| Argumentation provides a powerful means to achieve complex, non-demonstrative reasoning within a multi-agent system. However, the advantages gained by this form of automated reasoning are not without consequence. When arguments are constructed out of formulae in an underlying formal logical language, decisions such as argument evaluation involve deciding the consistency or logical entailment of the component formulae of arguments. Given that these decisions are generally intractable, and further that an agent's decisions need to be resource bounded, malicious exploitation is possible. By strategically expanding the syntactic content of its arguments, a malicious agent may exploit the resource bounds of its opponent's decision procedures, effectively manipulating their outcome towards its own ends. This type of exploitation presents an important vulnerability in the security of open multi-agent systems employing argumentation as a means of communicative interaction. It is therefore necessary to investigate and develop strategies for detecting and defending against this type of malicious exploitation. However, such defense strategies will necessarily be insufficient; it would be impossible to construct a perfect defense against malicious resource exhaustion strategies in open multi-agent argumentation systems. Nonetheless, research into efficient means of minimizing the risk of such exploitation is warranted, as is continued investigation into further methods of malicious agent strategies in practical argumentation systems. |
