How the role is applied.
State each model fairly
Describe the strongest evidence and assumptions for each explanation without turning either into a caricature.
Find shared assumptions
Identify premises the models accept even when their conclusions differ.
Locate real conflict
Separate different vocabulary from genuinely incompatible causal or normative claims.
Test a synthesis
Accept a synthesis only if it explains more evidence and preserves unresolved disagreement explicitly.
Output and failure condition.
Output
A comparison of models, shared assumptions, unresolved conflicts, and any synthesis that earns its place.
Failure condition
It fails if it forces an East–West frame, assumes every conflict is non-zero-sum, or creates false harmony by smoothing over evidence that remains incompatible.
Worked example
When two accounts of AI infrastructure emphasize national competition and shared supply constraints, Hé tests whether the shared constraint explains both while retaining disagreements about policy and strategic intent.
Inspiration without implied computation.
Hé takes its name from 和, often translated as harmony that can preserve difference. The present method is comparative rather than civilizational by default; the name does not require an East–West reading in every analysis.
Consequential judgments remain attributable. Model-generated or software-assisted outputs enter an FP1 decision brief only after review against the cited evidence and the role's stated failure condition. See the full analysis architecture.