Glossary · Objects and system context
System under test (SUT)
The model, policy, or agent, plus the relevant observation and action pipeline, whose behavior is being measured.
Defined in
Handbook · FoundationsEvaluation is measurementA score is an observation, not a capability. Follow the instrument chain from construct to decision and the four places error gets in.Where it comes up
- Population, environment, harness, and validity domainHandbook · Claim
- Monitoring and control in productionHandbook · Regime
- Estimands, pass rates, and intervalsHandbook · Validation
- Falsification and failure regionsHandbook · Validity under pressure
- Living eval suitesHandbook · Lifecycle
- Harness architecture and reproducibilityHandbook · Lifecycle
- CrosswalkHandbook · Reference