In open self-organising systems, interaction partners can change often due to changes in system structure or inclusion of new agents. Since benevolence of the agents can not be assumed, there is also asymmetric information about the intentions and strategies of other agents (Schillo et al. 2000). To deal with such situations, a reputation system can be used which requests and combines the opinions of other agents and generates recommendations (Mui et al. 2002). This enables cooperation between agents that do not know each other or have little experience with each other.

If a reputation system is used, there has to be a consensus among the agents about the meaning of trust and reputation values. It is thus necessary to have a common metric that is used to calculate trust values from experiences. However, each agent can still have a different threshold to decide whether a potential interaction partner is trustworthy or not. It is also possible to appraise a reputation value by discounting the value with the trust value of the agent that provides the reputation.