The person running the outage shouldn't touch the code

transcript

SEV 1. Checkout is down. Someone takes command. And their whole job is to never debug.

Without structure: panic, blame, several people changing things at once, and silence. With it: clear roles, coordinated troubleshooting, and everyone informed.

The incident commander coordinates and decides. They do not debug. The technical lead runs the investigation and proposes mitigations. The communications lead updates the status page and stakeholders.

Mid SEV 1, the tech lead is stuck on a database problem. Should the commander jump in and debug it?

No. Pull in a database expert. The tech lead gets unblocked. And the commander keeps coordinating, and communicating.

An IC who starts debugging stops coordinating. That's exactly when conflicting changes and missed updates creep in. So one change at a time, announced first. I'm changing X.

Structure and roles matter more than individual heroics. The commander's tool isn't the terminal. It's the team.

Full breakdown in the lecture.