So while that is true (and not to go too far off into the weeds on an analogy), in an emergency, people are trying to follow procedure under pressure and the odds of error in operation of an interface increase. You want the interface that is used in an emergency situation to either be well practiced or absolutely as intuitive as possible.
To destructure the analogy and give a concrete example, if I'm dying of allergic shock I don't want my doctor unable to access my medical history because somebody in the process can't remember how to "break glass" on the encryption on my medical records, even if there's a procedure to do so. I want my records in plain text format and as readable as possible.
Your concrete example would never happen the real world. If you were in anaphylactic shock, no doctor is going to go off looking for your medical records first, even if they were sitting on the table next to him. He's just going to stick you with epinephrine and then MAYBE look at your medical records later.
All that said, I get your point, but I'm not sure how it applies to this discussion anyway.
My point is there's a tradeoff between prevention and penalty. Sometimes, the best option isn't a locked door; it's a clear sign saying "trespassers will be prosecuted" and a security camera (i.e. auditable access, not preventative access denial). That way, people can get in if they need to to do something critical, and one can resolve the question of trespass later.
- If you normally keep the door locked, unlocking it with the thumb turn is what you do every time anyway
- In cases where there are many people expected to use doors that they're unfamiliar with, it's typically even simpler to exit (panic bar on business fire exits, automatically-unlocking deadbolt on hotel doors, etc)
