Adversarial Risk Analysis: Applications to Basic Counterterrorism Models
Interest in counterterrorism modelling has increased recently. A common theme in the approaches adopted is the need to develop methods to analyse decisions when there are intelligent opponents ready to increase our risks. Most of the approaches have a cle
- PDF / 195,826 Bytes
- 10 Pages / 430 x 660 pts Page_size
- 64 Downloads / 204 Views
Manchester Business School, UK Royal Academy of Sciences, Spain
Abstract. Interest in counterterrorism modelling has increased recently. A common theme in the approaches adopted is the need to develop methods to analyse decisions when there are intelligent opponents ready to increase our risks. Most of the approaches have a clear game theoretic flavour, although there have been some decision analytic based approaches. We have recently introduced a framework for adversarial risk analysis, aimed at dealing with problems with intelligent opponents and uncertain outcomes. In this paper, we shall explore how such framework may cope with two of the standard counterterrorism model formulations: sequential defend-attack and simultaneous defend-attack moves.
1
Introduction
Recent high-profile terrorist attacks have demanded significant investments in protective responses. This has stirred a great deal of interest in modelling issues to deal with counterterrorism decisions. Good accounts and introductions to this field may be seen in Parnell et al. (2008) and Bier and Azaiez (2008). Besides some reliability analysis studies based on tools such as fault trees, much of this literature has a distinct game theoretic flavour. Two examples include Zhuang and Bier (2007), who compute best responses and Nash equilibria as a basis for allocating resources against terrorism when the defender and attacker have different multiattribute utility functions, in situations of both simultaneous and sequential play; and Brown et al. (2006), who present max-min, min-max and min-max-min optimization models for defender-attacker, attacker-defender and defender-attacker-defender problems. Following Raiffa (2002), we remain skeptical about the relevance of such concepts in counterterrorism modelling, based on common knowledge assumptions that entail that parts have too much information about their counterparts, in a field in which secrecy tends to be an advantage. The other mainstream literature in the field has a decision analytic flavour. Among others, we mention Von Winterfeldt and O’Sullivan (2006), who use decision trees to evaluate Man-Portable Air Defense Systems countermeasures; and Pinker (2007), who applies influence diagrams to assess the deployment of various short-term countermeasures. Their recurrent (critical and criticised) problem is the need to assess the probabilities of the actions of the others, which is a key F. Rossi and A. Tsoukis (Eds.): ADT 2009, LNAI 5783, pp. 306–315, 2009. c Springer-Verlag Berlin Heidelberg 2009
Adversarial Risk Analysis: Applications to Basic Counterterrorism Models
307
issue of the Bayesian approach to games, see Kadane and Larkey (1982) or Raiffa (2002). Banks and Anderson (2006) provide a simple numerical comparison to both approaches to game theory within a smallpox threat problem. This tension between game theoretic and decision analytic approaches to decision making problems with adversaries is not exclusive of counterterrorism models but appears in other business and industrial areas, see e.g. v
Data Loading...
