Short-term friendly and long-term hostile?

Abstract

Several authors have suggested, and we are hearing some additional papers on the subject at this conference, that our computer systems should be "friendly" -- that the new user, or the infrequent user, should be able to use them quickly, without any special learning, and without any resort to written materials. My colleagues and I are responsible for a large analysis system [1, 3] that has been in active use outside its development group for about five years and which has several philosophically similar predecessor systems that go back another three or four years [4, 5, 6]. It is interactive in the sense that one of its reasons for existence is to permit the user to interact with data and tease results out of them in a variety of ways -- it has never been, nor is it derived from, a front-end to a batch system or batch thinking. Its users have ranged in skill and background from the beginning student to the professional statistician developing new techniques; from the academic researcher to the clerk in commercial environments. We draw, from this experience, some differing views on what kinds of system designs are friendly and what sorts of assumptions lead to "friendly" systems.