Contact tracing

Since I first posted this piece late last night, discussion of contact tracing appears to have gone viral (as the saying goes), partly in response to the rumour that the UK is about to launch something very similar to TraceTogether (first developed in Singapore and described below). I was going to add some links but you can find them everywhere. So, by all means skip the piece below; it adds very little to the general discussion. I do however recommend taking a look at the Youtube link at the end - if you haven't see it already. 

There appears to be a growing consensus that the only effective way to manage (and eventually eradicate) the Covid-19 pandemic is through a rigorous program of testing and contact tracing. This is the policy advocated by amongst others: the WHO’s Bruce Aylward and the UK former Health Minister, Jeremy Hunt. The merit of the approach is supported by evidence from SE Asian countries — most notably Singapore and South Korea — where testing and contact tracing were adopted from the outset with a measurable degree of success.

So how does it work? At the simplest level, public health bodies identify people either testing positive for the virus or showing symptoms and then, by means of interviews, ask them to remember with whom they have been in close contact. These people would be followed up in turn and offered testing and advice on limiting further spread.

It doesn’t take long to see the impracticalities of this method. For a start, people are unlikely to remember all of their close contacts and would be unable, in most instances, to identify individuals - for example, after travelling on a bus or visiting a supermarket. Also the resource requirements involved in tracking down and interviewing a widening network of contacts would be considerable.

So how was it done in Singapore and South Korea?

In South Korea the government texted people to let them know if they were in the vicinity of a diagnosed individual using location data from mobile phones. In any case, it is an approach that would be unlikely to gain acceptance in the UK on account of the privacy implications. It is also difficult to see how this would help trace contacts of asymptomatic individuals who later tested positive.

The method adopted in Singapore is more interesting. Here they encouraged people to install an app on their phones called TraceTogether. Each app was assigned a unique ID and the Ministry of Health maintained a database linking each ID to the user’s phone number.

The way it works is this: every time you come into close contact with another person with the same app installed on their phone the apps communicate (via Bluetooth) and each simply stores the ID of the other contact together with the current time. If you remain free of the virus there is no need for you to pay attention to the data being collected - in fact, it is virtually meaningless, consisting simply of a number of timestamped numerical IDs. If you test positive for the virus however, you are encouraged to contact the Ministry of Health and allow them to access the data stored on your phone. The Ministry is then in a position to inform those contacts whom they deem to be at risk and advise them to be tested.

Although this still requires the involvement of a government ministry, the TraceTogether app is designed in such a way as to reassure users with regard to privacy. In essence it is little more than a streamlined version of the contact tracing approach described earlier, the main advantage being that it does not rely on individuals being able to recall (or identify) the people they have been in contact with.

Before I heard about the TraceTogether I found myself thinking about something along similar lines. As it turned out, this was very similar to TraceTogether except for the fact that, in my own scheme, the records would be uploaded to a central body which would, as a consequence, be in possession of the data necessary to construct a detailed representation of contacts across the entire population. Not only would this allow those at risk of infection to be identified and contacted but it would also enable the co-ordinating body to evaluate the effectiveness of social isolation directives as well as assessing the consequences of contacts involving key workers (emergency services, medical staff etc.)

It is the sort of idea that seems wonderful for a while, until one realises it could also be a nightmare.

Many of these issues are discussed in a recent article by Jon Evans. In it he makes the observation:

More generally, at what point does the urgent need for better data collide with the need to protect individual privacy and avoid enabling the tools for an aspiring, or existing, police state? And let’s not kid ourselves; the pandemic increases, rather than diminishes, the authoritarian threat. 

The whole article makes very interesting reading. You can find it here:

https://techcrunch.com/2020/03/29/test-and-trace-with-apple-and-google/

The author comes to the conclusion that Apple and Google are capable of constructing a near-complete contact model within the time frame necessary to bring the virus under control and that furthermore they can be trusted more than governments not to misuse the information obtained.

I am not inclined to comment further other than to point out that such a proposal would be — at the very least — controversial.

Finally though, you should take a look at the following short presentation. It shows what is presently possible using location data from mobile phones.

It is a bit of an eye-opener.