I got a strong reaction when I criticized Gmail's interface. A few people mentioned that you can activate hot keys so you can just press the r key to reply. Fair enough. I burrowed into the Settings and activated the hot key feature. It was a good idea, but nothing happens when I press r. You'll tell me this is a case of operator error, which obviously it is, and that is my point. Why is it so easy for the operator to make an error?

Someone mentioned that clicking the big unlabeled white box under the message is one of several ways to initiate a reply. I had always wondered why that big white space was there. This invites the obvious question WTF? How is a user supposed to know that a big unlabeled white space is a reply button, especially when it is right below the button labeled REPLY that seems to do the trick all by itself? And that REPLY button is not to be confused with the other REPLY button at the upper right, which brings me to my next point.

You might think that having more than one REPLY button would make it easier to find at least one of them. After all, two is better than one. But that's not how your brain is wired. Allow me to give my favorite overused example from writing:

If you say, "The ball was hit by the boy," it means exactly the same as "The boy hit the ball." But the first example makes your brain work harder to untangle the meaning. You're wired to first figure out who is the subject, then figure out what that person is doing. As a writer, you try to be conscious of anything that makes the reader work too hard. Likewise, as an interface designer, you presumably do the same thing. To that end, I want just ONE way to reply to email, damn it. If there is more than one way, I have to make a decision. That's working too hard. I'm no interface design expert, but I have to think that putting two buttons with overlapping functions on the same page is a "don't."

One commenter on this blog confessed to being the designer for the Gmail interface and listed his credentials. Assuming that message came from a real person, and I think it did, I am forced to reassess my sweeping statement that no qualified person worked on the interface design. Now my assumption is that there was a management failure. Either Google didn't do usability testing on civilians, perhaps because of budget or timing issues, or management interfered with the design in some other way. That's my best guess.

A number of people snarkily noted that there are usability issues with Dilbert.com. Of course there are. We don't do formal usability testing, as it would be cost-prohibitive, and it could take months. So we fix the big bugs and save up the usability comments for the next revision. Inevitably the new version introduces new confusions while fixing the last ones. I imagine that 99% of web sites are designed without rigorous usability testing in a lab setting.

One of my corporate jobs, in my previous life at the phone company, involved working closely with Pacific Bell's Usability Testing Lab, so I got to see how useful that process was. A highly qualified interface designer can only get you halfway to where you want to be. You need usability testing for the second half. The Gmail interface looks half done to me.