Sam Morgan argues the Government should stop flogging the dead horse of the COVID Tracer app and instead adopt the bluetooth CovidCard to avoid another lockdown
Since late March I’ve been leading the private sector team responsible for CovidCard. We’ve been working alongside our assigned Government minders at the Department of Internal Affairs. We’re a team of epidemiologists, mathematicians, technologists, software developers, hardware engineers, communications people, industrial designers, and more. For the knockers: I’m a volunteer. None of the team are in this for financial gain.
The reason we’ve been focused on CovidCard is because a) Community transmission is probably inevitable and could occur any day between now and vaccine day; b) Manual contact tracing is vital, but it can’t stop community spread under normal social settings; c) Apps don’t work; and d) Lockdowns, while effective, are too economically and socially brutal to be frequently repeated.
We concluded that we simply must upgrade our tools if we’re serious about elimination.
Covid-19 is a tricky virus. About half of the people who get Covid-19 are asymptomatic and don’t feel sick enough to get tested. We only contact trace people after they’ve tested positive, of course. Some asymptomatic cases won’t spread it, but then someone will go and infect 50 highly social people at a nightclub.
As soon as Ashley Bloomfield says “we have a case that is not connected to travel”, you should prepare for a lockdown. Head to the shop and refresh your stocks of toilet paper if that’s your thing.
Community transmission may come from failures in quarantine, a false negative test, from returning aircrew, or perhaps from an infected customs officer. The source is not important and good people are trying hard to prevent that, but we should plan as if community transmission is coming any time.
Manual contact tracing is a critical part of our response, but we need to be realistic about what it can do – gold standard or otherwise. New Zealand was not saved by contact tracing. We were saved by the lockdown. Under lockdown, each case has about two contacts, mostly in the same house. Under such forgiving conditions, manual contact tracing works just fine.
Apps, specifically apps using Bluetooth for close contact proximity detection, have been much talked about, but are fatally flawed. There are three main issues: 1) They do not achieve the necessary level of adoption; 2) iPhones don’t detect each other reliably due to Apple’s security architecture; 3) Determining the actual distance between phones is horribly unreliable across the different makes and models of phones being carried in all manner of different ways.
Adoption rates are the real killer. If, optimistically, 40 percent of people download the app and only half the phones work reliably, you end up with 20 percent effective usage. With 20 percent effective usage, only four percent of contacts end up being detected, because 20 percent of people detecting 20 percent of people they meet (20 percent of 20 percent) is four percent.
The international experience has borne this out. Australia’s CovidSafe app, which has been downloaded 6.6 million times (about 26 percent of Australians) has not detected a single contact in the state of Victoria that wasn’t detected by the manual contact tracers, according to Victoria’s chief medical officer.
New Zealand’s COVID Tracer App is also on track to achieve nothing. Achieving 600,000 downloads is great, but it is not the relevant statistic to track. We need to look at whether people are scanning in. If 10,000 scans take place each day, it is the equivalent of each New Zealander scanning in once every 18 months. This approach was never going to work. We tried to tell them.
Our modelling suggests that even if wide usage is achieved it will contribute almost nothing over and above manual contact tracing. Experience from Singapore and Australia shows that the contact tracers will not use technology that doesn’t add to their work.
The Government can’t mandate a technology that requires a smartphone. Māori and Pasifika leaders are frustrated because they know it does not work for their people. Where was the consideration of equity?
The renewed flogging of the dead horse that is the COVID Tracer App is eroding any political capital that could be spent promoting technology that does work. They either know it can’t work or they are poorly advised in this area. It was a bad idea that was poorly executed. When it launched, no businesses had the codes up. More promotion won’t save it. Please stop.
The CovidCard is a dedicated Bluetooth wearable. You simply wear it in places of congregation – pubs, clubs, shops, on the plane, on the bus, at church, at the marae. That’s it. The Government would send you one for free. It’s envisaged that if we get community transmission, we would move up to Alert Level 2 and wearing your card would then be strongly encouraged or perhaps required.
Each card detects the cards close by (within a couple of metres for 15 minutes) and stores those encrypted card numbers on your card. There is no personally identifiable information stored on CovidCard. No contact data leaves your card. The card does not know where you are. There is no GPS or connection to the Cloud. Your close contacts would be stored for the last 14 days and would only be downloaded if you test positive for Covid-19.
In the rare event that you test positive, your contacts would be immediately notified, meaning they can immediately self-isolate, proactively protecting their loved ones even before the contact tracing teams connect with them.
CovidCard is not necessarily a slam dunk. You need to believe that New Zealand can achieve widespread usage. Achieving that might ultimately require a degree of mandating in places of congregation. The card is ideally worn visibly on your person and not tucked away in a handbag. We realise it isn’t a highly desirable fashion accessory but wearing it around your neck on a lanyard or clipped to your top makes it easy for venue owners to observe compliance and check patrons have their cards on them. It also ensures the card function optimally.
Most of our efforts on CovidCard have been to see if the technology actually works. We’ve satisfied ourselves on this score. We’ve also consulted broadly with businesses, Iwi, Pasifika groups, unions and so on. There is widespread support.
We’re now just waiting to see if the Government wants to do it. Singapore is already doing it after giving up on their app. Here’s a promotional photo of what it looks like from the Singaporean Government, with the brochure given out in a trial last month.
Deploying a Bluetooth card to five million New Zealanders will take six months and cost $100 million if we put the hammer down and go at it. The battery on the card will provide 12 months coverage – so we’d have cover for the whole of 2021 and could then decide whether we want to renew the cards.
$100m sounds like a lot but, for context, that equates to two days of the health budget. The cost of locking down the economy is billions per week.
The big questions
Is the Government prepared to run with the CovidCard as an insurance policy for New Zealand to sustain our strategy of elimination?
Do we believe New Zealanders would wear a CovidCard in the places they congregate if it meant we might stay out of lockdown?
New Zealand has a fantastic opportunity to lead the world by sustaining our strategy of elimination with upgraded digital tools. At present, we’re lucky to have no Covid-19 in the community. But if we see any community transmission right now, with the tools we’ve got, then we’re heading back into lockdown. It needn’t be that way.