In a recent Substack post, the Canadian journalist Paul Wells described what it was that made major newspapers and television networks so powerful in their pre-internet iteration:
- Because they were effectively the only place for advertisers to get the word out about their products, media companies generated huge revenues.
- Because media companies had such rich sources of revenue, they could hire teams of talented journalists and researchers who were often dedicated to specific beats, like City Hall or Parliament.
- Such “embedded” teams were able to build relationships with powerful people and institutions that often led to valuable insights into the ways those institutions worked. And they had the time and resources to deeply understand their subjects.
- Because those media companies had reliably large audiences, their insights travelled far, potentially sharpening our understanding and oversight of public institutions.
For reasons beautifully described in Wells’ post (and the others that made up the series), all that’s long gone. But I wondered if we could, to some degree, replace that valuable oversight by leveraging publicly available data resources. This article is a proof of concept for how that might work.
I chose Canada’s Supreme Court because most of us know next to nothing about it. Its justices are hardly household names, how and when justices are appointed to the court is largely a mystery, and what they do with themselves from day to day is hidden.
And because the justices are called upon to provide a last-resort address for deciding how the rule of law will be applied to our lives - both personal and public. What these people do is important.
The hidden and mysterious part of it is certainly not by design. By law, the Supreme Court is required to make their deliberations and decisions available to the public. And this they most certainly have done, providing a website containing a comprehensive record of 146 years of their history.
With a little work (translation: a lot of work), I was able to programmatically pull data representing 631 Supreme Court judgments between 2013 and 2023. I then tortured the poor data using standard analytics tools until it had no choice but to reveal everything I asked of it. Feel free to take a look at the code I used in the process.
For the rest of this article, I’ll share my observations. I can’t necessarily explain the significance of everything I’ve seen, but I think the big picture I can offer does illustrate at least the potential for this kind of…I don’t know…journalism?
My first observation involves changes to career longevity. That is, the average time on the bench for justices who served in the 19th century was 15.5 years. For those who served in the 20th century, that was 13.7 years. And for justices who served (and completed their service) in the 21st century it was 11.6 years. For whatever reasons, justices are retiring relatively sooner than in the past.
For context, 18th Century US Supreme Court justices lasted an average of 9.8 years, 19th Century justices stuck it out for 17.8 years and, through the 20th Century, the average tenure was 16.6 years. I believe that hanging on until the bitter end is something far more common in the US than in Canada.
Another easy observation was that the annual unanimous decision rates suddenly dropped around 2016 and still haven’t fully recovered. Of course, there’s nothing systemically wrong with internally divided courts. Robust debate is, after all, a hallmark of open societies. But it’s the trend that’s intriguing. Here’s how that looks:
What made this happen? Was it changes to the kinds of cases making it to the high court, or was it the evolving makeup of the court itself?
Specifically, between 2013 and June 2023 there were 93 split decisions overall, of which 17 (18%) concerned constitutional law. I’m focusing on constitutional law because that’s where things are most likely to get political. For this type of analysis, it’s the possibility that the court has become politicized that I think should most concern us.
To dig a bit deeper, I identified the justices appointed by Canada’s most recent two prime ministers, the Liberal Justin Trudeau, and the Conservative Stephen Harper. Here’s the breakdown:
- The Hon. Michelle O’Bonsawin
- The Hon. Mahmud Jamal
- The Hon. Nicholas Kasirer
- The Hon. Sheilah L. Martin
- The Hon. Malcolm Rowe
- The Hon. Russell Brown
- The Hon. Suzanne Côté
- The Hon. Clément Gascon
- The Hon. Richard Wagner
- The Hon. Michael J. Moldaver
- The Hon. Andromache Karakatsanis
- The Hon. Thomas Albert Cromwell
In an ideal world, all cases would be decided purely on their legal merits. There might be philosophical positions shared by some but not all justices, but the doctrinal differences should have a relatively mild on total case loads. However, if we discover a persistent pattern of conflict following along strictly party lines, then we’d be justified in suspecting that politics is playing a role. That’s not necessarily a evil thing, but we’d certainly want to know about it.
Here’s how I went about examining the data. I filtered all non-unanimous cases (whether the total set of cases, or just those concerning constitutional law) for the names cited in dissent. If none of the dissenting justices were from the list of Harper appointees, then I “concluded” that only Trudeau appointees disagreed and the judgment leaned to the right. If there were no Trudeau appointees in the dissent, then I assumed the decision leaned politically leftwards.
Based on that (obviously imperfect) methodology, I found that 61 of the 93 split decisions were “Liberal-friendly”, and just four were supported by “Conservative” justices. Of the 17 contested cases relating to the constitution, eight leaned left and only one leaned right.
I should note that those numbers included a few ambiguous cases where the dissent was voiced exclusively by justices appointed before Prime Minister Harper’s term. But the numbers wouldn’t have much impact on the big picture we’re looking at.
What we do see, however, is that dissent on the court does often follow “party” lines - as many as nine of the seventeen constitution-related split decisions seemed to follow that pattern. That’s certainly not the same as suggesting that the court is politicized, but it does offer us direction on where to look next.
And where is that? Well here are the case numbers for those eight “liberal-leaning” contested constitutional cases:
- 38663, 38781, 39116
And here’s the one that “leaned right”:
Head over to the Supreme Court site, look up the case descriptions, and decide for yourself.
What’s next? Wouldn’t it be fun to get your favourite large language model (like GPT) to digest and summarize some of this data at scale to look for more patterns? Believe me: that can be a lot easier - and productive - than you might think.