Getting personal. (editor's note).(computer systems for the graphics industry)(Editorial)

Ask any group of people you know to list the ways computers are changing, and chances are they would start by saying that systems are getting smaller, faster, and cheaper. After that, they might add that computers are becoming more ubiquitous and connected. And finally, they might mention that input and output devices are improving or perhaps even getting better at sensing their surroundings. In fact, this is the list most people would probably come up with, regardless of their age, education, or profession. And they would all be right.

But is the list as it should be? Is this the right order of research priorities for computer system developers? No way. In fact, the items on the list are in exactly the reverse order if computers are to better meet our needs and if the computing industry is to halt its fiscal decline. One industry visionary who promotes this view is Bill Buxton, chief scientist at Alias|Wavefront and a speaker on a panel titled "The Future of Computer Graphics" at the most recent Siggraph conference. The problem with focusing on making computers smaller, faster, cheaper, and so forth, he says, is that every one of these is a technological attribute, but technology for its own sake doesn't matter in terms of what's going to have social and economic importance.

Buxton is right. Rather than taking a techno-centered approach, developers need to begin creating computers that serve people, instead of continuing to require that people serve computers. What should be at the top of the priority list, instead of at the bottom, is the development of systems that can make sense of their environment, in terms of recognizing the identity, location, movement, and speech of the people around them. For example, understanding a person's identity and location is critical because it provides context, Buxton explains. "When I'm at the podium, you know I'm the speaker. You know that the other panelists are speakers, even though they're not speaking. We need to spend time making such semantics implicit for machines so we don't have to specify them explicitly." But this has not happened because we're still focused on technology rather than on people, he says. "And there's not a computer science department in the world where in order to get a degree in computer science it's necessary to have ever written a program used by another human being."

While that may be true, at least one university-based alliance is striving to create a new, "human centered" form of computing. Called Project Oxygen, the MIT-industry partnership aims to develop systems that will enable computers to recognize us, understand our gestures, converse with us in plain English (or other languages), and even interact with other machines and computers without our intervention, ...