A new experiment in crowdsourcing the law: flagging negative treatment on CanLII Connects

Posted on Updated on

Repost with Permission from The CanLII Blog

In this post from last year about new changes to CanLII Connects, we announced that our users can now mark cases as having received negative treatment, or as containing negative treatment about another case. Over the last year we have tweaked that feature based on user behaviour.

What does this mean exactly?

Just to get everybody up to speed, CanLII Connects is a companion site to CanLII.org where users can post summaries or commentaries about cases in the CanLII collection. Each decision in our collection has a page where our users can post comments or summaries. If you have never visited CanLII Connects, you can see it here, and we invite you to look for the numerous links and visual clues on CanLII.org.

Let’s explain this new feature with an example: when an important case like (in 2015) Carter v. Canada (Attorney General) is published, CanLII Connects (and the legal web in Canada, more generally) erupts with a storm of commentaries and summaries. One thing that our contributors didn’t fail to point out is that Carter was a departure from the SCC’s 1993 decision in Rodriguez v. British Columbia (Attorney General).

The following posts, among others, discussed the change from Rodriguez to Carter explicitly:

However, neither of these two commentaries were cross-posted as comments on the Rodrigez case’s page (just to be clear: contributors can do this, but we don’t expect it), where they are, in a sense, more necessary.

Indeed, imagine Alice, an hypothetical legal researcher who is absolutely new to the issue of assisted suicide and has no knowledge at all of the history or state of the law about this issue. Imagine that Alice starts researching today about assisted suicide, and for some reason stumbles on the Rodrigez case, but not Carter.

At the same time, Bob, another hypothetical legal researcher is also searching on assisted suicide (and is also completely new to this field of law) and finds Carter but not Rodriguez.

Who would benefit the most from a CanLII Connects post that informs the reader that Carter is a departure from Rodriguez? Alice of course, because she only found the older case and could certainly use a little heads-up about the more recent one. In contrast, by stumbling on Carter instead of Rodriguez, Bob has the full story since Carter refers to Rodriguez.

Theoretically, CanLII Connects contributors could publish the same post in both places (as a comment on Carter and as a comment on Rodriguez), but it may not be the case that the discussion of an earlier case is that central to the more recent one (there are 44 occurrences of the word “Rodriguez” in Carter).

With this new feature, CanLII Connects users can now use a new and streamlined form to “flag” cases in our collection as having received negative treatment in another case (and vice versa), as illustrated by this GIF:

When a case is flagged (as I did with the Rodriguez case) there’s a mention of that next to the case’s name in the case’s page on CanLII Connects:

Note that you don’t need to be publishing content to CanLII Connects in order to “flag” cases for negative treatment. Any CanLII Connects registered user can flag cases without writing about them as this is a separate form. If you are not a CanLII Connects registered user yet because you don’t want to publish summaries or commentaries (or lack the time), you now have a simpler way to contributing to the community, so don’t hesitate to get an account!

That being said if you are a regular CanLII Connects contributor, and you are posting a summary of comment about a case that is overturning (or, more generally, negatively treating) an earlier one, the form for posting your comment/summary now has a place to tell us just that, and we would be grateful if you could use it.

What types of negative treatment can be flagged?

Judging only from the nomenclature in the Supreme Court headers, cases can be reversed, overturned, overruled, distinguished, not followed, disapproved, criticized etc.

While it’s easy to determine that a case in the lower courts has been reversed by an appeal court in the same matter (and still, a lower court decision may only have been reversed in part), qualifying the type of negative treatment between two unrelated cases (where the older case’s applicability as a precedent is under consideration in the later case) is highly subjective. Carter and Rodriguez are good examples of this subjectivity: Carter is pretty much a complete turnaround on the assisted suicide issue, yet the headnote in Carter lists Rodriguez as having been distinguished (which seems like the Supreme Court’s way of saying that Rodriguez wasn’t wrong and is not to be overturned, but that there are changes “in the circumstances or evidence that fundamentally shifts the parameters of the debate” that justify reaching an entirely different conclusion on the central issue).

Therefore, our form does not ask you to qualify the type of negative treatment the earlier case received. We do ask for an explanation in the description field of the form (which could be as simple as pasting the relevant paragraph in the more recent case, as we did in the above GIF demo), but we won’t publish this information: it’s just to help us make sure that flags are valid. In case you are of the shy type: other users won’t see who indicated that a case received negative treatment, but we may one day create a hall of fame of those contributors who have flagged the most content and include you, with your consent.

In summary, we ask that users tell us when a case received negative treatment of any sort in a later case (or vice versa) as a heads up to our users that the earlier case should not be relied on without at least reading the more recent decision.

Where are we going with this?

We know that this feature (or perhaps we should say “experiment” really) will not spontaneously generate a comprehensive database of all cases having received negative treatment in the CanLII collection, but we hope to be able to gather enough references over time to make it a useful tool, or to help us build useful tools (e.g. if we use the results from this crowdsourcing experiment as a sample to a machine learning experiment). We’re also experimenting with ways to complement the information we obtain from CanLII Connects contributors with programmatic means of finding negative treatment. One day, indications of negative treatment may appear on CanLII.org.

In conclusion, if you stumble upon case that should be “flagged”, please “pay it forward” and tell us about it. Your fellow CanLII users, as well as members of the public, could greatly benefit from the exercise in the long run. More selfishly, just see this as a way to help your future self in the event that you (or someone in your team) stumble again on the same “negatively treated” case.

 

Tell us what you think

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s