the perfect measure of UI success, is based on the total efficiency of the system. If 10 million users have a problem with a driver that could be easily fixed

the perfect measure of UI success, is based on the total efficiency of the system. If 10 million users have a problem with a driver that could be easily fixed with one command line call. you still have: 10 million * (30 min on google + 1 hr reading docs + 1 min typing the command line) that is a 15.0166 Million person hour loss. Making the process more efficient, does not involve making it easier to Fix the problem for the user, it involves making the problem not existent. The number 1 rule of design is "if a user does not need to make a decision, then you should make it for them." When I was in Japan, you could put your subway ticket in any direction, and "IT WOULD JUST WORK," the machine did not spit it back out at you and FORCE the USER TO DO IT THE RIGHT WAY, because it was a waste of time to do so. Instead the machines are incredibly more complicated to account for someone putting the ticket in wrong. It made the whole subway experience work much more smoothly. Now people who can not read, make a mistake, or are completely foreign to the system, can JUST USE IT. They do not have to read docs, to do one simple thing, pay for the subway ride that should be fast, simple and easy. The whole software industry could use a little bit of time learning industrial design.