CovidCard contact tracing devices have been shown to work in controlled environments but would only be made compulsory as an “extreme last resort”, Government Digital Services Minister Kris Faafoi says.
The devices, which supporters say could be provided to all Kiwis at a cost of $100 million, have been examined by the Defence Technology Agency, Otago University experts, and the GCSB.
But Faafoi indicated the Government was still hedging its bets, saying a range of technology options was being considered to improve contact tracing “including Bluetooth app technology being used in Ireland, which has shown promise”.
If the CovidCard got the go-ahead from the Government all Kiwis – perhaps those aged over 16 – would be able to get one of the Bluetooth devices, which would be worn as a lanyard or attached to clothing to record people’s close contacts.
* Government plans roll-outs of Covid Card as part of big push to secure the border
* Sam Morgan’s CovidCard: The cases for and against
* Contact tracing: NZ ‘missed opportunity’ during 102 days without Covid
The cards are touted as being capable of recording “clinically significant contacts” with greater than 90 per cent accuracy (fewer than 10 per cent “false negatives”).
That is while keeping the number of spurious contacts that are wrongly identified as clinically significant (false positives) at fewer than 20 per cent.
Otago University’s public health department was contracted by the Department of Internal Affairs to oversee a trial of the CovidCard by 42 health workers in Nelson.
The next steps include a trial in Rotorua to gather feedback on how acceptable the card might be to the wider public.
Housing Minister Megan Woods announced CovidCards would also be piloted with staff at managed isolation facilities.
Otago University senior research fellow Tim Chambers said the Defence Technology Agency, a branch of the armed forces, had validated that the technology worked.
A Defence Force spokeswoman clarified the agency made some recommendations concerning “technical performance improvements and risk mitigation”.
It also recommended the system as a whole should be “conclusively demonstrated in a large-scale controlled trial” by independent experts, she said.
But Chambers, who has seen the agency’s high-level findings, said if it had reservations about the internal validity of the card it would be “commissioning much stricter field trials or not going ahead at all”.
All the health workers in Nelson had said they would be happy to wear the cards again, he said.
“You don’t have to wear the card everywhere you go, just when you go into public places.
“It is less obtrusive than wearing a mask.”
The decision on whether to take a case to Cabinet for the roll-out of CovidCards will partly fall to Faafoi.
Former UnitedFuture leader and internal affairs minister Peter Dunne said government technology programmes had not had a particularly positive history.
“While this one seems quite attractive and would certainly get around all of the problems with contact tracing at the moment, ministers would want to take a pretty cautious view I would imagine.”
The risk was not just of a “$100m white elephant” if it didn’t deliver, but of public ridicule, he said.
But he agreed there was also a danger of being too risk-averse.
For all those reasons, the Government might want to seek cross-party support if it did press ahead, he said.
Faafoi did not rule that out.
Opposition parties had supported urgent legislation and regulations that had been put in place to deal with Covid, he said.
“I would hope to see that kind of cross-party cooperation extending to whatever additional solutions emerge.”
Sources said National would be likely to lend support, particularly if it could pave the way for a faster and safer partial reopening of the border.
Supporters of the CovidCard include former prime minister Helen Clark, Trade Me founder Sam Morgan, BusinessNZ chief executive Kirk Hope, Animation Research founder Ian Taylor and Navman founder Sir Peter Maire who described it as “very simple, very reliable and so cheap to implement”.
Clark said the CovidCard was “a very important innovation which would help New Zealand manage the inevitable future Covid-19 outbreaks without hard lockdowns”.
Morgan, who has contributed financially to work on the CovidCard – but who says he is unpaid and has no commercial interest in it – said it was a loose coalition.
Facing them is an army of people asking questions.
Privacy commissioner John Edwards described the CovidCard in July as “untested and unproven”.
“The big concern that I have is there are so many unanswered question about the logistics of a roll-out – how do you ensure you have a register of every New Zealander and that you have a specific card registered against that?” he says.
“If you are establishing a system for the collection of interactions between every New Zealander and every other one, you have got a significant privacy issue, whether you are taking steps to mitigate it or not.”
Only if the cards made a “material contribution to contact tracing” might that intrusion be justifiable, he said.
Dean Armstrong, chief executive of Hamilton wireless engineering consultancy Virscient, which counts Samsung and Qualcomm among its clients, has driven the development of the CovidCard.
“The reason I got involved was because of the equity side of it,” he said.
“Any smartphone-based solution is for the richest half of the world’s population.”
While he didn’t rule out benefiting from commercial opportunities overseas if CovidCards advanced, that was “very much a secondary consideration” and at the moment his time was unpaid, he said.
For contact tracing to work effectively, people would need to give up some level of privacy, he agreed.
“It is a big task to consider delivering these small electronic devices to the population.
“It is not to be underestimated, and it hasn’t been underestimated.”
Armstrong said the cards sent out a Bluetooth radio signal every eighth of a second to advertise their presence and listened every 20 seconds for ID codes transmitted by other cards.
If they detected a signal above a certain strength and time threshold, they would record that code along with details of its signal strength and a time stamp, all in an encrypted form, on the card.
To improve privacy, the ID codes sent out by the cards change about every 15 to 20 minutes.
If someone caught Covid they would hand over their card to contact tracers who could match the codes to the cards that generated them, and hence to the people the infected person had been in contact with.
Each card can store details of 57,000 interactions but the data on the cards would be automatically wiped, perhaps every three weeks, again for privacy, Armstrong said.
“On the server side we generate a sequence of ‘hashes’ which means you can flush out old identifiers and have no way to go back to them because of the way the encryption is generated,” he said.
“So we can say hand-on-heart we can ‘forget history’ and have no way to go back to track people.”
“The appropriate folk in government” had reviewed the solution, he said.
While the Defence Technology Agency had been primarily responsible for that, the Government Communications Security Bureau “will obviously have looked at this as well”, he said.
“You cannot connect to a card unless you hold a button and there is digital handshake that provides ‘man-in-the-middle’ protection.
“To access a card you need a properly signed digital certificate from the health authority and to know the serial code for the card you are accessing – not that the data would be of any use to somebody trying to ‘sneak in’ as it is all encrypted.”
For the purposes of the Nelson trial, a “clinically significantly contact” was defined as a situation where people were within 2 metres of each other for 15 minutes over the course of a day – plus other scripted interactions between people that were monitored on video by clinicians and judged to be “significant”.
“We had human observers and videos to get a concrete assessment of which interactions met the clinical definition,” Armstrong said.
“We reviewed every minute between every pair of people; were they within 10 metres, 2 metres, or obviously in close contact because they were hugging or dancing or something like that.”
Health Minister Chris Hipkins announces a Covid-19 card trial.
Armstrong said the sensitivity of the system could be turned up or down by contact tracers even after cards were issued, by tweaking the algorithm used to analyse the data uploaded from the cards on the strength and duration of each recorded signal.
Increasing the sensitivity would reduce the number of clinically-significant contacts that were missed, at the expense of generating more false alarms, while decreasing sensitivity would have the opposite effect.
The most worrying interactions could be looked at first.
“This isn’t an automated system and is not designed to be,” Armstrong said.
“It is a ‘human in the loop system’ to help our contact tracers do their job faster.”
Some technologists have queried whether CovidCards would have a long enough battery life to be useful.
But Armstrong said that had been stretched out to 12 months by building in an accelerometer.
That sent the cards into “sleep mode” when they were stationary and hadn’t picked up a signal for a while, for example when people were asleep at night.
The cards have an Led light activated by a button that people would be encouraged to press perhaps once a day to check the battery wasn’t dead.
An automatic battery light was considered but ruled out as “nothing kills a battery quicker than a blinking Led at the end of its life”, Armstrong said.
The value of CovidCards would rise (and fall) exponentially based on their take-up, which does point towards mandatory use.
If only half of people carried CovidCards, for example, they could only record a quarter of the interactions that happened between people.
If take-up fell to a third, that figure would fall to just 11 per cent.
But in theory at least, CovidCards could be made available voluntarily at first and then mandated later.
Faafoi said the Government had discussed “the question of mandatory requirements around whatever Covid track and tracing methods were used”.
“At this point, the Government has no plans to make use of any technology mandatory and prefers to see New Zealanders willingly play their part in doing what’s required to help respond effectively,” he said.
Ian Taylor believed that what might seem one of the big drawbacks of the CovidCard – the fact that it would need to be worn visibly – meant compliance could be achieved through community peer pressure rather than anything much more heavy-handed.
“Ultimately it might have to be compulsory. ‘Compulsory’ is how you truly protect everyone,” Taylor said.
“But you imagine turning up at a supermarket and you have forgotten to bring yours and everyone else has one. It would be like ‘sorry, sorry’.”
Assuming compulsion would be inevitable eventually, another question might be whether that would be only when the virus was circulating in the community.
For the cards to be useful in “upstream” tracing to track down the source of a new outbreak they would need to be worn when there was no community transmission.
But an option might be to only mandate their use during an outbreak, say in a region affected by a level 3 alert, just to assist with the “downstream” task of working out who known carriers might have infected.
Armstrong agreed data might need to be kept on the card for a bit longer than three weeks to be of maximum help useful for upstream tracing in a situation similar to the current one, where there is a mystery about the chain that led to the new community outbreak in Auckland.
That is one decision that would need to be made before the cards were issued, as it is not something that could later be changed.
Armstrong said the cards could be distributed through a variety of organisations such as PostShops and marae and online and via the mail.
People would register the cards to themselves creating a database that would be managed by a new government agency set up for that purpose, which would only allow data on the cards to be accessed for pandemic contact tracing.
CovidCards could in theory be assigned to venues and attached to building entrances and exits so they could operate similarly to QR codes and work out who might have shared an indoor space.
But Armstrong said that would raise separate privacy issues, given the system as proposed would not need to record people’s location.
Faafoi said it was important to put the CovidCard through a community-wide trial in Rotorua.
“Let’s see what that trial shows in terms of how easy the CovidCard is to use, people’s willingness to use it, and how it is able to link in with the Health Ministry’s Covid response systems.”
The card has been tested as a lanyard.
“We are continuing to test it being worn elsewhere, like on a belt or in a pocket,” he said.
Armstrong said a drawback of not having the cards worn at about the same height was that they could then under-record the significance of brief, very close contacts, because the faces of the cards which housed their radio antennas could then be at an acute angle.
Sam Morgan estimated it would take five months for the cards to be deployed once a greenlight was given by the Government.
But Faafoi said he was unlikely to take a recommendation to Cabinet until “towards the end of the year at the earliest”.
That was given the need to assess feedback from the further trials and to ensure CovidCards would “interact successfully with the Ministry of Health’s wider Covid response”.
Morgan appeared unimpressed by that.
“They have said they are most keen to evaluate the question of social acceptance,” he said.
“I really don’t know how you could get a good answer to that – it really takes someone to make a brave decision.”
Armstrong echoed some of that sentiment.
“You have to lead with this stuff.”
By the time your hand is forced and you realise you have to do it, it is actually too late, he said.
“Sometimes we require our leaders to lead and we understand that they take risks with every decision they make.
“It’s a game for grown-ups. But I understand how challenging it is.”