If a rule needs constant exceptions, the rule may be describing reality poorly.

Any rule needing an exception every time it is applied is not describing what actually happens. Instead of admitting that, people add conditions, and the rule survives while getting less useful and more confusing. The exceptions are evidence, and they are usually being ignored to avoid revisiting a decision. The pattern is telling you the model is wrong.

When you notice repeated exceptions, treat them as data rather than nuisances. Ask what rule would make them unnecessary. Rewriting the rule is usually simpler than continuing to administer the exceptions.