The paper includes a series of letters exchanged between the author and the late Professor Feyerabend, concerning the best "method" for military intelligence, as a test case for the role of conceptual frameworks in philosophy of science. The letters deal with issues like: Is it possible to make an intelligence estimate without a conceptual framework? Does such a framework have any 'positive' role? If so, how should a conceptual framework in intelligence be built? What risks lurk within it? Is it (...) possible to indicate methods for intelligence estimates which are 'better' and more 'successful' than others? (shrink)