Cause-Aware Alarms: Stop Paging on Random Numbers

A developer proposes replacing output-count alarms with cause-aware alerts that distinguish between zero output due to failure versus benign quiet periods. The approach moves classification logic into the job emitter, eliminating false alarms without raising thresholds. The technique is demonstrated with a nightly matcher job on AWS, costing nothing extra beyond existing logs.