Many behavioral scientists and most designers of man-computer interfaces view communication in, a certain way. This viewpoint includes the implicit belief that communication from system A to system B essentially involves the encoding of some internal state in system A into an external statement for transmission to system B. System B decodes this message and changes its internal state. Communication is considered “good” to the extent that there is an isomorphism between the internal states of the two systems after the message has been sent. This paper argues that this view is inadequate both for an understanding of communication between two persons and as a theoretical foundation for any kind of man-computer interaction, particularly, in natural language. Empirical results supporting this proposition are reported. In addition, an alternative view of the communication process is outlined. This view stresses the game-theoretic aspects of communication, the importance of viewing message-building as a constructive (rather than translational) process, the importance of metacomments, the multiplicity of channels involved in human natural language communication, and stresses that, under certain conditions, the “vagueness”, “fuzziness” and ambiguity of natural language are assets, not liabilities. The paper concludes by discussing some ways these ideas could serve as possible guidelines for the design of man-computer interfaces. A major purpose of the paper is to encourage the expression of alternative views on these issues. © 1978, Academic Press Inc. (London) Limited. All rights reserved.