This paper compares two alternative computer architectures that have been proposed to provide the control mechanism that enables an intelligent tutoring system to decide what instructional action to perform next, i.e., discourse management networks and blackboards. The claim that an intelligent tutoring system controlled by a blackboard architecture can deliver significantly more effective and flexible instruction than one controlled by a discourse management network is supported by considering an architectural formalization of each architecture and the way in which its key components facilitate or hamper the dynamic planning of instruction. The relationship between this ability to plan instruction dynamically and to generate effective and flexible instruction is also supported. To ground these discussions, two examples of each architecture are provided: GUIDON and MENO-TUTOR exemplify discourse planners implemented as discourse management networks, while IDE-INTERPRETER and the BLACKBOARD INSTRUCTIONAL PLANNER exemplify planners implemented in the blackboard architecture. Application of each architecture to control an intelligent tutoring system to teach LISP is also considered to further illustrate differences in capability that result from choice of architecture. The text is supplemented by 7 figures, and a 13-item bibliography is provided. (EW)