I think the big question is whether you ever win the previous war.

If every problem adds permanent overhead ("add a check for that issue in the review process") then you're slowly losing ground. If problems get solved ("automate that step so no one forgets it"), you're making progress even if you don't anticipate issues.