Statistical Concepts of Deceit and Detection Statistical Concepts of Deceit and Detection

Publication Date
Volume
19
Issue
3
Start Page
22
Author(s)
Jonathan Sanborn - Brookhaven National Laboratory
File Attachment
V-19_3.pdf3.03 MB
Abstract
A mathematical model of a treaty-verification situation is presented in terms of statistical hypothesis testing and probability theory. A conflict situation is discussed in which the two players (arbitrarily chosen as male) are the party being inspected and the party doing the inspecting. In the case where the party being inspected is complying with the terms of the treaty, it is assumed that his reported declarations are truthful; in the case where he is violating the treaty, he is assumed to falsify his reported declarations to cover up the violations if possible. The model describes a situation in which evidence for treaty compliance or non-compliance is not conclusive, but where the ambiguities can be characterized by using statistical models. Reporting falsification strategies are described for the inspected party, and hypothesis- testing counterstrategies are described for the inspecting party. The concept of a perfect reporting strategy is discussed, and it is shown that in the absence of perfect reporting the data declared by the inspected party may be used against him. The ability of the inspected party to successfully falsify is shown to be dependent on his state of knowledge of the inspector's information. This can be affected, for example, by randomized strategies on the part of the inspector.
Additional File(s) in Volume
V-19_1.pdf5.36 MB
V-19_2.pdf3.34 MB
V-19_3.pdf3.03 MB