How Do You Know?

Many usability professionals believe that you cannot know whether an interaction is good unless you test it. That's why they are constantly asking, "How do you know?" But I have noticed something very curious. When they ask, they are not playing devil's advocate. They are asking for the simple reason that they really don't know good design when they see it.

At least four large companies that I work with have a long history with usability professionals. The companies decided to invest in usability. They hired professionals who built their labs, performed their studies, identified likely problem areas, and made a series of guesses about how to improve things. The programmers diligently made changes to their programs, and not much happened, except that the programmers had worked a lot harder. After a few cycles of this, the programmers simply gave up, and so did most of the managers. They could see that it was very expensive and time consuming, yet it wasn't solving the fundamental problem.

Interaction designers rely on their experience, training, and judgment to make an accurate evaluation. They have principles, idioms, and tools for every situation, and they triangulate these tools with other sources of information. How does a radiologist know that someone needs surgery from examining an X-ray? X-rays are so difficult to read that it is hard for a layperson to imagine reading one, but trained doctors do it all the time. How does a judge know whether a defendant is guilty? How does an investor know that now is the time to buy? These professionals might not be right all of the time, but they don't guess.

I have seen well-respected usability professionals blandly take potshots into the dark. They developed sophisticated tests to isolate users' reactions to existing software, and then they studied the tabulated results to find interactive rough spots. When their rigorous, scientific method uncovered a problem area, they would lapse into the most amateurish babbling about solutions: "Well, I guess we could move this control over to this dialog box," or "I suppose that if we added a button here the user could have better control."

It is fine to say, "I don't know" but very self-defeating to guess at answers. What's worst is that any gazing off into space and guessing will cause programmers the ones with skin in the game to quietly write you off as a quack.