Governing Question
The primary question a technical discipline, architecture, or control mechanism is designed to answer. Distinct disciplines are distinguished less by their tools than by the questions they were built to answer. Evaluating an architecture begins by identifying that question, before assessing how well the architecture answers it.
Term introduced by FERZ, Inc. in The Governing Question (2026).
Why the concept exists
Technical disciplines are usually compared by their tools, and tools are the least stable thing about them. Tooling migrates. Techniques migrate. Vendors converge.
What persists is the question each discipline was built to answer, and the architectural commitments that follow from taking that question seriously. An observability architecture is optimized for completeness of record and minimal interference with the running system. A credential architecture is optimized for the durable properties of actors. A model validation regime is optimized for statistical rigor across a population of decisions. Each optimization is correct for its own question, and each is why the architecture does not substitute for one built around a different question.
The Governing Question Framework
The Governing Question Framework evaluates any architecture in two ordered steps.
- Identify the governing question the architecture was built to answer.
- Evaluate how well it answers that question.
Inverting the order produces two predictable errors. An architecture assessed against a question it was never built to answer will appear deficient. An architecture assessed against its own question will appear complete.
Neither judgment tells you whether the architecture addresses the question you actually have.
Three diagnostic questions
Applied to a specific component, the framework reduces to three questions that can be answered in a sentence each by anyone who knows their own system.
Which question does this component actually answer?
Not which question its category name suggests, and not which question the deployment was intended to address.
Where does it sit relative to the effect?
Before the action becomes real, in its path, or beside it. Observing an action is a different position from standing in its way.
What happens when it is unavailable?
If unavailability degrades visibility, the component answers a visibility question. If it degrades the record, it answers a records question. If it halts the action, the component sits on the execution path. Failure behavior describes architecture more reliably than documentation does.
Applying the framework
The disciplines currently converging on AI governance each arrive with a different governing question, and each question is legitimate. The left column names where a question came from, not what it classifies. A single discipline routinely produces work answering more than one of these, and a component is classified by the question it answers and where it sits, not by the field it came from.
| Discipline | Governing question |
|---|---|
| Identity and access management | Who is requesting this, and what have they been granted? |
| Infrastructure | Where should control sit, and how does it behave when things fail? |
| Observability | What happened, and can it be reconstructed? |
| Compliance | Were required controls defined, applied, and documented? |
| Model governance | Is this model fit for purpose, and is it still fit? |
| Alignment | How does this system tend to behave? |
| Execution authorization | Should this specific proposed action execute, before it executes? |
The questions are complementary rather than competing. A serious deployment needs answers to all of them. What it does not get is one answer that covers the rest.
What this concept is not
The Governing Question Framework is an analytical method, not a governance mechanism. It authorizes nothing, enforces nothing, and produces no evidence. It is a way of reading architectures, including this one.
It is also not a claim that any discipline is deficient. An architecture doing well the thing it was designed to do is not thereby doing something else badly.
Related concepts
Execution-Time Authorization. the architectural implementation of ex-ante authorization at runtime, per action. Its governing question is the one this framework isolates as distinct.
Runtime Authorization Boundary. the structural location where authorization is enforced. Where a component sits relative to the effect is the second of the three diagnostics.
Three Problems Taxonomy. Visibility, Alignment, and Authorization as categorically distinct problems. The taxonomy is what the framework yields when applied to AI governance specifically.
Frequently Asked Questions
What is a governing question?
The primary question a discipline, architecture, or control mechanism was designed to answer. Architectures inherit the optimizations of the question they were built around, which is why an architecture built for one question rarely substitutes for one built around another.
How is this different from a use case?
A use case describes what someone wants a system to do. A governing question describes what the system was structurally built to determine. The two often diverge, and the divergence is where evaluation errors originate.
Why does the evaluation order matter?
An architecture assessed against a question it was never built to answer will appear deficient. The same architecture assessed against its own question will appear complete. Identifying the question first prevents both errors.
Does this framework favor any particular architecture?
No. It is a method for reading architectures, and it applies to FERZ's work on the same terms as any other. Its output is a classification, not a ranking.
References
Meyman, E. (2026). Authority versus Authorization: A Definitional Framework for AI Governance. FERZ, Inc. https://doi.org/10.5281/zenodo.21341907
Meyman, E. (2026). A Taxonomy of AI Governance Approaches: Distinguishing Visibility, Alignment, and Authorization. FERZ, Inc. https://doi.org/10.5281/zenodo.18275969
FERZ, Inc. (2026). Governing Question. https://ferz.ai/governance/concepts/governing-question
BibTeX
@misc{ferz2026govquestion,
author = {{FERZ, Inc.}},
title = {Governing Question},
year = {2026},
url = {https://ferz.ai/governance/concepts/governing-question}
}