I play around Rules as Code, which is the automation of reasoning over statutes and regulations, etc. Like smart contracts but for governments. I'm in the category of Rules as Code proponents for whom the lodestone is "the legal expert owns the code." In support of our argument are the examples of the Netherlands and Finland, both of which have had considerable success automating their national tax systems by using custom systems for making the encoding task something that their tax experts do, not their software developers. Frustratingly, neither were sharing the platforms. So the systems that were readily available for Rules as Code had never been deployed at scale, and the systems that had been successfully deployed were not available. That changed yesterday. ALEF (Agile Legal Execution Factory) has been released in open source under the EUPL. That's should be a game-changer for Rules as Code around the world. So I was very interested to see what they actually did. And the answer is "not much." The took JetBrains MPS, created a bunch of DSLs with text-based projectional editors (for data structures, rules, tests, nlg, etc.), and Java code generation, and then packaged them with JetBrains. I had been hoping they had gone further. I'm not a fan of projectional editing in text. It prevents you from making syntax errors, but it doesn't expose the syntax so that you can learn it, and you lose some of the things you expect text-based editing to do. They have also, frustratingly, made the entire thing Dutch-specific in a way that is going to be relatively complicated to un-do. On the plus side, the state of the art is going to be pretty easy to improve upon.