User perception of data importance and the economic feasibility of hardware and storage devices will determine the future direction of online systems. A retrieval system's functions are ultimately dictated by database design. Early online sytems were designed for bibliographic information limited to citations only using sequential files. When indexes and abstracts became desirable for users, their controlled vocabulary requirements could be economically met by inverted files and direct storage devices. The actual users, however, were information specialists familiar with the specialized vocabularies, rather than subject specialists. Now full text information, numeric information, and manipulative systems are being perceived by a class or users not interested in specialized vocabulary skills. These users are placing demands that will require sophisticated internal technology, improved and expanded storage capability, more sophisticated file structure, and transparent data manipulative devices. The economic potential of these demands will bring about the development of a new generation of multitiered databases that allow users at all skill levels to approach them. The point is that, as the user's need for types of data to be retrieved changes, so must the database technology change. (RAA)