Canada’s COVID Alert App Needs to be Shut Down. Here’s Why, Part II
Even if We Had Testing Capacity, the App Should Still Go Away
If you haven’t read the first post, please start there. And to avoid any of this getting too long and overwhelming, I’m going write a part III (maybe more?). The idea here is to do a bit of a post-mortem and lessons learned that also considers how this app performed (and didn’t) before PCR testing capacity was a problem.
I’ve been somewhat uncomfortable pushing the simple narrative that because we don’t have testing capacity in Ontario, the Covid Alert App is broken. Though this is true, if testing were to be returned, we should still get the app shut down. This post is an effort to expand on the testing capacity part to explore the broader why of that ‘shut it down’ statement - why it holds regardless of testing capacity.
But it’s also good for us to think in terms of the efficacy lines that we should have set up before the app was launched. It helps to think about what could have been set up, data and policy wise, so that we would not have to be making the argument to shut the app down now. If we agree that we need a policy framework for future tech interventions before they’re launched, then let’s try to answer the questions about what a framework to define efficacy might look like. Efficacy isn’t all of the framework, but it’s an important part. To that end, this post will zoom out a bit to try to understand the data that Health Canada set itself up to use with Covid Alert to make decisions about it.
Lastly, it also helps for us to look at the technosolutionist cultural elements that were at play here, how the tech industry and economic development forces influenced our pandemic response, how jurisdiction is a feature, and with it, governmental accountability.
The Jurisdictional Factor
To rewind a bit, and to define where things are at today, let’s add one piece of complexity in terms of the testing capacity/brokenness. Health Canada is the business owner of the app, and thus accountable for it, but the testing capacity piece of the puzzle differs across provinces. Here we have a jurisdictional factor that would have been apparent since day one.
It is also necessary to remember that Google/Apple mandated use of only one instance of their protocol per country. They began to create a narrative that this was the best way to support mass uptake. What this meant in Canada is that there was no way for the provinces to do their own thing using the Google/Apple protocol. Health Canada took this restriction on knowing well what it meant. Or at least we should be able to expect that they did.
I was also reminded last week that some provinces *did not* adopt the app. To get specific in terms of talking about the efficacy of the app, we have to start with who is using it and who is not.
The following provinces and territories never signed on to be part of the app: Alberta, British Columbia, Nunavut, Yukon. So put these to the side. Also, we should consider the jurisdictional politics and cultural dynamics within government that led these provinces and territories to make this decision. Side-note — Alberta has its own app. Pinned for another day : ) Back to Covid Alert.
This means that the following provinces and territory did join the program:
Saskatchewan, Manitoba, Ontario, Quebec, New Brunswick, Newfoundland & Labrador, Prince Edward Island, Nova Scotia, and the Northwest Territories.
We know Ontario has had limited testing capacity since January of this year, but what about the other 7 provinces and 1 territory? Here’s what I found, corrections and more details of course welcome. In short: PCR testing capacity has been an issue across the board. Said another way: the brokenness is everywhere.
Current Status on PCR Testing Capacity
Saskatchewan: No general public capacity since Feb 2022 source
Manitoba: No general public capacity since Jan 2022 source
Ontario: No general public capacity since Jan 2022 source
Quebec: No general public capacity since Jan 2022 source
New Brunswick: No general public capacity since Jan 2022 source
Newfoundland & Labrador: No general public capacity since Mar 2022 source
Prince Edward Island: Reduced general public capacity since Mar 2022 source
Nova Scotia: Capacity reduced then returned Feb 2022 source
Northwest Territories: No general public capacity since Jan 2022 source
So let’s think about the dynamics at play here. All the provinces should know that the app is broken, but Health Canada is solely responsible for the communications around the app. Last December there appeared to be some friction and walk-back action of note in the communications realm between Newfoundland & Labrador’s Health Minister and the federal government.
History of the App — Timeline Pieces
The software code for the COVID Alert App was written by Shopify volunteers. They wrote an app called COVID Shield. The COVID Alert app is the official Canadian version of COVID Shield. The app’s starting point was also part of a rhetorical framing that took hold culturally, as Apple and Google jointly announced in their protocol: “Software developers are contributing by crafting technical tools to help combat the virus and save lives.”
The COVID Alert app was launched in the province of Ontario on July 31, 2020. It became available in the other seven provinces and one territory by October 2020.
If you look at the high-level data on the Download COVID Alert app page, the last update on app usage in Canada shows data for February 1, 2022, despite Health Canada saying the app and dashboard is currently supported. The number of downloads of the app, at February 1, 2022, is listed as 6,893,423. The number of one-time keys used is 57,704.
To Understand Data About Efficacy We Have to Understand the Process
If we’re going to talk about the efficacy of the app, we need to focus on both downloads and the one-time keys used — the two metrics above. Many that were critical of these apps before they launched talked about the challenge involved in getting a large enough percentage of the population using the app, if their phone could even support it. This challenge was primarily understood as downloads. On this count, the experiment failed — it did not attain critical mass.
Then beyond the download piece is the more specific part that makes the app “functional” — the part tied to PCR testing capacity: the positive PCR test results that were uploaded to the app so others could be notified of an exposure. These are reflected in the number of one-time keys used. In this case, as at Feb 1, just under 58,000 total.
To be clear: those one-time keys represent the number of cases where someone tested positive for COVID via a PCR test AND was provided a one-time key AND they uploaded that key to the app. I have not gone through this process, and don’t know anyone that has, so if you did, please do share your experience.
So by February 1 2022 about 58,000 people in the nine provinces/territories had taken that step to upload their one-time key to the app. One thing that features consistently in efficacy reporting about the app over time is the process that involves the test result being uploaded — the one-time key process. As this CTV piece reported in December 2021, “COVID Alert app still active, but rarely used to record positive tests.” The one-time key is what ties us back to testing capacity. Without testing capacity there is no one-time key. Without the one-time key the app is unable to alert others of the positive cases.
Understanding the Data to Understand Efficacy
To properly understand efficacy, we have to go further than understanding the poor uptake on the keys and the related process around getting them/using them. We’d need to understand what happened with the people that received an exposure notification.
I haven’t looked at the data that is available much at all yet, so I’ll wrap up here for today. If you have information or have seen studies or reporting on this part of the equation, please do share them. I’m just beginning to understand what is available on this front in the GitHub repositories for the app. I’m early in that digging around, so if anyone can lend a hand with looking at that data to think about efficacy, please let me know too.
Finally, the core reasons for this app needing to be shut down as described in the first part still stand. There is no more data required to make the shut-down decision. It’s the right thing for Health Canada to do and it should have been done already.
More as I have it, including some more replies to people that have chimed in with support for the app, or with prompts to consider, such as, why can’t we evolve the app rather than shut it down? Thanks for those and please keep them coming : )