The Belief System Method — Framework

01. What It Is

Documentation tells you what to do.
This tells you what to do when the rules run out.

“Belief system” here doesn't mean ideology or faith. It means the structured set of claims, priorities, and constraints a person is actually operating from when they make a judgment call — whether or not they could state it if you asked.

Most attempts to formalize expertise produce a manual, a decision tree, a set of rules. That's useful, and structurally limited — rules cover the cases someone thought to write down. Real judgment gets exercised precisely in the cases nobody anticipated.

The Belief System Method extracts the layer underneath the rules: not what a practitioner does, but why — and what they'd do next if a situation shifted just enough that the existing rule no longer cleanly applied.

02. The Schema

Six fields, filled in once per practitioner.

Every extraction fills in the exact same structure, regardless of domain.

belief_system.schema

v1.0 Canonical

constitutive_claims:

what's non-negotiable

The foundational commitments that create the practice itself, rather than merely regulating behavior.

taboos:

what can't be said, or done, under pressure

Inviolable boundary lines that the system will never cross, regardless of operational urgency.

load_bearing_distinctions:

what actually carries weight in a live decision

The subtle, expert differentiators that separate real signal from noise when stakes are high.

sacred_tensions:

where two principles pull apart, deliberately

Incommensurable values preserved in permanent tension to avoid over-optimizing one at the expense of the other.

generative_test:

the question that reveals a stance on something new

A diagnostic scenario testing whether an unfamiliar dilemma resolves in alignment with the underlying conviction.

escalation_paths:

where judgment hands off before it breaks

Explicit deterministic boundaries where machine judgment yields to human authority before failure cascades.

The conceptual foundation of this schema — how beliefs generate judgment when rules run out, and the accidental ladder from checklists to protocols to beliefs — is explored in depth in Every AI Failure Looks Like a Capability Problem →

03. Grounded In

Two established lines of work, applied together in a new place.

Stafford Beer — Viable System Model (1972)

sacred_tensions mirrors Beer's S3–S4 homeostat: the built-in balance between optimizing the present and adapting to the future.

Shanahan, McDonell & Reynolds — Nature (2023)

The multiple-perspectives claim follows the established scientific account of an LLM as a simulator, capable of playing many roles at once.

04. How to Cite

Dhruv. “The Belief System Method.” Iceberg Tribe, 2026.
https://icebergtribe.com/thinking/the-belief-system-method

Documentation tells you what to do.This tells you what to do when the rules run out.

Six fields, filled in once per practitioner.

Two established lines of work, applied together in a new place.

Stafford Beer — Viable System Model (1972)

Shanahan, McDonell & Reynolds — Nature (2023)

Codify Your Institutional Belief System

Documentation tells you what to do.
This tells you what to do when the rules run out.