Lecture 1.4
A machine-executable rule says what follows when its conditions are met. It does not say who is entitled to cause that outcome, under whose delegation, against which version, or on what evidence. This lecture draws the line the rest of the discipline is built on: encoding a rule and authorising an actor to act under it are different institutional functions.
Supports Law 1 — Authority Must Be Explicit Before It Can Be Executed. →Learning objectives
After this lesson, the reader should understand:
Concept framework
Rule versus authority
Case study
A benefits system encodes an eligibility rule perfectly: the conditions are complete, the logic terminates, the outcome is correct for every input. A claimant challenges a refusal. What can the institution actually show?
That the rule was applied — and almost nothing else. It cannot show, from the rule alone, that this system was the one authorised to refuse rather than to recommend; that the caseworker's delegated limit covered this decision; that the version applied was the one in force on the day; or that the two facts the refusal turned on are facts the institution accepts as evidence. Each of those is a separate question, and a rule engine answers none of them. Law as Code produces the first artifact. The authority relationship — issuer, delegation, version, evidence, boundary — is the second, and it is the one a challenge actually tests.
Discussion questions
Exercise
Take one automated or semi-automated decision and separate its rule from its authority.
Research notes