3.1.4 Anticipating legal protection under data- and code-driven normativity

By Mireille Hildebrandt

Anticipating how legal protection may be transformed in the case of data- and code-driven normativity we need to be in the clear about what those terms mean. This is the subject of the second and third working papers. Here we provide a preliminary working definition:

A. data-driven normativity in law refers to the integration of machine learning approaches, such as prediction of judgments and search and optimisation methods into the practice and the study of law, leaving in the middle whether and if so under what conditions such ‘law’ actually qualifies as law

B. code-driven normativity in law originally referred to the integration of smart contracts and smart regulation (both of them based on blockchain technologies) into the practice of law, again leaving in the middle whether and if so under what conditions such ‘law’ actually qualifies as law. We have extended our research to other types of code-driven law, such as logic- and knowledge-based approaches, notably including ‘rules as code’ and ‘interoperable digital law’.

Based on the previous analysis we raise the following questions that should feed into the second and third COHUBICOL working paper:

1. What kind of protection, afforded by text-driven legal norms, may be diminished or transformed in the case of data-driven ‘law’?
2. What kind of protection, afforded by text-driven legal norms, may be diminished or transformed in the case of code-driven ‘law’?
3. To what extent and in what sense could ‘protection by design’ be qualified as articulating legal norms?
4. Does ‘legal protection by design’ mean that the legal norm itself is implemented in the architecture of the system, implying something like ‘legal by design’?
5. Could computing systems produce digital legal speech acts that express legal norms that in turn result in institutional legal facts such as a transfer of ownership?
6. If legal norms are written in both natural language and in computer code, which should prevail if the interpretation of the natural language differs from the interpretation developed in the code?
7. How can those subject to law contest the interpretation of the natural language if it was the legislature that enacted both the natural language and the code version of the relevant legal norm?
8. What may be the consequences of attributing legal effect to self-executing code or to legal norms written in code?
9. What does is mean for the mode of existence of legal norms if legal effect were to be given to predictions of judgment by data-driven systems?
10. How could computational legal search, based on ‘natural language processing’, affect the development of legal norms?

---