ACC Better Rules Collaboration report summary
Exploring Machine Consumable Code - ACC
Introduction
Supported by the Accident Compensation Corporation (‘ACC’), the Ministry of Business, Innovation and Employment (‘MBIE’) is exploring ways to modernise the Accident Compensation Act 2001 (‘the AC Act’). This includes exploring a structural rewrite to provide a modern and responsive piece of legislation that supports effective and efficient delivery of the accident compensation scheme.
While exploring opportunities to make the AC Act more accessible and fit for purpose from a structural rewrite, MBIE and ACC learned about ‘machine consumable’ legislation, and the work of the ‘better rules – better outcomes’ (‘Better Rules’) initiative and the Service Innovation Lab.
Machine consumable legislation means that its rules can be presented in a structured format (code) that can be processed by a machine (computer) without human intervention. It recognises that machines are ubiquitous, services are increasingly consumed digitally, and there are many uses for machine consumable legislation that are only beginning to be explored. The Better Rules initiative is examining ways to support this through the way New Zealand regulation is developed and implemented, and the Service Innovation Lab previously facilitated a similar experiment.
Approach
The cross-agency and multidisciplinary Accident Compensation Better Rules Discovery Team (‘the team’) undertook a ‘discovery’ (experiment) over six weeks of half-days based at the Service Innovation Lab starting March 2019. The team comprised legal, policy, legal drafting, business rules, operational, and software development experts from ACC, MBIE, the Parliamentary Counsel Office, the Department of Internal Affairs (‘Internal Affairs’), the Inland Revenue Department (‘Inland Revenue’), and the Service Innovation Lab.
The discovery aimed to develop a blueprint of rules for entitlements in the AC Act by developing concept models, decision models, rules statements, and code. The discovery was to support a potential structural rewrite of the AC Act through: creating a shared understanding of the rules for entitlements; exploring use of machines to identify gaps in logic of the rules; providing a practical example that shows how this approach can be applied; and developing tools and techniques that could be applied to future work.
The team recognised that it was unrealistic to complete this blueprint for all entitlements in this time, and developed a prioritised list of entitlements to work through.
Key findings
-
Rules statements are of significant value for legal drafters: Rules statements – a simple set of rules synthesised from complex legislation – that have been robustly developed and tested are expected to assist the Parliamentary Counsel Office (legal drafters) to undertake a more comprehensive and accurate rewrite that is more human and machine consumable.
-
Developing rules statements identified gaps: The process of developing rules statements identified gaps in the logic of legislation that had not been previously identified, which range from technical gaps that could be resolved by legal drafters through to gaps that require further analysis. By resolving these logic gaps, legislation can be more human and machine consumable, the risk of misinterpretation can be reduced, and what is implemented can better match policy intent.
-
Legislation may have longer rules and more of them: A prevailing view in the team at the beginning of the discovery was that a structural rewrite would result in fewer or shorter rules to make the AC Act more accessible and human consumable. To the contrary, the team found that ensuring there are no gaps in the rule logic can result in longer rules. However, these longer rules were clearer and would ensure legislation is more human and machine consumable. Furthermore, as legislation becomes machine consumable, and policy developers have tools to better understand and model its (positive and negative) effects, additional rules may be created to create more nuanced policy.
-
Developing rules statements and code at the same time most effective: The discovery initially explored handing rules statements to software developers for turning into code. However, it was found that writing code at the same time as rules statements produced better quality rules statements and code, and was more efficient. It improved shared understanding, reduced follow up questions, and allowed the code to validate the rules statements.
-
Concept models improved shared understanding: The use of concept models as the first step for any new topic improved shared understanding and team engagement. They showed not only what was in scope but also what was out and enabled focused and effective communication. Concept models proved to be a core first step for the multi-disciplinary team to start working on each entitlement.
-
A small multidisciplinary team can produce most outputs: While the discovery had a broader team to draw on (and who were all used), the outputs that would support a structural rewrite were produced with a small team (fewer than six) of experts across business rules, operations, policy, legal, legal drafters, and software development.
-
Tools increased efficiency and broadened use of code: During the discovery, the team identified areas where different tools may have been of value, including tools for capturing concept models, tools that can test policy, and tools for storing rules in a dataset form from which executable legislation as code could be derived. The team experimented with tools and found that they demonstrated value and (at least in one case) were complementary to each other, but neither is necessary for a structural rewrite alone.
-
Rules requiring human judgement can be written in code: The team explored developing code for an area of legislation requiring significant human judgement to test whether it is difficult to code, and found it was no different or more complex than other rules. It emerged that identifying these types of rules could be used to track the quantity of human judgement within a piece of legislation, which could provide a new perspective and way to measure how much a piece of legislation introduces.
-
Value of software developers is beyond code: The software developers brought a lot of value to the multi-disciplinary team that was beyond the code they create. They looked at rules and logic on behalf of the machine, and improved the logic and robustness of rules statements.
-
A multidisciplinary team improves implementation: Using a multidisciplinary team to develop rules statements improved connections of people involved in policy development through to those in operations, and was observed to reduce translation gaps and is expected to improve implementation. It also sped up the cadence of work as the experts were in the room at the right time, eliminating handover errors and delays that usually occur when experts from a different team provide input.
Opportunities
-
Develop rules statements for other parts of the AC Act: To support legal drafters to ensure a structural rewrite would result in more human and machine consumable legislation, MBIE and ACC should develop rules statements for all provisions in the AC Act that may benefit from this work. A pragmatic approach would be to develop a prioritised list of provisions that would benefit most from rules statements, and scope the effort (time and resources) to complete that work in the time available. This does not require another intensive discovery, but rather could use regular discrete blocks of time (eg, 4 hours per week over six months). To develop code that tests the logic, a software developer is required.
-
Use methods from the discovery for other policy and legislative changes: While the focus has been on a structural rewrite, the approach tested across the six-week discovery could also be used as part of everyday business when considering policy and legislative changes. While this is particularly true for the AC Act, which is complex and composed of many interrelated provisions, considering different policy and legislative proposals against rules statements and code would robustly test their logic and implications.
-
Improve macro view through concept models: MBIE and ACC should seek opportunities to improve the macro view of accident compensation scheme settings through concept models. This could give a transparent and consistent view of the accident compensation scheme, and provide a common foundation for cross-agency collaboration, in the same way the smaller and more targeted concept models did for the discovery.
-
Code the whole AC Act and make it public: To fully support the long-term ambition of machine consumable legislation in New Zealand, ACC, MBIE, the Better Rules initiative, and the Service Innovation Lab could explore coding the whole AC Act. Much of the AC Act would have been coded as part of the process of creating the rules statements. A key realisation of the potential of machine consumable legislation is publicly publishing the code in an executable form for the same reasoning governments publish their legislative text: to aid people’s comprehension of the law. Coding the AC Act would be dependent on the Better Rules initiative completing its planned work on developing New Zealand standards and architecture for machine consumable legislation.
-
Continue to explore using code already developed: Some code has already been developed as part of the process to develop rules statements (and more would be developed if the rules statements are developed). This code presents a unique opportunity for innovation. For example, it could be used for policy development (eg, modelling test cases), for consumers or advocates to develop an end-user interface, or for government to see how the accident compensation legislation interacts with other legislation. Working with the Better Rules initiative and the Service Innovation Lab, ACC and MBIE should continue to explore this so that the opportunity is not wasted.
NOTE
The full report about this discovery is available as a PDF; however we are currently working on an accessible version of this for future release. As always we would like to know your thoughts and comments about the Better Rules work so please make contact with us.
