AIM Framework glossary
Normative Entailment Principle (NEP)
Canonical definition
The methodological principle that where a descriptive result of the framework identifies a structural condition under which the system produces a determinate and specifiable harm, and where the same reasoning that establishes the result also specifies the condition whose alteration removes or bounds that harm, the construct that names that condition and the intervention that alters it are entailments of the descriptive result and not additions to it. Such entailed constructs and interventions carry the same epistemic status as the descriptive result from which they follow, subject only to the further premise that the specified harm is to be avoided, which the framework treats as the minimal condition of the analysis being conducted at all rather than as a contestable value commitment. The principle is triggered only where the intervention is specified by the same reasoning that specifies the harm; interventions requiring an independent value premise, a contested empirical conjecture, or a design choice not fixed by the derivation are merely consistent with the framework rather than entailed by it. The constructs so entailed are constructs of avoidance and management rather than of perfection, because the architectural features generating the relevant conditions cannot be removed from the system; the entailed normative goal is therefore Subcriticality Maintenance rather than elimination.
Explanation
Most theories describe how something works and then go quiet when the question turns to what should be done, on the view that any recommendation has to be argued for separately using outside values. This framework is different in a narrow way. Some of its recommendations are not brought in from outside at all, they are simply the second half of a sentence whose first half is a description. Once the framework shows that a particular structural condition reliably produces a particular harm, it has, in the same breath, told you what to change to contain that harm, because the thing to change is the condition it just named. This is the same kind of reasoning an engineer uses: describing an engine already tells you it needs oil and coolant, and saying so is not moralising, it is finishing the description. The only value you are asked to grant is that the harm is worth avoiding, and if you did not grant even that, there would be no reason to be studying the harm in the first place. That the interventions in this book are unfamiliar does not make them a different kind of thing, any more than a novel engine's novel maintenance requirement is optional because no one has stated it before.
Epistemic status
Framework-Specific Methodological Principle (logic-level, not empirically falsifiable).
Derivation
Chapter 3, Section 3.10.2; Chapter 3, Section 3.10.3; Chapter 3, Section 3.10.5.
Also in the entry
Source: Chapter 3, Section 3.10.