Alice. Alice Is Impatient

Marc Brooker, an engineer at AWS, explains the "inspection paradox"—why customers experience significantly longer latencies and outages than a service's metrics suggest. Users encounter a time-weighted distribution of delays; longer outages are sampled multiple times by customers arriving during that period, making their mean experience substantially worse than the service's calculated mean time to recovery or request latency. For example, a service with a 1-hour MTTR might show customers experiencing a 6-hour mean recovery time due to this statistical phenomenon.