On September 22, the Better Legislation for Smoother Implementation (BLSI) community held its September 2023 BLSI virtual breakfast! This event focused on 'digital-ready drafting,' a term introduced by Better Regulation Tool #28. As part of the discussion, we had the pleasure of welcoming Pia Andrews, Strategic Advisor of Public Services, and Liane Huttner, Assistant Professor, Université Paris-Saclay, both of whom shared their invaluable insights.
The key highlights of the event and a link to the recording can be found below.
Key takeaways
What is Digital-Ready Drafting? (The Digital-Ready Policymaking Team, DIGIT B2)
In this session, delved into the concept of Digital-Ready Drafting, as proposed by Tool #28 of the Better Regulation Toolbox.
- The European Commission adheres to the Better Regulation Guidelines, which encompass a toolbox for preparing new policy initiatives. This toolbox includes a dedicated tool for digital-ready policymaking (Tool #28), which emphasises digital-ready drafting as a crucial component.
- While there is no prescriptive definition of digital-ready drafting, Tool #28 advises policymakers to “set out clear rules in the legislative act while keeping those rules future-proof to technical development”.
- Prima facie, clear rules are obtained via simple, precise and concise wording. Nevertheless, digital-ready drafting may also involve several other elements: a proper description of the IT system; the reuse of existing concepts; the provision of an implementation timeline that is realistic from a technical point of view; the use of cutting-edge technologies; and allowing for the possibility of automatic implementation.
Digital Rules and Digital Drafting for a Digital Economy (Pia Andrews, Strategic Advisor of Public Services)
Pia Andrews guided us through the process of digital-ready drafting, aimed at achieving the effective, efficient, and traceable adoption of legislation/regulation.
- Currently, the regulation and legislation drafting process can be guess-driven, inefficient, inconsistent, ineffective, and may even result in untrustworthy outputs.
- To ensure effective, efficient and traceable adoption of legislation/regulation:
- Co-draft human and machine-readable rules. Having a digital version of legislation/regulation ensures that everyone can consume the rules in the same way. This also enables the generation of traceable records, the ability to test rule outputs for lawfulness and to monitor the usage patterns of said rules.
- While the human-readable version serves as the primary legal authority, the machine-readable version would function as a highly trusted reference implementation.
- To achieve better regulation, assembling multidisciplinary teams consisting of policy analysts, legislative drafters, service designers, and software developers is essential. These teams, guided by concept models, should work collaboratively, ensuring that legislative versions, pseudocode, and software code are drafted simultaneously.
- When determining which rules to encode, prioritise prescriptive rules while incorporating placeholders in code for principles that must be invoked. Additionally, ensure that emerging precedents are systematically codified over time.
- Co-draft human and machine-readable rules. Having a digital version of legislation/regulation ensures that everyone can consume the rules in the same way. This also enables the generation of traceable records, the ability to test rule outputs for lawfulness and to monitor the usage patterns of said rules.
- To design and continuously improve test-driven legislation/regulation:
- The process should commence with concept modelling, followed by co-drafting both human and machine-readable versions utilising a test-driven, human-centred approach. Ideally, this would also produce a test-suite. All these materials can then be used during the consultation phase. Once the legislation is enacted by Parliament, the API can be made immediately available.
- The ideal outcome of the isomorphic drafting process would be a ‘Policy Twin’ which would provide a shared digital interpretation of legislation/regulation as code.
- However, a significant obstacle lies in the need to establish a shared and strategic policy infrastructure. This requires reconciling the diverse tools, interpretations of policy, viewpoints, monitoring methods, and reporting systems utilised by different stakeholders.
- To build trustworthy systems in the age of AI:
- Rules as code provide a real guardrail for artificial intelligence.
- Decision-making systems may be grounded in machine learning or rule-based approaches. However, within the governmental context, machine learning-based automated decision-making presents higher risks due to challenges in adhering to Administrative Law and the Rule of Law.
- Consequently, effective regulatory interventions require consuming regulation as code and/or testing against regulation (as code). Ensuring fairness also mandates the measurement of these systems' impacts.
The Catala Project: Adapting Code to Law (Liane Huttner, Assistant Professor, Université Paris-Saclay)
Liane Huttner shared with us her experience working on the Catala Project, a new programming language that was specifically designed to allow a straightforward and systematic translation of statutory law into an executable implementation.
- Liane's presentation introduced an alternative perspective to Pia’s. The Catala Project has focused on law that is not digital ready, looking at how fair and transparent code may be produced from that.
- In France, laws have indeed been translated into code within various domains, including tax law and social security pensions. However, a common challenge has been the complexity and lack of clarity within the existing algorithms.
- To address this issue, the Catala project has undertaken the creation of a domain-specific computer language. This specialised language is designed with the intent of translating statutory law into code that is accurate, transparent, and comprehensible. It achieves this by mirroring the familiar legal structures and terminology that lawyers are accustomed to.
- Effective development of this human-readable code hinges on interdisciplinary collaboration, a method known as "interdisciplinary pair programming," which brings together lawyers and computer scientists.
- Overall, the Catala Project places a strong emphasis on aligning code with existing laws, rather than advocating for alterations to the laws themselves.
Presentation and recording
Please see the attachments for a downloadable version of the webinar's key takeaways, including the Q&A session and the speakers' presentations.
To access the recording of the session, please click here.
Next steps
Our Digital-Ready Policymaking Team looks forward to seeing you at our next virtual breakfast session! Until then, please subscribe to our newsletter and do not hesitate to contact us with any questions you may have.