IBM's service science

For years, IBM has been one of the world's leading research bodies when it comes to semiconductors, databases, electron microscopes and other "hard" sciences.

Now, Big Blue is getting into social sciences.

The company's Almaden Services Research group, a 22-employee outfit based in Silicon Valley, has set out on a mission to discover--and then hopefully exploit--quantifiable, predictive principles that underlie the delivery of technology services.

For many physicists and chemists, the social sciences often rank up there with numerology and voodoo.

In other words, IBM is combining anthropology, game theory and behavioral economics with technologies from its labs to see if it can make corporate processes run smoother. The first person recruited from outside IBM to join the group was, in fact, an anthropologist.

"A lot of social-science research has been performed on emerging, indigenous cultures," said Jim Spohrer, director of the group and the former chief technical officer of IBM's venture arm. "But the cultural variations inside of businesses are just as varied as those in the jungles or deserts."

The goal of the program essentially is to better marry the company's extensive research division to the fast-growing services group.

In 2003, the group worked on some 100 customer projects in the On Demand Innovation Services program. Assignments have included finding a way to implement speech-recognition systems so that customers or employees would actually use them.

But Spohrer admits that he has loftier ambitions, too. His fantasy goal for the group is to win a Nobel Prize in economics. In an early assignment, staff members had to read the major works of previous winners of that prize.

For many physicists and chemists, of course, the social sciences often rank up there with numerology and voodoo--amusing, but not really science.

Corporations are notorious for introducing technology without considering the human consequences.

An electron can't lose its charge, but die-hard progressives can, one morning, start watching the Fox network.

Anthropology, in particular, is suspect. It's the college major of every parent's nightmare. As a joke, a friend and I once wrote a treatment for a business book we would have called, "Management by Maya--How the Secrets of the Ancients Can Improve your Bottom Line." It provided lessons from ancient cultures (along with tips on handling a jade disemboweling knife.) We didn't get a publisher.

Still, the dynamics of human behavior, especially in large groups, can be encapsulated to a certain degree. One of the lessons learned in the movement toward industrial automation in the '70s was not to automate too much, Spohrer noted. Employees got bored, because they had little to do, and when a crisis arose, they couldn't handle it.

Over time, companies building those systems changed them so that there would be a more constant stream of human-machine interaction.

"How do you take social change and reduce the time to implement it?" is the question at the heart of this type of work, according to Spohrer.

Moreover, the need to get a better grasp of human organization has become pressing. In 1945, approximately 60 percent of the U.S. work force was employed in the manufacturing sector, according to statistics from the U.S. Department of Labor.

Now, roughly 50 percent of that employee pool has an office job. The dynamics of office work, however, are not completely understood. Companies often try to tweak the compensation plans for sales representatives, and then they find that they spend most of their time trying to swipe accounts from each other.

Corporations are notorious for introducing technology without considering the human consequences.

Eric Johnson, a professor at the Columbia Business School, has found that maximizing "stickiness"--or increasing the time a customer spends on a Web site--is actually a bad goal for e-commerce companies to pursue. Stickiness usually connotes that customers are getting lost on a Web site. By contrast, Amazon.com has seen revenue rise and stickiness decline, because easy navigation brings customers back.

Other studies have shown that the number of employees that participate in employee retirement plans changes drastically when default settings are slightly adjusted on the Web-based enrollment form.

"Defaults are probably the most important decision you can make on Web sites," Johnson said.

IBM's foray into the social sciences is only about a year old, but progress is being made fairly rapidly. In May, academics from the University of California at Berkeley, Harvard University, Northwestern University and elsewhere will converge at IBM in New York for a faculty summit on organization theory. Spohrer's group is also digging around IBM for interesting projects to tie into.

One promising area of development is the combining of these techniques with IBM's data-mining research. Ideally, systems could be developed that would allow humans and machines to detect bank accounts that are serving to fund terrorist activities.

Over time, institutions may be able to carve out enough data and concepts to give shape to academic disciplines focused on these areas.

"Humans are intentional agents, and intentional agents can resist or accelerate change," Spohrer said. But studying that problem "is exactly what we have to do."