With all due respect to fellow computing gurus around the world, the University of Maryland's Ben Shneiderman doesn't think speech will ever become the main way people communicate with computers. He's convinced our eyes will do better than our voices at helping us control the digital machinery of the 21st century.

His explanation seems so simple, so obvious, so testable, that you have to wonder why Microsoft, IBM and other research labs have been spending billions of dollars trying to let us talk to computers the way people did to HAL in "2001: A Space Odyssey."

"Hollywood and the image of HAL gave us this dream, this hope, this vision, but the reality is quite different," says Shneiderman, a computer science professor and well-known researcher, sitting in a College Park office more cluttered with books than computers. "It turns out speaking uses auditory memory, which is in the same space as your short-term and working memory," he adds.

What that means, basically, is that it's hard to speak and think at the same time. Shneiderman says researchers in his computer science lab discovered through controlled experiments that when you tell your computer to "page down" or "italicize that word" by speaking aloud, you're gobbling up precious chunks of memory -- leaving you with little brainpower to focus on the task at hand. It's easier to type or click a mouse while thinking about something else because hand-eye coordination uses a different part of the brain, the researchers concluded.

The upshot, Shneiderman contends, is that while speech may help blind and disabled people interact with computers, it's unlikely to become the dominant way people connect with them.

"It's the bicycle of user interfaces," he says of human voice, his own sounding a tad weary, as if he has said this many times before. "It gets you there . . . but it's not going to carry the heavy load that visual interfaces will."

Visualization, you see, is Shneiderman's thing. You can tell by how his voice springs to life when he starts talking about the visual interface projects underway at the University of Maryland's Human-Computer Interaction Lab, the research group he founded 19 years ago to develop novel ways of interacting with computers. When the public descends on his College Park lab for its annual symposium and open house May 30-31 (www.cs.umd.edu/hcil/soh), it will see an array of new graphical tools for exploring information.

Shneiderman thinks visual tools are what will let humans master computers. If he's right, the next-generation Internet may have fewer software "robots" than most pundits predict. Or if software agents do catch on, visual tools may be how we control them.

Control, after all, is what Shneiderman thinks is still missing from the computing experience. Computers and the Internet are too darned frustrating, he says, and the only way to put people back in control is through new software designs that are more human-centered, chiefly by leveraging our powerful visual sense.

His latest visual tool is the "timesearcher," a graphical box that lets people ask questions about massive amounts of data and see the answers visually. Instead of having to type each question using words and numbers ("show me all the stocks that rose in price more than 30 percent between January and April," for example), the timebox lets you drag around a box on the screen, shrinking or expanding it to explore complex relationships among data over time, with results displayed instantly in an adjacent panel.

Wall Street analysts are testing his timeboxes for technical stock analysis. "But our interest really is in genomic data, where you have 10,000 genes on a DNA microchip and you are looking for patterns over time," Shneiderman says.

His team of more than half a dozen researchers is led by Ben Bederson, a younger version of Shneiderman who has taken over as director of the College Park lab (colleagues call them Ben S. and Ben B.) and is equally committed to the power of visual tools.

Bederson developed software to browse the thousands of digital photographs he had taken of his 3-year-old daughter. The software, called PhotoMesa, lets people see a ton of image directories and thumbnails at once, then zoom in to get a bigger view of any particular image or group. After downloading the free software from the lab's Web site, I used it to find, in about 10 minutes, a photo I had searched for in vain for months among the thousands of digital photos I have stashed on my home computer.

The lab has had its share of commercial successes, notably with software that creates graphical sliders to let users quickly refine searches or try out what-if scenarios based on different variables. The so-called dynamic search query software was taken to market by a Massachusetts company called Spotfire and is being used by big drug companies and other businesses.

Many programs emerging from the lab (including funny-looking fish-eye menus where type starts small and gets bigger before turning small again) let users visually zoom in and out easily, the idea being that people grasp data better if they can associate it with the big picture. You can see for yourself at SmartMoney.com's online map of the stock market, which shows 500 stocks in a maze of overlapping rectangles, with green representing companies and sectors with rising stock prices and red depicting falling ones. It was developed with mathematical formulas for "treemaps" created in Shneiderman's lab.

Shneiderman is best known for inventing a form of hyperlinked text called "Hyperties" in the 1980s, a forerunner of the World Wide Web's hyperlinks. He was right about the power of visually linked text back then. There's no reason to think he isn't right now about how timeboxes, dynamic query sliders and similar graphical interfaces will one day let us discover startling truths -- much as Galileo shook up the 17th century when his telescope revealed craters on the moon and led to the discovery of galaxies far, far away.

"These tools are like telescopes and microscopes," Shneiderman says. "They are a new way of viewing things you couldn't see before."

For a Web video tour of the Human-Computer Interaction Lab, go to www.washingtonpost.com/walker. Leslie Walker's e-mail address is walkerl@washpost.com.

University of Maryland professor Ben Shneiderman sits in front of a "treemap," a method he devised for data visualization.