GUI Gaffes

Computerworld - Sometimes innovative GUI features don’t work out. User interface design experts weigh in on a few.

Adaptive Menus

Now you see them ... now you don't. Micro­soft Office's Intellimenus — an attempt to determine which menu items should appear based on usage patterns — only increased user confusion. A move to rafted tool bars, where two tool bars might share the same space and the computer decided which would appear based on use patterns, created similar problems.

“When you migrated from the short to the long menu, the [missing] items didn’t show up at the bottom of the list, so you had to rescan the whole list,” says Jensen Harris, principal lead program manager for Microsoft Office. “Not knowing what was on the menu confused people. What was on your menu was different from my menu, or if you went on vacation, things would disappear.”

Microsoft has pulled back from this idea. “We had design tenets for Office 2007 that said, “Don’t do adaptive behaviors,’” he says.

The Location-based Metaphor

Another failed attempt at a GUI innovation, Apple’s eWorld, “was too reliant on the metaphor for moving though space,” says Jakob Nielsen, a principal of Nielsen Norman Group. “Walking around on the screen is too much effort. In the physical world, you have to move, but in the computer world, why not just click on what you want? Why simulate something that has a more awkward interaction style?”

Microsoft's “Bob” offered a similar experience. “They used the rooms metaphor, and it was not helping people perform tasks and simplifying those tasks,” Nielsen says.
Clippie and other Anthropomorphic User Agents

Anthropomorphic user agents such as “Clippie,” the animated character used in the Office help system, evoked strong negative reactions from many users. “The trouble with Clippie is that it gets in your face,” says Don Norman, a principal of Nielsen Norman Group.

What’s more, the advice it offers often isn’t useful because it’s trying to infer the user’s intentions. “ [People] have enough trouble trying to infer another person’s intentions,” Normannotes. “Why would my computer be any good at it?”

“People like controllable, predictable, comprehensible and consistent user interfaces, not adaptive, anthropomorphic and agent-based [ones],” says Ben Schneiderman, founding director of the University of Maryland’s Human-Computer Interaction Lab. “It’s a distraction, and it takes control away from the user.”

But acceptance of anthropomorphic characters also hinges on cultural preferences, says Schneiderman. “In Japan, the use of avatars is pervasive. Even if you’re at a kiosk buying a subway ticket, there are little avatars bowing to you,” he says.