Judicial Analytics: Facing Uncomfortable Truths

For as long as there have been judges, people have tried to predict judges’ decisions. In so doing, they have always understood that judges are human beings. They are not calculators from an assembly line, each of which will display the same result if one punches in the same inputs. Thus, at any watering hole where litigators gather, it will be overheard that “Justice Smith comes down hard on drug offenders,” or “If your client has soft-tissue injuries, you had better hope that you don’t draw Justice Jones for the trial.”

In France, it would seem that such conversations are now illegal. Loi no 2019-222 du 23 mars 2019 forbids the use of information about the identity of judges in order to evaluate, analyze, compare or predict their actual or expected decisions.[1] The Assemblée nationale was worried by the emergence of artificial intelligence applications which crunch data about prior judicial decisions, and use it to predict decisions. An excellent forthcoming Dalhousie Jaw Journal article by Amy Salyzyn and Jena McGill argues persuasively against this ban. Salyzyn & McGill also identifying the challenges and opportunities posed by the emergence of mainstream “judicial analytics” technology.

France’s ban on judicial analytics is a bad idea, but there is a noble sentiment behind it. People aspire to live under the “Rule of Law,” and not the “Rule of People.” We want, and we should want, the outcome of legal disputes to depend exclusively on the facts and the law, and not on what individuals happen to adjudicate those disputes.

Undeniably, however, there are patterns in decisions which can only mean that the identity of the judges affects the outcomes. To take a Canadian example, Sean Rehaag has demonstrated that the outcome of a refugee case before the Federal Court of Appeal depends significantly on which judge happens to hear it. Modern judicial analytics simply brings these patterns to light more consistently and quickly than lawyers’ shop-talk did in previous generations.

In the words of Friedrich Hayek, “the law cannot rule. Only men can exercise power over other men.”[2] Much as we might wish it to be so, it cannot be reasonably maintained that the law itself always makes the all the decisions, with judges merely serving as its neutral mouthpieces. France’s ban on judicial analytics is an effort to suppress this uncomfortable truth and preserve what is, at best, a noble lie. It is incompatible with the freedom of speech, and the accountability of public officials.

In addition to being misguided, the effort is probably futile in this day and age. If there is demand for judicial analytics in France, then someone will meet that demand from outside the country’s borders. We would all do better to face the uncomfortable facts that judicial analytics reveals about the nature of judging. With eyes wide open, we can work to design a justice system which, as much as possible, delivers consistent and fair results to everyone.


[1] “ Les données d’identité des magistrats et des membres du greffe ne peuvent faire l’objet d’une réutilisation ayant pour objet ou pour effet d’évaluer, d’analyser, de comparer ou de prédire leurs pratiques professionnelles réelles ou supposées.”

[2] The Constitution of Liberty.


  1. This is of interest to me because of a follow-up question.

    I have some recollection of reading somewhere that Canadian counsel are sometimes able to find ways to avoid dealing with a particular judge on a particular matter. I don’t know if that is true.

    Speaking as a self-represented litigant I can say that it is not uncommon for SRLs to develop and express opinions about certain judges. If, as we are expected to believe, no party ever has a voice in determining which judge(s) a matter will be assigned to, can we be sure that the case assignment process itself is never subject to any bias? In partial answer to that question I note that, from what I’ve seen, certain high profile (I mean especially politically sensitive) cases tend to get assigned to certain judges.

    If there is a basic premise that anyone appointed to a bench is, from the day of that appointment, considered to be qualified to hear any matter, then the case assignment protocol should not allow for assignments that would tend to suggest a bias.

    Years ago I asked the Chief and Associate Chief Justices of the BC Supreme Court for an account of how they assigned cases. The reply I received was concluded with, “Neither parties nor counsel may choose the judge who presides in a given case; rather the assignment of judges to particular cases is an important aspect of judicial independence and information relating to case assignment and the scheduling of the judges’ rotas is confidential to the judiciary”. Right after that I and the person I was trying to assist found ourselves in a courtroom presided over by the one judge I had named in my letter asking about case assignment. I had not of course suggested who I wanted to see preside over that case, but rather had named a judge I particularly did not want the case to be assigned to.

    I note as well that some parties have had the experience of expecting to be in front of a particular judge and then finding themselves in front of someone else. That’s justified by the exigencies of running a courthouse. Well running a courthouse and seeing that scheduled hearings take place as scheduled is surely not much different than what many businesses have to do in managing resources. Businesses with competitors might understandably want to keep their resource management methodologies to themselves. But the courts have no competitors and they are paid for out of the public purse. So I think the secrecy is not justified.