Finding the “COVID Boundary” in CanLII Usage Statistics 🦖

I’m no palaeontologist, but being the father of a 7-year-old boy, I talk about dinosaurs more than the average human. My son doesn’t necessarily have that of a great interest in our prehistoric friends themselves, but his thirst for knowledge about the circumstances of their disappearance is seemingly insatiable.

Those father-son discussions about the extinction of dinosaurs led me down a Wikipedia rabbit hole only to discover the existence of such a thing as the Cretaceous–Paleogene boundary. This boundary is a line of dark dust with high concentrations of iridium, which, according to the leading hypothesis, came from the dinosaur-killing asteroid, and spread all over the world upon impact (even if the impact crater is in Mexico, the boundary can be found in rock as far as Europe).

I had a shower thought recently that if you can find a literal boundary in rock that marks the disappearance of dinosaurs thousands of kilometres away from the asteroid impact, you can find indices of past catastrophic events anywhere.

So I wondered if “data palaeontologists” from thousands of years from now (I suppose I could just call them historians but it’s less fun), given CanLII usage statistics and nothing else, would be able to guess what happened to us in 2020. In other words, can we see the “COVID Boundary” in our stats?

With apologies for the over dramatic introduction, here are a few questions that I asked myself, and their answers.

(Please be assured that when looking at CanLII usage statistics, that I only look at aggregate data that contain no personally identifiable information.)

Q. Can we find a “COVID Boundary” in usage statistics?

Yes, pretty clearly. Here’s how we can observe it most vividly:

If you take every week of 2020 so far and compare them to the same week in 2019 (week 1 2020 v. week 1 2019, week 2 2020 v. week 2 2019), you can see the most significant week-for-week drop in relative traffic happened on the week of March 16, 2020:

That 0.47% drop is more than two times the standard deviation for the population (0.18%). Only 4 other weeks in the year have a drop higher than the standard deviation. All of them are the weeks immediately following the start of the lockdown except the week of September 7, 2020 when I attribute the size of the drop to the fact that Labour Day fell during week 37 in 2020, but happened during week 36 in 2019.

Traffic climbed back up pretty steadily after this. So far we are up almost 14% for total visits in the year compared to 2019. Traffic had increased by about 21% in 2019 versus 2018, and we were on a pace to match this increase in 2020 had it not been for COVID. Still, October 2020 was our best month ever in terms of total visits. As Dr. Ian Malcolm would say, life, uh, finds a way.

Since the week of November 2 appears in the “biggest drop” rankings, I guess y’all watched a bit more CNN, and did a bit less CanLII searching, during that week didn’t you?

Q. Have there been changes in the most popular statutes or regulations?

Readers of the “What’s Hot on CanLII” column know that we spend a bit of time each week looking at the most popular cases on the site. However, we don’t usually spend that much time looking at top legislation rankings (I include regulations here), probably because we assume that they are pretty set in stone with the Criminal Code, Charter, Civil Code and other “mega statutes” (my expression) leading every month.

Still, in my mission to assist “data paleontologists” of the future, I was curious to see if COVID meant that there would be material changes in the 2020 rankings versus 2019, and if these changes allow us to tell a coherent story. The answer is yes.

I created 3 lists:

● List 1: Top movers among statutes and regulations that feature in both the 2019 and 2020 Top 1,000
● List 2: Highest ranked newcomers in the 2020 Top 1,000
● List 3 Highest ranked statutes and regulations in 2019 that dropped out of the 2020 Top 1,000


● The Companies’ Creditors Arrangement Act (aka CCAA) is unfortunately on top. An expected top mover no doubt considering how COVID impacted large Canadian retailers. Interestingly, the other statutory horse of the business apocalypse, the Bankruptcy and Insolvency Act, isn’t in the top movers (it moved down one non-material spot).

Ontario’s Police Services Act is the second-biggest mover. Not related to COVID, but certainly a reflection of the conversation on “defunding the police” this summer that included takes like this.

● British Columbia’s Insurance (Vehicle) Regulation completes the podium, no doubt as a result of the strong winds of change in BC around vehicle insurance described in this CRT explainer.

● The rise in popularity of the Patent Act and Medical Device Regulations is likely related to measures implemented in reaction to COVID including the amendments to the Patent Act that allowed governments to use patented inventions in their fight against COVID.


● This is 2020 in a nutshell. Canadians have been wondering about emergency management and public health, protecting the brand of the new online services they were launching, trying to get plane tickets reimbursed… and worrying about contractual obligations not being fulfilled, the ventilation system in their building, the safety of their work environment, and their matrimonial situation.

● The Riparian Areas Protection Regulation is perhaps the head-scratcher here, but it’s always good to see that, pandemic or not, the good folks of British Columbia are taking steps to protect their population of delicious Pacific salmon.


● I suspect that if it was a musical artist and these rankings were the Billboard, the Elections Act would have been secretly hoping for the minority government to fall in order to maintain its standing in the 2020 rankings, and its wish came pretty close to fruition! Let’s see if it makes a spectacular comeback in 2021.

● It also seems like our users are slowly getting to the end of the list of new questions from their clients with respect to cannabis regulations.

Q. Have there been changes in the technology used to access the site?

A little thing I noticed is that for the first time in… forever (with apologies to fellow parents and kids at heart for the earworm infection), Internet Explorer won’t finish the year in second place of the most popular browser rankings. As I tweeted a few weeks ago, I suspect this is because work desktop computers tend (sadly) to have legacy software.

Since many people haven’t touched their work keyboard in months, Internet Explorer is roughly in the same predicament as these abandoned office plants. I have much more sympathy for the plants than the outdated and insecure browser though.

Q. Are there noticeable changes in the top search queries?

Yes, and this is the icing on the cake of this post.

The COVID Boundary can be seen clearly from this animation generated from the weekly top queries on CanLII since the start of the year. Warning: you might not like that you have reached a point in your life where you are rooting for bars in a graph.

The animation is a vivid picture of the events of the year (unfortunately, tragic ones are overrepresented) and shows how they interrupt the usual rotation of top queries.

I put the COVID-related queries in bright red to distinguish them from the rest. Sit back, relax, and watch for the appearance of “force majeure” as a top query in late February.


  1. This is fascinating. I know that internet search data has been used to predict outbreaks, allowing quicker and more targeted public health responses. Has this sort of CanLII usage data been used by firms or other justice system actors to plan services/products?

  2. Hi Noel,

    Thanks for your comment.

    I can be wrong, but I’m not confident those stats have predictive value, at least not in a way that would allow us to spot new needs and trends before they become obvious to everybody else.

    It’s a little bit like the story (statistically true or not) about negative reviews on scented candles being correlated with the number of infections: It’s fun to imagine that this could have given us a clue that a new weird illness was around before the virus was first identified, but who would ever have predicted that this is what negative reviews of scented candles meant before hearing about the infections in Wuhan? We could of course monitor this data point going forward, but I’m sure there better early indicators of upcoming waves of infections than bad reviews of scented candles.

    That “force majeure” trended as a keyword before the start of the lockdowns is interesting, but there was also a lot of law firm newsletters that covered this topic early in the pandemic when COVID was seen more as a threat to the supply of goods coming from China than an actual health threat.

    Sarah has been able to find opportunities to develop content matching queries that had a high bounce rate, but I’m not sure this can be considered as predictive.

    We also tend to thread carefully with respect to the usage stats given that one of the principles of the Montreal Declaration on Free Access to Law is that the LIIs should provide anonymous access to their content.

    But maybe this is all a lack of imagination on my part.