Interaction, Usability, and the Dancing Elephant
By Michael Nygard, Wed Aug 02 00:00:00 GMT 2000

(August 2, 2000) We all know mobile terminals are assuming many roles formerly reserved for desktops. But we often fail to recognize the not-so-obvious trade-offs, and find it hard to change our ways.

Mobile Internet applications are still in the dancing elephant stage. At first, it's pretty amazing that the elephant dances, then it's pretty impressive to see that much mass moving, but after a while, it's just not very pretty at all. These applications suffer from the worst usability since the days of the teletype. Let's consider the deficiencies of WAP apps with typical desktop applications. What are the biggest differences in usability between desktop and mobile applications? Yes, the screen is smaller. Yes, one of them can go wherever you do. OK, now that those are out of the way, let's dig a little deeper. Usability is about more than screen size or bitmapped buttons. It is about how easily a user can get the job done. It is about the system helping, instead of impeding, users' interactions.

(By "desktop", I am referring to the WIMP--windows, icons, mouse, and pointer--environment and its complement of applications. I am not speaking specifically about Windows, Macintosh, or X desktops, but rather, the modes of user interaction employed by those popular environments. Of course, it's easy to point to the worst offenders against each of these principles. Nevertheless, the principles are valid, even if there are violators.)

In the language of patterns, a solution balances forces in a context to produce a resulting context. (One hopes that the resulting context is better than the original!) If you change the initial context or the forces at work, the solution must change, too.

Desktops have evolved to respond to a certain set of technological forces: stationary use, large bitmapped screens, multi-channel input, and huge computing resources. By balancing these forces in the context of human psychology, desktops have evolved several attributes that aid usability.

Spatial Recall

The human mind excels at finding things again. This form of memory is called "spatial recall". Spatial recall helps a person find something by providing the general location. Once in the vicinity, that person can find the target for fine location. This works for hunters returning to camp as well as users repeating an action. Once a user moves beyond the "rank novice" stage, he often uses spatial recall to start moving toward the target. You can see this by watching someone move their mouse before their eyes move to the target region. The less a target control moves around the screen, the stronger this effect is. The Windows "Start" menu or the Macintosh Apple menu are so tightly bound that a user can throw the mouse to them without looking at all. (As an experiment, try moving a Windows user's task bar to the top of their screen and count how often they send the pointer to the lower left corner.)

Of course, spatial recall applies best when spatial navigation means something. Even so, selecting a menu item can be aided by spatial recall.

Spatial recall doesn't work if a control is overloaded. If the same two buttons perform every kind of accept, reject, forward, back, and submit action and their meaning depends on what is on the screen, spatial recall will not help at all.

Persistent Application State

If I am reading a news site at my desktop, and I see a good report about FutureMegaCommTech, I can flip to another window and buy that stock. With the mobile version of these two applications (news, stock trading), I cannot. The mobile applications do not maintain their state as I move from one to the other. (Strictly speaking, there is only one application at work here in each case: the web browser. The user perceives each task as a separate application, however, and reasonably expects to come back to the same place in application 1 after doing something entirely different in application 2.) It frustrates me to go through the same menus from the beginning to get back to where I was.

This doesn't happens because the screen is small, but because there is no way to tell the minibrowser that I want to come back here after doing something else. Bookmarks are too permanent, and the "back" key, even if present, is too difficult.

Broad Input Channel

A keyboard has 104 keys. Add the control, alt, shift, and "windows" modifiers, plus various button clicks and double-clicks, and you have around 270 different input "gestures". For all except the hunt-and-peck typists, these gestures can be activated about 5 per second. Common gestures like Ctrl-S can be activated even more rapidly. Even compared to T9 text entry, this is a virtual boulevard of input. Surely, this broad input channel contributed to the development of very wordy, text-oriented applications.

Multi-Modal Feedback

From the blinking caret to dialog boxes, from progress bars to hourglass pointers, desktops provide feedback through many channels: visual, aural, textual, and even tactile. The desktop and its applications cannot provide too much feedback. In fact, a common complaint from users and designers alike is, "We need more and better feedback!" Communicating context requires a lot of effort. On a commercial web site, 25 to 40 percent of the area is devoted to communicating the user's static context. In the realm of huge real estate, the sacrifice of that space not only acceptable, but well worth it. On the opposite end of the spectrum, a typical WAP application devotes no effort at all to communicating context, static or dynamic.

Can these attributes be duplicated with mobile devices? Probably not. Moreover, duplicating the desktop is the wrong goal. Putting icons or a taskbar on a phone will not grant it the abilities of a desktop. These all evolved from a different context. Mobile applications will evolve, too. They will not evolve by adding desktop features but by balancing the forces that impress upon our devices and our users. Let us not waste our effort and imagination in fruitless emulation, but instead spend our time in new invention!

Michael Nygard is Chief Scientist at Javelin Solutions, where he analyzes emerging trends in the network economy. His experience runs the gamut, covering scientific, military, financial, educational, banking, and manufacturing applications. Michael is focusing on true integration of wireless devices in the enterprise.