The Brookings Institution has modeled all 350 million Americans, their genders and ages, their location by zip code and their daily travel habits. The large-scale agent model lets researchers program rules of behavior and additional demographic information into each agent, rendering them pretty fair representatives of the people they represent. These are pretty rudimentary as computerized stand-ins go, nothing like the 3D, full color, dressed, buffed and made up avatars strolling the streets of Second life. Brookings' agents are just little dots of color, even less from a distance.
Nevertheless, the model marks a huge advance toward the day when government officials will be able to model how well or badly different programs and policies address the problems they are intended to solve. The model is big enough to just about encompass the current world population of 6.7 billion—it can house 6 billion agents--offering the possibility of testing responses to global pandemics, the worldwide effects of alterations in trade or monetary policy, and the human cost of natural and political disasters, among many other things.
The ability to grow artificial societies allows government policymakers and officials to watch how social, economic, biological, and civil events develop and then to see the effects of governmental and other interventions on the outcomes. Such agent-based modeling took hold in the social, biological sciences and economics in the early 1990s. But only now have they become big enough and flexible enough to model and predict human behavior on a large scale.
The Brookings Institution’s Center on Social and Economic Dynamics is home to the LSAM, where it resides on eight computers. It was developed under the auspices of the Homeland Security Department University Center of Excellence on Preparedness and Catastrophic Event Response (PACER) at The Johns Hopkins University.
The large-scale model creates easily understandable visual representations of vast events and includes the vagaries of human behavior. Programming into the model the release of a pandemic flu bug in Los Angeles and modeling rate of infection based on a set number of family, work and school interactions among people, produces a spreading scarlet stain across the map of the continental United States as zip code after zip code turns red, signifying that more than five percent of the population has become infected. Blue signifies death, removal from the area or immunity.
The model also can represent a full panoply of human responses, with some agents refusing to be vaccinated, for example, as would a significant portion of the U.S. population in the event of a true epidemic. Adding such realism helps improve the model’s predictive power and its depiction of real-world outcomes.
In the flu scenario, the model showed that reducing interactions among agents by 75 percent for a month prevented the outbreak from becoming an epidemic. Instead, it fizzled for lack of carriers. In real life, a government proscription against attending school, going to work, shopping—what is known as a nonpharmaceutical intervention called “social distancing”—would seem Draconian and be difficult, if not impossible to carry out.
So what about a 50 percent reduction over six months? Well, many more people die, but the longer period at least buys time, perhaps enough to develop a vaccine.
The goal here is to improve policymakers' understanding of the dynamics of epidemics and therefore help them make better choices of preventive strategies. And adding information to the model can make the predictions better. For example, Brookings plans to add the location and capacity for handling a surge of patients of every hospital and emergency room in the United States and then to model how well or badly such resources are distributed. The ultimate plan is to model the population of the Earth.
Brookings is working with other members of PACER to combine its agent-based model with other computer models to produce richer depictions. With Bharat Soni of the University of Alabama at Birmingham mechanical engineering department, Brookings is examining transportation options in response to chemical contaminant releases in cities. The combined model shows how a toxic plume from a river barge would spread across New Orleans and how people in city buildings would react.
Without government intervention, many people would die as a result of panic. They pour out of buildings to escape and become stuck in the congested streets under the cloud. Slowly but surely, the little yellow dots turn red. The simulation also will model human behavior during disasters, including resistance both to evacuation and shelter-in-place, the keen desire to locate and join family members, concern for property and the belief that authorities are unreliable.
As exciting as news of Brookings' large-scale model is, there are plenty of caveats. The biggest one is that the use of models and simulations as predictive devices has been so discredited by misuse and manipulation. And it's true, the Brookings team admits, that they only are as good as the information put into them. Brookings uses Census and other data deemed pretty reliable, and researchers, not federal agencies, choose what to study. But all that means is that consumers of the model's revelations must trust the researchers.
Joshua Epstein, director of the Brookings center, who also heads up global epidemic modeling for the National Institutes of Health Models of Infectious Disease Agent Study, is adamant that he will not allow agencies to commandeer LSAM, even though it was funded by DHS dollars. “If the government wanted to use it to figure out better means for urban warfare, Brookings could refuse,” he says.
It's a sad irony that what would propel the use of such modeling in spite of the doubts would be just the sort of calamity the models are being used to prevent. If disaster struck in, say, the form of the release of a pandemic flu virus, and Americans found out the government could have modeled such a thing and then acted on the findings to prevent it, the political results would be devastating and modeling just might be adopted.
Meanwhile, on March 11, the Brookings center received the 2008 Modeling & Simulation Award for Outstanding Achievement in Analysis from the National Training and Simulation Association for the computational feat of creating a mammoth model suited for so many research projects.