Another Way to Fix Intelligence

Michael Schrage in the Washington Post has an interesting proposal to improve intelligence analysis. He argues that analysts should add numerical probabilities to their estimates and assessments (e.g., there is 20 percent chance that people in Iran will rise up against the mullahs). I think this a great idea. If investment banks can quantify the risks of its investments (world events certainly play a role in the risks of its investments), then why can't analysts at the CIA do the same?