Operations

Repeatable rigor.

A Standard Operating Procedure exists to answer one question in advance: what does a competent person do here, every time, regardless of mood, urgency, or how many times they have done it before? Aviation maintenance runs almost entirely on SOPs for exactly this reason: a mechanic’s judgment is valuable, but a mechanic’s memory, under time pressure, on the fortieth inspection of the week, is not something any serious operation should rely on alone.

The hard part of writing a real SOP is not the writing, it is deciding what belongs in it. Write down too little and the procedure fails to catch the actual failure modes it exists to prevent. Write down too much and nobody follows it because it reads like a legal document instead of a working tool. The genuinely useful SOPs in aviation, and in software operations, tend to be short, sequential, and built around the specific steps people have been observed to skip under pressure, not a generic list of best practices copied from somewhere else.

What stays loose matters as much as what gets written down. A procedure for what to check before a production deploy should be rigid. The judgment call about WHETHER to deploy on a Friday afternoon should stay a judgment call, informed by context an SOP cannot fully capture. Confusing the two, over-proceduralizing the judgment calls or under-proceduralizing the mechanical steps, is where most SOP efforts go wrong.

The operational discipline behind this thinking already runs on the Piloting Mindsets page: systems borrowed from the cockpit, where a missed checklist step is never just a paperwork problem.