blameless postmortem

term

ops culture

Reviewing an incident to find the conditions that allowed it, rather than the person who acted within them.

The argument for it is practical rather than kind: people who expect blame report less, later and less accurately, and an organisation that cannot see its near misses is flying on the subset of events too large to hide. The discipline is asking why the action made sense at the time - what the operator saw, what the tooling suggested, what the pressure was - because a system where one wrong command causes an outage has a design problem that will outlive the person who ran it.

Also known as: postmortem

All glossary entries