The most hypeable technologies are composed of a routine side whose function is instantly understandable combined with a miracle part that promises to make the routine perform way beyond experience. Cars that fly! Doors that recognize faces! Ads that present themselves only to serious prospects! The greater the contrast -- the more ordinary the routine part or the more exotic the miraculous part -- the greater the technohype potential, and the more carefully buyers need to keep their hands on their pockets.
In the late 1980s the drums began to roll for a technology intended to extend computing to situations where keyboards were impractical, often because the user was walking around. The idea was to provide a display surface the user could write on, like a piece of paper on a clipboard, except that in this case the paper would be able to associate patterns of strokes with specific letters. This would allow the device to store and process handwritten text as a computer-readable file -- in simplest terms, to recognize handwriting.
The routine part of pen computing (the pen itself) was as old as civilization. The miracle part (the handwriting recognition) was one of the defining problems of artificial intelligence. As a result, the technohype potential proved considerable. In April 1992 we reported that 78 percent of CIOs thought that pen computing would be of immediate value to their corporations, although no serious products were then available. By May we had jumped into the pool ourselves. "If you start your pen-computing evaluation and pilots now," CIO said optimistically, "you'll be ready for deployment when the products are ready."
Only a few months later, however, in the wake of the bankruptcy of a major vendor (Momenta Corp.), we started to have second thoughts. "Few existing pen applications have yet to graduate from their pilot programs," we admitted, a polite way of saying that most such programs had failed.
It turned out that pen computers were expensive (the function required powerful machines, which made equipping a field force pretty painful), power hungry and physically clumsy. Developers were limited to a very small applications library. Most importantly, the recognition worked poorly. This was a major blow, since people generally notice the miraculous part of a technology first. If that fails, the products often don't get a second look. "[Handwriting recognition] sank pen computing," observes Conrad Blickenstorfer, editor in chief of Pen Computing Magazine.
Look deeper, however, and our articles reflect glimmers of another approach to pen computing: using pens to steer through a series of forms, pick lists and check boxes. But this approach -- object selection as opposed to text entry -- did not seem to excite anyone much. The concept dated back to the 1950s, when radar operators began using light pens to track objects. Pick lists themselves went back to Apple's Macintosh. Furthermore, form-oriented applications came with a conspicuous price tag since they had to be designed anew for each application.
This was a disappointing retreat from the romance of handwriting recognition, with its promise of a single product that could handle any kind of data-entry application. Object selection seemed like a passing phase in the technology, as a way for pen computing to make a little money while it learned to read.
Some pragmatists discovered, however, that pick lists seemed ideal for field ooperation. Navigating through menus automatically associated each entry with a rich context of predefined categories. Handwriting recognition couldn't do that. Skilled users could cut through the process much faster than they could write the same information on a tablet. Finally, pick-list handling was easy on resources: Applications could run on a $200 handheld instead of a $2,000 laptop. It seemed that even if handwriting recognition had delivered exactly as promised, pick-list computing might have been preferable anyway.
Over the last five years, pick-list-based pen computing has spread to most categories of mobile workforces, from delivery to insurance to warehousing. Even though handwriting recognition is much improved (partly because of Jeff Hawkins' Graffiti, a special script that makes recognition easy for machines), the technology has been pushed to the bottom of the development tool kit. "A lot of what we do is make it possible for users to avoid [handwriting recognition]," says Ivan Philips, CEO of Pendragon Software, a form design tools company.
"Everything is done with clicks," CEO Charles Koo of iMedica Corp. says proudly of his company's wireless data-entry tool for doctors. Pen computing has more than fulfilled its original promise; it just had to forget about using the most fancy new technology to do so.
This story, "Write on Target" was originally published by CIO.