The Country. We Might Get Apps Instead.
April 29, 2020 Caroline Haskins, Stephanie M. Lee, Megha Rajagopalan, Pranav Dixit
3 May Update: The first version of this article that I published incorrectly characterised the Manx Government as dismissing apps. The headline and parts throughout have been corrected to remove that characterisation