Operational Boundaries¶

Some runtime surfaces are safe operational tools. Others depend on stronger scientific judgment.

If the first question is which public surface to use at all, start from the surface selection guide before applying these operational boundaries.

flowchart TD
    A[Need to run something] --> B[Deterministic execution task]
    A --> C[Interpretive scientific task]
    B --> D[Operationally safer]
    C --> E[Needs stronger review]

Operationally Safer Surfaces¶

tree validation and inspection
deterministic formatting and reporting
explicit bundle and artifact packaging
dataset preparation where the main question is structural correctness rather than biological interpretation

Surfaces That Need Stronger Review¶

comparative model interpretation
ancestral inference interpretation
external-engine results used in publication claims
native benchmark interpretation when one difference may reflect model assumptions rather than one generic success or failure
mixed-model or posterior claims where the repository documents the boundary more strongly than it reproduces the underlying lecture workflow

Practical Rule¶

If the output mainly answers "did the runtime execute the declared procedure correctly?", the surface is closer to operational. If the output mainly answers "what biological conclusion should I publish?", the surface needs stronger scientific review.

That distinction keeps operational convenience from turning into false scientific certainty.