What Is Incident Severity, but a Lie Agreed Upon?
summary
Argues severity is a negotiated artifact, not a truth: a spreadsheet used a percentage of events unavailable within 24 hours threshold that teams could not measure, an internal tool (T incidence default-to-S1 bug) auto-escalates unknown fields to S1 and burns out on-call staff, and merging products at Twilio exposes incompatible severity models. Recommends adopting Fisher Rule (declare incidents, decide severity later), make severities measurable and actionable, track when teams spend hours arguing rather than resolving, and iterate with responders.
AI-written summary · about
chapters (9)
- 0:00The myth of severity
- 1:45Defining incident impact
- 5:08The problem with math
- 8:28Priority and subjectivity
- 10:22Incident mechanisms
- 16:05Organizational root causes
- 20:19Communication expectations
- 24:51Socio-technical approach
- 30:40Empowering the sharp end
related talks
13mThe Incident Is The Way: Using Your Incidents to Win...
19mHow to handle incidents at scale
24mWe should all be declaring more incidents
37mManaging incidents in microservices
27mLessons from Incident Management and Postmortems at Atlassian
13mNavigating SRE/Incident Management
33mDon't Panic! Effective Incident Response
24mThe Un-Incident: Extracting Value from the Gray Area of...