5 Technologies That Will Change the World

That's where Ho's sick servers come in. By creating mathematical models of diseases such as diabetes, obesity, asthma, or arthritis in a computer, researchers can run virtual tests of their new drug candidates -- much in the way that an aeronautical engineer uses a computer simulation to imagine how an airplane design will perform once it's built. Often called "biosimulation," the approach compiles everything that is known about a given disease -- even down to the activity that takes place inside a single cell. And the computer models can be updated as scientists learn more about how the diseases work.

Researchers can anticipate bad reactions before they give a drug to animals or humans, and they can run many more tests on a computer than they could run in the real world. Ideally, biosimulation will help Johnson & Johnson and other pharmaceutical companies focus their efforts on the drug prospects that are most likely to succeed.

"Even by the time a drug candidate gets to Phase III clinical trials -- the last stage before it reaches the market -- the failure rate still approaches 50%," says James Karis, the CEO of Entelos, the Foster City, California, company that supplies biosimulation technology to Johnson & Johnson. "That's after eight or so years of research. Those failures are very expensive." (Today, the standard figure for bringing a drug to market is between $800 million and $1 billion.) Simulation software will make those failures less painful and help pharmaceutical companies find useful drugs sooner.

Ho's team used biosimulation to reduce the amount of time and the number of patients required for the first phase of clinical trials of a new, as-yet-unannounced drug for Type II diabetes. (Ho estimates that the software, in its first outing, saved between six and eight weeks in trials.) Using sick computers as a stand-in for sick humans is still a new idea that will have to prove its value by contributing to the development of important new drugs. But eventually, Ho predicts, "this will become commonplace. It's a tool we never had before."

Nagui Halim believes that there are few constants in the world of computing. Chips get faster and more powerful, storage gets cheaper, and communications bandwidth keeps increasing. But one thing doesn't change: Our computers are still horrible at coping with problems.

"People are excellent at handling changes in the environment, or in their own body," says Halim, the director of distributed computing at IBM's Watson Research Center in Hawthorne, New York. "If you have too much work, you know how to prioritize and do the work that matters most. If you're feeling sick, you might lie down for a while." But even the most sophisticated computers aren't self-aware enough to know how to handle stress, or react to their own health problems.

Halim is part of a group at IBM that's working on what the company has termed "autonomic computing": developing computers that are smart enough to configure themselves, balance intense workloads, and know how to predict and address problems before they happen. At IBM, the leader in the field, the annual research budget for autonomic computing approaches $500 million. And the quest to develop systems that take care of themselves isn't just an abstract research initiative: Its fruits have begun creeping into Big Blue's product line.

"We've already got storage management software that can tell you when a storage device will fail before that failure happens," says Alan Ganek, vice president of IBM's autonomic-computing initiative. "And we're selling database software that can recommend a configuration based on the hardware environment you're running it in. Most database administrators had previously done that by trial and error."

The long-term promise of self-aware computers and software is greater reliability with fewer human baby-sitters. Right now, Ganek says, IT staffs at large companies are swamped with the tasks involved in "managing, maintaining, upgrading, and the care and feeding of their systems. That work squeezes out any innovative projects that they'd like to be doing to establish a competitive advantage."

Imagine, Halim says, a system that is smart enough both to see that online orders are spiking as the holiday shopping season approaches and to temporarily commandeer a bit of extra processing power from the human-resources server so that it can handle the influx of orders. Personal computers might know when a software upgrade became available and install it themselves. But as Adam and Eve discovered, self-awareness and sin often go hand in hand. A key challenge for IBM will be imposing restraints on this smarter generation of computers, so that your PC doesn't go out and spend $100 upgrading to Windows 2005 without your permission.