Earlier quoted context omitted.
We shouldn't underestimate what we don't know about building machines. Maintenance should accommodate time to check for wear on parts we don't expect to, to check things that are different from what we expect them to be. The problem was there for decades and one inspection in one unit over that period would probably have evidenced the issue and added one more item for the maintenance checklists. It's not like the man…
> Maintenance should accommodate time to check for wear on parts we don't expect to How much time do you spend performing exhaustive checks for problems that don't exist?
You absolutely should spend time looking for things you expect not to find, because sometimes you will be wrong.
When you're maintaining an aircraft, a nuclear plant, or anything sufficiently complex or life critical, you should spend a certain amount of time looking for ghosts. You _will_ find them. Your model of an engineered system should include a finite rate of unknown bugs.