Aspect-oriented programming could mitigate those problems. Xerox Corp.'s Palo Alto Research Center subsidiary this month released an update to version 1 of its AspectJ software, a Java development tool written with a grant from the Defense Advanced Research Projects Agency. IBM Research's HyperJ aspect-oriented tool for Java apps is available in alpha form on the company's Web site, and the company is developing a follow-up tool, code-named Cosmos, that creates a hyperindex of a program's requirements documents, drawing-board diagrams, and code as a reference for programmers. Cosmos could be ready for release by March.

At Sun, Gosling leads a research team that's trying to build development tools that represent programs' source code as graphical models instead of text. Semantic models that generate source code could make it easier for programmers to add features to and debug large systems, Gosling says.

The most valuable programming reflects a design's complexity, and that should include nonprogrammers' expertise, Intentional Software co-founder Simonyi says.

In September, Charles Simonyi, a senior technologist at Microsoft for the past 21 years, who while at PARC in the '70s invented a pioneering word processor, Bravo, that became Microsoft Word, and Gregor Kiczales, a researcher who developed AspectJ at PARC in the '80s and '90s, quit their jobs and founded Intentional Software Corp., an independent company funded with Simonyi's personal fortune. Intentional will develop tools for programmers that help capture and reuse nonprogrammers' expertise. In some cases, the tools open the world of programming to specialists such as engineers or chemists, helping generate code from their vocabulary of symbols, tables, and equations.

"Design is the sum of the contributions of the stakeholders that surround a software project," Simonyi says. "The value of software is in the reflection of that complexity."

If aspect-oriented programming catches on, it could improve software quality and lower IT development and maintenance costs by automating more of the dialogue between developers and businesspeople. That's important as systems become larger and more complex. "The strategic value of software should be that it's changeable," Kiczales says. "Programmers using Java or .Net today can write programs that only the best could write 10 or 15 years ago. People are really struggling for the next step after object-oriented programming."

Maintaining, changing, and repairing software consumes about 70% of a project's time, says William Griswold, an associate professor at the University of California, San Diego, who studies software change. "The majority of costs are lodged in those changes and updates," he says.

Meanwhile, software quality remains a nagging problem at most companies. According to a May InformationWeek Research Web survey of 800 business-technology professionals, nearly 90% say bugs have cost their company lost revenue, higher costs, or both. More than 60% say the computer industry does an inadequate job of ensuring that commercial software is bug free.

"Aspect-oriented tools force you to think at a higher and more concrete level," says Juri Memmert, whose development shop, JPM Design, has designed software for companies such as Volkswagen AG and Dresdner Bank AG. "If you don't do that, then you're basically hosed, no matter what you're using."