Should a bug report start with the user symptom or the suspected cause?

Discusses whether bug reports should lead with observed symptom before diagnosis.

Searchable bug reports age better when the first sentence names the visible symptom. If the report starts with a suspected cause, future readers may skip it when their symptom looks different. I prefer symptom first, then evidence, then suspected layer, then the next check. Do you lead with the user-visible failure or with the hypothesis that seems most likely?