Everything you wanted to know about Handwriting Recognition, but were afraid to ask

by David G. Propson


These days, the rollout of new technology resembles the type of psychological experiments scientists used to get in trouble for. Companies fling interesting but imperfect products into the world, and leave us to deal with them as best we can.


“Beta versions” of software have long been made available to die-hards who want to help programmers find the bugs in their systems, but many so-called final releases remain incredibly bug-ridden or, worse, just don’t work. We are all beta-testers now.


A new breed of applications now filtering into common use has already caused much consternation. Anyone using a handheld computer now uses handwriting-recognition software, and voice-recognition packages have made it onto some PCs. These innovations are meant to make communication between a person and a computer more like interactions between two people. I write a letter, you recognize it. I say a word, you respond.


So far, though, these interactions have become more human in just one way: My computer now misunderstands me just as much as everyone else does.


Come Again?
You see, even other people don’t always understand what I’m saying. I slur my words, I swallow them, I talk into my hands. And anyone at the office can confirm that my handwriting is sometimes inscrutable. By contrast, my keyboard lets me pour my thoughts and words into electronic form with almost perfect fidelity. That’s why, right now, voice-recognition is a good option only for people with vision limitations, repetitive stress injury sufferers, or sadomasochistic tendencies.


Handwriting recognition, on the other hand, is already quite common — at least, a version of it is. Graffiti, a program included in the Palm operating system, allows users to enter information in something resembling normal handwriting. Millions of people use it, and others use one of Microsoft’s PocketPC devices, an older Apple Newton, or a Palm with a handwriting-recognition program other than Graffiti.


One reason behind the commercial failure of the Newton — one of the first handheld devices — was its inadequate handwriting-recognition capabilities. The technology has improved much, thanks to the many handheld users who spent the late 1990s as paying guinea pigs, but it still isn’t perfect. I can almost imagine Palm guys in white coats watching me when I first played with mine.


Subject repeatedly pokes at the tiny screen with the even-tinier stylus, growing frustrated. After failed attempts to create a semicolon, he buries the pen deep in his tympanum.


Writing On The Wall
A program intended to recognize people’s own handwriting as they normally write it — that is, without a special alphabet — must pay attention to many factors. The shape of the character you create with the stylus, the speed at which the strokes are made, even the rate at which that speed changes. Tricks like these let computers recognize an individual’s writing with a high-degree of accuracy. (A few companies have even released signature-recognition security packages that replace all your typed passwords with a John Hancock entered through a special pad.)


Palm’s Graffiti doesn’t truly recognize users’ handwriting. It requires that every user learn a specialized version of the alphabet and distinctly enter each letter in a small square entry field. Creating a space requires its own keystroke, and all pieces of punctuation require two. In its own way, Graffiti is as rigorous and restrictive as using a keyboard.


However, requiring people to meet the technology halfway allows the program to be fast and relatively accurate: Palm has put its customers to work for it, and has ambitiously and audaciously attempted to re-teach them how to write. Imagine if the same attempt were made with voice-recognition technology: If people had to change the words they used to make themselves understood, would they do it?


I don’t know. But clearly we only have begun to realize the subtle and pervasive effects of the technological experiment we are all conducting.

Small Business Computing Staff
Small Business Computing Staff
Small Business Computing addresses the technology needs of small businesses, which are defined as businesses with fewer than 500 employees and/or less than $7 million in annual sales.
Previous article
Next article

Must Read

Get the Free Newsletter!

Subscribe to Daily Tech Insider for top news, trends, and analysis.