Impossible-to-Use Software

Edmund X. Dejesus

THE TROUBLE WITH COMPUTERS: USEFULNESS, USABILITY, AND PRODUCTIVITY, Thomas K. Landauer, MIT Press, ISBN 0-262-12186-7, $25

The entire computer industry -- including this publication -- is predicated on the proposition that computers are useful: They do things people cannot do, or they help people do things better. In The Trouble with Computers: Usefulness, Usability, and Productivity , Thomas K. Landauer convincingly calls that proposition into question.

Landauer explains why computers have not increased business productivity significantly. He then addresses the main reasons for this and offers some solutions. He presents an impressive statistical analysis demonstrating only negligible productivity gains -- and actual productivity losses in many instanc es. In the sections expounding statistics, the going gets pretty dry, and I'm glad. I would far rather see a complete and thoughtful analysis of the statistical inferences, contextual caveats, and procedural minutiae than a one-page "Stats Lite" summary. The subject is too important to be treated cursorily. No one can dismiss this book as superficial or lacking depth of detail.

Happily, the text is leavened throughout with anecdotes of Landauer's encounters and misadventures involving computers. One of my favorites involves a point-of-sale system in an upscale store, with an "intuitive" icon-and-touchscreen interface. No one could figure out how to ring up a sale for candles. In this system's menu structure, "candles" were under "vases" under "platters" under "mixers" under "chairs."

In Landauer's analysis, phase one used computers for tasks that people cannot do (e.g., CAT scans, MRIs, and telephone switching equipment), or replaced people with computers for very simple or repetitive tasks (e.g ., automatic lathes and milling machines) or dangerous tasks (e.g., controllers for chemical processes). Understandably, phase one of computerization showed tremendous gains in productivity.

Under phase two, we use computers as tools to augment our activities: writing, drawing, communicating, organizing, administering, and so on. Phase two has shown little evidence of gains in productivity. In fact, according to Landauer, "investments in computer technology have yielded significantly lower returns than investments in bonds." He writes that the gains from such an investment in computers ($4 trillion since 1960; now over $300 billion annually) should be obvious, but they are not.

Naturally, the step from analyzing symptoms to hypothesizing causes is a difficult one, and it's vulnerable to criticism. Landauer unhesitatingly lays major blame on software, in particular the problems of usability and usefulness. He believes that software developers are -- mirabile dictu -- computer experts, people not likely to know, understand, or appreciate the difficulties faced by the mere mortals who must actually use their software. "Programmers and software designers are too smart," he writes.

User testing and user input are nearly nonexistent in the industry. "Many designers seem to be genuinely uninterested in testing." When companies want to beta-test their products, they turn to computer managers and support personnel who are -- mirabile dictu -- computer experts, not ordinary users. Managers bemoan the high cost of training users. However, the corollary is that software must be too hard to use if it requires such effort and training to perform the simplest tasks.

Landauer sees the solution to these problems in user-centered engineering. He has the "embarrassingly simple" notion that if we want users to achieve more, they should be considered from day one in the design, development, and deployment of software. He marshals considerable evidence to show that in the studied cases where this is done (a nd there aren't many, according to Landauer), the results are impressive.

Landauer has written the manifesto and planted a stake in the sand. We can keep dumping computer resources into a bottomless pit of unusabilty, or we can support and demand the effort to turn things around. Everyone in the computer industry now has to make a choice.

---

Edmund X. DeJesus is a senior technical editor located in BYTE's Lexington, Massachusetts, office. You can contact him on the Internet or BIX at edejesus@bix.com .