Problem setting

Imagine you went to court fifty years ago to obtain a paper copy from a court clerk in a case of child abuse. Your purpose was to learn more about the case law on the subject. At that time, a human physically took a court decision from a record, copied it, and handed it over. Imagine now you go to court today with the same question and the same purpose. Chances are you will instead e-mail the court or visit its website and fill in your request. Most probably you will obtain an electronic form of a decision, e.g. a PDF. The requester can obtain a copy from thousands of court decisions on child abuse in a short period of time at a low cost, but it is highly likely that the copy is in its essence still conceived in the same way as a paper document in a physical register. Now consider that the courts understand the computational turn with the full potential and drawbacks of digital technologies for them. They store the decision in such a digital way that it affords different uses compared to the paper copy from a physical register. New technologies allow us to conceive of a court decision as an assemblage of (meta)datapoints, which can be split in a way a decision on paper cannot be split, and rejoined in totally different combinations. For instance, in thousands of decisions the requester can with the help of natural language processing and machine learning identify the type of injuries the child suffered and establish what the relationship is with e.g. the employment status of the parent(s). Also, the data about the measures taken by the juvenile court could be combined with the school results of the child during the ten years following the incident or with the medical data about the injuries. These are simplified examples; in reality the technologies enable the examination of cases across a myriad of dimensions.

New technologies are transforming the way law exists, which entails, among many others, the question of how to conceptualise the relationship between court (meta)data, case law, and the rule of law. Linda Mulcahy and Emma Rowden argue that ‘the design of invisible information highways has now become as much a component of court design as the architecture of physical spaces’ (p.330). Brazil is one of the few countries in the world where the courts have fully grasped this and adopted a ‘Resolution on ethics, transparency and governance in the production and use of AI within the Judiciary and other provisions’ (see also Gianmarco Gori’s interesting discussion of the Resolution). Important literature on these so-called information highways of the courts discusses questions about what data should be captured, how it should be stored, how it should be processed and linked to the data of other data ‘owners’, and who can access the data and for what purposes. Yet, this literature is based on the hidden assumption that court data are created, stored, and managed within the court system, whilst there are examples where this is clearly not how the court data are organised. For instance, in New Zealand the Ministry of Justice publishes court decisions, alongside NZLII, which is an initiative of universities and benevolent donors. Belgium has gone a step further: all court data with regard to insolvency procedures are captured, stored, and managed by the Bar Associations. Hence, the question arises as to which institution or person should create and maintain a court data infrastructure under the rule of law (as distinct from the important question of public access to those data).

Hereafter, I examine some of the requirements for the collection, storage, management, and use of courts’ (meta)data for adjudication under the rule of law. I argue that respect for the rule of law in a data- and code-driven society requires that the courts’ data-infrastructure is governed by the courts and should not be (co-)governed by other branches of government, the office of the public prosecutor, an institute for continued legal education, or the private sector. The first section discusses the role of adjudication in the rule of law. A definition of what the data infrastructure of the courts comprises follows in the second section. This analysis leads in the third section to a discussion of the court as a legal-technical ecosystem. The following section draws on the first three, and claims that a modern interpretation of the rule of law requires that the data infrastructure of the courts be created and managed by the judiciary.

Adjudication and the rule of law

An essential element of the rule of law is a division of government powers into the legislative, executive, and judicial branches, with a system of mutual checks and balances between them to prevent the concentration of all powers in one person or a small group of persons. The three branches are accountable to the members of society for their powers in different ways. The legislature has to be re-elected in a cycle of a number of years. In addition, the executive branch is accountable to the legislature, as Parliament can summon a minister and discuss her policies. Moreover, citizens can take the executive to court in order to enforce their rights vis-à-vis the executive. Finally, the court is a space of contestation; differences in views and opinions are translated into a legal format and a more or less formalised and ritualised discussion takes place. Thus, debate in court contributes to societies and their formation. The judiciary is accountable for its decisions through the reasoning expressed in its decisions, which have to be made within the boundaries set by the law. The judicial duty to give reasons (at p. 85) allows control of this exercise of judicial discretion in individual cases (at p. 1128), and allows society and parties in a dispute to contest a decision (at p. 67). Furthermore, the reasoned decision allows the embedding of the decision in society. Also, the (in principle) public character of the hearing and the decision makes it possible to scrutinise judges and courts and to hold them to account. Moreover, the judge has to adjudicate independently from the legislative and executive branches, as well as from members of society and business. The rule of law also requires the judge to decide impartially and assure a fair trial.

I claim that the purpose and function of these checks and balances remain essential in a modern society; new technologies should adapt to preserve the countervailing powers, not vice versa. In order to determine how code- and data-driven law should adapt to these important features of the rule of law, we must first decide how to conceptualise the data infrastructure of the courts in accordance with it.

Definition of data-infrastructure

For current purposes, data is defined as an item of information or related items of information taken together and used for e.g. calculation, reference, or analysis. Infrastructures in general are complex, heterogeneous assemblages; they are (immaterial and material) things and also relations between things, embedded in society, (legal) culture and economy. The courts’ data infrastructure comprises several elements:

  1. All elements contained in the procedural documents related to a specific court case.
  2. The datapoints in the submitted evidence in each specific case.
  3. The code to make legal text machine readable and to tag it with metadata.
  4. The code used by the courts to inform the public of their rights and obligations with an aim to promote ‘legal health’ and to contain and avoid disputes.
  5. The code developed to assist the judges in their decision-making tasks (In line with the cited Brazilian example cited above).
  6. The domain-specific best practices applied by judges.
  7. The relations between the courts and other instances, such as e.g. juvenile institutions and juvenile court, or databases with economic information and the business courts.
  8. The information relating to the organisation and the management of the court.

In sum, court data infrastructures constitute an entire set or collection of (meta)data points. In contrast with printed court reports, digital technologies afford sharing of (meta)data without e.g. geographical, temporal, or significant financial constraints. Moreover, the (meta)data can be split and recombined in new applications. The concept of a court data-infrastructure could be limited in the printed paper tradition to a register of court decisions, which should not be done for the reasons explained above. However, even then the decisions can be split in (meta)datapoints which can be recombined. With new technologies, the courts’ data-infrastructures are jurisgenerative in a different way than was the case with words printed on paper (at pp. 78, 104, 277, 307-326). They recalibrate adjudication and contribute to changing the relations between their human and non-human entities and activities, transforming courts into legal-technical ecosystems.

Data- and code-driven technologies cause courts to evolve into legal-technical ecosystems, in which humans and digital and computational technologies become interdependent. They mutually co-create each other. Legal-technical ecosystems are analogous to sociotechnical systems in which ethics, values, and interests are built into (digital and computational) technologies. The latter take shape through interactions with human subjects, whilst at the same time enabling or constraining ‘the potential behavioural outcomes in a particular context’. In such a legal-technical ecosystem, the court data infrastructure, of which court decisions form part, must be considered as a concept in its own right, with its own requirements. Following the computational turn, the existing concepts of a docket with files and of printed court documents held in a register no longer suffice to conceptualise court data and case law.

The Rule of Law requires that the organisation and management of the court data infrastructure is vested in the courts

Under the paper paradigm courts organised and managed their data infrastructure within the constraints established by the legislature. They ought to continue to do so, but in a way that is adapted to the digital paradigm. New technologies could facilitate a distributed management of the database of court decisions and by extension of the entire court data infrastructure. However, this technological possibility does not constitute a sufficient, let alone a legal, ground for the abandonment that would ensue of legal foundational principles such as the rule of law, fair trial, and the independence and impartiality of the human-combined-with-machine-judge. To the contrary, it follows from the above discussed concepts of (1) the role of the court in the exercise of countervailing powers in society, (2) the court data infrastructure, and from (3) the court as legal-technical ecosystem that it is in the remit of the court to organise its digital data infrastructure and to manage it. Since designing and developing data infrastructures is not neutral but requires choices which have normative implications, it undermines the independence of the judge to leave these tasks to other branches of government or to the private sector. Furthermore, to the extent a judge becomes dependent on the design choices of the executive or of private parties, her impartiality may be affected. Finally, the participation in the organisation and governance of the data infrastructure of the courts by the executive branch of government and by the public prosecution implies that at least two powers become joined, whereas they should be separated. This undermines the exercise of countervailing powers. Only to the extent that the legislature decides on the rules on civil and criminal procedure, should this branch of government indirectly intervene in the organisation and management of the court infrastructure. In other words, courts have an enduring constitutional duty to remain the forum for the adjudication of disputes. The fulfilment of this duty in a 21st century society requires that the courts design and manage their (meta)data infrastructure themselves.

Geneviève Vanderstichele
Judge in the court of appeal of Ghent (Be)
DPhil student at the University of Oxford (UK)

The author wishes to thank Dr. Elise Maes, Judge Dr. David Harvey, Prof. Dr. Mireille Hildebrandt and Dr. Laurence Diver for their valuable comments on earlier iterations of this blogpost