Peter Bryant

In building current computer systems, we tend to break them down into “levels” of control, command and communication; in using the system, we break our problems down correspondingly. The continued use of such a structure raises questions about its effects on the usefulness of future systems, particularly with regard to such trends as time sharing, parallel programming, and, eventually, systems which learn. In this essay some of these questions are posed, and the general attitude we must take in pursuing the problem further is discussed.