ABSTRACT
This paper applies Bayesian theories to critically analyse and offer reforms of intelligence analysis, collection, analysis, and decision making on the basis of Human Intelligence, Signals Intelligence, and Communication Intelligence. The article criticises the reliabilities of existing intelligence methodologies to demonstrate the need for Bayesian reforms. The proposed epistemic reform program for intelligence analysis should generate more reliable inferences. It distinguishes the transmission of knowledge from its generation, and consists of Bayesian three stages modular model for the generation of reliable intelligence from multiple coherent and independent testimonial sources, and for the tracing and analysis of intelligence failures. The paper concludes with suggestions for further research, the development of artificial intellignce that may measure coherence and reliability of HUMINT sources and infer intelligence following the outlined general modular model.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 The probability of any 7 digit number chosen at random is 1:1,000,000; the probability of both witnesses choosing that same number is 0.00000012; but of both choosing randomly some same number out of 1,000,000 numbers requires multiplying 0.00000012 by the 10,000,000 possible pairings of two randomly chosen numbers, hence 1:10,000,000.
2 Formally, analysts need a wide gap between:
And
where T1, T2, … , Tn represent informationally coherent units of testimony; C represents some common information source; S1, S2, … , Sn, are separate information sources, and B is background knowledge. The first line represents the likelihood of the coherent testimonies given some common information source; the second line, their likelihood, given separate information sources.