Modelling the Law: Turning Intent Into Code

When the letter of the law isn’t specific, how can you build digital systems to implement it?

In the Norwegian Labour and Welfare Administration (NAV), we build case-working systems to help us decide which citizens can get benefits. Benefits law is very concise, so first we need to interpret the law’s intent, which we can then turn into explicit rules. This is hard! Legal specialists, working in an area where nothing is ever black and white, must commit to specific interpretations and implementations. Developers and business analysts must quickly become subject-matter experts. Integrating complex domain material from multiple sources has a high cognitive load. And creating structured, usable documentation of the process and results is its own challenge.

We’ve been developing a tool to help our cross-functional teams collaborate more closely as they turn legislation into rules ready to implement. We’ll demo our tool (still a work in progress), and show you:

• How we went wrong right at the start by assuming the data domain (legislation) was most important, when really it was the process domain (analysis).
• How we then integrated process and data domains in the tool.
• How our models of these domains are evolving as we learn about different users’ needs, and how the tool is changing to reflect this.
• What design decisions we’ve taken to try to reduce cognitive load for users.
• How we’ve turned user input into structured, queryable documentation.
• What business benefits we believe the tool offers.