Cooperative clans

Abstract

Purpose

To provide a mechanism for agents to form, maintain, and reason within medium‐term coalitions, called clans, based upon the notions of trust and motivation.

Design/methodology/approach

The model is based upon the notions of trust (representing an agent's assessment of another's honesty and reliability) and motivations (which represent an agent's high‐level goals). The paper describes the motivational factors that can lead to clan formation, a mechanism for agents to form a clan or join an existing clan, and subsequently how clan membership influences behaviour (in particular though sharing information and acting on behalf of other members). Finally, describes the conditions under which agents leave a clan.

Findings

The proposed mechanism shows how agents can form medium‐term clans with trusted agents based on motivations that are essentially self‐interested. It is shown how this mechanism can be used to reduce missed opportunities for cooperation, improve scalability, reduce the failure rate and allow sharing of trust information (i.e. establish a notion of reputation).

Originality/value

Proposes a new approach to coalition formation based on the notions of trust and motivation, which allows self‐interested agents to form medium‐term coalitions (called clans) to increase their own (motivational) returns.

Citation

Griffiths, N. (2005), "Cooperative clans", Kybernetes, Vol. 34 No. 9/10, pp. 1384-1403. https://doi.org/10.1108/03684920510614722