What About the Precedent: An Information-Theoretic Analysis of Common Law

Abstract

In common law, the outcome of a new case is determined mostly by precedent cases, rather than by existing statutes. However, how exactly does the precedent influence the outcome of a new case? Answering this question is crucial for guaranteeing fair and consistent judicial decision-making. We are the first to approach this question computationally by comparing two longstanding jurisprudential views; Halsbury's, who believes that the arguments of the precedent are the main determinant of the outcome, and Goodhart's, who believes that what matters most is the precedent's facts. We base our study on the corpus of legal cases from the European Court of Human Rights (ECtHR), which allows us to access not only the case itself, but also cases cited in the judges' arguments (i.e. the precedent cases). Taking an information-theoretic view, and modeling the question as a case outcome classification task, we find that the precedent's arguments share 0.38 nats of information with the case's outcome, whereas precedent's facts only share 0.18 nats of information (i.e., 58% less); suggesting Halsbury's view may be more accurate in this specific court. We found however in a qualitative analysis that there are specific statues where Goodhart's view dominates, and present some evidence these are the ones where the legal concept at hand is less straightforward.

On the Role of Negative Precedent in Legal Outcome Prediction

“…To find a better explanation of the performance asymmetry, we now turn in our discussion to the legal perspective. In precedential jurisdictions, of which ECtHR is one (Zupancic, 2016;Lupu and Voeten, 2010;Valvoda et al, 2021), the decisions of a case are binding on future decisions of the court. Two cases with the same facts should therefore arrive at the same outcome.…”

“…We modify the ECtHR corpus, a popular legal NLP dataset, for our experiments. Specifically, we use the scrape of the corpus from Valvoda et al (2021). An ECtHR case can be separated into two sections: facts and arguments.…”

“…While legal outcome prediction models are interesting in the context of academic research (Valvoda et al, 2021), any real-world deployment of Legal AI necessitates a certain level of understanding of how the system works (Valvoda et al, 2023b). After all, it is the reasoning towards the outcome of the case, rather than the outcome itself, that is of tantamount importance in legal outcome prediction.…”

“…While ECtHR interacts with civil law jurisdictions, its judges rely on precedent(Valvoda et al, 2021).9 See Appendix B for examples from our dataset or HUDOC for all the ECHR caselaw.…”

LexGLUE: A Benchmark Dataset for Legal Language Understanding in English

“…The rapid growth of the legal text processing field is demonstrated by numerous papers presented in top-tier conferences in NLP and artificial intelligence (Luo et al, 2017;Zhong et al, 2018;Chalkidis et al, 2019a;Valvoda et al, 2021) as well as surveys (Chalkidis and Kampas, 2018;Zhong et al, 2020b;. Moreover, specialized workshops on NLP for legal text (Aletras et al, 2019;Di Fatta et al, 2020;Aletras et al, 2020) are regularly organized.…”