If you have two security models that provide identical actual security, and one of them is invisible to the user and the other one is outright user-hostile like

If you have two security models that provide identical actual security, and one of them is invisible to the user and the other one is outright user-hostile like the TSA, yes of course the invisible one is better.