We distinguish between static (SIS) and dynamic (DIS) information states of a dialogue agent. The dynamic state is the part of the information state which can change during the course of the dialogue, while the static state stays the same. In the static state we can include rules for interpreting utterances, updating the dynamic information state, and selecting further moves. Also, we can include dialogue move definitions, plan libraries, dialogue game definitions (e.g. in the form of Finite State Transition Networks) and domain databases, insofar as these knowledge sources do not change during the dialogue. If e.g. the database can be updated during the dialogue by information from the user or in any other way, or if the system is capable of learning new rules, these resources should be included in the dynamic state.