March 2, 2005
Reaching a decision with colleagues who disagree is a common challenge for federal executives. But a new computer program is designed to make that task a little easier.
Steven Popper, professor of science and technology policy at RAND Graduate School in Santa Monica, Calif., and his colleagues have developed a way to make decisions without first agreeing on a set of assumptions about how the world works--a useful tool, especially if the decision-makers come from diverse backgrounds.
The software, called CARs, takes advantage of the ability of computers to make millions of calculations in a few seconds. Can't agree on whether the population will hold steady or rapidly expand, or whether environmental concerns should trump economic growth? No problem. CARs -- the first three letters stand for computer-assisted reasoning -- will work out what can happen under each set of assumptions.
Most modeling programs lack this flexibility and force users to share assumptions, such as the rate of population growth or likely government regulations. Popper and his team wanted to create a modeling program that didn't fail when users disagreed on the basics.
"It became clear that in order to achieve this vision, you need a gizmo, a piece of software," he said. CARs has been used by the National Defense University, the Defense Advanced Research Projects Agency and Volvo Car Corp.
Desmond Saunders-Newton, a visiting associate professor at the University of Southern California, used the software when he coordinated an examination of unstable countries with experts from the U.S. Agency for International Development and the State and Defense departments last September. The team wanted to figure out which countries were most likely be unstable in the hopes of preventing the need for U.S. military action down the road.
While military personnel tend to accept models, AID and State workers usually distrust them, Saunders-Newton said, because "they think they reduce the complexity and richness of the world." But because CARs allows for a range of assumptions on factors such as demographics and politics, it was appealing to all the participants, he said.
"The ability to play with this really caught their attention," Saunders-Newton said. "If you make tools of this type available…you're actually opening up a whole new line reasoning and inquiry for decision-makers."
Popper offered the San Francisco Bay as another area ripe for CARs analysis. Interest groups have different ideas about the balance between environmental protection and economic growth and would not likely resolve their differences by sitting in a room and debating. With the CARs model, however, groups could come together and simulate what could happen to the area based on different assumptions, such as a certain level of economic growth, temperature changes, government regulations and how quickly pollutants were flushed out the system.
The computer then generates different future based on those assumptions. While participants might not agree on which is the "best" future, they will end up with more information about the effects of certain factors.
Without the flexibility of incorporating a variety of diverse assumptions into a model, participants can be derailed before they even start debating decisions.
"The current set of tools are inappropriate for what we need to do to face tough problems," said Saunders-Newton.
March 2, 2005