Skip to main content

Counting as a Human Being in the Era of Computational Law

Periodic Reporting for period 2 - CoHuBiCoL (Counting as a Human Being in the Era of Computational Law)

Reporting period: 2020-07-01 to 2021-12-31

This project investigates how the prominence of counting and computation transforms the assumptions, operations and outcomes of the law. It targets two types of computational ‘law’ that both involve AI systems in the sense of the EU AI Act (art. 3.1 jo Annex I): data-driven systems (e.g. machine learning), and code-driven systems (e.g. code as rules initiatives). We put ‘law’ between inverted commas to highlight that one of the research questions is whether the integration of these systems into legal practice should result in qualifying the output of such systems as law, giving them legal effect. To understand and address the transformations this will entail, modern positive law will be analysed as text-driven law, followed by a comparative analysis of text-, data- and code-driven normativity. The overarching goal is to develop a new hermeneutics for computational ‘law’, the intermediate goals are an in-depth assessment of the nature of legal protection in text-driven law, and of the potential for legal protection in data- and code-driven ‘law’. The new hermeneutics will enable a new practice of interpretation on the cusp of law and computer science. The research methodology is based on legal theory and philosophy of law in close interaction with the theory of computer science, integrating key insights in both the affordances and limits of relevant AI systems into legal theory and legal methodology, thus achieving a pivotal innovation of legal method.
We have

- STUDIED THE THEORETICAL BACKGROUND that will enable lawyers to identify key assumptions of text-driven normativity: interpretation theory, speech act theory, postphenomenological philosophy of technology, theory of language and action and of rules and norms and developed a dedicated bibliography that underpins our inquiry (https://www.cohubicol.com/about/bibliography/);

- CONDUCTED AN IN-DEPTH STUDY OF A SET OF 10 FOUNDATIONAL LEGAL CONCEPTS that are constitutive of both the vocabulary and the grammar of modern positive law, resulting in A DEDICATED ‘LEGAL VOCAB’ (https://publications.cohubicol.com/vocabularies/law/);

- PUBLISHED THE FIRST COHUBICOL WORKING PAPER ON ‘TEXT-DRIVEN NORMATIVITY AND LEGAL PROTECTION’ (https://publications.cohubicol.com/working-papers/text-driven-normativity/) which demonstrates how the framing concepts of the project (mode of existence, affordance and legal protection by design) contribute to a better understanding of how the text-driven normativity of current law enables legal protection;

- ENGAGED IN GROUND BREAKING COLLABORATION BETWEEN THE LEGAL AND THE CS TEAM, centred around the core concepts of machine learning, kicking-off the development of the CS vocab, that will feed in the second and the third working papers (on code- and data-driven normativity and legal protection);

- PUBLISHED THE FIRST EVER DEDICATED TEXTBOOK ‘LAW FOR COMPUTER SCIENTISTS AND OTHER FOLK’, which is exemplary for COHUBICOL’S refusal to infantilize knowledge shared across disciplinary borders, offering the first comprehensive in-depth and nevertheless accessible textbook of law for CS, introducing law at both the foundational and the practical level (https://www.cohubicol.com/about/publications/law-for-computer-scientists-and-other-folk/);

- PUBLISHED AN IN-DEPTH INQUIRY INTO THE CROSS-OVER BETWEEN THE PHILOSOPHY OF TECHNOLOGY AND LEGAL THEORY. Digisprudence: Code as Law Rebooted, which extends the treatment of affordance in legal theory scholarship, proposing an ecological view of legality that necessitates certain design standards be met by any code- or data-driven system – including those involved in legal practice. The proposed set of ‘digisprudential affordances’ are contextualised through real case studies, thus bridging the gap between legal theory and design practice in a novel way (https://www.cohubicol.com/about/publications/digisprudence-code-as-law-rebooted/);

- DEVELOPED THE PROJECT WEBSITE, that should survive the end of the ERC funding and become a major hub for high quality content on the relevant subject matter. The accessibility and longevity of the project’s outputs have been kept in mind at every stage of its development; by using free and open source technologies and methods the fruits of the project will be available and accessible well into the future (www.cohubicol.com);

- ORGANISED THREE PHILOSOPHICAL SEMINARS with leading lawyers, philosophers and computer scientists, on the topics of Text-driven normativity and the rule of law (2019), Interpretability issues in machine learning (2020) and Legal effect of code-driven law (2021) (https://www.cohubicol.com/about/philosophers-seminars/).
DOING LEGAL THEORY WITH COMPUTER SCIENTISTS (AND VICE VERSA) implies a novel cross-disciplinary methodology to ‘do’ legal theory, with the active involvement of computer scientists.

We set up the JOURNAL OF CROSS-DISCIPLINARY RESEARCH IN COMPUTATIONAL LAW (CRCL), based on a novel understanding of the relationship between law and computer science, grounded in mutual respect for ‘the other’ discipline, avoiding attempts to mix them into one big ‘soup’ while also avoiding the colonisation of one discipline by the other. The prime objective is to help lawyers and legal researchers to better understand the affordances and limitations of computational ‘law’, based on a better understanding of what CS has on offer. Simultaneously, the Journal hopes to contribute to a better understanding within CS of what law stands for, how it connects with the rule of law and how CS can help to sustain and reinvent legal protection in the realm of computational ‘law’ (www.journalcrcl.org).
We are developing A TYPOLOGY OF LEGAL TECHNOLOGIES that will be made available via the website (July 2022), with various filters to obtain a quick overview of different types of technologies. The idea behind the typology is that the theoretical research on the side of both law and computer science should be nourished by a study of concrete examples of ‘legal tech’ (legal search engines, prediction of judgments, argumentation mining, rules as code technologies etc.), whilst, in turn, the study of such concrete systems will trigger new or different theoretical questions. The typology should be available via the website mid-2022.
The typology of legal tech will be made available to a set of stakeholders/experts/legal practitioners/tech developers/others for feedback, as part of the scheduled expert meetings. Such feedback will be a crucial resource to finalise the first version of the typology that goes public.

We will be CO-AUTHORING (PART OF) THE FIRST WORKING PAPER WITH AN NLP SYSTEM, most probably GPT-3, to get a first-hand ‘feel’ of what these systems are capable of and to show the reader what it means for an NLP system to contribute to a scholarly text.

We will organise FOURTH PHILOSOPHERS’ SEMINAR ON ‘A PHILOSOPHY OF TECHNOLOGY FOR COMPUTATIONAL LAW’ (2022)

We are now preparing the WORKING PAPERS ON CODE- and DATA-DRIVEN NORMATIVITY AND LEGAL PROTECTION (2022).

We are preparing TWO MAJOR INTERNATIONAL CONFERENCES IN 2022 AND 2023 on code- and data-driven legal technologies (2022) on a new hermeneutics for computational law.

The overarching aim of the project is the development of A NEW HERMENEUTICS FOR COMPUTATIONAL ‘LAW’, raising the issue of the performative effects generated by the use of legal technologies. After the final conference this will result in a monograph by the PI.
COHUBICOL logo