In the context of comp-sci, I'd say it applies to requiring a Strict Serializable consistency model merely because it's easier to reason about it, even if some other model would be a better fit to the problem at hand.

Double plus if you use that model because you don't know of the existence of the others.