Quiet under load

There’s a particular silence that lives inside an active incident. The room itself is loud: alerts, people typing, the hum of an over-energized Slack channel. The silence I mean is internal. The small space inside the head of whoever is running the room. The half-second pause before the next sentence. The decision not to react.

Most of the value an incident manager produces is in those pauses.

I used to think incident response was about speed. Move quickly, decide quickly, communicate quickly. There’s something to that. Every minute a system is degraded costs someone something. But the more of this work I do, the more I think it’s actually about pace. The right pace. Which is sometimes faster than feels comfortable, and almost always slower than the rest of the room wants to go.

When everything is on fire, the shape of leadership is mostly: don’t add to the fire. Don’t get spun up. Don’t re-ask the question that just got answered. Don’t loop back through context the team already has. Hold the operating picture in your head, hold a few decisions back until you have the right context to make them, and absorb the panic so the engineers next to you don’t have to.

This is harder than it sounds. The thing nobody tells you about incident management is that calm is a depletable resource. You can run a 90-minute incident on calm. You cannot run six of them in a week. The real work is figuring out what kind of system makes calm easier to keep. For you. For your team. For the next engineer who has to walk into the room.

That’s what most of the writing on this site is going to be about. The systems behind the calm. The on-call rotation that lets people sleep. The escalation path that doesn’t require improvisation. The postmortem workflow that turns three days of writing into thirty minutes of editing. The runbook that turns last time’s wishes into this time’s instructions.

There’s a reason the cloud animation behind this page moves slowly. The work I care about looks like that.