Methods

Correctness claims should stay bounded by evidence.

OperatorWorks is designed around explicit assumptions, deterministic transformations, support boundaries, reproducible artifacts, and fail-closed outcomes where correctness cannot be established.

Method lifecycle

From a scoped claim to a reviewable artifact.

1. DefineState the expression, claim, domain, and intended review question.

2. DeclareRecord assumptions, conventions, operator families, and support boundaries.

3. TransformApply deterministic, inspectable steps without silently widening scope.

4. EvaluateReturn pass, fail, inconclusive, or unsupported.

5. PreservePackage trace, environment, outcome, notes, and hashes.

Outcome discipline

Four outcomes—not one confidence theater.

PASS

Established within scope

The stated claim is supported under the recorded assumptions and method.

FAIL

Contradicted within scope

The stated claim does not hold under the recorded assumptions and method.

Inconclusive

The available method or evidence does not establish either pass or fail.

Unsupported

The requested workflow falls outside the implemented and evidenced support boundary.

Research-grade posture

Reproducibility requires more than an exported expression.

A reviewable result should preserve the inputs, assumptions, transformation steps, software context, support boundary, outcome, and stable identifiers. Where these are unavailable, the artifact should say so.

OperatorWorks does not treat tags, interface labels, or a plausible-looking result as proof. Consequential claims should be traceable to the evidence that supports them.

claim recordminimum fields

claim_id
input_expression
assumptions[]
transformation_trace[]
outcome
support_boundary
environment
artifact_hashes[]
review_notes[]

Fail-closed principle

When correctness cannot be established, stop.

The method should not silently substitute a nearby problem, drop a noncommutative condition, infer an unstated assumption, or promote an unsupported result into a verified one.

Deterministic behavior

The same supported input and context should produce the same canonical result and evidence structure.

Visible boundaries

Unsupported domains and inconclusive checks remain explicit outcomes.

Independent review

Evidence should be structured so a technically qualified reviewer can inspect and challenge it.

Versioned methods

Scientific behavior changes should be tied to tests, release records, and updated evidence.