Bottleneck or Crossroad? Problems of Legal Sources Annotation and Some Theoretical Thoughts

So far, in the application of legal analytics to legal sources, the substantive legal knowledge employed by computational models has had to be extracted manually from legal sources. This is the bottleneck, described in the literature. The paper is an exploration of this obstacle, with a focus on quantitative legal prediction. The authors review the most important studies about quantitative legal prediction published in recent years and systematize the issue by dividing them in text-based approaches, metadata-based approaches, and mixed approaches to prediction. Then, they focus on the main theoretical issues, such as the relationship between legal prediction and certainty of law, isomorphism, the interaction between textual sources, information, representation, and models. The metaphor of a crossroad shows a descriptive utility both for the aspects inside the bottleneck and, surprisingly, for the wider scenario. In order to have an impact on the legal profession, the test bench for legal quantitative prediction is the analysis of case law from the lower courts. Finally, the authors outline a possible development in the Artificial Intelligence (henceforth AI) applied to ordinary judicial activity, in general and especially in Italy, stressing the opportunity the huge amount of data accumulated before lower courts in the online trials offers.