Link to current text
A.3.3:8 - Common Anti-Patterns and How to Avoid Them
| Recognizable failure | Repair |
| Inferring a transition law from procedure text or a workflow diagram’s layout | Recover the actual assertion under :4.8, then test for both a state space and a transition law under :4.1. |
| Treating telemetry as a law | Declare the proposed law, derive the observed coordinates through :4.5 and compare its consequences with the telemetry. |
| Using dashboard labels as state coordinates without their meanings | Recover the characteristics, units, scales and comparability rules, and check whether the chosen coordinates retain the needed predictive information. |
| Treating a simulation as release approval | Check its predicted result against the receiving decision’s conditions in :4.6, then obtain that decision. |
| Extrapolating beyond established model conditions | State the applicable region, source-currentness condition and lowering condition; use :4.7 if the use requires transfer. |
| Relying on a learned prediction without its domain and error conditions | State its training domain, observation relation, error or uncertainty policy and applicability window before reliance. |