When researcher Karen Clark developed the first probability-based model for measuring the threat of natural disasters in the U.S. in 1987, almost no one cared. Clark, then 30, started her own company in Boston and used tens of thousands of data points–from the wind speeds of hurricanes to the lengths of fault lines–to help insurance firms estimate how often a disaster might strike and how much harm it might do. Then, in 1992, Hurricane Andrew struck, wreaking more havoc than anyone–except Clark and her small team at AIR Worldwide Corp.–had ever imagined possible. As the toll climbed past $15 billion, AIR’s phones began ringing. Today probabilistic modeling technology is so well accepted that after the attacks of Sept. 11, 2001, AIR’s clients immediately called for a new model that would capture the risks of terrorism. The model, completed in 2002, assesses the likelihood and cost, in human life and dollars, of different kinds of attacks in every part of the country. It’s not perfect, but it’s smart. “The risk of terrorism is everywhere,” says Clark. “The question is, How much risk?” This time the insurance industry quickly accepted modeling as the basis for figuring out how much terrorism coverage should cost. International terrorism, as most experts will tell you, is not as unpredictable as it feels. Terrorists follow patterns. And while we can’t read the minds of zealots, we can get a good idea of what kind of damage they could do in any given location. We can estimate the cost of an attack on a port in Los Angeles vs. an attack on a port in Prince William Sound. We can calculate where a nuclear blast of a given force would kill 500,000 people as opposed to 50,000. These are the logical estimates that insurers and investment banks are seeking as they try to quantify the risk they face. But while all this strategic thinking is going on in the private sector, the government has responded to terrorism in a less rational way. Since the Sept. 11 attacks, about $13.1 billion has surged into state coffers from the Federal Government–sorely needed money that has gone for police, fire and emergency services to help finance equipment and training to prevent and respond to terrorist attacks. That is a 990% increase over the $1.2 billion spent by the Federal Government for similar programs in the preceding three years. But the vast majority of the $13.1 billion was distributed with no regard for the threats, vulnerabilities and potential consequences faced by each region. Of the top 10 states and districts receiving the most money per capita last year, only the District of Columbia also appeared on a list of the top 10 most at-risk places, as calculated by AIR for TIME. In fact, funding appears to be almost inversely proportional to risk. If all the federal homeland-security grants from last year are added together, Wyoming received $61 a person while California got just $14, according to data gathered at TIME’s request by the Public Policy Institute of California, an independent, nonprofit research organization. Alaska received an impressive $58 a resident, while New York got less than $25. On and on goes the upside-down math of the new homeland-security funding.