hope is not a strategy
expressionops culture
The SRE reminder that wishing a system stays up is not the same as engineering it to.
Popularized by Google's SRE culture: reliability comes from measured objectives, error budgets, and rehearsed failure, not optimism. If the plan for a risk is 'it probably won't happen', there is no plan.
Hope is not a strategy is the observation that an unverified assumption is not a plan, and it names a specific failure: believing something will work because nobody has evidence that it will not. It is a staple of site reliability engineering for a good reason.
The instances are everywhere once you look for them. Backups that have never been restored are hope. Failover that has never been exercised is hope. A runbook written once and never followed is hope. Capacity that has never been load tested is hope. In each case the belief may well be correct, and the point is that nobody knows, and the moment of discovery is scheduled by circumstance rather than by you.
The discipline that follows is testing the assumption deliberately and on your own timetable: restore a backup on a schedule and verify the contents, fail over during business hours with everyone watching, run the runbook as an exercise, load test to the point of failure so you know where it is. Every one of those converts a hope into either a fact or a repair, and the cost of finding out on purpose is invariably lower than the cost of finding out otherwise.