This disconnect is the number-one threat to science librarianship today—perhaps to all academic librarianship. How can science libraries persist when scientists haven't the least notion that libraries or librarians are relevant to their work?
When I think about the library "catalog" I think that it will contain more than just metadata about library materials but also the materials themeselves. Think mostly books, journal articles, encyclopedia articles, definitions, images, data sets, etc. Moreover, the library "catalog" will enable people to do things with the items in the collection. It is more than just find and get.