The first ten minutes of an outage are a coordination problem

Most incidents don't run long because the bug is hard. They run long because four competent people start fixing at once. Here's the process layer, with no tooling in it.