In the aftermath of the September 11^th attacks, there were a series of reforms put in place to make the intelligence community operate more effectively. Subsequently, there has been a great deal of debate about whether the reforms are sufficient and whether they are succeeding in their objectives. The idea has been to increase communications between various agencies, to streamline the community’s leadership, and to coordinate counter-terrorism activities.  There are varying opinions about how successful the reforms have been in achieving these goals. While these maybe admirable goals, if the thought is that such actions will result in the prevention of future attacks, their pursuit will likely end in disappointment. Why can’t we achieve a high probability of success through such reforms? Because in order to prevent attacks it is necessary for someone to be able to observe ongoing events and reliably report back, and the reforms don’t do much to eliminate the gap that exists in that regard. There is no substitute for human intelligence. Sun Tzu said that “Advance knowledge cannot be gained from ghosts or spirits, inferred from phenomena, or projected from the measures of heaven, but must be gained from men…” At the time, Sun Tzu’s ideas might have been cutting edge because they disregarded the role of oracles, soothsayers, and spiritual advisors in favor of direct observation. Today we are not prone to accept the word of any proposed practitioners of extrasensory perception, but there is an almost equally fallacious belief by a small number of people that useful information can be gained from risk analysis of various potential targets. When I say that such methods are not of value, I mean practical value in determining how, when, and whether to devote resources to mitigate the risk in question. Knowing whether a building has a one in a billion versus a one in a trillion probability of being attacked in a given period (along with knowledge of the buildings value) may be useful in assigning insurance premiums. However, the fact of the matter is that the target set is too large, the decision making process sufficiently uncertain, the terrorist event too rare, and the adaptability of terrorist is too great to make target prediction a realistic probability no matter how much computing power one has at hand. If one were successful in determining a meaningful probability for a given target’s attack; by hardening that target, its appeal to terrorist would be reduced and the probability would decline.  It is very difficult to get inside information from within these organizations for a number of reasons. One must rely on cultivated agents who are willing and able to come forth from within these close-knit and fairly dispersed organizations, because the long-term commitment required to slowly work into place is both unacceptably time-consuming, requires a level of sacrifice that is incomprehensible, and for which it is difficult to recruit willing participants. In medieval Japan, there were people who were quite good at this type of behind-the-lines activity but there example may be of little illustrative value because: a.) it was a highly homogenous society; b.) war was almost a constant fixture such that the perspective of long-term was much elongated even compared to our present Global War on Terror viewpoint; c.) the process required that part (or all) of the spy’s family be held hostage as insurance that the spy would not switch sides.  None of these criteria is applicable to the situation at hand. Trusting individuals already on the inside can be costly, and the non-hierarchical structure of these organizations may mean that the individual is compartmentalized out of the areas in which they might be most useful. If the Global War on Terror is really going to be the long-run affair that is anticipated, then it might serve us to look at how to re-envision the nature of human intelligence operations.