In the beginning of RaC…
Rules as Code has been the ‘hot new thing’ for digital transformation for some time now. In fact, an article on RaC in The calls it ‘decades in the making’.
In the more recent past (2020), the Organisation for Economic Co-operation and Development (OECD) released a report ‘Cracking the code: Rulemaking for humans and ’. That report flagged the potential of RaC for governments. A subsequent OECD report ‘Global Trends in Government Innovation ’ talked about ‘significant progress’ in RaC. However the example projects it gave were actually quite sparse, and two of them were projects Salsa was involved in (BenefitMe and a COVID vaccination proof of concept).
So, is RaC moving as quickly as it should? To answer this question we interviewed two RaC pioneers, Pia Andrews and Hamish Fraser. Pia’s first RaC project was in 2016, working with the Anti-Money Laundering and Counter-Terrorism Financing Act (AML/CTF Act). And Hamish’s first project was in 2017, with NZ’s Better Rules .
RaC expectations
When you started working with RaC, what were your expectations?
Pia: At that time (2016), I was testing the hypothesis that we could make it easier to both adopt new regulation and easier to monitor and regulate. Banks and regulators had worked in the RaC space, but had found the maintenance burden too large. I knew drafting Rules as Code would help with this, but it hadn’t occurred to the players at the time. This was a huge missing opportunity and it was a necessary part of modern digital infrastructure. I wasn’t surprised drafting RaC wasn’t normal yet, but I could see it had huge potential to change things for the better.
Hamish: My expectations were unclear. It took me some time to properly understand the implications of what we were embarking on. The immediate wins were obvious but I was concerned about the possible negative effects that might yet be realised. During the Legislation as Code research project I was involved with, I was able to resolve those concerns and at that point my expectations for RaC were high, while my expectations in seeing such practices adopted were, shall we say, realistic.
Has RaC lived up to these expectations?
Do you think RaC has lived up to its potential and your expectations?
Pia: Its uptake, and the benefits and problems of implementing RaC, is what I’d expected when I first got involved in the space. The take up in government is slower, but I understand that. The concept has taken off much faster in academia than government, but part of the challenge in this domain is the question of who owns Rules as Code? It's a new opportunity, which falls somewhat between the cracks of traditional lines of responsibility in public sectors.
Hamish: In a way it’s exceeding its potential — it’s exceeding expectations in terms of what it’s delivering — the uptake is slow but perhaps quicker than I thought it’d be.
Barriers for implementing Rules as Code
What do you think the barriers are?
Pia: I think one of the key blockers is that people who are responsible for drafting rules don’t have the tech skills to explore this space. Other barriers for government include a lack of ownership and the gap between policy design and delivery, and the cadence of policy drafting.
Hamish: I agree. I think it’s about finding a hook for people, to help them see the potential and invest in the processes.
How to bring clients on the RaC journey
What are the hooks?
Both Pia and Hamish have recently been focusing on some new ‘hooks’ to help reduce the barriers to RaC uptake.
Pia: One hook for me is the opportunity and concept of policy . RaC as part of the shared infrastructure needed to manage the entire policy cycle end-to-end gives government (and others) the means to measure, monitor, continuously improve and optimise policy and public outcomes. Another hook is around improving drafting processes and requirements. For example, in the EU they’re passing new requirements for all regulators to produce a reference implementation of new regs as code as part of the drafting requirements. If you update drafting guidelines to implement RaC then you have to do interpretation early, which means they’re tested before passing to parliament, and able to produce a test suite for all new rules (legislation/regulations) that can be used to validate all interpretations, as well as to model the impact of change.
Hamish: I’ve had recent success with RaC uptake focusing on reporting needs. If you have a regulatory need to report, and you want to fulfil that need digitally as data (rather than pdf), then the government department or other body has to interpret its rules to actually build the data-reporting application.
Artificial intelligence (AI) and Rules as Code
How do you see AI shaping or changing RaC implementation going forward?
Pia: I see Rules as Code as a necessary guardrail for AI, to test against, to consume, and to ensure no decisions/actions are taken that aren’t contradictory to the legislative and regulatory rules of that government entity. In other words, RaC provides a necessary, but not sufficient, means to ensuring legal compliance of AI usage in government. When you complement RaC with test suites, impact monitoring, and all the rest, you get a powerful way to ensure all systems, not just AI, are having the right impact on an ongoing basis. Even automated decision-making can be trustworthy if the systems are explainable and traceable back to their legal authority, which Rules as Code can help enable.
Hamish: So far it’s proved more a distraction and a derailing of legitimate RaC investigations. I believe it could end up being quite helpful but predominantly in a manner that complements the precise effort that RaC promotes. I can see us using AI to generate tests and to examine document structures. I’m not yet convinced it will generate RaC in any useful way as the act of generating RaC is perhaps more valuable than the end product.
About Pia and Hamish
Pia Andrews is a serial public sector transformer and reformer, usually working within the machine to drive adaptive policy management, policy infrastructure, participatory governance, high trust and legitimate systems and operating models (inc AI), citizen-centric design, service agility and real, pragmatic continuous innovation in the public sector and beyond. She is on a self-mission to transform public sectors to be more humane, responsive and effective in the 21st century, at the speed of continuous change and urgency.
Hamish Fraser struggles with the rules. Originally a designer, he led his own software development company ( ) for 20 years during which he also ended up elected to a municipal body as a councillor. In his time there he was fascinated by the future positive and negative impacts he could see technology would have on our democratic institutions. This led him to the New Zealand capital (via 6 years in a bus) to pursue his interest in institution design and technology.
He spends his time nerding out on the policy, legislation and public service machines and meeting the people operating them.