Hard Problems We Handle in Incidents but Aren't Recognized
summary
Identifies four concrete incident activities—diagnostic work, therapeutic actions, recruiting people, and status reporting—and shows how familiar tools (Slack/Zoom, Postgres, nginx, backups, cross‑DC replication) create specific tradeoffs: bringing experts up to speed costs diagnostic time, delegating tasks causes context switches, killing slow queries or disabling encryption are explicit sacrifice decisions, and simultaneous unrelated outages create the parallel incidents dilemma. Cites Laura McGuire’s dissertation and the NYSE July 7 2015 shutdown as concrete examples and argues these phenomena are real, recurring, and worth naming.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:30Title
- 3:07Diagnostic Work
- 5:58Recruiting
- 7:24Status Reporting
- 8:41Costs of Coordination
- 10:04The Dilemma
- 13:42Divide Conquer
- 18:57Sacrifice Decisions
- 21:54hindsight bias
- 22:39parallel incidents
- 24:26chat transcript
related talks
19mStorytelling as an Incident Management Skill
34mTaking Human Performance Seriously in Software
48mThe Secret Lives of SREs - Controlling the Costs of Coordination across Remote
47mThe Four Agreements of Incident Response
51mAnatomy of an Incident
27mIncidents: the customer empathy workshop you never wanted
14mThe Legend of the Arsonist Firefighter
48mAnatomy of an Incident