I ride the bus a fair bit and I've had a chance (since I'm just sitting there) to do some User & Task analysis.
User Goal: Get off the bus.
Task: Open the door.
Task Completeness: I'd say 3 out of 10 users are able to accomplish the task without 'incident'.
User scenarios: Touch the door, bang on the door, wave their hands furiously at the door, yell at the bus driver, run to the front of the bus, kick the door, swear.
Calgary Transit bought some new buses awhile ago that have this engineering miracle. It uses motion sensors above the door frame to activate the hydralics to open the doors. Only it doesn't work very well for a couple of reasons:
1. Riders are 'used' to the old doors. The old doors have a crash bar on them that is a well-known affordance - push on the bar with some force and the door opens. The new doors are asking the user to abandon an established mental-model in favour of something else and it doesn't work very well.
2. The new doors aren't 'engineered' to be usable. Oh, they've been engineered alright - to work the way the engineers intended, but without thought of how a user will interact with them. Rather than push on the door (which seems to lock up the process), the user has to lightly touch or wave their hands (like an automatic hand dryer). Then, they have to WAIT for the door mechanism to cycle and open the door automatically. If the user pushes on the door prematurely, again it freezes up.
Even when familiar with the doors workings (as an experienced user), I still encounter difficulties with the mechanism. The engineering is not user-centric and I would daresay, faulty.
Jacob Nielsen use to say that any user testing is better than no user testing. I wonder if the folks that built these doors even bothered, or didn't like what they saw...