Column

Vavilov & Dunsmuir: Looking for Signals in the Precedential Game of Thrones

As we all know, administrative law nerds (their own expression) received a nice Holiday present on December 19 when the Supreme Court issued its long-awaited decision in Vavilov. This important event in Canadian law isn’t just fun for the admin law crowd, it’s also an occasion for us, legal information geeks, to live in real time another game of “precedential game of thrones.”

I’ve been interested for a long time in finding signals that could indicate that a case is no longer good law (or at least no longer to be cited without caution). There are ways to spot those using the language of a superseding case (and then again, it’s not trivial), but I’ve always been curious about other types of signals, including variation in citation counts.

Watershed cases, like those that set a new standard of review, are interesting because of the volume of citations they generate. It’s hard to find meaningful statistical signals looking at citations for a decision that is cited only a handful of times every year, but when looking at cases destined to accumulate thousands of citations, it’s easier to spot interesting patterns. To help anticipate what will happen with Vavilov, let’s look at what happened when Dunsmuir was issued in 2008.

While Dunsmuir is “only” the third most cited case in the CanLII collection, the two cases above Dunsmuir in the “most cited ever” rankings (Gray and Crane) are cited more than 30,000 times each, but almost exclusively by the same tribunal (e.g. Gray is cited 30,963 times, but only 49 times by courts). So Dunsmuir is definitely sitting on the throne as the case having the most impact on later court decisions (cited 16,732 times total, including 14,330 times by courts).

Vavilov will no doubt be a worthy challenger to Dunsmuir’s throne but there’s no certainty that it will ever beat Dunsmuir’s total citations. First, Dunsmuir has a spectacular head start, and second, Dunsmuir will continue to be cited, despite memes suggesting otherwise (the rise of the legal meme culture in law schools would probably be worth a post some day). The latter is clear from one of the first cases citing Vavilov, in which Justice Graesser of the Alberta Court of Queen’s Bench wrote:

“As for the reasonableness standard applicable to this application, assistance from cases such as Dunsmuir and Newfoundland Nurses and Alberta Privacy Commissioner v ATA still offer guidance in assessing the reasonableness of a particular tribunal’s decision under the newly-articulated standard.”

Peter v Public Health Appeal Board of Alberta, 2019 ABQB 989 (CanLII)

Considering this, Vavilov will have to stand as the leading case on standard of review for a very long time to eventually top Dunsmuir’s total citations, and I don’t know enough about admin law to opine on whether it stands on solid enough grounds in law (and, I should say, ease of application) to accomplish that.

If you’ve never been a fan of Dunsmuir and are desperately looking for something to root for in your life (but perhaps you should instead look for a hobby), you can root for Vavilov to be quicker than Dunsmuir at reaching 500 citations. It took Dunsmuir only 114 days to reach this plateau, which is quite the feat considering that only about 0.2% of all cases reach 500 citations in their “lives”. The ubiquitous R v. Jordan, in comparison, took 402 days to reach that milestone.

Where it gets interesting is when you look at the impact of new decisions on the citation frequency of the decisions it “replaces”. Before Dunsmuir, leading cases on standard of review included Southam, CUPE v. NB Liquor Board and Pushpanathan. The impact of Dunsmuir on the citation frequency of these cases was instantly felt (post-Dunsmuir numbers are in the yellow shaded section):

[Click image to see the larger graphic.]

There’s a clear drop from the moment that Dunsmuir is issued for both Southam and Pushpanathan. CUPE is an older decision, having been issued in 1979 while the other two are from the late 90s. The post-Dunsmuir drop for CUPE is harder to see from the graph but it was cited about 38 times per year on average from 1998 to 2007 and that dropped to 30 times post-Dunsmuir. So a drop there is, but probably not something that’s statistically significant.

Dunsmuir was issued on March 7, 2008, so there were two months in that year where decisions issued by courts were unaffected by the jurisprudential Thanos that was to come and “snap” their citation numbers. Just for fun, let’s zoom in and check if the pre/post trends can be observed on a monthly basis. Taking Pushpanathan as an example, we can see how pretty instantaneous the Dunsmuir effect was (post Dunsmuir is in blue shade):

[Click image to see the larger graphic.]

Sure, there’s a little spike in May and June (we’re dealing with smaller numbers here than when looking at yearly stats, so some variance is to be expected) but 4 of the last 5 months of 2008 have single digit citation counts, when it never dipped below 15 citations prior to Dunsmuir. Also note that the citation counts during the May-June “spike” are lower than the 5 months prior to Dunsmuir (in other words, the spike wouldn’t look like a spike if there wasn’t such a big drop in March).

Just as we’re likely to say about Dunsmuir in 20 years, Southam, CUPE and Pushpanathan are still cited to this day. Indeed, 2019 is not on the first graph because there’s still decisions coming our way from 2019 at this time of the year, but so far they were cited respectively 61, 14 and 22 times. Accordingly, if our heuristic-in-the-making to identify decisions that are not the freshest law was to wait for them not to be cited at all, we wouldn’t get far.

Let the games begin!

Comments

  1. I’m not a lawyer and therefore not surprisingly I won’t call myself an administrative law nerd. But the game that includes Dunsmuir and now Vavilov is of interest to me.

    My opinion is one with which I doubt any lawyer will openly agree. I think the entire administrative law standard(s) of review story is a Shakespearean tragedy. I’d like to think there will be a final act and scene and that I’ll be around to witness it.

    I’ve made use of CanLII and other information technology tools in only the most rudimentary ways, but I’d like to suggest to the author, given his comments about precedents, that he take a look at two decisions issued by a B.C. tribunal, each of which was written to be cited. The first one was written in 1994 and the second one in 2003, and they both recite a section of B.C.’s Labour Relations Code. Sometime between 1994 and 2003 that section changed (though the legislature did not revisit it after the initial debate in 1992). Two instances – in two successive lines – of a hybrid Latin / English legal term with which all lawyers are familiar were replaced by ostensibly “plain English” words.

    Search CanLII on “Terry Norris”, the decision issued in 1994, on “James W.D. Judd”, the 2003 decision, and on both of them together. The Norris “reconsideration” decision itself is not on CanLII because the CanLII database for the BCLRB decisions only goes back to the year 2000. The BCLRB’s database of decisions goes back to 1990, so the Norris decision, B156/94 is found here – http://www.lrb.bc.ca/decisions/B156$1994.pdf.

    The 2003 Judd decision makes not mention of the Norris reconsideration decision (though one of the three authors on each decision is the same person). But read carefully what paragraph 99 of Judd asserts about the intention of the legislature. What constitutes legislative intent and how adjudicators can decipher it has been the subject of much debate. But if there is a decipherable intent for section 13 then surely it is found in the Hansard record from November 26, 1992 – https://www.leg.bc.ca/documents-data/debate-transcripts/35th-parliament/1st-session/19921126pm-Hansard-v6n24. The text of the debate on that one section, with the hybrid Latin / English highlighted, is found here – http://www.uncharted.ca/images/users/ssigurdur/hansard_on_sect_13_2.pdf.

Leave a Reply

(Your email address will not be published or distributed)