Naming is probably the single hardest part of writing scalable, legacy code. Cache invalidation is a cakewalk in comparison. I believe this is in part because naming is a miniature exercise in empathy for future readers particular of a piece of code.