Some Large Questions about Very Large Data Bases

There are now in use many data bases, each containing billions of bytes, which have evolved with no reference to any of the topics discussed at length in data base conferences. Examples are the U.S. Census, UTLAS, PRESTEL, PLATO, MEDLARS, ORBIT and SITA, to name a few. What can be done to prevent data bases from going the way of programming languages, where most of the world continues to use Cobol and Fortran? Perhaps evolutionary models are needed; perhaps data base specialists have to pay more attention to communications. Certainly they have to do more experimental work on operational systems and establish credibility with managers of existing data bases.