maios-software-diagnosislisted

# Diagnose the mechanism that changed the result Connect expected behavior, observed behavior and the conditions of the case. A visible symptom can be far from the mechanism that produced it. Preserve the selected object and distinguish diagnosis from implementation when only understanding was requested. ## Establish the relevant observation Use the reported trace, source and actual environment. Determine what was observed directly, what was reported by another actor and what remains inferred. A reproduction is useful when it can discriminate the cause or support the confidence needed for a change; do not claim it happened if the required environment is unavailable. Separate an observation that detects the stated failure from a check that always succeeds. When an executable probe is selected, make it sensitive to the mechanism and run it in an appropriate, authorized environment. A source inspection can also settle a particular relationship without a new test. ## Compare explanations that could change the correction Consider materially different mechanisms supported by the evidence, rather than generating a fixed number of hypotheses. For a live candidate, know what should be seen if it is right, what would contradict it and which available observation best distinguishes it from alternatives. Trace the path from the symptom toward the first incorrect transformation. Distinguish the trigger, the causal mechanism, the way the error propagates and conditions that amplify