The software is written using the high level architecture (HLA) programming standard prescribed by the Department of Defense. This protocol enables developers to mix programming languages and operating systems and create interoperable "war game" simulations, which are used by a number of organizations including NATO and the Institute of Electrical and Electronics Engineers.

Using HLA, Sandia has created what they call the Enterprise Modeling Architecture, which houses a database of information such as a list of diseases and their treatments plus demographic statistics of a given municipality's population. The program filters these details of the simulation. "We can have any number of clients participate in this simulation and interact with these different models but only based on their role," says Johnson.

For example, a user can log in as a public health official and receive detailed information on both Anthrax and Cipro, whereas someone logged in as a first responder-a fireman or ambulance EMT-might only see emergency room reports, which contain a bare minimum of facts. Both people can work through the same incident, and see how the decisions of each affect the other.

"It's a valuable tool," says Michael Nacht, dean of the Goldman School of Public Policy at University of California Berkeley, who tested the program last summer. "It gives you a flavor of the uncertainty in a crisis environment in the event of a domestic weapons of mass destruction attack."

But according to Nacht, who also chairs a panel on the U.S. security department's Threat Reduction Advisory Committee, some things are impossible to anticipate, let alone simulate. For example, a major Anthrax attack on a metropolitan area will create a crisis demand for Cipro. If there are limited places where the drug is available, people may line up literally for miles, a situation that might lead to panic. "A lot of things that are plausible are not necessarily included in the model," says Nacht. "The secondary and tertiary consequences of the crisis don't tend to get built in."

Currently, the project is internally funded, but other government agencies, which Sandia declines to name, are interested in creating larger bio-defense and nuclear counter-terrorism models for other metropolitan areas. "It is my belief that these problems will play out at regional levels," Hirano says, "yet they'll have national consequences."

And because these consequences are still so unknown, simulation programs are crucial. "We need to do it through simulation," says Nacht, "because we have no data for the actual event-thank God."