QA systems are useful as models because they permit partial and incomplete answers, as well as themodelling of nonsensical answers, when the question is insufficient or when the answer is vague. They also help modelling of questions that are not possible with a currentKMsystem(or even with the current state of the art), but which could be provided by an enquiry of human resources or generalised expertise (e.g. in a library). A QA systemalso permits us to reserve a portion of the role for the enquirer in interpretation – we cannot assume that the details that are delivered by the tuple returned are going to necessarily provide the final answer – it may require reprocessing by another system, or combination with other answers to make up an answer in a hypersystem. What we are modelling is the entirety of the knowledge resources of an enterprise, not just the portion of it that is computerised, let alone encoded and stored in a database.