The use of empirical observations in the development of formal models

The Xerox 800 is a very simple word processing machine. At Xerox, however, we are finding that a large number of people are calling up their service representative and saying, "Take this machine out - I don't want it." This, of course, causes a great deal of concern. The main reason for recalls of this machine is not because it breaks down, not because it produces results that are unsatisfactory, but because it is virtually impossible to use, and to learn to use! One of the reasons this problem occurs is that, for the type of people who are using these very low level, simple kinds of computer systems, there is an extremely high turnover rate in the offices. In the Los Angeles area, the turnover rates for secretaries in law offices (for example) can be anywhere from 50 per cent to 200 per cent per year; that means that a person stays for an average of six months or less. Most of the people who are using these kinds of systems and who only stay around for a short period of time can be described as naive or inexperienced users of the system. If they are having this kind of trouble with a very simple system like the 800 typing system, I think we can imagine the kinds of problems that we can expect in much more complex systems, such as fully automated text editors, graphics packages, forms handlers, data base inquiry systems, management information systems, etc.