Topic overview · Experiential knowledge
Expertise becomes automatable not by writing it down, but through testable interplay.
An expert system copies no person. It makes one bounded part of the decision work visible: signals, rule, exceptions and the consequences of error.
Experience has four layers
Experts do not act like a rigid if-then program. They consider sequence, context and interactions. Only the combination explains why the same observation can lead to different decisions in two cases.
01SignalA recognisable observation that triggers a decision.
02RuleA reasoned if-then with provenance and purpose.
03ExceptionThe similar case that must not reach the same outcome.
04ConsequenceA measure of what an error costs.
01 · DecisionStart with a bounded question, not a profession — trigger and outcome must be clear.
02 · ContrastsNormal, borderline and prohibited cases give the rule its shape.
03 · RolesThe model interprets, deterministic logic decides — failures stay localisable.
Core method
The decision module
Decision questionWhich bounded question should be answered?
Required inputsWhich information must exist, and which source governs?
Professional ruleIf …, then …, because …
Validity rangeApplies when … / not when … / uncertain when …
Deterministic checkWhich required fields, thresholds or combinations?
Escalation & versionWhen does it stop, and which tests rerun?
Choose a recurring professional decision. Build not a product but a testable module: rule with provenance, three contrasting cases, validity and two stop signals.