I like to say that the very best user interface is the one that does not need to exist at all because its function is performed automatically. That would certainly apply to telling any lunar lander designed today that it has indeed landed on the moon.
*cough* Mars Polar Lander *cough*
Ah hem, I assume you're referring to the microswitch design on the legs...well, inasmuch as there was no pilot on board, there couldn't be a pilot/machine interface, could there?
Edited to add: but this could be seen as a reasonable argument for another feature of the Apollo user interface mentioned in Aldrin's spiel: "Descent engine command override off". That's a really cool-sounding buzzphrase referring to a set of manual switches in parallel with the computer-controlled relay that fired the descent engine. They didn't totally trust the computer to not shut down the engine early, so this essentially required the astronauts to actively agree with the computer before this could happen. Since an early engine shutdown is what killed MPL, it might have made a difference -- had there
been a pilot.
Nothing in my argument excuses flying untested buggy designs. There were so
many problems with MPL, and so little data, that the board could not say that the microswitch problem actually killed it. They could only say that
if the mission actually made it that far, then it would
definitely have failed at that point.
A moderate amount of human supervision to detect and hopefully intervene when major problems appear in an otherwise automated system is a perfectly reasonable thing. But a design that
requires a steady stream of highly arcane and error-prone commands -- like manually poking values into a computer memory -- before it can even work at all is a different story. It's all about having machines do what they do best, and having people do what people do best, and not forcing one to do the other's job.
