Looking Back--April 1999

Looking Back—April 1999

In order to learn from our mistakes, we wanted to analyze the results throughout. One observation that really stood out was the notion of “burden” or “load.” Many subjects complained that they had to “think all the time” and “search for the keys.” We had started with the knowledge that most users are familiar with the character grouping in a standard phone keypad, and assumed that this comfort level would be conducive to learning the one-row layout. Looking back after our two user tests, it seemed clear that this hypothesis was perhaps not altogether false, but at least overly optimistic.

We realized that the subjects often spoke in spatial terms. They would describe a certain key as being “on the left,” or in a “northeast” direction. Some users had become very quick with common sequences, for instance, short articles like “the” or frequent word endings such as “-ing.” They were typed almost automatically. This strongly indicated to us that the users’ mobile phone keypad skills were more “in the muscles” than “in the eyes.” So, instead of constantly gazing at the keys, skilled phone keypad users make use of their physical, procedural abilities in locating keys.

Another reason was also identified, also having to do with the spatial aspect. Whereas the standard 12-key phone keypad is two-dimensional, consisting of four rows and three columns, the one-row uses only one dimension, the horizontal one. This sounds like a small change at first glance, but it makes a big difference. In the phone keypad, the location of each key can be unambiguously identified by its orientation relative to a fixed origin. For example, if we take the middle key, the number key 5, as an anchor point, then each key has a unique orientation relative to this point. In the one-row keyboard this principle cannot be applied—all the keys are just “more or less to the left or right.” The one-row keyboard, therefore, allows users to leverage their spatial ability only to locate the keys’ approximate position; the final search has to be done using the visual domain. Therefore, although we had probably made users’ lives a bit easier by retaining the character grouping, we had broken a connection that was even more essential: location in two dimensions. The character D that used to be “in the northeast” was not there anymore.

This second reason is a serious concern, since it might mean that the one-row keyboard is inherently slower than the phone keypad. Whereas some very heavy users of mobile phones can enter text messages even with their eyes closed, they might not be able to perform this feat with the one-row keyboard, which requires the additional step of visual search.

Although the one-row keyboard can be used with several fingers, each successive key press has to be verified with a single pair of eyes!

The issues mentioned here are fundamental—so fundamental that, although we could identify them, there was nothing we could do about them without abandoning the core idea. The idea was dead. The one-row keyboard would never be a marketable product. The text input issue was so central to the concept that the user interface built around the one-row format was invalidated along with it. The design team started looking at new ideas.