Arguing that understanding what lies behind the apparent usability of direct manipulation style interfaces might not only help in building better interfaces, but can also draw attention to possible side effects, this paper uses examples from a prototype data manipulation system to pursue a characterization of direct manipulation interfaces as forestalling the need for procedural reasoning, a class of skills that people appear to have difficulty mastering. Two issues are raised: (1) whether direct manipulation is a design philosophy that will inhibit the development of procedural reasoning skills, and (2) the significance of procedural reasoning proficiency for users in the future. The design philosophy is then examined in terms of the reasoning skill and the reasoning skill in terms of the design philosophy, with the intent of shedding some light on both. It is concluded that direct manipulation techniques could be useful in explicitly teaching procedural reasoning skills; that procedural reasoning is just one alternative perspective; and that some of the issues raised here might look different as more complex direct manipulation systems are developed, and their resolution will have to be based on empirical work. Thirteen references and 60 technical reports by members of the Institute for Cognitive Science are listed. (Author/MES)